i
UNIVERSITY OF SOUTHAMPTON FACULTY OF ENGINEERING AND THE ENVIRONMENT
Civil, Maritime and Environmental Engineering
The vibrations of a exible planing craft: hydroelasticity, boat motion and noise.
by
Peter K Halswell
Thesis for the degree of Engineering Doctorate
March 2015
ii
Blank page
iii
UNIVERSITY OF SOUTHAMPTON
ABSTRACT
FACULTY OF ENGINEERING AND THE ENVIRONMENT
Civil, Maritime and Environmental Engineering
Engineering Doctorate
THE VIBRATIONS OF A FLEXIBLE PLANING CRAFT: HYDROELASTICITY, BOAT MOTION AND NOISE
by Peter K Halswell
The Royal National Lifeboat Institution (RNLI) is the charity that aims to save lives at sea. The RNLI D-class is a ve metre in atable lifeboat that is used near the shore in waves and surf. Anecdotal evidence indicates that the D-class has improved performance due to its unique, exible, fabric structure, and this exibility is highly likely to aect the vibrations generated by the D-class. The boat motion is experienced by the on-board crew, and the air and water borne noise are heard by the on-board crew and the wildlife. This thesis aims to measure these two types of vibration, predict the perception of these vibrations and measure the eects of hydroelasticity on both the vibration and perception. Three aspects of hydroelasticity were identi ed within the D-class: hydroelastic slamming, hydroelastic planing surfaces and global hydroelasticity. which to view the eects of hydroelasticity.
This gives a new perspective with
A four stage full-scale holistic hydroelastic ex-
periment was performed with each stage aiming to trigger one aspect at a time.
The four
stages were: static tests, at water trials, drop tests and wave trials. The D-class was tted with 52 sensors to measure the boat motion, engine thrust, sponson and keel pressures, deck hinge angles, deck panel de ections and the fabric hull deformation. The static trials measured the shape of the D-class under only buoyancy and weight forces. The at water trials measured the eect of a hydroelastic planing surface on the forward speed and investigated a phenomenon termed the pulsing motion. The drop tests were performed at full-scale and quasi-2D, and they measured the eect of hydroelastic slamming on the peak acceleration and predicted the Whole Body Vibration (WBV). The open-water wave trials investigated the global hydroelasticity. The static tests showed that the shape of the D-class was more dependent on the keel pressure than the sponson pressure. The at water trials proved that a exible planing surface decreases the forward speed by 0.44 knots.
The pulsing motion surprisingly exhibited the
highest forward speed and it is hypothesised that the structure achieved an unstable equilibrium position of minimal potential energy. The full-scale and quasi-2D drop tests demonstrated that hydroelasticity can aect the peak accelerations and WBV, but the trend was inverted when the drop height was varied from 0.5 m to 1 m. It is believed that the keel is the dominant component during the at water trials and drop tests, and this is coupled with the fabric hull.
iv
No statistical dierence was found in the wave trials results but this was explained through the drop test results. The predicted WBV from the wave trials does emphasises the need for a new WBV reduction strategy and incorporating an element of hydroelasticity along with other reduction methods could make a signi cant impact on the WBV. The airborne noise of the D-class was measured using ISO 14509. The airborne noise was above the limits set out by the European directive 2003/44/EC. A method was developed to measure the water borne noise of small High Speed Craft (HSC) in shallow waters. The water borne noise propagation was modelled using an Image source Transmission Loss (ImTL) model. The perception of the air and water borne noise by a harbour seal was predicted and it showed that the D-class is unlikely to cause damage to the auditory system at one metre but will de nitely be audible to the seal at 20 m. The horizontal and vertical transmission loss through a shallow water channel was investigated.
Contents 1 Introduction
3
1.1
Approach to boat motion
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
1.2
Approach to boat noise
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10
1.3
Aims
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10
1.4
Thesis structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12
1.5
Novel contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13
2 Hydroelastic literature review
15
2.1
Rigid in atable boats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15
2.2
D-class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16
2.2.1
Construction
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16
2.2.2
D-class literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20
2.2.3
Structural coupling
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22
Boundary tension membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23
2.3.1
Material properties
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23
2.3.2
Form nding
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
2.3.3
In atable cylindrical beams
. . . . . . . . . . . . . . . . . . . . . . . . .
26
2.3.4
Experimental methods . . . . . . . . . . . . . . . . . . . . . . . . . . . .
28
2.3
2.4
Fluid domain
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
28
2.5
Hydroelastic slamming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
2.5.1
Problem de nition
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
30
2.5.2
Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31
2.5.3
Critique of modelling methods
. . . . . . . . . . . . . . . . . . . . . . .
32
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
33
2.6
2.7
Hydroelastic planing surfaces 2.6.1
Problem de nition
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
33
2.6.2
Literature review
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
34
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
34
2.7.1
De nition of longitudinal hydroelasticity . . . . . . . . . . . . . . . . . .
35
2.7.2
Literature review of global hydroelasticity . . . . . . . . . . . . . . . . .
36
Global hydroelasticity
2.8
Coupled hydroelasticity
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
36
2.9
Perception of boat motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
37
2.10 Summary
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
39
vi
CONTENTS
3 Hydroelastic methodology 3.1
3.2
3.3
3.4
3.5
41
Proposed numerical methods for individual aspects . . . . . . . . . . . . . . . .
41
3.1.1
Modelling of boundary tensioned membranes
. . . . . . . . . . . . . . .
42
3.1.2
Slamming numerical models . . . . . . . . . . . . . . . . . . . . . . . . .
42
3.1.3
Other numerical models
43
. . . . . . . . . . . . . . . . . . . . . . . . . . .
Proposed experimental methods for individual aspects
. . . . . . . . . . . . . .
43
3.2.1
Quasi-2D fabric wedge impacting a free surface
. . . . . . . . . . . . . .
44
3.2.2
Fluid impacting an inclined fabric plate
. . . . . . . . . . . . . . . . . .
44
3.2.3
Flat fabric planing surface
. . . . . . . . . . . . . . . . . . . . . . . . .
45
3.2.4
Scale model of a fabric vee-shaped planing surface
. . . . . . . . . . . .
45
3.2.5
Segmented scale model without sponson in waves
. . . . . . . . . . . .
46
3.2.6
Segmented scale model with sponson in waves
. . . . . . . . . . . . . .
46
. . . . . . . . . . . . . . . . .
47
3.3.1
Towing tanks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
47
3.3.2
Open water trials
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
49
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
Proposed experimental methods coupled aspects
Quasi-2D drop test method 3.4.1
Equipment
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
3.4.2
Parameters
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
53
3.4.3
Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
53
3.4.4
Post Processing
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
54
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
54
3.5.1
Aims . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
55
3.5.2
D-class equipment
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
56
3.5.3
Stationary tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
63
3.5.4
Drop tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
63
3.5.5
Flat water trials
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
65
3.5.6
Wave trials
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
66
3.5.7
Calibration
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
67
3.5.8
Post Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
68
Full-scale test method
4 Hydroelastic results under vertical loads 4.1
4.2
4.3
Static tests
73
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
73
4.1.1
Eect of internal pressures with no crew on structural deformation
. . .
73
4.1.2
Eect of internal pressures with three crew on structural deformation . .
75
4.1.3
Eect of keel pressure on structural deformation
. . . . . . . . . . . . .
78
4.1.4
Eect of crew loading on structural deformation . . . . . . . . . . . . . .
81
Quasi-2D drop tests
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
84
4.2.1
Accelerations
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
84
4.2.2
Structural deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . .
92
4.2.3
Error analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
92
Full-scale drop tests
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
94
CONTENTS
4.4
4.5
vii
4.3.1
Accelerations
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.2
Frequency spectra of acceleration . . . . . . . . . . . . . . . . . . . . . . 100
4.3.3
Structural deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.3.4
Error analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Hydroelastic discussion Part 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 4.4.1
Whole body vibration of drop tests . . . . . . . . . . . . . . . . . . . . . 113
4.4.2
Root cause to hydroelastic slamming . . . . . . . . . . . . . . . . . . . . 114
Summary
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5 Hydroelastic results under horizontal loads 5.1
94
Flat water trials
121
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.1.1
Eect of internal pressures on forward speed . . . . . . . . . . . . . . . . 121
5.1.2
Dynamic motion
5.1.3
Pulsing motion
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.2
Wave trials
5.3
Hydroelastic discussion Part 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
5.4
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
5.3.1
Whole body vibration of wave trials
5.3.2
Root cause to hydroelastic planing surfaces
5.3.3
A hydroelastic planing craft
Summary
6.2
6.3
6.4
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Airborne noise
171
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
6.1.1
International standards
6.1.2
Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Water borne noise
. . . . . . . . . . . . . . . . . . . . . . . . . . . 172
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
6.2.1
International standards
. . . . . . . . . . . . . . . . . . . . . . . . . . . 174
6.2.2
Critique of international standards
6.2.3
Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
7.2
. . . . . . . . . . . . . . . . . . . . . 176
Perception of noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 6.3.1
Perception by humans
. . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
6.3.2
Perception by wildlife
. . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
Summary
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
7 Noise methodology 7.1
. . . . . . . . . . . . . . . . 164
. . . . . . . . . . . . . . . . . . . . . . . . 168
6 Noise literature review 6.1
. . . . . . . . . . . . . . . . . . . . 160
187
Water borne noise methodology
. . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.1.1
Criteria
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.1.2
Course layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
7.1.3
Instrumentation
7.1.4
Measurement requirements and procedure . . . . . . . . . . . . . . . . . 189
7.1.5
Post processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Air and water borne noise trial method . . . . . . . . . . . . . . . . . . . . . . . 191
viii
CONTENTS
7.2.1
Equipment
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
7.2.2
Procedures
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
7.2.3
Post processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
7.2.4
Propagation models
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
8 Noise results 8.1
Overall air and water borne noise . . . . . . . . . . . . . . . . . . . . . . . . . . 201 8.1.1
8.2
199
Maximum sound pressure level
One-third-octave air and water borne noise
. . . . . . . . . . . . . . . . . . . . 202
8.2.1
One-third-octave spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
8.2.2
Hydroelastic noise
8.2.3
Sound pressure level at one metre . . . . . . . . . . . . . . . . . . . . . . 205
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
8.3
Error analysis
8.4
Perception of air and water borne noise
8.5
. . . . . . . . . . . . . . . . . . . . . . . 201
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 . . . . . . . . . . . . . . . . . . . . . . 207
8.4.1
Overlay method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
8.4.2
Critical transmission loss method . . . . . . . . . . . . . . . . . . . . . . 211
8.4.3
Background transmission loss method
Summary
. . . . . . . . . . . . . . . . . . . 214
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
9 Conclusion
217
Bibliography
224
A Hydroelastic in atable boats: Relevant literature and new design considerations 237 B Measuring the hydroelasticity of in atable boats to reduce injuries in harsh waters 249 C An experimental investigation into the whole body vibration generated during the hydroelastic slamming of a high speed craft. 257 D Measuring the stress-strain relationship of rubber-coated fabrics
299
D.1
Aims . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
D.2
Method
D.3
D.4
D.5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
D.2.1
Test samples
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
D.2.2
Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 D.3.1
DTex 1670 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
D.3.2
DTex 1100 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302 D.4.1
Dierence between DTex 1670 and DTex 1100
D.4.2
Eect of lamination
. . . . . . . . . . . . . . 302
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
CONTENTS
ix
E Waterline de ection experiment
307
E.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
E.2
Aims . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
E.3
Method
E.4
E.3.1
Loading conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
E.3.2
Hysteresis eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
E.3.3
Measurement method
. . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309 E.4.1
E.5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
Error analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
Analysis and discussion
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
E.5.1
Symmetrical de ection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
E.5.2
Hysteresis eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
E.5.3
Central loading conditions . . . . . . . . . . . . . . . . . . . . . . . . . . 310
E.5.4
E.5.3.1
Vertical de ection
. . . . . . . . . . . . . . . . . . . . . . . . . 310
E.5.3.2
Length de ection . . . . . . . . . . . . . . . . . . . . . . . . . . 311
E.5.3.3
Width de ection . . . . . . . . . . . . . . . . . . . . . . . . . . 311
Crew loading condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312 E.5.4.1
Vertical de ection
. . . . . . . . . . . . . . . . . . . . . . . . . 312
E.5.4.2
Length de ection . . . . . . . . . . . . . . . . . . . . . . . . . . 312
E.5.4.3
Width de ection . . . . . . . . . . . . . . . . . . . . . . . . . . 313
E.6
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
E.7
Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314 E.7.0.4
Keel and sponson pressure
. . . . . . . . . . . . . . . . . . . . 314
E.7.0.5
Vertical, length and width de ections
E.7.0.6
Fabric position . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
. . . . . . . . . . . . . . 314
F Compact RIO programming
315
G Flat water trial videos
317
G.1
Graphs at 2 psi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
G.2
Graphs at 3 psi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
G.3
Graphs at 4 psi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
G.4
Graphs of the pulsing motion
H Drop test videos
. . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
319
H.1
Video of drop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
H.2
Graphs at 2 psi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
H.3
Graphs at 3 psi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
H.4
Graphs at 4 psi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
x
CONTENTS
I Full-scale drop tests - additional graphs
321
I.1
Time and frequency domain of hull de ection
. . . . . . . . . . . . . . . . . . . 321
I.2
Time and frequency domain of deck panel strain
. . . . . . . . . . . . . . . . . 324
J Transmission loss predicted by ImTL model
335
K MatLab code for Savitsky prediction
341
List of Figures 1.1
RNLI B- and D-class.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5
1.2
Hydroelastic cycle.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
2.1
Main components of the D-class.
2.2
In atable sponson.
2.3
Laminated fabric hull.
2.4
In atable keel running along the centre line of the vessel.
. . . . . . . . . . . .
19
2.5
Four composite deck panels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19
2.6
Deck hinges. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20
2.7
Total resistance of the D-class, Dand et al. (2008).
21
2.8
Hull distortion of the D-class at 19.4 knots, Dand et al. (2008).
2.9
Construction of rubber-coated fabrics, Hot Ribs (2011).
. . . . . . . . . . . . . . . . . . . . . . . . . .
17
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
18
. . . . . . . . .
21
. . . . . . . . . . . . .
24
2.10 Discretised continuum used in the dynamic relaxation method, Lewis (2003).
.
25
2.11 Jet of water impacting a inclined plate, Saunders (1957). . . . . . . . . . . . . .
29
2.12 Longitudinal pressure distribution on a planing at plate, Dand (2003c). . . . .
29
2.13 Flexible components within a vertically impacting IB.
. . . . . . . . . . . . . .
30
2.14 Air pocket formation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31
2.15 Drop stitch technology, Yakangler (2011).
. . . . . . . . . . . . . . . . . . . . .
34
2.16 Reducing the vertical boat motion through longitudinal hydroelasticity. . . . . .
35
2.17 Hydroelastic design cycle.
37
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1
Additional iteration in nding the elastic hull shape due to BTMs.
. . . . . . .
42
3.2
Natural air pockets due to the shape of the D-class hull
. . . . . . . . . . . .
43
3.3
Quasi-2D drop test rig. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
52
3.4
Using the cRIO as a data logger with and without a real time controller. . . . .
57
3.5
Location of accelerometers during the full-scale drop tests (all dimension in metres).
.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
58
3.6
Deck strain gauge locations (all dimension in mm). . . . . . . . . . . . . . . . .
60
3.7
Deck hinge bridge design.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
60
3.8
Hull string potentiometer locations. . . . . . . . . . . . . . . . . . . . . . . . . .
62
3.9
Full scale drop test set-up. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
64
3.10 Hinge angle calibration charts.
. . . . . . . . . . . . . . . . . . . . . . . . . . . xi
71
xii
LIST OF FIGURES
4.1
Static deck hinge angles due to varying the internal pressures with no crew.
4.2
Static deck micro-strains due to varying the internal pressures from S2.25 K2 to S4.25 K4 with no crew.
. .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
74
75
4.3
Static hull shapes due to varying the internal pressures with no crew.
. . . . .
76
4.4
Static hinge angles due to varying the internal pressures with three crew. . . . .
77
4.5
Static deck micro-strains due to varying the internal pressures from S2.25 K2 to S4.25 K4 with three crew.
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
78
4.6
Static hull shapes due to varying the internal pressures with three crew.
. . . .
79
4.7
Static deck hinge angles due to varying only the keel pressures with no crew. . .
80
4.8
Static deck micro-strains due to varying only the keel pressure from S3.25 K2 to S3.25 K4 with no crew.
4.9
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Static hull shapes due to varying only the keel pressure with no crew.
80
. . . . .
82
4.10 Static deck hinge angles due to varying the crew loading. . . . . . . . . . . . . .
83
4.11 Static deck micro-strains due to varying the crew loading.
83
. . . . . . . . . . . .
4.12 Static hull shapes due to varying the crew loading at S3.35 K3.
. . . . . . . . .
85
4.13 Example frequency spectrum of the quasi-2D drop test with a 250 Hz low pass lter (deadrise angle = 5°, drop height = 1 m).
. . . . . . . . . . . . . . . . . .
86
4.14 Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 15°, drop height = 0.5 m. . . . . . . . . . . . . . . . . . . . . . . . . . .
87
4.15 Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 5°, drop height = 0.5 m.
. . . . . . . . . . . . . . . . . . . . . . . . . .
88
4.16 Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 15°, drop height = 1 m. . . . . . . . . . . . . . . . . . . . . . . . . . . .
89
4.17 Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 5°, drop height = 1 m.
. . . . . . . . . . . . . . . . . . . . . . . . . . .
90
4.18 Mean peak accelerations measured during the 2D drop tests. . . . . . . . . . . .
91
4.19 Transverse images of 2D drop test with a fabric hull under 0 N/m (15 degrees deadrise angle, 1 m drop height).
. . . . . . . . . . . . . . . . . . . . . . . . . .
93
4.20 Examples of the acceleration-time history from the crew position during the full-scale drop tests with a drop height of 0.5 m. . . . . . . . . . . . . . . . . . .
95
4.21 Examples of the acceleration-time history from the crew position during the full-scale drop tests with a drop height of 1 m. . . . . . . . . . . . . . . . . . . .
96
4.22 Examples of the acceleration-time history from the bow during the full-scale drop tests with a drop height of 0.5 m. . . . . . . . . . . . . . . . . . . . . . . .
97
4.23 Examples of the acceleration-time history from the bow during the full-scale drop tests with a drop height of 1 m. . . . . . . . . . . . . . . . . . . . . . . . . 4.24 Mean peak accelerations measured during the full-scale drop tests. 4.25 Mean power spectra of all the full-scale drop tests.
. . . . . . .
98 99
. . . . . . . . . . . . . . . . 102
4.26 Mean power spectrum of the quasi-2D drop test with a MDF hull, 1 m drop height and 5° deadrise angle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 4.27 Mean time and frequency domain of the sponson pressure.
. . . . . . . . . . . 105
LIST OF FIGURES
xiii
4.28 Mean time and frequency domain of the keel pressure.
. . . . . . . . . . . . . . 106
4.29 Mean time and frequency domain of the deck trim angle.
. . . . . . . . . . . . 107
4.30 Mean time and frequency domain of the aft deck hinge angle (B1).
. . . . . . . 109
4.31 Mean time and frequency domain of the middle deck hinge angle (B3). . . . . . 110 4.32 Mean time and frequency domain of the hull sensor H9.
. . . . . . . . . . . . . 112
4.33 Quanti cation of WBV during the quasi-2D drop tests. . . . . . . . . . . . . . . 115 4.34 Quanti cation of WBV during the full-scale drop tests. . . . . . . . . . . . . . . 116 5.1
Water conditions during the at water trials.
5.2
Aect of the internal pressures on the at water speed. . . . . . . . . . . . . . . 122
5.3
Hinge angle variation during the at water trials. . . . . . . . . . . . . . . . . . 124
5.4
Hinge angle variation during the at water trials. . . . . . . . . . . . . . . . . . 125
5.5
Change in the quasi-steady planing micro-strains in the deck due to the internal pressures.
. . . . . . . . . . . . . . . . . . . 121
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
5.6
Change in quasi-steady planing hull shape during the at water trials.
. . . . . 130
5.7
Mean strain on the outboard engine trim pin.
5.8
Mean deck trim angle.
5.9
Examples of the deformation of the sponson and keel during the at water trials.133
. . . . . . . . . . . . . . . . . . 132
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
5.10 Static and quasi-steady hull shape during the at water trials. 5.11 Change in the hull shape during the at water trials.
. . . . . . . . . 135
. . . . . . . . . . . . . . 136
5.12 A dynamic shape of the hull during the at water trials at 2 psi (fourth repeat). 137 5.13 Strain on the outboard engine trim pin while pulsing. 5.14 Deck trim angle while pulsing.
. . . . . . . . . . . . . . 138
. . . . . . . . . . . . . . . . . . . . . . . . . . . 139
5.16 Change in deck micro-strains from static C0 S2.25 K2 to the mean pulsing run. 139 5.15 Deformation of the sponson and keel during the pulsing run of the at water trials at 2 psi. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 5.17 Mean quasi-steady planing hull shape during the pulsing at water run. 5.18 Hull shape prior to the dynamic pulsing motion.
. . . . 142
. . . . . . . . . . . . . . . . . 143
5.19 Transom acceleration peak counting in head seas comparing the performance at 3 psi and 4 psi internal pressure.
. . . . . . . . . . . . . . . . . . . . . . . . 150
5.20 Crew acceleration peak counting in head seas comparing the performance at 3 psi and 4 psi internal pressure.
. . . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.21 Transom acceleration peak counting in head seas comparing the performance at 2 psi and 3 psi internal pressure.
. . . . . . . . . . . . . . . . . . . . . . . . 152
5.22 Crew acceleration peak counting in head seas comparing the performance at 2 psi and 3 psi internal pressure.
. . . . . . . . . . . . . . . . . . . . . . . . . . . 153
5.23 Transom acceleration peak counting in following seas comparing the performance at 3 psi and 4 psi internal pressure.
. . . . . . . . . . . . . . . . . . . . . 154
5.24 Crew acceleration peak counting in following seas comparing the performance at 3 psi and 4 psi internal pressure.
. . . . . . . . . . . . . . . . . . . . . . . . 155
5.25 Comparison of weighted VDV measured at the transom during the wave trials comparing dierence signi cant wave heights.
. . . . . . . . . . . . . . . . . . . 164
xiv
LIST OF FIGURES
6.1
Air and water borne noise spectra of two outboard motors measured by Young and Miller (1960) (left - 7.5 HP Johnson AD-12, right - 18 HP Evinrude).
6.2
. . . 171
Hydrophone geometry suggested by ANSI S12.64-2009 (left - grades A and B, right - grade C).
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
6.3
Hydrophone geometry suggested by the DNV silent class notation.
6.4
Noise measurement method from vertically below the vessel, Bubliü et al. (2008).179
6.5
Illustration of the human auditory system, Moore (2012).
. . . . . . . . . . . . 180
6.6
Equal loudness contours of the human ear, ISO 226:2003.
. . . . . . . . . . . . 181
6.7
A, B and C weighting response characteristics, ANSI S1.4-1971. . . . . . . . . . 181
6.8
Typical frequency bands of sounds produced by marine mammals and sh, Gotz et al. (2009).
6.9
. . . . . . . 176
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
Theoretical zone of noise in uence, Richardson et al. (1995). . . . . . . . . . . . 183
6.10 Perceived sound level (dBht ) of sh and marine mammals for an example noise spectrum, Nedwell et al. (2007).
. . . . . . . . . . . . . . . . . . . . . . . . . . 184
6.11 Zones of in uence for salmon and herring caused by piling noise, Nedwell et al. (2012).
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
7.1
Mariner 50 HP outboard motor, Oshoremarine (2013).
7.2
Drifting pass-by course layout.
7.3
An example of the deformed CPA pole. . . . . . . . . . . . . . . . . . . . . . . . 196
8.1
Navigation chart showing locations of the noise trials and weather stations.
8.2
Maximum S-weighted overall SPL against the vessel speed.
8.3
Mean one-third-octave noise spectra measured at 3 psi comparing the port and starboard aspects.
. . . . . . . . . . . . . 189
. . . . . . . . . . . . . . . . . . . . . . . . . . . 193
. . 201
. . . . . . . . . . . 202
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
8.4
Mean one-third-octave noise spectra comparing the aect of hydroelasticity.
8.5
Mean one-third-octave noise spectra propagated to one metre.
8.6
Audiogram of a harbour seal, see Nedwell et al 2004 page 237 and 238.
8.7
SPL of the D-class vs the audiogram of a harbour seal.
8.8
Threshold of hearing the D-class by a harbour seal, range 20km.
8.9
Threshold of hearing the D-class by a harbour seal, range 1km.
8.10 Audible threshold based on the background noise.
. . 206
. . . . . . . . . 207 . . . . 209
. . . . . . . . . . . . . 210 . . . . . . . . 212 . . . . . . . . . 213
. . . . . . . . . . . . . . . . 215
D.1
Tensile test experimental set up . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
D.2
Stress-strain graph of single ply DTex1670.
D.3
Stress-strain graphs of single ply DTex 1100. . . . . . . . . . . . . . . . . . . . . 303
D.4
Comparison of the stress-strain graphs for single ply DTex 1100 and DTex 1670. 304
D.5
Eect of lamination on the stress-strain relationship of DTex 1100.
E.1
Static de ection experiment; top shows a unloaded IB1 and bottom shows an
. . . . . . . . . . . . . . . . . . . . 302
. . . . . . . 304
IB1 loaded with 1.75 tonnes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 E.2
Load positions (all dimensions in mm)
. . . . . . . . . . . . . . . . . . . . . . . 308
E.3
Change in sponson beam during hysteresis check
. . . . . . . . . . . . . . . . . 310
LIST OF FIGURES
xv
E.4
Vertical de ection of the IB1 due to centrally placed loads . . . . . . . . . . . . 310
E.5
Linear relationship between maximum vertical de ection and central load
E.6
Change in length against load for central loading conditions . . . . . . . . . . . 311
E.7
Change in width against load for central loading conditions
E.8
Vertical de ection of the IB1 due to crew placed loads
E.9
Change in length for all loading conditions . . . . . . . . . . . . . . . . . . . . . 313
. . . 311
. . . . . . . . . . . 311
. . . . . . . . . . . . . . 312
E.10 Change in width for all loading conditions . . . . . . . . . . . . . . . . . . . . . 313 I.1
Mean time and frequency domain of the hull sensor H11. . . . . . . . . . . . . . 321
I.2
Mean time and frequency domain of the hull sensor H14. . . . . . . . . . . . . . 322
I.3
Mean time and frequency domain of the hull sensor H15. . . . . . . . . . . . . . 323
I.4
Mean time and frequency domain of the hull sensor H17. . . . . . . . . . . . . . 323
I.5
Mean time and frequency domain of the deck panel strain D1x.
. . . . . . . . . 324
I.6
Mean time and frequency domain of the deck panel strain D1y.
. . . . . . . . . 325
I.7
Mean time and frequency domain of the deck panel strain D2x.
. . . . . . . . . 326
I.8
Mean time and frequency domain of the deck panel strain D2y.
. . . . . . . . . 327
I.9
Mean time and frequency domain of the deck panel strain D4x.
. . . . . . . . . 328
I.10 Mean time and frequency domain of the deck panel strain D4y.
. . . . . . . . . 328
I.11 Mean time and frequency domain of the deck panel strain D5x.
. . . . . . . . . 329
I.12 Mean time and frequency domain of the deck panel strain D5y.
. . . . . . . . . 330
I.13 Mean time and frequency domain of the deck panel strain D6x.
. . . . . . . . . 331
I.14 Mean time and frequency domain of the deck panel strain D6y.
. . . . . . . . . 332
I.15 Mean time and frequency domain of the deck panel strain D7x.
. . . . . . . . . 333
xvi
LIST OF FIGURES
List of Tables 1.1
A summary of the RNLI RIB and IB.
. . . . . . . . . . . . . . . . . . . . . . .
6
2.1
D-class parameters.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16
2.2
Summary of DTex 1100 and DTex 1670 data sheets.
2.3
Fabric properties vs. aluminium properties.
. . . . . . . . . . . . . . .
24
. . . . . . . . . . . . . . . . . . . .
39
3.1
Comparison of experiment methods for individual aspects of hydroelasticity. . .
48
3.2
List of sensors for open water trials.
. . . . . . . . . . . . . . . . . . . . . . . .
50
3.3
Hull string potentiometer locations. . . . . . . . . . . . . . . . . . . . . . . . . .
62
3.4
Mass of crew members during wave trials (kg).
. . . . . . . . . . . . . . . . . .
67
3.5
Uncertainty of sensors.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
67
4.1
Internal pressures of the sponson and keel during static tests with no crew. . . .
73
4.2
Static sponson rotation due to varying the internal pressures with no crew. . . .
74
4.3
Internal pressures of the sponson and keel during static tests with three crew. .
77
4.4
Static sponson rotation due to varying the internal pressures with three crew.
.
77
4.5
Internal pressures of the sponson and keel during static tests with varying keel pressure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
78
4.6
Static sponson rotation due to varying only the keel pressure with no crew.
80
4.7
Internal pressures of the sponson and keel during static tests with varying crew loading.
. .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
81
4.8
Static sponson rotation due to varying the crew loading. . . . . . . . . . . . . .
81
4.9
Student's T-test of the peak accelerations measured during the 2D drop tests. .
92
4.10 Mean static internal pressures during the drop tests.
. . . . . . . . . . . . . . .
94
4.11 Student's T-test of the peak accelerations measured during the full-scale drop tests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 4.12 Mean peak internal pressures during the drop tests. . . . . . . . . . . . . . . . . 104 4.13 Mean peak trim angle during the drop tests. . . . . . . . . . . . . . . . . . . . . 104 4.14 Peak hinge angles during the drop tests.
. . . . . . . . . . . . . . . . . . . . . . 108
4.15 Peak X-axis micro-strains during the drop tests. . . . . . . . . . . . . . . . . . . 109 4.16 Peak Y-axis micro-strains during the drop tests. . . . . . . . . . . . . . . . . . . 110 4.17 Mean minimum-peak hull de ection during the drop tests. . . . . . . . . . . . . 111 xvii
xviii
LIST OF TABLES
4.18 Deformation of the sponson and keel related to the change in internal pressure during the drop tests. 5.1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Student's T-test of the top speed measured during the at water trials (excluding pulsing run).
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.2
Mean internal pressures while at rest during the at water trials.
. . . . . . . . 123
5.3
Internal pressures whilst quasi-steady planing during the at water trials.
5.4
Sponson rotation (mm) whilst quasi-steady planing during the at water trials.
5.5
Aect of the internal pressures on the planing trim angle and hinge angles.
5.6
Change in X-axis micro-strains from static C0 S2.25 K2. . . . . . . . . . . . . . 127
5.7
Change in Y-axis micro-strains from static C0 S2.25 K2. . . . . . . . . . . . . . 127
5.8
Mean deadrise angles from all the hull sensors during at water trials.
5.9
Distance between the hull and the deck.
. . . 123 124
. . 126
. . . . . 128
. . . . . . . . . . . . . . . . . . . . . . 131
5.10 Drop in keel pressure as the D-class accelerates. . . . . . . . . . . . . . . . . . . 134 5.11 Aect of the internal pressures on the planing trim angle and hinge angles.
. . 138
5.12 Change in X-axis micro-strains from static C0 S2.25 K2 for the pulsing run. . . 139 5.13 Change in Y-axis micro-strains from static C0 S2.25 K2 for the pulsing run. . . 141 5.14 Wave data recorded by Lymington Wave Buoy. 5.15 Wave data recorded by ChiMet.
. . . . . . . . . . . . . . . . . . 144
. . . . . . . . . . . . . . . . . . . . . . . . . . 145
5.16 Crest factors from the wave trials.
. . . . . . . . . . . . . . . . . . . . . . . . . 146
5.17 RMS values from the wave trials.
. . . . . . . . . . . . . . . . . . . . . . . . . 147
5.18 Unweighted VDV from the wave trials. 5.19 Seated VDV from the wave trials.
. . . . . . . . . . . . . . . . . . . . . . 148
. . . . . . . . . . . . . . . . . . . . . . . . . 161
5.20 Standing VDV from the wave trials.
. . . . . . . . . . . . . . . . . . . . . . . . 162
5.21 VDV comparison between the D-class and literature. . . . . . . . . . . . . . . . 163 5.22 Deformation of the sponson and keel related to the change in internal pressure during the at water trials.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
5.23 Savitsky prediction input valves and predicted drag forces.
. . . . . . . . . . . 167
5.24 Summary of the advantage and disadvantages of a hydroelastic slamming approach to reducing WBV.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6.1
Maximum SPL (dB(A)) for an outboard engine depending on engine power. . . 182
6.2
Comparison of noise exposure level and duration for the same cumulative 90
LEP,D
Noise Dose, Nedwell et al. (2007).
. . . . . . . . . . . . . . . . . . . . . 184
7.1
Background noise correction value.
7.2
Input properties for the ImTL model.
8.1
Main parameters and environmental conditions measured during the noise trials. 200
8.2
Maximum AS-weighted SPL.
8.3
Maximum S-weighted SPL.
8.4
Characteristic frequencies of the Mariner 50 HP outboard engine, see Mariner website.
. . . . . . . . . . . . . . . . . . . . . . . . . 195 . . . . . . . . . . . . . . . . . . . . . . . 198
. . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
LIST OF TABLES
xix
8.5
Water borne barely-audioable threshold of a harbour seal.
. . . . . . . . . . . . 211
D.1
Force and elongation at breakage of DTex 1100, DTex 1670 and two ply DTex 1100. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
xx
LIST OF TABLES
Declaration of authorship I, Peter K Halswell, declare that this thesis and the work presented in it are my own and has been generated by me as the result of my own original research. The measurement and perception of the vibrations of a RNLI D-class in atable lifeboat: Hydroelasticity, boat motion and noise I con rm that:
1. This work was done wholly or mainly while in candidature for a research degree at this University; 2. Where any part of this thesis has previously been submitted for a degree or any other quali cation at this University or any other institution, this has been clearly stated; 3. Where I have consulted the published work of others, this is always clearly attributed; 4. Where I have quoted from the work of others, the source is always given.
With the
exception of such quotations, this thesis is entirely my own work; 5. I have acknowledged all main sources of help; 6. Where the thesis is based on work done by myself jointly with others, I have made clear exactly what was done by others and what I have contributed myself; 7. Parts of this work have been published as:
(a) Halswell, P. K., Wilson, P. A., Taunton, D. J., and Austen, S. (2012). Hydroelastic in atable boats: Relevant literature and new design considerations.
Journals of Small Craft Technology, 154 (Part B1):
International
39-50. See appendix A.
(b) Halswell, P. K. (2013). Measuring the hydroelasticity of in atable boats to reduce injuries in harsh waters.
National Instruments Case Study.
See appendix B.
(c) Halswell, P. K., Wilson, P. A., Taunton, D. J., and Austen, S. (2013).
An ex-
perimental investigation into the whole body vibration generated during the hydroelastic slamming of a high speed craft. pendix C.
Signed: Date:
xxi
Ocean Engineering,
pending. See ap-
xxii
LIST OF TABLES
Acknowledgements I would like to rst acknowledge the supervision of Prof. Philip A. Wilson and Dr. Dominic J. Taunton for their immense support, help, advice and knowledge that has helped me reach the end of this thesis. In particular, thanks to Philip for all his help whenever I needed it, even if it was a just silly, simple question. I would also like to acknowledge the support from Steve Austen who facilitated this oncein-a-lifetime opportunity to work on such a unique and interesting project. Thank you also for the advice and encouragement. Another thanks to Chris Burton who helped with all the RNLI o ce based tasks. I am highly appreciative of the guidance and time from Dr. Ian Dand who went out of his way to convey his knowledge and experience of the D-class to me. The same also goes for John Barnes. Thanks you. I would like to recognise the support of all the sta at the Inland Lifeboat Centre. Without their help it would have been tenfold harder to perform all the experiments. They provide advice and tools for tting the sensors and setting up the D-class. D-class during all the tests.
They also helmed the
A special thanks to Dave for his incredible accuracy with the
crane, and to John Barnes, Peter Chandler and John Urry. I am especially in debt to Prof.
Victor Humphrey and John Dixon for their extensive
knowledge regarding both air and water borne noise. Without their help (and equipment) it would not have been impossible to complete the noise trials. My nal appreciation goes to all those outside the project who have supported me. The EngD team for the o ce banter. night get away.
The Talking Heads crew for the Friday and/or Saturday
And nally, an obvious huge thanks to my parents for their through-life
support.
xxiii
xxiv
LIST OF TABLES
Nomenclature 4
Received SPL - background SPL (dB)
a
Absorption loss coe cient
Aatm
Attenuation due to geometric spreading (dB)
Adiv
Attenuation due to atmospheric absorption (dB)
ASEN
Adjustment for miscellaneous hydrophone sensitivity (dB)
c
Speed of wave or sound (m/s)
CP A
Closest point of approach (m)
d
Water depth (m)
DW L
Data window length (m)
DW P
Data window period (s)
ε
Strain
Ewave
Wave energy (J)
F
Tip load capabilities (N)
Fr
Froude number
F rD
Froude depth number
g
Gravity (9.81
H
Wave height (m)
HE
Coe cient of hydroelasticity
kp
Compressional sound attenuation (dB/m/Hz)
log
Log base 10
l
Length (m)
L
Vessel length (m)
m/s2 )
xxv
xxvi
LIST OF TABLES
Ln
SPL of background noise (dB)
L0p
Background noise adjusted SPL (dB)
L00p
Sensitivity-adjusted SPL (after background noise adjustment) (dB)
LpHF
SPL of hull form (dB)
Ls+n
SPL of signal and background noise (dB)
Ls (r)
Power-averaged SPL for two hydrophones for all runs (dB)
Ls (r, h1)
SPL of shallow hydrophone for all runs (dB)
Ls (r, h2)
SPL of deep hydrophone for all runs (dB)
LW L
Waterline length (m)
Mz
Mean grain size (log base 2)
N
Geometric loss factor
p
Internal pressure (psi)
ρ
Density (kg/m )
Pn
Sound pressure of background noise (µP a)
Ps+n
Sound pressure of signal and background noise (µP a)
Pref
Reference sound pressure (µP a)
Prms
RMS sound pressure (µP a)
r
Radius (m)
rc
Radius of curvature (m)
rcc
Centre of curvature to centre of chord (m)
R
Sound speed ratio
SP L1m
SPL at one metre (dB)
SP LR
SPL at distance, R, metre (dB)
v
Vessel speed (m/s)
vkn
Vessel speed (knots)
Vp
Compressional sound speed (m/s)
Vs
Shear wave speed (m/s)
x
Distance to neutral axis (m)
3
LIST OF TABLES
Acronyms ADC
Analogue to digital converter
ANSI
American national standards institute
BEM
Boundary element method
BSI
British standard institute
BTM
Boundary tensioned membrane
CPA
Closest point of approach
cRIO
Compact reprogrammable input output device
CTD
Conductivity, temperature and depth
DGPS
Dierential global positioning system
DIC
Digital image correlation
DNV
Det norske veritas
DTex
Decitex
DVM
Digital voltage meter
DWL
Data window length
DWP
Data window period
EA16
Original version of the D-class
EAV
Exposure action value
ELV
Exposure limit value
FDM
Finite dierence method
FEM
Finite element method
FPGA
Field programmable gate array
FRP
Fibre reinforced plastic
GPS
Global positioning system
HP
Horse power
HSC
High speed craft
IB
In atable boat
xxvii
xxviii
LIST OF TABLES
IB1
Inshore lifeboat one (current version of the D-class)
ICOMIA
International council of marine industry association
ILC
Inshore lifeboat centre
ImTL
Image source transmission loss model
ISO
International organization for standardization
ISVR
Institute of sound and vibration research
ITTC
International towing tank conference
LCG
Longitudinal centre of gravity
LOA
Length overall
LWL
Waterline length
MDF
Medium density breboard
NI
National instruments
OSPAR
Oslo and Paris convention for the protection of the marine environment of the north-east Atlantic
PTS
Permanent threshold shift
RIB
Rigid in atable boat
RNLI
Royal national lifeboat institution
RMS
Root mean squared
RAO
Response amplitude operators
SOG
Speed over ground
SOW
Speed over water
SPL
Sound pressure level
SWOT
Strengths, weaknesses, opportunities and threats
TL
Transmission Loss
TTS
Temporary threshold shift
VDV
Vibration dosage value
WBV
Whole body vibration
ZoI
Theoretical zones of noise in uence
LIST OF TABLES
Full-scale experiment notation C#S#K# Crew loading - sponson pressure - keel pressure (static and at water trials) H#S#K# Drop height - sponson pressure - keel pressure (drop tests) A#
Accelerometer from A0 to A2
B#
Deck bridge strain gauge (deck hinge angle) from B1 to B4
D#x/y
Deck strain gauges from D1x to D7y
H#
Hull string potentiometer from H1 to H17
K#
Keel pressure transducer from K1 to K3
O#
Outboard engine strain gauge from O1 to O2
S#
Sponson pressure transducer from S1 to S7
SR#
Sponson rotation string potentiometer from SR1 to SR2
1
2
LIST OF TABLES
Chapter 1
Introduction 1 within the
This thesis will explore the eect of transport on the environment and visa versa eld of planing vessels.
Planing is a term used to describe; when the majority of a boat's
mass is supported by the hydrodynamic lifting forces caused by the angle of attack of the hull, and when the Froude number is greater than one, see Faltinsen (2005). Planing vessels are essential in coastal regions for fast transportation, including but not limited to: leisure, shing, search and rescue, and increasing interest as marine renewable-energy support vessels. Speed is always important for transportation but as the speed of planing vessels have increased it has led to problems. One issue is the eect of boat motion caused by waves on the helm and crew; the other is the eect of boat noise on wildlife. The eect of boat motion on the crew explores the eect of environmental factors on transport and the eect of noise on the wildlife studies the eect of transport on the environment. The boat motion of planing vessels can cause many health risks because the crew are exposed to high levels of acceleration. The range of health risks is large; from chronic and acute, to physiological and psychological, see Townsend et al. (2012b). Ensign et al. (2000) reports that the physiological injuries include; spinal and abdominal injuries, damage to internal organs (kidneys), torn ligaments and, broken ankles and legs; and that the psychological injuries include; annoyance, fatigue, anxiety, loss of visual accuracy and reduced hand-eye coordination. In 2005, it became a legal requirement to regulate the exposure of workers to physical vibration through the Control of Vibration at Work Regulations, see Pond (2005); however, planing vessels still exceed the limits set out by the regulations. Townsend (2008), Allen et al. (2008) and Myers et al. (2011) showed that planing crafts can easily exceed the Exposure Limit Value (ELV). The ELV is the upper limit of Whole Body Vibration (WBV) that a worker can be exposed to under the 2005 legislation. Allen et al. (2008) measured the WBV on a Royal National Lifeboat Institution (RNLI) Atlantic 75 at speed of 15 to 20 knots in sea states 2 to 3. Two trials were performed and they exceed the ELV by 23% and 131%. This shows that even at relatively low speeds (Myers et al. (2011) performed their trials at 40 knots) health risks can occur. The highest accelerations are normally linked to when the planing vessel losses contact with the water surface and then impact vertical downward, called
1 The Industrial Doctorate Training Centre at the University of Southampton, under which this thesis was completed, is titled Transport and the Environment. 3
4
CHAPTER 1.
INTRODUCTION
slamming. New technology is being developed to reduce the WBV, such as suspension seats (Coats et al. (2003); Coe et al. (2009); Olausson (2012)), suspension decks (Townsend et al. (2012a) and even porous hulls (Coats et al. (2009)). Townsend et al. (2012a) highlights some disadvantages to the current methods; for example, suspension seats increase the weight of a craft, raise the centre of gravity, occupy vital space and only function if the crew is seated. No single new technology appears to solve the problem and, Coats et al. (2003) and Coe et al. (2013) both concluded that a combination of solutions will be required to su ciently reduce the WBV to meet the legislation. Therefore, the stakeholders of these health risks are still searching for new solutions. Anthropogenic noise from maritime vessels can impact the wildlife in a variety of manners. Gotz et al. (2009) reports that the eects of water borne noise on marine wildlife include but are not limited to: masking, behavioural disturbance, hearing loss (temporary or permanent), discomfort, stress, injury or even death. This is cause for concern especially when considering underwater noise over airborne noise because humans rarely hear underwater noise and have little comprehension of its potential impact.
The eects of water borne noise from planing
vessels can be magni ed in shallow water because planing vessels are: a) faster than marine species so the wildlife cannot out run loud, dangerous noises, and b) operate close to the shore where the wildlife are trapped in areas such as harbours, river mouths or lakes. There is a clear division in the environmental noise of planing vessels; even the conventional reporting format is divided with airborne noise being reported at the measurement distance but water borne noise measurements being propagated back to one meter from the source. Furthermore, the eect of boat noise from planing vessels on the environment is divided in terms of standards and regulations. The division is ampli ed in shallow water because water depth has minimal eect on airborne noise but it dominates the sound propagation through a shallow water sound channel.
Airborne noise is well understood with ISO measurement procedures (ISO 14509)
suitable for planing vessels and a European directive (2003/44/EC) aimed at small recreational crafts including planing vessels; there is even a database of airborne noise measurements, see SoundBoat (2005).
On the other hand, water borne noise is less appreciated.
There are
measurement procedures for large ships (ISO 17208-1:2012 and ANSI S12.64-2009) but these are not suitable for smaller planing craft in shallow water primarily due to the shear dierence in scale between a hundred meter or more container ship and (say) a ve meter planing vessel, see 6.2.2 for more details. Therefore, an applicable method is required to measure the water borne noise of planing vessels so that an accurate environmental assessment can be made. There have been a number of water borne noise measurements of planing craft, see Young and Miller (1960); Kipple and Gabriele (2003, 2004b,a); Amoser et al. (2004) but the Sound Pressure Level (SPL) was frequently reported at the measurement distance with minimal consideration for the shallow water sound channel.
The shallow water channel can amplify
or attenuate the SPL due to re ections from the water surface and the channel bottom, see Loeser (1999). The measurement procedure for ships uses deep water to remove re ection of the channel bottom, which allows the transmission loss caused by propagation of noise through water to be accurately predicted. Accurate prediction of the transmission loss is essential for
5
environmental assessments so that the SPL at a given distance from the noise source can be predicted; therefore, the boundaries conditions and variables of the shallow water sound channel must be known and a suitable propagation model is required to accurately predict the environmental impact of a planing craft in littoral waters. This Engineering Doctorate is supported and partially funded by the RNLI. The RNLI is the charity that aims to save lives at sea all around the coasts of the UK and Ireland. They design, build, maintain and operate a range of 10 vessel classes for almost any situation possible in the UK waters. For the RNLI, planing vessels are the primary choice in littoral waters and they own the largest eet of In atable Boats (IB) and Rigid In atable Boats (RIB) in the UK. The primary RIB and IB used by the RNLI are summarised in table 1.1 and it emphasises range of RIB and IB and their uses; there is even a 3 m Y-class IB used as a tender boat for the Severn and Tamar classes. The D-class is a 5 m IB with a single outboard engine capable of achieving 25 knots. The D-class and the B-class, an 8.5 m RIB, are shown in gure 1.1. The D-class operates in harbours, bays and very close to the shoreline due to its shallow draft, manoeuvrability and outstanding performance in large breaking waves (reported by the RNLI). The construction of the D-class is unique compared to other planing vessels because it has a fabric hull and a hinged deck which allows it to ex and deform considerably more than conventional wood, metal or bre reinforced plastic hulls. Moreover, this deformation can be aected and loosely controlled by the internal pressures of the sponson and keel. The RNLI have reported that the keel pressure aects forward speed and boat motion in waves of the D-class. This introduces an unusual and rare parameter that should aect the performance of a planing vessel, exibility.
Figure 1.1: RNLI B- and D-class.
Boat motion and noise are both types of vibration and a vibration is directly aected by exibility. The simplest vibration is simple harmonic motion and it can be mathematically described using equation 1.1, where exibility is the spring constant (k), m is the mass and x
6
CHAPTER 1.
INTRODUCTION
Vessel
Arancia
D-class
B-class
E-class
Also known as
-
IB1
Atlantic 85
Mk2
Length (m)
3.88
5
8.44
10.5
Beam (m)
1.73
2
2.85
2.9
Draught (m)
-
0.277
0.53
0.7
Displacement (kg)
165
400
1800
5900
Speed (knots)
26
25
35
40
Range
20 miles
3 hours
2.5 hours
3 hours
30 hp
50 hp
2 x 115 hp
435 hp
outboard
outboard
outboards
inboard
Number of crew
2
2-3
4
4
Number of casualties
5-6
5-6
20
20
Fabric
Fabric
Composite
Composite
Propulsion
Hull type Operational area
shallow-V
shallow-V
deep-V
deep-V
Near beaches
Littoral water
Coastline
Thames
Table 1.1: A summary of the RNLI RIB and IB.
is the displacement. This demonstrates that exibility will aect the motion of a vibration.
M x + kx = 0
(1.1)
This thesis intends to demonstrate whether or not exibility can be used to solve current and future challenges within the performance of planing craft. Performance is de ned by boat motion, which is a current challenge, and boat noise, which is a future challenge because the environmental impact should always be minimised and ISO are currently trying to develop a methodology to measure the water borne noise of marine vessels in shallow water, see ISO 17208-1:2012. The perception of a vibration by living organisms must be included to fully consider these problems. There can be signi cant dierences between a measured vibration and the perception of the same vibration. For example, the ranges of audible frequencies to humans is 20 Hz to 20 kHz, see Robinson and Dadson (1956); therefore, it is possible to measure a loud noise at 40 kHz but it would not be audible or perceivable to human. Perception is not new and some aspects have been well researched normally in conjunction with the relevant standards and regulations of both vibration problems. Human's perception of airborne noise has been very well studied; appropriate weighting schemes have been developed and are applied in the European directive (2003/44/EC), see section 6.3.1. Conversely, water borne noise of planing vessels does not have any applicable European directives. Audiograms, which are the frequency dependent sensitivity of a species auditory system, have been recorded for a number of marine species, see Nedwell et al. (2004), but there can be large variability between dierent species; consequently, this makes it di cult to truly consider the perception of water borne noise. The perception of WBV is considered in the Control of Vibration at Work Regulations and uses weighting methods to account for the frequency dependent response of human body. The weight scheme is applicable to repeated shock and that applies to a planing vessel moving
1.1.
APPROACH TO BOAT MOTION
7
forward; however, the slamming characteristics of a planing craft are often investigated using drop tests or 2D models. Only an individual shock is measured during a drop test or a 2D model and the perception of WBV is di cult to interpret. The perception of the vibrations will be considered within this project but they are not the focus. The overall goal of the thesis is to measure or predict the eect of exibility on the two physical vibrations: boat motion and boat noise. Then the thesis wishes to predict the eect of exibility on the perception of the two vibrations.
1.1 Approach to boat motion The rst and possibly the most important questions to ask when considering the eect of exibility on boat motion is; how much exibility is required to noticeably aect the boat motion?
Townsend et al. (2012a) used 2D numerical models to predict whether the hull
stiness could aect the slamming accelerations (i.e. vertical acceleration caused by the hull vertically impacting the water surface). The Young's modulus of the hull was reduced from 69 GPa to 6.9 GPa but this had minimal eect on the accelerations. This suggests that the hull stiness must be reduced further to have a noticeable eect. The D-class will now be used as an example boat because it has been reported that exibility aects the motion. The stressstrain relationship of the rubber-coated fabrics used on the D-class was not known so tensile tests were performed as an appendix experiment of this thesis to help answer this question, see appendix D. The tensile tests showed that the Young's modulus of the rubber-coated fabrics ranged from 0.28 GPa to 0.41 GPa, which is consider hyper-elastic compared to conventional materials.
This provides a benchmark magnitude of exibility required to noticeably aect
the boat motion. A new phenomenon occured when the hull stiness was reduced to the levels on the Dclass and the RNLI called it the pulsing motion, see Dand et al. (2008). The pulsing motion stopped the D-class from continuously planing on at water because a pressure wave under the hull built up and was released periodically, leading to a pulsing motion, see section 2.2.2 for more detail. The impact of the pulsing motion means a purely 2D investigation of the hull stiness on the slamming acceleration is not satisfactory on its own because it only considers the vertical motion; the horizontal motion may be detrimentally eected by exibility, thus stopping a planing-vessel planing.
The pulsing motion was removed using a care trial and
error process by Dand et al. (2008). The term exibility is no longer su cient to describe the deformation of a planing craft because it does not include the vital interaction between the structure deformation and the uid forces.
The term hydroelasticity is used to de ne this uid-structure interaction, see
Bereznitski (2001). The uid applies a force to the structure which causes the structure to deform; simultaneously, the deformation of the structure causes a change in the uid forces, see gure 1.2. Current hydroelasticity theories break the problem down into global and local hydroelasticity. Global hydroelasticity looks at the interaction of the overall deformation of the vessel with waves by comparing the ship's structure to an I-beam, see Bishop and Price
8
CHAPTER 1.
INTRODUCTION
(1979); whereas, local hydroelasticity investigates the deformation of the local hull panels and is often simpli ed using 2D models of a transverse slice through the ship, see Faltinsen et al. (2004). This thesis refers to local hydroelasticity as hydroelastic slamming, which is mainly concerned with the vertical accelerations generated by the vessel impacting the water surface. The D-class shows that both local and global hydroelastic aspects are important because the fabric hull of the D-class is loose in places leading to large local hull deformations and the bow visibly de ects due to interactions with waves, seen by the author during early investigations. However, neither conventional local or global hydroelastic theories can describe the pulsing motion exhibited by the D-class.
This is because the I-beam analogue using in global hy-
droelasticity does not consider local hull deformations and hydroelastic slamming commonly uses 2D simpli cations. Therefore, a third and very new aspect of hydroelasticity has been introduced called hydroelastic planing surfaces. Only one 2D analytical model of a hydroelasticity planing surface has been found, see Makasyeyev (2010), which shown that this aspect is very much in its infancy. Furthermore, it is hypothesised that local and global hydroelasticity cannot be considered separately for vessels that are exible enough to noticeably aect the boat motion because the deformations are so large that the aspects of hydroelasticity interact. This can be demonstrated by considering how global hogging and sagging of a vessel will aect deformation of local hull panels. Hogging forces the keel and hull in compression thus allowing the local hull deformation to be ampli ed; whereas, sagging will pull the bottom of a vessel into tension and reduce the local hull deformation. It is also thought that hydroelastic planing surfaces cannot be considered separately to local or global hydroelasticity due to the large deformations. The interaction between the hydroelastic planing surfaces and, local and global hydroelasticity can also be proved by considering the forward speed.
There is a non-linear
relationship between forward speed and boat motion of RIB and IB, see Townsend (2008) and Dand (2004), so if a hydroelastic planing surface aects the forward speed then this will aect the boat motion and in turn change the local and global hydroelasticity. The consideration of all three aspects at once could be termed holistic hydroelasticity.
Figure 1.2: Hydroelastic cycle.
This leads to the question, how can this problem be simpli ed and modelled? The following is a list of criteria that must be included to accurately model the boat motion of a hydroelastic planing vessel: planing, 3D, hyper-elastic materials and hydroelasticity. Hydroelasticity can be expanded to include slamming phenomena, such as: air entrapment, air compressibility, bubble
1.1.
APPROACH TO BOAT MOTION
9
generation and compressibility of water, see Kapsenberg (2011). Kapsenberg (2011) also says only a computational method properly including all these eects will give an accurate answer; also model test will not be capable of doing this, if only because the methods to extrapolate the results of model to full scale are not yet developed.
Kapsenberg goes on to discuss
global hydroelasticity separately to local hydroelasticity, whereas, both need simultaneously considering in this problem. This shows it is very di cult to simplify this complex problem and the most suitable method is full scale experiments because numerical methods and scaled models are not suitable to fully explore the problem. In this instance (this project) there is a suitable planing craft that can be used to investigate the eect of exibility on boat motion and that is the D-class.
Therefore, full scale experiments will be used to learn whether or
not exibility (and hydroelasticity) can be used to help solve the current problem of WBV in planing vessels. The full-scale experiments will include static tests, drop tests, at-water trials and open-water wave trails; each sub-experiment was picked to trigger one aspect of hydroelasticity at a time but include the interaction of all three aspects. The internal pressure of the sponson and keel can be used to loosely control the overall stiness of the D-class. This project has the unique opportunity to measure the full scale deformations of a hydroelasticity planing craft and, by performing the experiments at full scale, no simpli cations have to be made that might overlook critical interactions.
That being said, numerical methods or
scale models will still be explored. Numerical methods will be studied through the literature review to provide the full background of the problem and future numerical methods will be suggested. Scale models are also reviewed and six are suggested for studying individual aspects of hydroelasticity.
Finally, a quasi-2D model of a fabric wedge impacting a free surface is
employed to measure the eect of only hull stiness on peak vertical acceleration. There is no benchmark or data to compare the results of the full-scale experiments with, so the quasi-2D model provides this reference point and hone in on one component of hydroelasiticity that is likely to aect the slamming characteristics and WBV.
The Control of Vibration at Work Regulations provide applicable methods for predicting the perception of boat motion for repeated shocks.
There are two common methods for
repeated shock, Root Mean Squared (RMS) referenced to eight hours, A(8), and Vibration Dosage Valve (VDV), and these are applicable when a vessel is moving forward because the shock is repeated. These methods can be used in full-scale open-water trials. However, the author could not nd any prediction methods for an individual shock, such as drop tests or 2D models. The current methods consider the accumulative eect of shock over time, which cannot be considered during an individual slam; however, the regulations do recommend using the frequency weightings in International Standard ISO 2631-1:1997.
Thus, using A(8) or
VDV for an individual shock can account for the frequency dependent response of the human body but not the accumulative eect.
10
CHAPTER 1.
INTRODUCTION
1.2 Approach to boat noise There is a vast dierence in standards, regulations and knowledge regarding airborne and water borne noise of planing vessels, as mentioned earlier. There is su cient knowledge about the propagation of airborne noise that standards have been developed to repeatably measure the airborne noise of planing craft and there are regulations limiting the maximum SPL a vessel can generate. Therefore, the standard method can be used to measure the eect of exibility on the airborne boat noise and the regulations can be used to predict the perception. On the other hand, water borne noise does not have a standard method; instead a method has to be developed within this thesis before the eect of exibility on the water borne boat noise can be measured. Then the perception of water borne noise needs to be predicted but there are no regulations and there is an extensive variety of marine wildlife where the perception will vary between species. So, the thesis focuses more on the water borne noise than the airborne noise due to the need to develop a measurement method and a method to predict the perception. The rst step in understanding water borne noise is to critique the current water borne noise measurement methods developed for ships in deep water; this will demonstrate what needs to be altered to make it suitable for planing vessels in shallow water. Then a method can be developed to account for the dierences between ships and planing vessels, and between deep and shallow water. Once a method has been developed the perception must be considered. Due to the large variety of marine wildlife, a case study will be performed on one species to predict their perception of the boat noise. This requires an audiogram of that species and the harbour seal was chosen because air and water borne audiograms are available. This provides a unique opportunity to predict the perception of both air and water borne noise by the same species. The airborne noise regulations use a weighting scheme based on the human audiogram and not wildlife, which leads to the assumption that the SPL limits are based on the human audiogram as well. Therefore, the environmental impact of airborne noise from planing craft is also considered. The eect of hydroelasticity on perception of planing craft noise can be explored once the method to measure water borne noise and the method to predict perception have been developed.
The D-class is still a suitable vessel to measure the eect of hydroelasticity on
boat noise for the same reason explained in the approach to boat motion (i.e.
the D-class
is considerably more exible than conventional vessels). The same parameters, the internal pressures of the sponson and keel, can be varied by the same magnitude as the full-scale boat motion experiment. The knowledge gained during the boat motion trials might also provide insight in to the eect of hydroelasticity on boat noise, if hydroelasticity aect boat noise.
1.3 Aims The overall aim of this project is to gain more knowledge regarding the main vibrations (boat motion and sound) found within D-class by exploring the perception of the vibrations and the eect of hydroelasticity on the perception. First of all, this involves stating the current level of knowledge regarding these vibrations through the literature review. Next, accurate
1.3.
AIMS
11
and repeatable methods for measuring these vibrations must be found. These measurements can then be used to estimate the magnitude of perception and eect of hydroelasticity on the perception. From the academic perspective, one aim is to prove whether or not hydroelasticity can minimise the WBV. Another aim is to develop a method for measuring the water borne noise of high speed vessels in shallow water. The nal aim is to quantify the noise of the D-class. From the perspective of the RNLI, the primary aim of this project is to prove whether or not hydroelasticity improves or worsens the performance of the D-class, where the performance is de ned as planing performance (forward speed and the pulsing motion) and the sea keeping performance. If hydroelasticity is proven to improve the performance then results should be written up into design guidelines. The secondary aim is to measure the air and water borne noise to determine whether or not noise reduction methods should be considered in the future.
1. Develop a method to measure the rigid body motion and the hydroelastic vibrations of the D-class.
2. Measure the rigid body motion of the D-class and the eect of hydroelasticity on this rigid body motion.
(a) Measure the performance and deformation of the D-class during stationary tests, drop tests, at water trials and wave trials. (b) Gain further knowledge regarding the deformation of a hydroelastic fabric planing surface.
3. Predict the magnitude of perception of the rigid body motion and the eect of hydroelasticity on the perception of the rigid body motion.
(a) Prove whether or not hydroelasticity is a novel method for reducing the WBV.
4. Develop a method to measure the water borne noise of RIBs and IBs using international standards and previous literature as a guideline.
(a) The method is aimed at small planing craft in shallow water. (b) ISO 14509 provides a method for measuring the airborne noise.
5. Measure the air and water borne noise generated by the D-class, and the eect of hydroelasticity on the noise.
6. Predict the magnitude of perception of air and water borne noise and the eect of hydroelasticity on the perception of air and water borne noise.
12
CHAPTER 1.
INTRODUCTION
1.4 Thesis structure There are two halves to this thesis; boat motion and boat noise.
They are both types of
vibration and there are many similarities; for example both require a method to be developed, then measurements to be made and nally a prediction of the perception. Furthermore, the same boat is investigated and the same parameter (the internal pressure of the sponson and keel) is varied; however, there is no literature that is relevant to both side of the thesis. Therefore, it has been decided that the two halves will be discussed separately. First, the boat motion and hydroelasticity will be covered, and then boat noise will be discussed. Chapter 2 reviews the literature relating to boat motion and hydroelasticity. It starts by reviewing RIB and the D-class, including the construction and previous experiments performed on the D-class. Then boundary tensioned membranes are covered as this provides an insight in to hyper-elastic materials.
Next, the three aspects of hydroelasticity are de ned and a
literature review is provided; followed by an examination of the perception of boat motion. Chapter 3 discusses in more detail the alternative methods to investigating boat motion and hydroelasticity, and provides more insight into the nal chosen method. First, potential numerical methods for studying individual aspects of hydroelasticity are reported followed by the experimental methods for studying individual aspects. The second half of chapter 3 fully describes the quasi-2D drop test and full-scale open-water experiments that will be performed as part of this thesis. The results of the quasi-2D drop test and full-scale open-water experiments are split into vertical and horizontal loads. Thus, full-scale static tests, quasi-2D drop test and full-scale drop tests are analysed and discussed in section 4 (vertical loads). The drop tests provide the rst insight into the eect of hydroelasticity on boat motion. The at water trials and wave trials are analysed and discussed in section 5 (horizontal loads). The at water trials discover how hydroelasticity aects planing performance (forward speed) of the D-class and the wave trials give the second understanding of the eect of hydroelasticity on boat motion. Chapter 6 discusses the International Standards and literature relating to boat noise. First it reviews airborne noise then is examines water borne noise. The International Standards for water borne noise are thoroughly critiqued. The end of the chapter considers the perception of noise by humans and wildlife. Chapter 7 begins by outlining how the International Standards need to be altered to accommodate planing crafts in shallow water, including: course layout, instrumentation and post processing. Then, the chapter describes the method used to simultaneously measure the air and water borne noise of the D-class. The results of the air and water borne noise trials are analysed in chapter 8 using a maximum SPL (in accordance to the airborne noise regulations) and one-third-octave spectrum (suggested by the water borne noise standards). The perception of the boat noise is considered through a case study of a harbour seal and uses three dierent methods. Chapter 9 draws the two halves together by concluding how exibility and hydroelasticity aect boat motion and boat noise. This chapter also reiterates whether or not hydroelasticity
1.5.
NOVEL CONTRIBUTIONS
13
could be used to solve the current and future problems within planing vessels.
1.5 Novel contributions The research in this thesis has provided novel contributions to the following areas:
A detailed breakdown of the three aspects of hydroelasticity in exible planing vessels (namely hydroelasticity slamming, hydroelasticity planing surfaces and global hydroelasticity) caused by the addition of hydroelastic planing surfaces, which invoked the development of an experimental method to trigger individual and combined aspects of hydroelasticity.
An experimental investigation into the eects of hydroelasticity on the peak slamming accelerations during drop tests, both quasi-2D half-scale and full-scale models.
An experimental investigation into the eects of hydroelasticity on the accelerations and WBV using full-scale open-water wave trials.
A simpli ed method to operate the National Instruments CompactRIO without a real time controller.
A development of a method to simultaneously measure air and water borne noise generated by planing vessels in shallow water.
Analytical methods for predicting the environmental impact of anthropogenic boat noise on an individual species and a generic method based on background noise.
The research has been presented and/or published in the following mediums:
1. Halswell, P. K., Wilson, P. A., Taunton, D. J., and Austen, S. (2012). Hydroelastic in atable boats: Relevant literature and new design considerations.
of Small Craft Technology, 154 (Part B1):
International Journals
39-50. See appendix A.
2. Halswell, P. K. (2013). Measuring the hydroelasticity of in atable boats to reduce injuries in harsh waters.
National Instruments Case Study.
See appendix B.
3. Halswell, P. K., Wilson, P. A., Taunton, D. J., and Austen, S. (2013). An experimental investigation into the whole body vibration generated during the hydroelastic slamming of a high speed craft.
Ocean Engineering, pending.
See appendix C.
14
CHAPTER 1.
INTRODUCTION
Chapter 2
Hydroelastic literature review 2.1 Rigid in atable boats In 1981, 1998 and 2005, three International Conferences were held in the UK to discuss the design and development of RIBs, RINA (1981, 1998, 2005).
However most of the evidence
was anecdotal and there was little scienti c proof using experimental or numerical methods. The topics covered included; history, development, construction techniques, propulsion, problems with model test, self-righting issues, example boats, safety, in uence of the helmsmen, electronics and equipment.
Stevens (1981) presented work on the di culties opposing the
testing of scale model IBs. The scaling of the overall structural stiness (bending, tensional and torsional) and the stiness characteristics of the fabric needs to be carefully considered but the scaled fabric properties are limited by the products currently being manufactured. Stevens noted that outboard motors also aect the testing technique because the thrust line is below the water. Moreover, the major problem arises with the scaling of atmospheric pressure because the internal pressure of the in atable components is dependent on the atmospheric pressure, which is the same at full and model scale. A bellow and spring system can be used; however, this increases the weight which is already a problem for scale model experiments. Haiping et al. (2005) undertook experiments into the eect of sponson type on the sea keeping performance of RIB. It was found that an in atable sponson had a lower Response Amplitude Operators (RAO) in heave and pitch than a foam sponson in the two load conditions. This suggests that the more exible sponson (i.e. the in atable sponson) improved the ride comfort and sea keeping performance. The only known numerical model for a RIB was developed by Lewis et al. (2006). The numerical model was used to predict the boat motion of a wave piercing RIB and the RNLI B-class in regular and irregular waves. The numerical model, which was based on non-linear strip theory, was compared with experimental results and, although the results looked promising, the results from the numerical model over predicted the experimentally measured results. The computational model was a hydrodynamic model meaning the structure was assumed rigid and hydroelasticity was not considered. Townsend et al. (2008a,b) performed a multitude of experiments to characterise the sea keeping performance of the RNLI Atlantic 75 RIB. Townsend et al. (2008a) studied the in uences of speed, ballast, 15
16
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
wave height, encounter frequency, and tube pressure on the boats motion of the Atlantic 75. This shows that minimal scienti c research into the performance of RIBs has been completed; however, the D-class has been redesigned and gone through a systematic trial and error process to improves its performance. This will be discussed in the following section.
2.2 D-class The basic parameters of the D-class found during the literature review are shown in table 2.1. Parameter
Value
Unit
m m m m
Length overall
5.12
Length on waterline
4.051
Breadth on waterline
1.903
Maximum static draught
0.277
Deadrise angle
0 to 15
O
Static trim
2.1
O
Lightship minus crew
436
kg
LCG (from transom)
1.32
m
Block coe cient
0.36
Prismatic coe cient
0.669
Midship area coe cient
0.609
Static wetted surface area
6.4
Transom immersed area
0.093
m2 m2
Table 2.1: D-class parameters.
2.2.1 Construction The D-class has a unique design and it is vital to understand the design and construction to be able to understand how the boat can deform. There are four main components; the sponson, hull, keel and deck, shown in gure 2.1. The unique design will now be shown through the construction process. Then it will be shown that each part is coupled together so that the deformation of one component aects the deformation of other components. The sponson is the in atable tubes that surround the boat, see gure 2.2; it is constructed from Hypalon®/Neoprene coated polyester fabric and is in ated to a pressure of 3.25 psi (224 mbar). There are seven internal compartments divided by ba es so that if one or more sections are punctured the boat is still functional. The sponson is the rst step in the construction process. The fabric hull is a rubber-coated fabric sheet currently constructed from two sheets of Hypalon®/Neoprene coated polyester fabric laminated together, see gure 2.3.
When the
original manufacturers of the rubber-coated fabrics stopped production it caused the RNLI many unexpected issues because other similar fabrics did not provide the same performance, even though they were reported to have very similar properties. Many fabrics were tested and nally it was found that the lamination of two sheets provided the best planing performance.
2.2.
D-CLASS
17
Figure 2.1: Main components of the D-class.
The second step is to glue the transom and bow board in between the sponson, shown in gure 2.4. The third step in the construction process is to glue the hull to the underside of the sponson and the transom. The transom and all the deck panels are laid up to form a 30 mm thick glass bre reinforced epoxy sandwich panel. The keel is a tapered in atable tube that runs along the centreline of the fabric hull, see gure 2.4. It is constructed from Hypalon®/Neoprene coated polyester fabric and is in ated to a pressure of 3 psi (206 mbar). It is glued at speci c places to the centreline of the hull for the fourth step of the construction process. The individual deck panels are the stiest structural components in the boat. They are manufactured from a 30 mm thick glass bre reinforced epoxy sandwich panel.
The deck
is sectioned into four parts to intentionally allow exibility and each deck hinge has its own rotational stiness due to the friction within the hinge and the bending stiness of the sponson and keel. The deck panels are called; the bow board (1), thrust board (2) and main deck (3 & 4), see gure 2.5. The aft hinge between the two aft most deck panels has a physical bond to hold the two parts of the deck together; however, the two front hinges act like a pivot point so that compression is required to hold them in place. The fth stage in the construction is to slot the thrust board and the main deck in place; there is no physical bonding process. The three hinges are shown in gure 2.6.
Hinge one is in between the bow board and
thrust board. It is a simple aluminium tongue, shown in the lower gure, which slots into a matching aluminium groove. Hinge two is in between the thrust board and the main deck. It is again a simple tongue and groove hinge manufactured from plastic. Hinge three holds the main deck together and it is constructed from several layers of machine pulley belt which are glued and screwed into place (this is the only hinge that can resist tension). The keel is in ated during the last stage of construction, which forces the hull downwards
18
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
Figure 2.2: In atable sponson.
Figure 2.3: Laminated fabric hull.
2.2.
D-CLASS
19
Figure 2.4: In atable keel running along the centre line of the vessel.
Figure 2.5: Four composite deck panels.
20
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
Figure 2.6: Deck hinges.
to de ne the shape of the planing surface and it forces the deck upwards. This stresses the structure and only now does the structure have the stiness and shape to function properly. It is (anecdotally) reported that if the keel is over in ated by only 0.25 psi it causes the boat to hog and reduces the performance.
2.2.2 D-class literature Dand (2002a,b, 2003b,a,c, 2004) performed a number of experiments to explore the performance of the original D-class, the EA16. Dand rst studied the open-water propeller characteristics and strut drag of the Mariner 40 HP outboard engine. In 2002 and 2003, Dand measured the model and full scale resistance of the EA16 in a towing tank and calm water sea trials. During the towing tank experiments, Dand also explored the eect of the longitudinal centre of gravity (LCG) on the hydrodynamic performance. Finally, Dand used the scale model to measure the sea keeping performance in head seas. Austen and Fogarty (2004) documented the re-development of the D-class, from the EA16 to the IB1 (the current version of the D-class), because new materials and construction methods were used. When introduced into service, the D-class suered performance problems due to the fabric oor, ventilation and cavitation. Dand et al. (2008) used a careful trial and error process to increase the speed from 20 knots to 25 knots. The important knowledge gained from this literature is now summarised. Figure 2.7 shows the measured and predicted total resistance of the EA16, see Dand et al. (2008).
Planing begins at around 10 knots and then there is a plateau which causes an
instability in the boat speed, allowing it to jump from 10 knots to 16 knots. predictions were derived from the results of a rigid quarter-scale model.
The model
It was expected
that there would be a signi cant dierence between resistance coe cients of the hydroelastic (full scale) and rigid (1/4 scale) models, however, this was not the case and the resistance coe cients were surprisingly similar. Dand noted that the rigid scale-model showed forms of blister spray which adhered to the sponson and entered the boat. This was resolved by adding longitudinal spray strips along the underside of the sponson and transverse strips across the end of each sponson in similar positions to the reinforcing patches of fabric on the D-class.
2.2.
D-CLASS
21
Figure 2.7: Total resistance of the D-class, Dand et al. (2008).
This indicates that some of the reinforcing patches on the boat have secondary functions. Dand (2003b) further explored the deformation of the fabric hull by exploiting underwater photographs as the boat passed at various speeds, see gure 2.8. It was shown that the hull did distort and at 12 knots or more the fabric touched the rigid deck and this became the eective planing surface. It also appears that the in atable keel distorts as well as the fabric hull. The similarities between the elastic and rigid model, see gure 2.7, implies that the hull and keel distortion does not aect the resistance of the hull. The photographs did not show whether the distortion was static or dynamic.
Figure 2.8: Hull distortion of the D-class at 19.4 knots, Dand et al. (2008).
22
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
Dand (2004) measured the sea keeping performance (mainly the pitch and heave response) of the rigid 1/4 scale model of the D-class in regular and irregular waves at various speeds. The behaviour of the model became non-linear at high speeds (> 18 knots) and a secondary response was found.
The peak accelerations in regular waves were higher than in irregular
waves. The highest vertical acceleration measured in the crew's position was 3.9 g and 2.5 g in the helm's position. During the coastal evaluations of the D-class in 2004, Austen and Fogerty found the boat performance quickly reduced with the incorrect keel pressure. It was corrected by acquiring more accurate pressure gauges but this highlights the importance of the keel pressures. Dand et al. (2008) discussed the deterioration of performance in the D-class, which aected both the sea keeping performance and the speed of the vessel. The sea keeping performance was aected by heavy spray not being thrown clear of the boat, similar to the issue with the scale model. The speed issue was a consequence of a pulsing motion in the pitch direction, similar to porpoising. When the boat nearly reached full speed, it appeared that a pressure wave would form under the bow along with a drop in speed. The wave would pass under the boat and observations showed that the sponson rotated, and then the speed was recovered. This would occur approximately every 30 seconds. If the crew were located o-centre then the pressure wave would be released on one side causing the boat to kick sideways. This aect only occurred in at water, i.e. when the water was poppled or in the presence of waves the pulsing was absent. Sea planes have been known to have a similar problem when they stick to the water's surface; von Kármán (1929) and Wagner (1932) studied this. The re-development of the EA16 into the IB1 demonstrated the dependency of the planing performance on the boundary conditions, pre-tensioned stresses and material properties of the hull.
Austen and Fogarty (2004) changed the deck of the D-class from marine plywood to
bre reinforce plastic (FRP) sandwich panel to increase the overall (bending, tensional and torsional) stiness; this reduced the deck deformation leading to a stier fabric planing surface and better performance. Dand et al. (2008) found the composite deck was being manufactured outside the design tolerances which caused the boundary conditions and pre-tensioned stresses in the hull to change. During the re-development the fabric manufacturer was changed and the new fabrics had dierent material properties and thus dierent static and dynamic responses. Austen and Fogarty (2004) used a novel material testing method to nd the most suitable material combination for the fabric hull.
2.2.3 Structural coupling The structure of the D-class is highly complex and not even one component of the D-class can be considered rigid. Even the complete deck would collapse if it was used as a simply supported beam because two of the hinges cannot resist tension. A BTM is de ned because it cannot support out-of-plane bending moments and the boundary conditions allow the membranes to support these loads, see section 2.3 for more detail on BTMs.
It is the same for the D-
class; it is the boundary conditions imposed by the interacting components that gives the structure its stiness and ability to resist loads. The complexity of the D-class arises because
2.3.
BOUNDARY TENSION MEMBRANES
23
the deformation of one (of the four) components is dependent on two other components, then the two other components are dependent on the fourth component. For example: the tension in the hull is dependent on the internal pressures of the sponson and keel but the sponson and keel are dependent on the shape of the deck and deck hinges. This shows that each component is coupled together and the structure is likely to be statically indeterminate.
2.3 Boundary tension membranes The deck of the D-class is constructed from a composite sandwich panel but the hull, sponson and keel, like many other IBs, are constructed from a rubber-coated fabric. Sandwich panel structures have been well studied in comparison to the rubber-coated fabric used on the Dclass. Information on composite sandwich panels can be found in Gere and Goodno (2009). A fabric has no out-of-plane bending stiness, see Lewis (2003). It is the combination of the boundary conditions and the tensile properties of the fabric that enable it to be used as a structural component. These fabric structures are called BTMs. The sponson is a BTM and it is constructed from sealed membrane tubes to form in atable cylindrical beams, however, the cylindrical symmetry of the tube means that they provide their own boundary conditions and the internal pressure places the bres in tension. They can withstand compressive and shear forces; therefore, they can be analysed using dierent methods. The static and dynamic behaviour of BTMs and the material properties of the rubber-coated fabrics are not thoroughly understood so this section will investigate them in more detail.
2.3.1 Material properties The fabric used on the D-class is Hypalon®/neoprene coated polyester fabric manufactured by Aerazur and the construction of the fabric is shown in gure 2.9. The external coating is a blend of Hypalon and Neoprene to optimise the weathering, abrasion and ozone resistance. The internal coating is a Neoprene compound to provide good adhesion and air permeability. The in atable sponson and keel are constructed from a fabric with a yarn count of Decitex (DTex) 1670 and the hull is constructed from two sheets with a yarn count of DTex 1100 laminated together. The dierence between the two fabrics is the thickness, the weave density and the bre construction.
24
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
Figure 2.9: Construction of rubber-coated fabrics, Hot Ribs (2011).
The data sheets provided by manufacturers typically only provide a minimum force at failure (and possibly an elongation) and a linear relationship has to be assumed. A summary of the data sheets for the DTex 1100 and DTex 1670 fabrics by Aerazar can be seen in table 2.2. Property
Unit
DTex 1100
DTex 1670
Material
-
Polyester HT
Polyester HT
Yarn Count
DTex
1100
1670
Type of Weave
-
Plain
Plain
Woven Weight
g/m2
250
355
Construction
Ends + Picks 1cm
10.5 x 11
10 x 10
Min. Tensile Strength - Warp
kg/50mm kg/50mm mm g/m2
306
510
Min. Tensile Strength - Weft Gauge Coated Weight
306
470
0.85
1.25
980
1500
Table 2.2: Summary of DTex 1100 and DTex 1670 data sheets.
Lewis (2003) states that these rubber-coated fabrics have a physical non-linear response so it is important to understand the stress-strain relationship of these fabrics to understand their response. Main et al. (1994) showed that at high pressure (i.e. high stresses) the change in material properties due to biaxial loading had a noticeable eect. The level of knowledge about the fabrics used on the D-class is not adequate for the current numerical techniques; therefore, the material properties of these fabrics should be obtained experimentally. This will provide new results for the RNLI and help further our knowledge regarding the response of the fabrics. The International Standard ISO 1421:1998 (Rubber- or plastic-coated fabrics - Determination of tensile strength and elongation at break) sets down a method for measuring the tensile strength and elongation at the point of failure. The principle of this method involves
2.3.
BOUNDARY TENSION MEMBRANES
25
extending a test coupon at a quasi-static constant rate of extension until it fails. This method can be used to measure the quasi-static stress-strain relationship of the fabrics used on the D-class. There are no international standards for testing the high strain rate of rubber-coated fabric to the author's knowledge. High strain rates are important because the time scale of these events can be very short; the duration of a slamming impulse is typically less than 0.2 s, see Faltinsen et al. (2004).
The most relevant standard in relation to high strain rates is
ISO 26203-1:2010 which describes how to perform high strain rate experiments with metallic structures. This standard states that for metallic test samples the strain rate should vary from
10−3 s−1 to 103 s−1 . The stress-strain relationship of DTex 1100 and DTex 1670 was measured using ISO 1421:1998 to investigate the biaxial properties, the eect of thickness and the eect of lamination. The results of this experiment can be found in appendix D.
2.3.2 Form nding Lewis (2003) provides a good review on BTMs and states that the static shape of a BTM cannot be de ned through any simple mathematical function; instead a form- nd process is required. The form- nding process can be achieved using physical scale models but there are noticeable errors in scaling. There are several computational form- nding methods discussed by Lewis including; the transient stiness method, the force density method and the dynamic relaxation method. Lewis adopted the dynamic relaxation method as the main numerical method because it is widely recognised as the most applicable and successful method for membrane structures, see Lewis (2003) page 57. The dynamic relaxation method depends on discretising the mass of the membrane into concentrated nodes on the surface, see gure 2.10. Each node is given a mass and then oscillated around the equilibrium position because the forces are out of balance. A damping force is used to reduce the oscillations. An iterative scheme is used to converge on the equilibrium position. Either viscous or kinetic damping forces can be used but viscous damping is a two stage solution whereas kinetic is only one stage, therefore kinetic damping is the preferred method. Kinetic damping operates by stopping the iteration when the oscillation reaches its maximum kinetic energy, then it is restarted from that position with zero initial conditions. Kinetic damping relies on the fact that, in simple harmonic motion, the maximum kinetic energy relates to a minimum potential energy, see Lewis (2003) page 67.
Figure 2.10: Discretised continuum used in the dynamic relaxation method, Lewis (2003).
26
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
Lewis (2003) concludes that tensioned membranes represent a considerably more complex aeroelastic problem than that encountered in aircraft wing design. In wing design the de ection is small and the uid ow is attached so the aeroelastic models can be linearised. Tension membranes have large deformations and ow may not be attached. Finally, Lewis notes that considerably more research is needed into the interactions of the surface geometry, wind loads and structural response. The dynamic response of a membrane is unique because of their low thickness and typically low modulus. Jenkins and Korde (2006) says that the communication of bending information spatially is very weak due to the resultant vanishing exural stiness and compares a membrane to a link chain strung between two supports. Simply, this means that local deformations in the centre of a membrane are weakly coupled with the global deformation and do not necessarily cause a global deformation. Lewis (2003) highlighted the di culties with predicting the static de ection of a BTM so the dynamic vibration is excepted to be more complex and time consuming, however, the dynamic vibrations of membrane wings will be reviewed in section 2.6.2.
2.3.3 In atable cylindrical beams In atable cylinders are used for the sponson and keel of the D-class and, even though they are BTM, the symmetry of their geometry means a simple mathematical function can be used to describe their static deformation. In atable cylinders have two states when loaded; a taut state, where all the fabric is in tension, and a wrinkled state, where the fabric buckles due to compressive loads, Comer and Levy (1963). Leonard et al. (1960) derived an equation for the maximum tip loading capabilities (F ) of an in atable cylinder acting as a cantilever beam, see equation 2.1. This simply shows that the load capability of an in atable cylinder is proportional to the internal pressure (p), where
r
is the radius and
l
is the length.
F = πpr3 /l
(2.1)
Comer and Levy (1963) performed work on in atable cantilever cylindrical beams for space structures and analytically found the tip de ection and maximum stress by comparing it to an Euler-Bernoulli beam. Fichter (1966) derived a set of non-linear equations to de ne the deformation from the principles of total potential energy of an in atable cylindrical beam element under bending, twisting and stretching; they can be compared to Timoshenko's beam theory. Webber (1982) extended Comer and Levy's theory to include the eect of torsion in the beam. Main et al. (1994) also furthered Comer and Levy's theory and concluded that the fabric properties changed with the change of internal pressure. Main et al. (1995) improved their theory to include this variation in material properties and this gives a better correlation over a wider range of internal pressures. Wielgosz (2002) showed that the plane sections do not remain plane and that shear deformation could not be ignored, which reinforces the work of Fichter
2.3.
BOUNDARY TENSION MEMBRANES
(1966).
27
Thomas and Wielgosz (2004) and Le Van and Wielgosz (2005) expanded Wielgosz
(2002) work to account for the eect of internal pressure both on the stiness and the geometry of the beam until there was very good correlation between the analytical solutions and the numerical results.
Cavallaro et al. (2003) showed that the resistance to shear deformation
depends solely on the rotational friction developed through pressure induced contact among the overlapping tows. Veldman et al. (2005) developed a load de ection model for orthotropic materials based on thin-shell theory, whereas all previous theories were based on membrane theory.
The new model correlated with the experimental results better than the previous
models. Essentially it is the same model as Comer and Levy except there is an added term for the compressive strength of the material. This highlights the importance of correctly choosing either membrane theory or thin-shell theory.
A considerable amount of work has been performed into the dynamic response of in atable dams, which uses a similar material to that found on the D-class. Early work started with Plaut and Fagan (1988) who calculated the 2D modal frequencies and shapes of in atable cylindrical beams and Plaut and Leeuwrik (1988) used two Galerkin's methods to derive the 2D non-linear equations of motion.
Hsieh et al. (1989) studied the 2D vibrations of a
liquid-in ated cylinder using Finite Dierence Method (FDM) and Boundary Element Method (BEM) and then Hsieh and Plaut (1990) considered the eect of impounding water on one side of the in ated dam. Wu and Plaut (1996) investigated the vibrations in an in atable dam under over ow conditions. Choura and Arabia (1997) showed that structural vibration of an in atable dam can be suppressed by correctly varying the internal pressure. Plaut and Cotton (2005) considered the consequence of in atable dams resting on a deformable foundation and compared it to a rigid foundation. It was found that the frequencies of vibration increased with an increase in the foundation spring stiness but decreased with an increase in the foundation's shear modulus. Dakshina Moorthy et al. (1995) performed the rst 3D investigation into the vibrations of an in atable dam using a Finite Element Method (FEM) by modelling it as an elastic shell element.
Mysore and Liapis (1998) analysed the 3D dynamics behaviour of a
single-anchored in atable dam using BEM and FEM. The eect of internal pressure, external water head and parallel owing water were investigated. It was found that the rigid foundation increased the frequency but the presence of water decreased the frequencies. Finally, water owing parallel to the dam also reduces the vibration frequencies.
The most relevant work performed on the vibration of in atable tubes with the application of RIBs was performed by Townsend (2008). The modal frequencies and shapes were found for the sponson of a RNLI Atlantic 21 through nite element analysis. The variables considered included: air in ated and foam sponson, internal tube pressure and hull attachment.
The
air-in ated sponson was modelled as a shell element so that the compressive strength was proportional to the in ation pressure.
The rst modal frequency of air-in ated sponson at
5 psi was 4.7 Hz, the next two were between 9.5 Hz and 10.6 Hz. The sponson attachment to the hull was found to have a greater in uence on the vibration frequencies than internal pressure. The presence of external water was not considered.
28
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
The literature review of numerical models has revealed that there are methods to predict the static shape of BTMs, but not the dynamic shape, and that there are methods for predicting the static and dynamic response of in atable cylindrical beams; however, it is not clear which method is most accurate (i.e. membrane theory or thin-shell theory). Also the work of Cavallaro et al. (2003); Veldman et al. (2005) showed the importance of the material properties which are currently not know, see section 2.3.1.
2.3.4 Experimental methods Jenkins and Korde (2006) provides a historical review of membrane vibration experiments and notes that there are signi cantly less experimental membrane vibration studies than numerical studies. The low number of experimental studies is due to the experimental di culties caused by the low mass and thickness of membranes. This means that standard experimental methods for measuring the modal shapes and frequencies, such as using accelerometers and shaker strings, cannot be used because these techniques noticeably change the mass and stiness of the membrane. Early work on membrane vibrations by Faraday (1831), as reported by Jenkins and Korde (2006), used a horsehair lament to excite the membrane and a ne lycopodium powder to expose the membrane nodes.
Bourget (1866) used an organ pipe to acoustically
excite the membrane and ne powder to expose the node positions. These appear to be the main experimental techniques until the introduction of laser vibrometer by Jenkins (1999) and Jenkins and Tampi (2000). A laser vibrometer measures the velocity of the membrane by measuring the shift in frequency between the incident and re ected light due to the relative motion of the surface (i.e. Doppler shift). The system uses a set of rotating mirrors so that it can scan an area rather than a single point.
2.4 Fluid domain The D-class and all the RNLI RIBs and IBs are planing vessels.
A vessel is considered to
be planing when the length Froude number (Fn) is greater than 1.2 (Fn > 1.2); however, a Froude number of one is considered the lower limit, see Faltinsen (2005). The Froude number is de ned in equation 2.2, where;
v
is velocity,
g
is gravity and
L
is Waterline Length (LWL).
A planing craft is de ned by its ability to support the weight of the craft on the hydrodynamic lifting force and the buoyancy force is reduced. A simple analogy of a planing vessel is a jet of water impacting an inclined plate and this is shown in gure 2.11, Saunders (1957). If it is assumed that the eects of gravity are negligible, the liquid is ideal, zero friction and ambient pressure is present at all edges of the plate, then Bernoulli's theorem can be applied, and the velocity (U0 ) is the same in all regions of ow. This implies that the force acting normal to the plate can be expressed as;
F = ρtU02 sinθ.
The pressure distribution on a planing at plate
is shown in gure 2.12, Dand (2003c).
p F n = v/ gL
(2.2)
2.5.
HYDROELASTIC SLAMMING
29
Figure 2.11: Jet of water impacting a inclined plate, Saunders (1957).
Strip theory is a common numerical method used to predict the planing performance of a vessel, Lloyd (1998). First, strip theory uses discrete transverse slices of the vessel to predict the hydrodynamic lift on each slice, then; it is integrated along the length to predict the overall performance, Lloyd (1998).
Figure 2.12: Longitudinal pressure distribution on a planing at plate, Dand (2003c).
2.5 Hydroelastic slamming Planing craft in rough waters expose the crew to high levels of vibration and the highest accelerations normally occur during a slam. Chuang (1966) de nes slamming as the impact of any portion of a moving ship upon the surface wave but this most commonly occurs at the bow of a ship. In the case of planing craft this can involve the entire hull losing contact with the water's surface. A slam was de ned by Ochi (1964) if the relative motion exceeds the local eective draught and the relative velocity at impact, see Lloyd (1998) page 292. Slamming results in a local and global response of the hull and can be classi ed into three categories; localised response, transition and overall response, Chuang (1966). The transition and overall response link to global hydroelasticity (which is discussed in more detail later) and this shows that slamming is a 3D problem, however, this section will consider the 2D problem of slamming because it allows the problem to be simpli ed and is in keeping with strip theory.
30
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
Figure 2.13: Flexible components within a vertically impacting IB.
2.5.1 Problem de nition The problem addressed within this section is regarding the eect of hydroelasticity on the loads and accelerations of a transverse 2D wedge vertically impacting a free surface. This aspect could also be called transverse hydroelasticity. An IB has three main exible components in the vertical direction which are the fabric hull, the in atable sponson and the in atable keel, see gure 2.13. In reality these three components act together and will aect the response of each other. However, for an initial investigation each could be studied individually. By considering a slamming event as hydroelastic it allows the possibility of changing the impact characteristics.
The main characteristics that can be changed, from a boat motion
perspective, are the peak acceleration and impact duration. It will also aect the structural loading but this project will not explore that side of the problem, see Faltinsen (1997) for more details. The new parameters for the hull are fabric material properties and pre-tensioned stresses and the new parameters for the in atable keel and sponson are material properties and internal pressure. Note that changing the internal pressure is the same as changing the pre-tensioned stresses, see Lewis (2003). The other important variables are impact velocity, deadrise angle and inertia. It has been proposed but not validated by many authors including Natzijl (1998) and Pike (2003) that the sponson absorbs energy during a slam. Townsend (2008) did investigate this concept but the internal pressure reduction was shown to have no eect. It is worth noting that the Atlantic 85 investigated by Townsend (2008) had a hull shape which caused the sponson rarely to come into contact with the water which is not the case for the D-class. The
2.5.
HYDROELASTIC SLAMMING
31
experiment proposed for the wedge sections with sponson will answer this question and allow an investigation into the eect of material properties and internal pressure. Other variables that will aect the amount of energy absorbed by the sponson include; sponson diameter, sponson overhang and sponson attachment.
2.5.2 Literature review Faltinsen et al. (2004) provides a good review of this problem and discusses the challenges within it when the hull is constructed from conventional materials. Here is a list of particular eects that may require consideration:
gravity, viscosity, air cushions, air pockets, air to
bubble generation, water compressibility, air compressibility, ow separation and membrane behaviour. Gravity can normally be neglected in this problem because the uid acceleration associated with the initial impact is much larger than the gravitational acceleration, Faltinsen et al. (2004). Viscosity is also commonly neglected when a vessel has a hard chine because viscosity does not aected the ow separation and the ow seperation can be modelled.
Although,
Faltinsen (2005) stated the round bilge ow separation is di cult to handle and here viscosity may need to be included. Air cushions and compressibility were initially ignored, but Bereznitski (2001) showed the importance of including them, especially at low deadrise angles. Air pockets can occur when the structure is very exible and can excessively deform vertically past the corner of the wedge, as shown in gure 2.14. Faltinsen et al. (2004) noted that the breakdown of air cushions into bubbles requires better understanding and the eect of this is unknown. The time scale of water compressibility is, typically, signi cantly smaller than the time scale of the local structural response so water can be assumed incompressible, Faltinsen et al. (2004). However, the time scale of the fabric deformation has not currently been identi ed so the assumption needs validation. Finally the membrane behaviour is signi cantly dierent from that of conventional materials with non-linear behaviour due to the interaction of the weave and weft, Lewis (2003).
Figure 2.14: Air pocket formation.
Faltinsen et al. (2004) divided this problem into two time scales.
The initial time scale
is that of the structural inertia phases where the large hydrodynamic forces lead to large
32
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
accelerations of a small structural mass. This phase is very short compared to the second time scale. The second scale is the free vibrations phase and is the highest wetted natural period of the structure.
The response is the free elastic vibration of the structure with the initial
conditions obtained from the rst phase. The maximum stresses occur in the free vibration phase. Faltinsen (1999) discusses the importance of hydroelasticity as a ratio between the rst period of natural vibration of a wet beam and the duration of the impact. It is quanti ed in terms of nondimensionalised parameters. Bereznitski (2001) uses the same ratio except that it uses the natural vibrations of a dry beam, see equation 2.3.
Bereznitski says that if the
ratio is greater than two then hydroelasticity does not play a signi cant role. Increasing either the material properties or pre-tensioned stresses in the fabric will alter the period of vibration therefore aecting the importance of hydroelasticity.
Ratio = (Duration.of.Impact)/(P eriod.of.V ibration)
(2.3)
Cooper and McCue (2011) were the rst to study the deformation of a exible membrane wedge impacting a free surface. It was found that during the free vibration phase the membrane vibrated at frequencies very near to its natural frequency, which depended on the pre-tensioned stresses.
2.5.3 Critique of modelling methods The problem of water entry of 2D bodies started in a purely hydrodynamic sense for a rigid body with the work of Wagner (1932) and von Kármán (1929). This work was advanced by many researchers but it was not until the work of Kvålsvold and Faltinsen (1995) that the local hydroelastic eects were considered. Using theory alone, Kvalsvold in 1995 studied the slamming-induced local stresses in the wetdeck of a multihull vessel for a doctor of engineering thesis and jointly published the results in Kvålsvold and Faltinsen (1995). The structure was modelled using a 2D Timoshenko beam and the uid was modelled using Wagner's theory. It assumed the uid to be incompressible and irrotational; and air entrapment and cavitation were not included.
This solution was
complex and simpli ed by Faltinsen (1997). Experimental results from Faltinsen et al. (1997) and Kvålsvold and Faltinsen (1995) agreed well with both theoretical solutions.
Faltinsen
(1997) used the numerical solution of Kvålsvold and Faltinsen (1995) to study the water entry of a wedge including the forward speed of the vessel by solving the coupled non-linear equations by a Runge-Kutta 4th order scheme. Korobkin et al. (2006) demonstrated that it is possible to couple a FEM for the structural domain directly with Wagner's theory for the uid domain. The results were compared with a modal method using a beam model and the results showed good correlation. Lu et al. (2000) used BEM for the uid and FEM for the structure. The non-linear free surface boundary condition was satis ed and the jet was properly treated. Good agreement was found with the results of Zhao and Faltinsen (1993).
2.6.
HYDROELASTIC PLANING SURFACES
33
Bereznitski (2001) published an important paper on the role of hydroelasticity in the 2D slamming problem and uses four methods for solving the problem.
The rst is a Wagner's
solution for a rigid body and this can be compared to the work of Faltinsen (1997) for an elastic body.
Bereznitski also used a self-developed code plus two commercial codes called
MSC Dytran and LS-DYNA. Bereznitski commented that the most suitable methods were either MSC Dytran or LS-DYNA because they can both deal with the coupled hydroelastic interaction and include air cushion modelling.
It is worth noting that MSC Dytran and
LS-DYNA are quite similar and the equations for the state of water and air are the same, Bereznitski (2001).
LS-DYNA has been used to study this problem by Bereznitski (2001);
Le Sourne et al. (2003); Stenius (2006). Stenius used nite element analysis based on multimaterial arbitrary Lagrangian-Eulerian formulation and a penalty contact algorithm and the hydrodynamic loads correlated well with experimental results.
2.6 Hydroelastic planing surfaces 2.6.1 Problem de nition The planing surface of an IB is normally constructed from fabric which has signi cantly less out-of-plane bending stiness than conventional metal or composite hulls. This will allow the planing surface to deform considerably under dierent loading conditions, see gure 2.8. The problem is to nd the shape of the fabric when it is in steady-state planing and the eect of this deformation on the planing performance.
The parameters of a fabric hull are material
properties and the pre-tensioned stresses. These parameters de ne the out-of-plane bending stiness of a fabric therefore as they are increased the material becomes stier and comparable to a conventional planing surface. A better understanding of a hydroelastic planing surface could lead to an increase in forward speed. Experiments by Dand (2002a, 2003b) were performed on an EA16 at full scale and model scale to measure the resistance, sinkage and trim. The full scale boat was exible and the fabric hull was able to deform but the scale model was rigid. The comparison of total resistance, see gure 2.7, showed that the full scale exible boat had slightly higher resistance than the rigid scaled model. Dand et al. (2008) attributed this to the change in trim angle due to the fabric hull deforming and causing a concave camber at the aft of the hull. They also found an instability when the boat was accelerating on at water which was described as a pressure wave slowly passing under the boat.
It caused a pulsing motion primarily in pitch and
heave. Whether the deformation was static or dynamic is unknown. The rst limitation of the at water performance is the pulsing motion instability found in the D-class. One hypothesis is that the reduced out-of-plane bending stiness of the hull allowed the concave camber to form. This causes the pre-tensioned stresses in the fabric to change as a camber forms and also results in a change in the hydrodynamic forces on the hull.
As the fabric stresses change, the deformation moves aft.
The deformation causes a
change in hydrodynamics which gives the operator the feeling of the pressure wave.
It has
also been reported that as the pressure wave passes under the hull the sponson can be seen
34
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
to de ect which indicates high lateral forces and fabric deformation. When this deformation reaches the transom the pressure is released and the cycle begins again.
This motion is
only found on at water; waves cause the cycle to be broken. So there is a limitation in the minimum out-of-plane bending stiness of the fabric hull to ensure this instability does not occur and this requires quanti cation. This belief was con rmed through trial and error when the EA16 was developed into the IB1. During the redesign it was found that the fabric had been permanently deformed and low quality control during construction led to a variation in fabric tension. The pulsing motion disappeared once this had been taken into account and the fabric tension was increased.
2.6.2 Literature review The most relevant literature is an analytical model developed by Makasyeyev (2010) to describe the planing performance of a 2D planing elastic plate. However this model requires validation and the structural domain deals with conventional materials not membranes. No literature directly related to a membrane planing surface has been found.
However
this uid structure interaction could be compared with the aeroelasticity of a membrane aerofoil, such as sails and membrane wings. Newman (1987) noted skin friction can change the membrane tension and in an inviscid ow it is constant. A strong coupling between the frequency of the membrane oscillations and vortex shedding frequency has been shown by Song et al. (2008); Rojratsirikul et al. (2009); Gordnier (2009). Gordnier importantly showed that the Reynolds Number caused the motion of the membrane aerofoil to change from a standing wave vibration to a dynamic vibration similar to travelling waves. None of the afore-mentioned literature contains a free surface which is vital for the planing uid forces. It is of interest to note that many new tender-boat designs now employ drop stitch technology for the hull. Drop-stitch technology involves two layers of fabric that are sealed together at the edges.
Then threads are weaved perpendicular to the layers of fabric to control the
shape when in ated, see gure 2.15. When the two layers are in ated it forms a sti panel that could be compared to a composite sandwich panel, Bagnell (1998).
Figure 2.15: Drop stitch technology, Yakangler (2011).
2.7 Global hydroelasticity This section investigates the global hydroelasticity of an IB by viewing the boat as a whole and studying the longitudinal bending and torsional twisting that exists. It has been observed
2.7.
GLOBAL HYDROELASTICITY
35
Figure 2.16: Reducing the vertical boat motion through longitudinal hydroelasticity.
that as the D-class passes over an oblique wave the deck bends and twists which provides a smoother ride. This could be regarded as a conventional hydroelastic response and theories such as the ones described by Bishop and Price (1979). The exibility of the boat will aect the wave induced dynamic response of the vessel which in turn aects the boat motion. Global hydroelasticity is a 3D problem but this does not provide a simpli ed 2D problem that is in line with the three aspects of hydroelasticity.
Within global hydroelasticity
is longitudinal hydroelasticity and this is used to de ne the third simpli ed 2D aspect of hydroelasticity.
2.7.1 De nition of longitudinal hydroelasticity Figure 2.16 shows how longitudinal hydroelasticity can reduce the vertical motions of a deformable vessel. In conventional vessels there is a coupled interaction between the heave and pitch as a vessel see-saws over a wave; however, this interaction will change if the boat is able to bend over the wave.
The rst advantage of this is reduced boat motion leading to
improved ride quality. A reduction of the boat motion potentially means that less energy from the propulsion device is absorbed through the vertical motion, which might reduce the added resistance in waves. This could allows either a higher top speed to be achieved or a smaller, lighter, propulsion device to be tted. The nal advantage is that the boat will be more stable, in pitch and heave, when stationary because the pitching motions will be reduced. An IB has many inter-connected parameters that will aect the global vibrations which include; deck properties (material properties and thickness), deck joints (number, position and stiness), sponson and keel properties (material properties and internal pressures), fabric hull properties (material properties and pre-tensioned stresses), mass (centre of gravity and inertia) and construction technique.
In the early stages of this thesis, a simple waterline
de ection experiment was performed to explore these inter-connecting parameters.
The D-
class was loaded with up to 1.75 tonnes and the waterline was measured. The deformation was calculated from the waterline. The report is including in appendix E. The results showed that the dominant parameters in the de ection of the D-class are the number, position and
36
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
stiness of the deck joints.
2.7.2 Literature review of global hydroelasticity Global hydroelasticity has been studied by many authors starting with the work of Bishop and Price (1979). Bishop and Price developed theories to describe symmetric and anti-symmetric hydroelasticity of ships, but these ships were displacement vessels and not planing vessels. There are numerical models capable of predicting the vertical motions and wave loads on a High Speed Craft (HSC) such as Santos et al. (2009); Chiu and Fujino (1989). al.
Santos et
modelled a fast patrol boat which had a planing hull form, but it was noted that the
approach used was not suitable for planing vessels.
They found large dierences between
the full scale measurements and the numerical model results. This shows that there are no validated numerical models currently available to model the hydroelastic performance of the D-class or other highly exible, planing craft. The D-class has distinct deck joints to allow the boat to hinge in certain points. These deck joints will aect the conventional theories of global hydroelasticity. Newman (1994) developed an analytical method to predict the motions of a hinged barge. Hamamoto et al. (1996) used a 3D coupled FEM-BEM model to predict the motion of module linked large oating structures. Numerical models like these might be useful in the future to account for the dynamic response of deck joints.
2.8 Coupled hydroelasticity Hydroelasticity slamming and global hydroelasticity have been researched and documented before but the hydroelastic planing surface aspect has not. Moreover, there are methods that couple local slamming loads with a global response but there are no theories that consider a hydroelastic planing surface. For example, a slam occurs when the vessel is normally moving forward and the frictional forces between the hull and the water surface will place the hull in tension.
So the hydroelastic planing surface will change the pre-tensioned stresses in a
hydroelastic slam. This means it is unclear how to consider these three hydroelastic aspect together. Strip theory provides an order in which to assess all the aspects of hydroelasticity. First the transverse slices of the vessel are assessed and this relates to the 2D problem of hydroelastic slamming.
Then the hydrodynamic lifting forces can be integrated over the
length to study the deformation of a hydroelastic planing surface, which primarily relates to performance on at water. Finally the hydrodynamic lifting forces can be used to predict the global de ection of the vessel structure, called global hydroelasticity and this relates to the performance in waves. This cycle is shown in gure 2.17. This breaks the complex 3D problem of an entirely hydroelastic vessel down into manageable problems and provides a perspective from which the performance of the D-class can be understood, Halswell et al. (2012). example of the application of the hydroelasticity aspects can be found in section D.4.1.
An
2.9.
PERCEPTION OF BOAT MOTION
37
Figure 2.17: Hydroelastic design cycle.
2.9 Perception of boat motion In 2002, the European directive 2002/44/EC was passed on the minimum health and safety requirements regarding the exposure of workers to physical vibration. This was included in UK legislation in 2005 through the Control of Vibration at Work Regulations, see Pond (2005), and again in 2007 via the Merchant Shipping and Fishing Vessels (Control of Vibration at Work) Regulations, see MCA (2007). The European directive sets the Exposure Action Value (EAV) for WBV to 0.5
ms−2
r.m.s (or 9.1
ms−1.75
Vibration Dosage Value (VDV)) and the
−2 r.m.s (or 21 Exposure Limit Value (ELV) to 1.15 ms
ms−1.75
VDV). The VDV is used
instead of the Root Mean Squared (RMS) when the crest factor is above six; the crest factor is de ned by the peak acceleration divided by the RMS acceleration. Planing vessels in rough waters expose the crew to non-linear vibration that regularly exceed these values, which will now be demonstrated. The highest accelerations occur during a slam and in the case of HSC this can involve the entire hull losing contact with the water's surface before impact. Dand (2004) experimentally tested a rigid scale-model of the D-class where accelerations of up to 4 g in the crew's position were measured in regular waves, with a full-scale wave height of 0.55 m and full-scale speed of 19.4 knots. Townsend et al. (2008b) showed that the RNLI Atlantic 85 RIB exceeds the EAV in 30 minutes at 32 knots with approximately 0.4 m signi cant wave height. Allen et al. (2008) measured the vibrations on the RNLI Atlantic 75 in two trials at speeds of 15 to 20 knots. The crest factors were above 6 which meant the VDV should be used instead of RMS and the values were 25.90
ms−1.75
and 48.51
ms−1.75
in sea states two and
three, respectively. Myers et al. (2011) measured the accelerations of a military HSC at 40 knots in a sea state of two to three and the VDV for a 3 hour transit was 57.05
ms−1.75
on the
38
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
deck. This is clear evidence that the vibration within these craft regularly exceeds the ELV and a solution, or combination of solutions, needs to be found. Whilst there is considerable debate in the marine community over the validity of applying the European directive to HSC, the RNLI are investigating methods to demonstrably mitigate the exposure of their crews and trainers to vibration. The human exposure to vibration can have many eects; from chronic and acute, to physiological and psychological, see Townsend et al. (2012a). Physiological injuries have been reported by Ensign et al. (2000) to include; spinal and abdominal injuries, damage to internal organs (kidneys), torn ligaments and, broken ankles and legs. Ensign et al. (2000) also reported that the psychological injuries include; annoyance, fatigue, anxiety, loss of visual accuracy and reduced hand-eye coordination (the latter two could be considered a combination of both physiological and psychological eects). Myers et al. (2011) demonstrated that a three hour transit in a 40 knot HSC would reduce the physical performance of the crew (including run distance and vertical jump height). So, by reducing the boat motion and the WBV it can reduce the risk of injury, provide a better working environment and increase the crew's eectiveness during and after transit. Researchers have explored many technological solutions to this problem; however, no single solution appears to be completely successful. Coe et al. (2013) concluded that a combination of solutions will be required to reduce the WBV enough to meet the legislation, which was backed up also by Coats et al. (2003). Suspension seats are one solution that is currently been heavily researched, see Coe et al. (2009, 2013); Coats et al. (2003); Olausson (2012); Cripps et al. (2004); although, Townsend et al. (2012a) points out their many draw backs. Townsend et al.
suggested a number of other strategies to reduce the WBV: suspended decks (also
discussed by Coe et al. (2013)), active and passive ns, trim tabs, interceptors, gyrostabilisers, exible hulls and elastomer coated hulls. Coats et al. (2009) discussed the use of a porous hull to reduce the impact loads and spread the energy over a longer time period, and they showed a signi cant reduction in impact loads. The rst component in the D-class is the sponson and it has been proposed by Pike (2003); Natzijl (1998), and discussed by others in the maritime community, that the sponson are able to absorb the energy during a slam but there is no scienti c evidence. Haiping et al. (2005) undertook an experiment into the eect of sponson type on the sea keeping performance. It was found that an in atable sponson had a lower RAO in heave and pitch than a foam sponson in two out of two load conditions. This suggests that a exible sponson can improve the ride comfort and sea keeping performance. Townsend et al. (2008a,b) performed experiments into the sea keeping performance of the Atlantic 75 RIB and they showed that the internal pressure of the sponson had minimal eect on sea keeping. It was supposed that the sponson did not contact the water enough to have an eect on the sea keeping performance. So the sponson of the D-class could be responsible for the anecdotal evidence of improved performance because they are fully in contact with the water during operation; however, there are three other exible components (deck, keel and hull) within the D-class that could also aect the sea keeping performance and the slamming characteristics. Townsend et al. (2012a) numerically
2.10.
SUMMARY
39
explored the eect of decreasing the hull stiness to isolate the humans from the vibration. They reduced the hull stiness from 69 GPa to 6.9 GPa and found that it had little eect on the response. This would suggest that the hull stiness has minimal eect on the slamming characteristics; however, in table 2.3 is a comparison of the properties of the rubber-coated fabrics, in the transverse direction of the hull (weft), to the properties of aluminium. Firstly, the density of the fabric 57 % less than aluminium and the Young's modulus is nearly 250 times lower; although, the ultimate tensile strength of the fabric is only 27 % lower than aluminium. This reveals that the fabric used on the D-class hull is 250 times more elastic but it is nearly as strong as aluminium. So it is anticipated that this degree of elasticity will aect the slamming characteristics. Property
Hull fabric (weft)
Aluminium
3 Density (kg/m )
1152
2700
Young's modulus (GP a)
0.28
69
Ultimate tensile strength (M P a)
> 80
110
Table 2.3: Fabric properties vs. aluminium properties.
2.10 Summary The hydroelastic literature review started with the construction of the D-class and it showed that the structure is highly complex and statically indeterminate. Then, this chapter reviewed all the literature that directly relates to the D-class, which primarily included six technical reports by Dand et al.
between 2002 and 2004.
The main material used to construct the
D-class is a rubber-coated fabric so the material properties were experimentally investigated and the results are shown in appendix D. Next, the form- nding process by Lewis (2003) was reviewed and the dynamic relaxation method was found to be most appropriate. The static and dynamic deformation of in atable cylindrical beams were nally reviewed because the symmetry of their geometry simpli es the problem. Hydroelastic slamming was reviewed and it showed that this has been well researched; although, the research is primarily aimed at assessing the loads during hydroelastic slamming and not the accelerations. The numerical modelling methods for hydroelastic slamming were critiqued and this revealed that there are numerical methods but none include BTMs. Hydroelastic planing surfaces are a novel area of research and only one paper was found on the subject, which included a numerical model of elastic planing plates. Membrane aerofoils were reviewed as a parallel to membrane planing surfaces but none included a free-surface which is vital for planing.
Global hydroelasticity was reviewed next and there has been consider-
able research into this area; however, the vessels considered were displacement vessels and not planing vessels. Therefore, any numerical models are not applicable to the D-class. The idea of longitudinal hydroelasticity was introduced. Finally, the hydroelastic literature review showed that the structural components of the D-class and the three aspects of hydroelasticity are all coupled together.
40
CHAPTER 2.
HYDROELASTIC LITERATURE REVIEW
Chapter 3
Hydroelastic methodology Strip theory facilitates the division of an entirely hydroelastic boat down into three manageable aspects of hydroelasticity called: global hydroelasticity.
hydroelastic slamming, hydroelastic planing surfaces and
However, it does not consider the interaction of the three aspects.
Any one aspect will change the parameters and boundary conditions of the other two aspects and the eect of this is not known. Therefore, to truly explore an entirely hydroelastic boat the interactions of the aspects must be considered. The literature review showed that there are no theories or models that directly apply to any individual aspect of the D-class because of the fabric structure; let alone the interactions of the three aspects.
It also showed that
there are major di culties involved in scaling individual aspects and these di culties are enhanced when the interactions are considered. This implies that the best route forward for this project is to perform full scale experiments on the D-class to measure the performance and deformation to explore the eects of hydroelasticity.
The three individual aspects of
hydroelasticity still hold high value because they provide simpli ed problems that have been considered in the past.
There are numerical models that, in the future, may be capable of
predicting the performance of the individual aspects; such as the models by Kvålsvold and Faltinsen (1995); Lu et al. (2000); Santos et al. (2009); Makasyeyev (2010) and commercially available models MSC Dytran and LS-DYNA. The project has explored possible methods to study individual aspects and to prove that full scale experiments are the most suitable route forward to study a hydroelastic planing craft. In this chapter, the discussion will rst focus on the potential methods for studying the individual aspects of hydroelasticity, both numerical and experimental methods. Then, the discussion will cover the full scale holistic methods including the use of a towing tank. Finally, the two chosen methods will be described in detail which includes the quasi-2D drop test method and full scale holistic method, found in sections 3.4 and 3.5.
3.1 Proposed numerical methods for individual aspects There are currently no numerical theories or models capable of studying the interactions of the three aspects of hydroelasticity or the individual aspects. Although, if more work is performed 41
42
CHAPTER 3.
HYDROELASTIC METHODOLOGY
there are a few numerical models that may be suitable for studying the individual aspects once they are validated against experimental results.
3.1.1 Modelling of boundary tensioned membranes The structural domain of the D-class primarily consists of BTMs which means that numerical models become very complex and computationally demanding. This is due to the number of iterations required to predict the 3D deformation of BTM within a hydroelastic event. The solution of a hydroelastic problem require an iterative solution because the uid forces will change depending on the de ection of the hull and the hull de ection will depend on the uid forces. Lewis (2003) suggests a number of methods for predicting the de ection of BTMs but all the methods require iteration. There is no simple formula for the deformation of a BTM, see Lewis (2003).
So, introducing a BTM into a hydroelastic problem doubles the number
of iterations required, see gure 3.1, meaning that larger amounts of computational time and power are required. There are a few techniques that can be used to simplify the problem of BTMs, such as; using the symmetry of an in atable cylindrical beam, see Comer and Levy (1963), and quasi-2D assumption like 2D slamming, see Wagner (1932). This increases the potential for numerical models in the near future.
Figure 3.1: Additional iteration in nding the elastic hull shape due to BTMs.
3.1.2 Slamming numerical models Numerical methods could be used to investigate the 2D slamming problem and these were reviewed in section 2.5.3. The rst computational method that could be used is membrane theory coupled with Wagner theory in a similar manner to Korobkin et al. (2006); Kvålsvold and Faltinsen (1995). BEM and FEM could also be coupled together to solve this problem such as Lu et al. (2000) and ANSYS. The best method would most likely involve using LSDYNA to explicitly couple the domains. LS-DYNA has been used and validated in the past
3.2.
PROPOSED EXPERIMENTAL METHODS FOR INDIVIDUAL ASPECTS
43
plus most of the considerations can be included, see Bereznitski (2001). However, there are still unknowns within the problem when the D-class is considered. It is unknown whether thin shell or membrane theory is more accurate for modelling the rubber-coated fabrics. It is also unknown whether air pockets form because this means air compressibility has to be included in the model. Figure 3.2 shows a picture of the underneath of the D-class hull and the green horizontal line matches a ruler placed across the hull. The green vertical line emphasises the natural air pocket due to the shape of the hull and sponson.
.
Figure 3.2: Natural air pockets due to the shape of the D-class hull
3.1.3 Other numerical models The work of Makasyeyev (2010) showed that it is possible to numerically model a 2D hydroelastic planing surface; although, this model still requires validation and does not consider a fabric planing surface.
Santos et al. (2009) showed that it is possible to model the hy-
droelastic vertical motions of a fast patrol boat. However, this model requires a large amount of work to be capable of predicting the hydroelastic response of the D-class.
3.2 Proposed experimental methods for individual aspects Five experimental methods will now be discussed that show good potential to explore the individual aspects of hydroelasticity. There will be a short description of each method, including the main transducers required and the likely outcomes of the method.
Then, a Strengths,
Weaknesses, Opportunities and Threats (SWOT) analysis is performed on each method and summarised in table 3.1. The strengths and weaknesses are internal to the experiment and relate directly to the method, whereas the opportunities and threats are external and relate to the application of the results.
44
CHAPTER 3.
HYDROELASTIC METHODOLOGY
3.2.1 Quasi-2D fabric wedge impacting a free surface This method involves a quasi-2D fabric wedge impacting a free surface, where the wedge cross-section is geometrically similar to a transverse slice of the D-class. When the wedge is released it will free fall along two rails to ensure the fall is vertical, until the wedge impacts a channel of water representing a free surface.
Accelerometers can be used to measure the
peak acceleration and impact duration of a slam. The primary outcome of this method is to investigate the eect of hydroelasticity on the accelerations of a slamming event.
Strengths and weaknesses This method provides great control over the parameters and the environment and has been shown to provide repeatable, reliable results in the past, see Lewis et al. (2010).
Dierent
wedge cross-section can be manufactured to study the eect of individual components, such as the hull, sponson or keel. However, this method neglects forward speed and the edge eect (the dierence between quasi-2D and 3D), plus the scaling laws still need to be considered in this experiment. Although, this method can be performed quickly and easily (because the project has access to a drop test rig with pre-manufactured wedges).
Opportunities and threats This method could provide a unique and novel way to investigate the eect of sponson on the slamming accelerations. To the author's knowledge this has not been performed before; however, if the sponson is ignored and a wedge with hard chines is investigated then it may be possible to compare the results to a numerical model. This will help with answering whether thin-shell or membrane theory is more applicable for the fabric used on the D-class.
This
method is also cheap to perform (because of the access to a drop test rig). On the other hand this method is quasi-2D so the results have low applicability to the RNLI because they cannot be directly related to the performance of the D-class.
3.2.2 Fluid impacting an inclined fabric plate A simple analogy of a planing surface is a uid jet impacting an inclined at plate, see section 2.4. A fabric sheet could be wrapped around a metal frame to form a at BTM plate instead of a at rigid plate, and then a jet of water would be pumped at the at BTM plate. A jet of water is normally turbulent which causes an issue because stationary water is initially laminar until the boat disturbs the water. The eect of this is unknown but this method could prove to be highly useful. This method can explore the shape of a simpli ed fabric planing under quasi-static loads.
Strengths and weaknesses The simplicity of this experiment and low number of variables means that the cause of any measured change can easily be established. It could also potentially provide a very controlled
3.2.
PROPOSED EXPERIMENTAL METHODS FOR INDIVIDUAL ASPECTS
45
environment; however, normally a water jet is turbulent whereas stationary water is laminar. On the other hand the run time of this approach is theoretically in nite allowing repeated pulsing motions to be studied, which is a unique property of this experiment. Another disadvantage to this method is that the trim angle and sinkage are not included and these could be linked to the pulsing instability, like porpoising. Finally scaling laws still need considering.
Opportunities and threats The simplicity of this method makes it unique and novel, plus the costs are comparatively low compared to the cost of towing tanks; nevertheless, the results cannot be compared to the real world and they are of minimal direct use to the RNLI.
3.2.3 Flat fabric planing surface The shape of the planing surface could be simpli ed to a at plate and towed along a towing tank to provide a simpli ed problem in a very controlled environment.
In this controlled
environment it will be feasible to use optical experimental techniques to measure the dynamic deformation of the fabric plate and its eect on planing performance.
Strengths and weaknesses This method will provide a controlled environment and good variability over the parameter, plus it has been shown to work in the past, see Shuford (1954). The results will be reliable and comparable to a at planing surface, although this means it neglects the cross- ow eect found on vee-shaped hulls. The other weaknesses include; a limited run time, trim angle and sinkage are not included because a plate has no buoyancy and the scaling laws still require consideration.
Opportunities and threats The planing surface is at so it could be possible to compare the results to the numerical model of a at hydroelastic planing surface, see Makasyeyev (2010), and it could be used as a benchmarking method for future numerical models. Nevertheless it still does not represent a vee-shaped hull, and trim and sinkage are not included so it cannot compare to the D-class.
3.2.4 Scale model of a fabric vee-shaped planing surface This is an extension of the previous method ( at fabric planing surface) by altering the fabric planing surface to a realistic hull shape with a deadrise angle. A rigid frame could be manufactured which will represent the keel and chines of a vee-shaped hull (i.e. no sponson) and a special fabric hull could be wrapped around the frame. This model could be made watertight so that it has buoyancy.
46
CHAPTER 3.
HYDROELASTIC METHODOLOGY
Strengths and weaknesses This method will provide a very controlled environment but there are likely to be di culties with the model and the control over the parameters. Lewis (2003) highlighted the di culties forming a smooth BTM and the experience with the D-class further emphasises the fact that BTMs are very complex structures. The hull shape would be realistic and watertight so the tests can be performed using standard International Towing Tank Conference (ITTC) methods. However, there is still a dierence between the thrust line of D-class with an outboard and model towed using a tow post, which will cause an error and possibly change the pulsing characteristics. The method still has limited run times and the model requires scaling.
Opportunities and threats The model is realistic so the results could be useful to the RNLI providing the BTM hull performs correctly, but the sponson are still ignored. The cost of manufacturing the model and using a towing tank are relatively expensive.
3.2.5 Segmented scale model without sponson in waves A rigid model of a simple deep-vee planing hull with a constant deadrise angle can be cut into segments and tted along a longitudinal backbone to control the segment dispersion and longitudinal stiness, see Lavro et al. (2007) . Then the model can be towed along a towing tank in regular waves and irregular wave spectra. The eect of longitudinal hydroelasticity on the vertical boat motion can be studied using this method.
Strengths and weaknesses The parameters and environment can be precisely manipulated. The segment dispersion and longitudinal stiness can be accurately varied using backbones of dierent stinesses and dierent sized segments; however, previous experimenters have found it surprisingly di cult to get reliable results, see Streeter et al. (2009). The hull segments can also lead to irregular uid ows, and there are still the scaling problems and thrust line issues.
Opportunities and threats The lack of the sponson means that the results are applicable to a large number of planing vessels and it could be possible to compare the results to potential future numerical models. However, the fact it lacks the sponson means that it does not directly relate to the D-class.
3.2.6 Segmented scale model with sponson in waves This method is very similar to the previous method (segmented scale model in waves) except the model will include the sponson. The RNLI own a rigid 1/4 scale model of the D-class, this could be divided into segments and tted along a backbone.
3.3.
PROPOSED EXPERIMENTAL METHODS COUPLED ASPECTS
47
Strengths and weaknesses The strengths and weaknesses of this method are identical to the previous method (segmented scale model in waves) although scaling can be more complex due to the round bilges of the sponson.
Opportunities and threats The presence of the sponson will change the opportunities and threats by directly relating the results to the D-class and increasing their applicability to the RNLI; however, it complicates the problem and there are no experimental results or numerical models to benchmark against.
3.3 Proposed experimental methods coupled aspects 3.3.1 Towing tanks Towing tanks provide a highly valuable method for studying the behaviour of a planing craft because the parameters and environment can be accurately controlled and the data acquisition system on the towing tank is already present and functional. The D-class is fortunately small enough to be accommodated by some larger towing tanks and Dand et al. (2008) used this approach in the past to test the D-class. The QinetiQ towing tank at Haslar is 270 m long and is capable of accommodating the D-class but the reliability of the results needs careful consideration before this route forward can be justi ed. The QinetiQ tank could be used to study two key performance indicators of the D-class; the planing performance (and the pulsing motions) in at water and the vertical boat motions in waves, and these will now be discussed. The D-class is capable of achieving 25 knots so to be able to study the full range of planing performance the towing tank must be capable of 25 knots; the QinetiQ tank is only capable of 22 knots. Next the run time need to be considered because Dand et al. (2008) reported that the period of the pulsing motion was approximately 30 s. Therefore, to study a complete pulsing motion the run time must be at least 30 s long. 270 m at 22 knots (11.3 m/s) takes 23.9 s (not including the acceleration and deceleration stage of the tow carriage) so it is not possible to study a complete cycle of the pulsing motion.
Finally, the pulsing motion is a
coupled interaction between the propulsion and resistance of the craft because the thrust is xed by the outboard engine and the resistance changes due to hull deformation. However, a towing tank is designed to drive the craft at a xed speed and not a xed thrust. There are also concerns regarding the aerodynamic blockage eect because the D-class will almost completely ll the space between the tow carriage and the water surface.
This shows that
the QinetiQ towing tank will not be capable of providing the required results for the planing performance of the D-class. The next performance indicator that can be studied in the QinetiQ tank is vertical boat motions in waves. Although the D-class is small enough to t within the tank there is minimal clearance between the boat and the towing carriage.
When forward speed and waves
are introduced into the problem, a collision is very probable and QinetiQ decided that this
Real boat (no scaling laws, trim & sinkage included), real environment, proven to work in the past, long run times
Low control over parameters and Real results, novel, evidence on environment, adverse effects the effects of hydroelastic on when changing parameters, large planing craft performance sensor network and complex data acquisition system
Sponson & keel pressure (leading to hull pre-tension and longitudinal stiffness), forward velocity, environmental conditions (wave height and drop height)
Boat motion (surge, heave, pitch, roll, yaw) environmental conditions (waves, wind, tide), Full Scale Holistic structural deformation (sponson & keel pressure, deck Approach deflection, deck hinge deflection, hull deflection)
Realistic model so useful for the RNLI
Low accuracy and lack of control over parameters, high costs
High costs
Ingores presence of sponsons, high cost
Applicable to almost all planing vessels, possible comparison to numerical models (due to lack of sponsons)
Hull shape is segmented leading to irregular fluid flow, previous experiment have shown difficulties, thrustline issues
Ingores presence of sponsons, high cost
Realistic model so useful for the RNLI
CHAPTER 3.
Vertical and horizontal forces, Longitudinal stiffness, number Good control over parameters of segments, forward velocity, and environment Segmented Scale vertical accelerations, longitudinal deformation, trim mass, LCG, deadrise angle Model with Sponsons in Waves and sinkage
Difficulties in achieving a smooth fabric planing surface, limited run time, thrustline issues, scaling laws Good control over parameters Hull shape is segmented leading and environment, easy to change to irregular fluid flow, previous segments and longitudinal experiment have shown stiffness difficulties, does not include sponsons, thrustline issues
Good control over environment, standard ITTC methods available
Material properties, pretensioned stresses, forward velocity, mass, LCG, deadrise angle Longitudinal stiffness, number of segments, forward velocity, mass, LCG, deadrise angle
Possible comparison to numerical Neglects cross-flow effect (i.e. 2D models simplication of a realistic hull shape) so low applicability to RNLI
Good control over parameters Neglects cross-flow effect, trim and environment, proven to work angle and sinkage are fixed, in the past limited run time, scaling laws
Fabric Flat Plate Planing Surface
Not realistic so low applicability to RNLI
Opportunities Weaknesses Controlled and novel work on Quasi-2D so low applicability to sponsons, possible comparison to RNLI numerical models, low cost
Material properties, pretensioned stresses, forward velocity, trim angle, sinkage
Strengths Weaknesses Good control over parameters Neglects forward speed, edge and environment, proven to work effects, scaling laws in the past, study individual components (hull, keel and sponsons), quick to perform
Vertical and horizontal forces on planing surface, deformation of planing surface, pulsing motion Vertical and horizontal forces Scale Fabric Model on planing surface, of a Deep-Vee deformation of planing Planing Surface surface, pulsing motion Vertical and horizontal forces, vertical accelerations, Segmented Scale longitudinal deformation, trim Model without and sinkage Sponsons in Waves
Variables Material properties, pretensioned stresses, internal pressure, impact velocity, mass, deadrise angle
Material properties, preVery few variable so cause of any Water jets are normally turbulent, Simplification of reality, novel tensioned stresses, jet velocity, measure effects can easy be trim and sinkage are not method, low cost angle of inclination establish, controlled environment considered, scaling issues and parameters, infinite run time
Performance Indicators Slamming accelerations and duration, deformation of structure
Vertical and horizontal forces on planing surface, Fluid Impacting an deformation of planing Inclined Fabric Plate surface, pulsing motion
Quasi-2D Fabric Wedge Impact Free Surface
Method
48 HYDROELASTIC METHODOLOGY
Table 3.1: Comparison of experiment methods for individual aspects of hydroelasticity.
3.3.
PROPOSED EXPERIMENTAL METHODS COUPLED ASPECTS
49
experiment was too risky. It is worth mentioning that an experiment with zero forward speed in wave was considered. This method means that the real structure deformation of the D-class can be measured in a controlled environment. However, this ignores the hydrodynamic forces and does not provide the RNLI with applicable results. A recirculating water tank could be used to test the D-class instead of a towing tank. The major advantage of a recirculating water tank is the in nite run time; however, the water's surface is not smooth. The recirculating water tanks available are not large enough to accommodate the full-scale D-class without the aecting performance. A scale model could be used, and then a recirculating water tank could be used with all the experiments discussed in section 3.2. Although, the irregular water's surface will cause random error and the error will increase as the scale decreases because the surface waves will become proportionally larger.
3.3.2 Open water trials So far this section has shown that a holistic approach must be taken to consider the interaction of the three aspects of hydroelasticity. It has also shown that there are no numerical or scaled methods that are capable of modelling this boat.
Finally, it was shown that the QinetiQ
towing tank is not suitable to further the work of Dand et al. (2008). The nal method to explore the D-class is full scale open-water trials. The open-water trials involve measuring the boat performance, the structural deformation and the environmental conditions of the full scale D-class in a range of real test conditions. The boat performance is de ned by the boat motion (primarily pitch and heave motions) and the planing performance (forward velocity and pulsing motion). The major structural components to be measured are: fabric hull deformation, deck de ection, deck hinge angle, sponson and keel pressures, and sponson rotation. The range of test conditions includes: static tests (i.e. zero forward speed), drop tests, at water trials and wave trials. The internal pressures of the sponson and keel are the essential parameters in the experiment to vary the hydroelasticity because, not only do they change the response of the sponson and keel, they also change the pre-tensioned stresses in the hull, the longitudinal stiness and the sponson rotational stiness. The underlining purpose of this experiment is to explore each aspect of hydroelasticity individually whilst still including the interactions of the three aspects.
This is achieved by
triggering one aspect at a time, using a number of specially selected test conditions, but measuring the structural deformation and boat performance in all three aspects.
So the
stationary test condition has been selected to measure the static undeformed shape of the boat (as this is not accurately known) and provides a baseline of data for the next test conditions. The drop tests will trigger the hydroelastic slamming aspect and provide an insight into the deformation caused by the planing forces (added mass) and slamming events during the wave trials.
The at water trials provide a quasi-static planing force to induce the hydroelastic
planing surface aspect and to give an indication of the global structural deformation without waves.
The nal wave trials aim to tie all the aspects together and stimulate the global
hydroelastic aspect. Each test builds on the previous test in a similar manner to strip theory. There are two coupled problems within the D-class that are worth making clear.
The
50
CHAPTER 3.
HYDROELASTIC METHODOLOGY
Measured parameter
Number of sensors
Linear boat motion
3 x tri-axial accelerometer
Angular boat motion
3 x rate gyro scope
Engine thrust
2 x strain gauge
Boat speed
1 x GPS
Sponson pressure
7 x pressure transducers
Sponson rotation
2 x displacement tranducers
Keel pressure
3 x pressure transducers
Deck panel deformation
8 x strain gauges
Deck hinge deformation
3 x angle transducers
Hull deformation
15 x displacement tranducers
Table 3.2: List of sensors for open water trials.
three aspects of hydroelasticity are coupled together and they were discussed in section 2.8; however, there are also the interactions of the structural components discussed in section 2.2.3. The aspects of hydroelasticity are the reason for the four stage holistic experiment and the interacting structural components are the reason for large structural sensor network. The success of this experiment relies on the sensor network and data acquisition system to accurately and reliably record the boat performance and structural deformation of all three aspect of hydroelasticity. It was initially estimated that more than 50 sensors will be required to measure all of the boat performance and boat deformation, which includes; boat motion, trim angle, engine thrust, forward speed, keel pressure, sponson pressure, sponson rotation, deck panel de ection, hinge angle and hull deformation (see table 3.2). This should be captured at a sampling frequency above 1000 Hz, see Townsend (2008). The sensor network and data acquisition system must also be a stand-alone (no host PC or mains power), compact, shock proof (up to 10 g per 0.1 s) and waterproof.
Strengths and weaknesses The major strength of this method is the realism; the boat is full scale and the environment is real. This means that there are no scaling laws to be considered and performance indicator such as trim and sinkage are automatically included. Although the realism still causes issues to the accuracy and reliability of the results because the control over the parameters and environment are lower. The environment is characterised using averages rather than reporting the exact wave pro le.
The parameters of the D-class can be changed using the internal
pressure of the keel and sponson, but this can change the shape and performance of the entire boat. For example, if the keel pressure is increased this will cause the boat to hog and reduce the top speed of the craft. Nevertheless, full-scale methods have been used countless times in the past and are the generally preferred method to study boat performance.
Plus the
experiment can be set up to have a run time over 30 seconds (period of the pulsing motion). The leading weakness of this method is the sensor network and data acquisition system because of the large range of variables that must be measured to fully understand the complex hydroelastic D-class. It is estimated that over 50 sensors are required to measure the important
3.4.
QUASI-2D DROP TEST METHOD
51
structural components and environmental conditions. Previous experience has shown that a set up similar to this can be problematic and time consuming. Also the system has to work in a very harsh environment including high shocks and salt water; this means that some of the accuracy is exchanged for a rugged and watertight system.
Opportunities and threats The method provides a unique and excellent opportunity to further our understanding about the performance of hydroelastic planing crafts by generating real and novel results. The results will hopefully give evidence to whether or not hydroelasticity improve the performance of a planing craft. Due to the realism of the experiment the results will be highly applicable to the RNLI. On the other hand, the lack of control over the parameters and low accuracy of the experiment reduces its academic credit. The costs associated are high (> ¿10 k).
3.4 Quasi-2D drop test method It is hypothesised that the pre-tensioned stress in the fabric hull of the RNLI D-class will aect the slamming characteristics, especially the peak acceleration and the peak duration.
3.4.1 Equipment The quasi-2D drop test rig is shown in gure 3.3 and is the same rig used by Lewis et al. (2010).
A pulley system was used to adjust the drop height.
The wedge was tted to two
vertical poles via four bearings to ensure the wedge fell vertically. The accelerometer was tted to the top of the wedge near the bearings (to avoid submersion). A quick release shackle was used as a release mechanism. The steel framed wedge section was originally designed to test composite panels at variable deadrise angles but a special fabric hull was wrapped around to form a fabric hull. The deadrise angle could be adjusted to 5°, 15° and 25°. The fabric hull was manufactured by the RNLI from two laminated sheets of DTex 1100 rubber coated fabric, the same material and lay up found on the D-class. A turn-buckle mechanism was used to adjust the pre-tension stresses in the fabric hull. The wedge was dropped into a water tank that was 5.8 m long and 0.75 m wide, and the water height was 0.5 m. In the centre of the tank there was a side window (0.825 m
Ö
0.485 m).
A Crossbow CXL-HF accelerometer was used to measure all three axis of acceleration. The sensitivity was
±
10 mV/g and it had a range of
±
100g. A DaqLab 2000 series data
acquisition system was used with a sampling frequency of 5000 Hz and an accuracy of 0.1 mV. This system together gave an accuracy of for storage and processing purposes.
±
0.01 g and it was connected to a laptop
A MotionPro X high-speed camera manufactured by
Integrated Design Tools Inc. was used to record the event at a rate of 2000 frames per second. Lights were required to provide enough contrast.
52
CHAPTER 3.
HYDROELASTIC METHODOLOGY
Figure 3.3: Quasi-2D drop test rig.
3.4.
QUASI-2D DROP TEST METHOD
53
3.4.2 Parameters The main parameters aecting the slamming characteristics of a quasi-2D drop test are the deadrise angle, drop height and mass. The deadrise angle of the D-class varies from 0° at the transom to a maximum of 15°; therefore, the variable deadrise angle wedge was tested at 5° and 15°. A 0° deadrise angle was not tested because other phenomena (such as air cushioning) occur at deadrise angles below 4°, see Bereznitski (2001). The maximum drop height of the rig was 1.2 m so 0.5 m and 1 m were tested, which correspond to an impact velocity of 3.13 m/s and 4.43 m/s, respectively.
The mass and geometry of the wedge were not varied but
they were scaled in relation to the D-class.
The mass of the D-class is 655 kg, taken from
Dand et al. 2008, and 5 m in length which is 131 kg/m. The wedge had a longitudinal depth of 0.735 m and a mass of 50.2 kg results in a mass per length of 68.3 kg/m. The transverse width of the wedge, diagonally from keel to chine, is 0.501 m so at a deadrise angle of 5° and 15° the horizontal width of the wedge (chine to chine) is 0.998 m and 0.968 m, respectively. The D-class has a width of two metres. This means that this wedge was approximately 1/2 scale for both mass and geometry. The main variable of this experiment is the hull stiness so three stiness conditions were chosen to replicate this eect: approximately rigid and, 1000 N/m and 0 N/m of pre-tensioned stress. The approximately rigid condition was represented by a six millimetre sheet of medium density breboard (MDF) (six millimetres thickness was the correct panel thickness for the frame) and the Young's modulus of MDF is approximately 4 GPa.
MDF was chosen for
its light weight and isotropic material properties. Lewis (2003) stated that varying the pretensioned stress within a fabric will have the same eect as varying the material properties, so only the pre-tensioned stress was varied. The real pre-tensioned stresses in the D-class' hull are unknown but, in places, the tension is very low because the fabric is very loose; therefore, one stiness condition was used to represent zero hull tension (0 N/m). It was unclear how much more tension would result in a measurable dierence so three times the magnitude was used (1000 N/m). The pre-tension stresses were only applied in the transverse direction and were measured using the strain in the fabric. The non-linear material properties of two ply DTex 1100 showed that a tension of 1000 N/m produced a strain of 0.005.
It was found
that even under the 1000 N/m the fabric de ected downwards due to gravity; therefore, this de ection was also measured to increase the repeatability of the experiment.
The central
vertical de ection at 1000 N/m was 6 mm and at 0 N/m the de ection was 11 mm.
The
boundary conditions of the fabric sheet along the chine and centreline were a pin joint but along the two transverse edges the fabric was eectively a roller, only restricting the out-ofplane de ection. The boundary conditions of the MDF were very similar but along the chine and centreline the panel was bolted in place causing a fully clamped condition.
3.4.3 Procedure 1. The deadrise angle and pre-tensioned stress were set. 2. The wedge was raised to the desired drop height.
54
CHAPTER 3.
HYDROELASTIC METHODOLOGY
3. The water was allowed to settle so that the surface movement was less than
±
5 mm.
4. The data acquisition system and high speed camera were started.
5. The wedge was released.
6. The data acquisition system and high speed camera were stopped.
7. The procedure was repeated from step two to acquire three repetitions.
3.4.4 Post Processing The DaqLab 2000 data acquisition system saved the voltage data from the accelerometers. The voltage was converted into acceleration using equation 3.1. The acceleration was zeroed using the mean of the rst 0.1 s of the recording before the wedge was released.
Acceleration = V oltage/0.01
(3.1)
The time domain signal was transformed into a frequency domain signal using a Fourier transformation. The fast Fourier transformation function in MatLab was utilised with a windowing length of the next power of 2 greater than the number of samples in the time domain. An example of the FFT MatLab script can be seen below.
m = length(dataset1); n = pow2(nextpow2(m)); y = fft(dataset1,n); freq1 = (0:n-1)*(fs/n); power1 = y.*conj(y)/n; The time signals were ltered using a Butterworth lter to remove any unwanted frequency data. The Butterworth lter function in MatLab was utilised with an order of 2 (n) and a cut-o frequency (Wn).
An example of the Butterworth lter MatLab script can be seen
below.
[b,a] = butter(n, Wn, 'low'); set1 = filtfilt(b, a, set1);
3.5 Full-scale test method It has been shown that the structural components of the D-class are coupled together and that the three aspects of hydroelasticity will also interact due to this coupling. This has led to a four stage holistic experiment being chosen where the four sub-experiments build upon one another to construct a picture of the entirely hydroelastic D-class. The four sub-experiments are; static tests, drop tests, at water trials and wave trials.
A basic description for each
sub-experiment was given in section 3.3.2 and it addresses how each sub-experiment builds upon the last using both the performance and deformation measurements.
3.5.
FULL-SCALE TEST METHOD
55
The primary parameter of the experiment is the structural stiness of the D-class which will be varied using the internal pressures of the sponson and keel.
The internal pressures
are coupled with all the other components and will change multiple other structural stiness variables within the D-class because this is a holistic approach, see section 2.2.3 for more detail. The other variables that are aected include: the pre-tensioned stresses in the hull, the deck panel longitudinal stiness, deck hinge stiness and the sponson rotational stiness. However, there is very little understanding about the coupling of these parameters and variables, and their aects have not been quanti ed. So the other variables must be measured at the same time as the performance.
This means that the experimental D-class must be tted with
numerous sensors to measure all these variables. The three internal pressure conditions will be 2 psi and 2.25 psi, 3 psi and 3.25 psi, and 4 psi and 4.25 psi for the keel and sponson, respectively. They will now be referred to as 2 psi, 3psi and 4 psi for simplicity. The internal pressures will be varied by the same degree in each sub-experiment so that the eect of hydroelasticity can be compared between the sub-experiments.
3.5.1 Aims The overall aim of this experiment is to explore the eect of hydroelasticity on the performance, de ned by the boat motion and forward speed, of the D-class and to prove whether or not hydroelasticity can improve the performance. The stationary tests aim to study the structural deformation due to only the weight and buoyancy forces and to quantify the de ections. These results will be used as a baseline to measure the de ections of the proceeding experiments. It will provide precise dimensions of the D-class in water and further our understanding of the coupled structure. It can also be compared to results of the waterline deformation experiment that is included in appendix E. The aim of the drop tests are to measure the eect of hydroelasticity on the slamming characteristics, primarily de ned by the peak acceleration and peak duration, due to only the vertical forces, i.e. ignoring the eect of forward speed. The slam characteristics will aim to quantify the change in WBV due to hydroelasticity. These results will be compared to the stationary lightship results to calculate the structural deformation due to only vertical forces. These results will be compared with the results of the quasi-2D impact wedge experiment also performed during this project. The at water trials aim to measure the structural deformation due to weight, buoyancy and quasi-static hydrodynamic lifting forces caused by quasi-steady planing. These deformation results will be compared to the stationary three crew results to calculate the deformation due to quasi-steady planing. This experiment primarily measures the eect of hydroelasticity on the fabric hull deformation and planing performance.
It aims to further our knowledge
about the pulsing motion found on the D-class and can be compared to the results of Dand (2002a, 2003b). The drop tests and at water trials focused on one performance indicator, either boat motion or planing performance, but the wave trials aim to pull everything together by exploring both performance indicators at once. The boat motion results intend to quantify the eect of
56
CHAPTER 3.
HYDROELASTIC METHODOLOGY
hydroelasticity on the WBV. These results can be compared to the last three sub-experiments and the work of Dand (2004).
3.5.2 D-class equipment The same D-class will be used throughout the sub-experiments with the same sensor network and data acquisition system, plus the parameters are varied by the same degree in each subexperiment. So the data acquisition system, sensor network, power source, parameters and equipment within the primary D-class will be described in this sub-section.
In the follow-
ing sub-sections the parameters, equipment and procedures for the sub-experiments will be discussed.
Data logger A sensor generates an analogue signal and this must be converted to a digital signal and then saved to a memory unit. There are many ways to perform these steps during an experiment but the data logger for these trials must meet the following speci cations:
Durable acceleration of at least 20 g were expected, see Myers et al. (2011).
Battery powered for minimum 3 hours.
Waterproof, or can be operated from inside a waterproof case.
Sample frequency greater than 180 Hz.
The WBV weighting has an 80 Hz low pass lter, then doubled for anti-aliasing.
52 channels of voltage from accelerometers, strain gauges, etc, see table 3.2.
There are very few data loggers that meet this speci cation; partly due to the number of channels and also the relatively high sample frequency. One data logger that does meet this speci cation is the National Instruments (NI) Compact Reprogrammable Input Output (cRIO) device. The cRIO is a very exible and rugged device that is designed to be an embedded control and monitoring system. Due to its exibility, ruggedness and availability it is the best data logger for this experiment. Eight C-series modules can be tted to the cRIO to control the inputs and outputs through a Field Programmable Gate Array (FPGA) chip. Four types of C-series module will be required for this experiment; strain gauge modules, voltage input modules, accelerometer modules and a SD card module. The strain gauge C-series module (NI 9236) completes a quarter-bridge Wheatstone circuit with a 350 ohm strain gauge. Each module has eight channels with individual ltered dierential ampli ers and Analogue to Digital Converters (ADC). The digital signal has a resolution of 24 bits. The module also includes a shunt calibration function. The voltage input C-series module (NI 9205) is capable of reading voltage ranges from
± 200 mV to ± 10 V from 32 single
ended measurements (or 16 dierential measurements). The voltage module can acquire the signal from all the type of sensors considered in the experiment except the strain guages and
3.5.
FULL-SCALE TEST METHOD
57
acccelerometers (hence the seperate C-series modules). A multiplexer is used to sample the 32 channels before the signal is passed through a ltered dierential ampli er and then an ADC. The digital signal has a resolution of 16 bits. The accelerometer C-series module (NI 9234) is able to sample 4 channels simultaneously.
The voltage input range is
±
5 V. The
module also includes an anti-aliasing lter and an ADC. The nal module to be used is the SD card C-series module (NI 9802). It is capable of simultaneously reading or writing to the two SD cards at a maximum rate of 2 MB/s and has a total memory of 4 GB. The data must be written directly from the FPGA chip and not the real time controller to achieve the maximum rate. All the sensors were sampled at 2500 Hz; the same sampling frequency used by Townsend (2008) to capture the boat motion of an Atlantic 75. The conventional method for using a cRIO as a data logger is to feed the output signal from the FPGA chip into a real time controller where the data is then saved to; the internal ash memory, a USB stick or a SD card C-series module. However, the SD card module is accessed through the FPGA chip (to provide it with the highest possible reading/writing speed of 2 MB/s) which allows the real time controller to be removed for this arrangement and is shown in gure 3.4. The author has not found any examples of a data logger application where the real time controller is not used, so this method is potentially novel. The bene ts for the method include; reliability, simpler coding and reduced compilation time. The code can be seen in appendix F. This method has been published as a NI case study, see Halswell (2013), and the case study can be seen in appendix B.
Figure 3.4: Using the cRIO as a data logger with and without a real time controller.
The mass of the data logger and associated batteries, voltage regulator, impact protection and waterproof case was 36 kg and it was located 2.03 m from the transom.
The same
customised system was used on both boats. The eect of the data logger on the LCG was predicted to move the location of LCG forward by 132 mm to 1.452 mm from the transom, using equilibrium of bending moments. The LCG was reported by Dand (2003c) to be 1.32 m from the transom, see table 2.1.
58
CHAPTER 3.
HYDROELASTIC METHODOLOGY
Boat motion The performance of the D-class is rst de ned by the boat motion.
It is conventional to
measure the boat motion using accelerometers and has been successfully used many times in the past, see Townsend (2008).
Dand et al. (2008) measured a peak of 4 g during a scale
model experiment of the D-class in waves, so a suitable accelerometer must have a range of at least
±
5 g and have an accuracy of at least 0.05 g.
Three CFX USCA-TX tri-axial
accelerometers were used to measure the accelerations. They had a range of
± 20 g with a DC
to 200 Hz at frequency response. Above 200 Hz the accelerometers had a -6 dB response. The British Standard 6841:1987 (Measurement and evaluation of human exposure to whole-body mechanical vibration and repeated shock) is concerned with the transmission of mechanical vibration to the human body within the frequency range of 1 Hz to 80 Hz, which ensures the accelerometers capture the complete human response. fed into a NI 9234 C-series accelerometer module. of
±
5.75
µg
The accelerometer signals were
This combined system has an accuracy
and a 2500 Hz sample frequency was used. The location of the accelerometers
is shown in gure 3.5.
The rst accelerometer (A0) was tted to the transom next to the
helm and the second accelerometer (A1) was tted to the deck between the knees of the crew. These accelerometers measure the vibration of the structure near the point of contact between the helm or crew to provide an estimation of the accelerations experienced by the helm or crew. The nal accelerometer (A2) was tted near the bow of the boat to investigate how the vibration changes along the length of the D-class. Three angular rate gyroscopes were also used to measure the angular rotation of the D-class to provide a more accurate estimate of the trim angle and pitch motion. The gyroscopes are model CRS03, manufactured by Silicon Sensing, they have a range of
±
100°/s and an accuracy of
±
6.6
°/ms.
The trueness and
precision of all the sensors used in the full scale tests are summarised in table 3.5.
Figure 3.5: Location of accelerometers during the full-scale drop tests (all dimension in metres).
Engine output The hydrodynamic resistance is harder to measure in open water than in towing tanks because there is no xed point of reference, but can be determined using the power or thrust of the engine.
The revolutions of the engine could be measured but this does not measure
the power output, it measures the potential power output.
Dand (2002a) used a load cell
carefully placed in between the transom and the outboard engine which produced good results. Although, discussions with Dr. I. W. Dand revealed that tting the load cell required complex
3.5.
FULL-SCALE TEST METHOD
59
modi cations to the outboard engine and the load cell can be easily damaged from impulse loads, plus they are an expensive option. A load cell measures the load by indirectly measuring the strain on a component inside the load cell. The properties of this component are accurately known so that the load can be calculated from the strain. The strain is normally measured using a strain gauge inside the load cell. The theory of a load cell can be applied to a component on the outboard. Then the measured strain on this outboard component can be equated to a known thrust on the engine. Almost all outboard engines use metal rods to adjust the trim angle of the propeller, called trim pins, and this is a suitable component for this idea because a x proportion of the force is transmitted through this pin. This method eectively makes the structure of the engine the load cell and the author does not know of this being used on outboard engines before so it may be a novel method. The strain experienced by the trim pin and the strain gauge must be matched to ensure this method has the maximum accuracy and precision while still performing within the elastic limits of the trim pin. The maximum strain that a standard strain gauge can withstand is three percent so the trim pin must not experience strains higher than three percent at the maximum load output from the engine. Three percent strain is still within the elastic limit of the stainless steel used to manufacture the trim pins. The thickness of the trim pins were calculated using simple beam theory, see Gere and Goodno (2009), to maximise the accuracy of the strain gauges. The trim pins regularly gets soaked in water so the strain gauges and lead wires must be waterproofed. A suitable strain gauge is the WFLA-3-350-11-5L from Techni Measure, which has an accuracy of
± 2.3 × 10−8 .
Panel de ection Currently the most common method for measuring the deck deformation is to use a strain gauge.
The strain gauges can be bonded directly to the deck in the areas of interest and
they provide a quick, cheap and eective method. Both the X and Y axis deformations can be measured using a bi-axial strain gauge. A 350 ohm quarter-bridge strain gauge must be used because the input to the strain gauge C-series module (NI 9236) is limited to a 350 ohm quarter-bridge circuit. Furthermore, a half or full bridge circuit cannot be used because the strain gauge cannot be tted to the underside of the deck panels. The second mode of vibration can only be measured using two strain gauges and the results from the waterline deformation experiment showed signs of the second mode of de ection, so two strain gauges will be used. A suitable strain gauge is the GFCA-3-350-70-5L from Techni Measure, which have a trueness of
± 2 × 10−4
and a precision of
± 2.3 × 10−8 ,
see table 3.5. The exact locations of all the
strain gauges can be seen in gure 3.6.
Deck hinge angle The deck hinge angle is a complex parameter to measure because there are few o-the-shelf sensors capable of ful lling the role. The major problem with measuring the angle is the need for a mechanically moving part which reduces the waterproo ng of the system. Therefore, it has been decided that the best method to measure the angle of the hinge is indirectly using a
60
CHAPTER 3.
HYDROELASTIC METHODOLOGY
Figure 3.6: Deck strain gauge locations (all dimension in mm).
strain gauge, in a similar manner to the outboard engine thrust, which provides a waterproof and inexpensive system.
The hinges will be bridged by a 3 mm thick strip of aluminium
which will deform as the hinge bends, see gure 3.7. One end of the bridge strip will be fully clamped and the other will be a roller joint; this causes the bridge strip to only experience pure bending moments so the curvature is constant along the strip. The deformation can be measured using a strain gauge and then the strain can be converted into a hinge angle.
A
suitable strain gauge is the FCA-3-350-11-5L from Techni Measure, which have a trueness of
± 2 × 10−4
and a precision of
± 2.3 × 10−8 ,
see table 3.5.
Figure 3.7: Deck hinge bridge design.
Hull deformation The next component to be measured is the fabric hull. Jenkins and Korde (2006) noted the importance of using non-contact methods to measure the deformation of membrane structures due to the potential increase in added mass and stiness. So the ideal measurement method would involve optical tools such as laser vibrometry or Digital Image Correlation (DIC), however, applying these methods to a D-class is not feasible. The rst reason is that an optical tool is normally placed at a distance from the membrane that is greater than or equal to the width or length of the component. In the case of the D-class this would be over one metre from the hull but then the deck blocks the view of the optical tool.
The other issue is the
3.5.
FULL-SCALE TEST METHOD
61
presence of water inside the hull because the angle of incidence between the optical source and the water surface de nes whether the light is refracted or re ected. The continuous irregular motion of the water inside the hull will cause this to change irregularly causing many errors in the results. If an optical method is not feasible then a contact method must be used. Contact methods must not be used on extremely thin and light-weight membranes, for example the membrane used in Jenkins and Korde (2006) experiment is 25.4
μm thick and has
2 a density of 35.6 g/m . Whereas, the fabrics used on the D-class are 1.25 mm and 1.7 mm thick and have a density of 355
g/m2
and 500
g/m2 .
Plus the added mass of the surrounding
medium needs to be considered and in the case of Jenkins and Korde's work the medium was air (density of air is approximately 1.2
kg/m3 ).
On the other hand, the surrounding medium
on one half of the hull is water (density of water is approximately 1030
kg/m3 ).
This reveals
that the mass and added mass of the hull are considerably higher than that of membrane structures considered by Jenkins and Korde. Therefore, this project will assume that if the added mass of a sensor is signi cantly lower (magnitude of 100 or more) than the mass and added mass of the hull then the errors caused by the presence of the sensor will be negligible. This allows more practical methods to be used including contact methods. Jenkins and Korde (2006) demonstrated that the local transverse deformations of a membrane are weakly coupled to the global deformation of the membrane structure. This implies that if a strain gauge was bonded to the hull then the measured strain is only the local strain and not the global strain; therefore, a strain gauge is not a suitable sensor. The next contact method involves using linear displacement transducers. A grid network of transducers could be tted through the deck and attached to the hull to provide the global deformation, only if the added mass of the transducer is signi cantly lower than the mass and added mass of the hull. Linear displacement rods are not suitable because they are designed to measure one xed axis so if the hull was to move perpendicular to the measurement axis it could damage the sensor. The most suitable linear displacement transducer is a string potentiometer with a low cable tension to reduce the added mass of the transducer. String potentiometers are not waterproof due to the string inlet, which is a major disadvantage to this method. There is no way around this problem so extra protection will be used. An appropriate string potentiometer is the JX-PA series by Uni Measure. The range of the potentiometer can be picked depending on the expected range of the hull movement. The resolution of a potentiometer is theoretically in nite but is actually de ned by the resolution of the data logger. The cRIO will save the voltage in a 16 bit format leading to a resolution of 0.01 mm. The wire tension is 0.41 kg will be signi cantly less than the mass and added mass of the hull so this contact method should have minimal eect on the response of the hull. The locations of the hull string potentiometers can be seen in table 3.3 and in gure 3.8.
Sponson deformation The sponson is di cult to measure because it exhibits the same problems when using contact methods as the hull deformation and the deformation is 2D (whereas the hull is quasi-1D). String potentiometers are no longer suitable because multiple potentiometers would be required
62
CHAPTER 3.
HYDROELASTIC METHODOLOGY
Sensor
H1
H2
H3
H4
H5
H6
H7
H8
H9
X-axis
493
770
772
955
953
1126
1122
1433
1426
Y-axis
494
276
484
276
484
276
479
276
468
Sensor
H10
H11
H12
H13
H14
H15
H16
H17
-
X-axis
1611
1603
1801
1802
2083
2072
2256
2243
-
Y-axis
276
459
208
313
119
445
208
434
-
Table 3.3: Hull string potentiometer locations.
Figure 3.8: Hull string potentiometer locations.
to measure the circular deformation of one 2D cross-section of the sponson and there is not enough space within the D-class for these sensors. The sponson deformation could be measured indirectly by measuring the internal pressure. This does not provide the exact sponson shape but the pressure is proportional to the shape, so if a pressure uctuation is measured then this indicates that there is a structural uctuation. It has been decided that the most suitable method to measure the deformation of the sponson is to use a combination of approaches; pressure sensors will be used to measure the internal pressures of each ba e and two string potentiometers will be used to measure the rotation of the sponson. The internal pressure can be measured using IMP pressure sensor manufactured by Impress Sensors & Systems. This sensor can measure the absolute pressure with a range of 0 to 14.5 psi (1 bar) and an accuracy of 0.03625 psi. The sponson rotation will be measured using the same string potentiometers that are used for the hull deformation.
Keel deformation The keel deformation is probably the hardest component to measure accurately.
The same
problems that were discussed with the hull and sponson still exist and the keel is virtually inaccessible once the deck has been tted.
So the best option will be to measure the keel
deformation indirectly using an internal pressure sensor in the same manner as the sponson pressures. Three pressure sensors will be tted along the length of the keel to measure the pressure uctuations along its length. The same IMP pressure sensors will be used.
3.5.
FULL-SCALE TEST METHOD
63
Power source The power source for the data logger and sensors was supplied by two 12 volt batteries in series to give a 24 volt potential dierence. The 24 volts must be regulated to provide dierent voltage levels for the data logger and the sensors. The required voltages are; 20 V for the cRIO, 12 V for the accelerometers, the pressure sensors and the string potentiometers, and 5 V for the gyroscopes. The voltage regulator system was manufactured by the Institute of Sound and Vibration Research (ISVR) workshop. The voltage regulator system also includes; fuses, a low voltage indicator and a power switch.
3.5.3 Stationary tests It has been hypothesised that the static shape of the D-class will change depending on the internal pressures of the sponson and keel, and the loading condition, so the parameter for the stationary tests were the loading conditions and the internal pressures.
Three loading
conditions were tested; lightship, two crew and three crew (all standard operating equipment was in place). The D-class was loosely secured to the pontoon in the Inshore Lifeboat Centre (ILC) marina to provide a relatively at water surface. The experimental procedure was: 1. Secure boat loosely to pontoon. 2. Set internal pressures of sponson and keel. 3. Load boat with required number of crew. 4. Record static shape over one minute. 5. Repeat steps 2 to 4 for each loading condition and internal pressures.
3.5.4 Drop tests It has been hypothesised that the internal pressures in the sponson and keel will aect the slamming characteristics of the D-class, especially the peak acceleration and the peak duration. The experimental set-up of the full-scale drop tests can be seen in gure 3.9. The D-class was lifted with a crane and a simple drop height gauge (a marked length of rope) was tted to the transom of the vessel to measure the drop height. The drop height gauge had an accuracy of
±
10 mm. A slip hook was used as a quick release mechanism which could be activated
from the shore. The bungee cord was required to stop the heavy shackles (used for trim angle adjustment) from impacting the deck and causing unwanted structural vibration. The trim angle was adjusted using various boat harnesses of dierent lengths. A GoPro Hero 2 was used to record the drops at 120 fps. Finally a remote trigger was used to start the data acquisition system. The main parameters for this experiment were the internal pressures of the sponson and keel, the drop height and trim angle. The drop heights of the full-scale drop tests matched the
64
CHAPTER 3.
HYDROELASTIC METHODOLOGY
Figure 3.9: Full scale drop test set-up.
3.5.
FULL-SCALE TEST METHOD
65
drop heights used in the quasi-2D drop tests; 0.5 m and 1 m. The boat was always released at the same trim angle of 4.25° and this was as close to the running trim angle that could be achieved with the various boat harnesses. Dand et al. (2008) measured a running trim angle of four degrees. The trim angle was measured with a calibrated spirit level that had an accuracy of
±
0.25°.
The experimental procedure was:
1. The internal pressures and trim angle were set.
2. The boat was raised to the desired drop height.
3. The boat and water surface was allowed time to settle so that the surface variation was less than
±
10 mm (measured using the drop height gauge).
4. The data acquisition system and video recorder were started.
5. The D-class was released.
6. The data acquisition system and video recorder were stopped.
7. The procedure was repeated from step two to six for ve repetitions.
3.5.5 Flat water trials It has been hypothesised that the at water speed will be aected by the internal pressures of the keel and sponson. The at water trials simply involved driving the D-class along the river Medina, on the Isle of Wight, in a straight line. The boat was driven at full throttle to minimise any errors from throttle control. The top speed was measured with the on board GPS unit which had an accuracy of each internal pressure condition.
±
0.1 knot. The experiment was repeated 10 times for
The D-class was loaded with three crew and all standard
equipment. The experimental procedure was:
1. The internal pressures were set.
2. The boat was lined along the river.
3. The data acquisition system was started.
4. Three seconds later the helmsmen accelerated at full throttle and continued for approximately 45 s.
5. Steps two to four were repeated to achieve 10 repeats, ve in each direction along the river to average any eects of tide, wind or waves.
6. Step one to ve were repeated for each internal pressure condition.
66
CHAPTER 3.
HYDROELASTIC METHODOLOGY
3.5.6 Wave trials It is hypothesised that the internal pressures of the sponson and keel will aect the sea keeping performance of the D-class. The parameters for the wave trials are the wave height and period which will be related to the sea states associated with a Beaufort force three to four, with both head and following seas. The stochastic behaviour of real waves means that no two runs will encounter the same wave pro les, therefore, a minimum of 100 wave encounters will be captured for repetition, see Lloyd (1998) page 170.
Also two D-classes will be driven side-
by-side so they then encounter very similar wave pro les and can then be compared to each other (one D-class will measure the performance and deformation, while the other D-class will measure only the performance) as recommended by Lloyd (1998) page 166. The boats will be driven at least two boat lengths apart so that the wake of one boat will not interfere with the other boat. Both boats will be loaded with three crew and all standard operating equipment. The wave pro le was measured using a local wave buoy called Lymington wave recorder. ChiMet and BrambleMet will also be used to measure the local weather conditions and wave pro le. The experimental procedure was: 1. The internal pressures were set to 3 and 3.25 psi in the sponson and keel, respectively, in both boats. 2. The boats were lined up in the direction of the waves (either head or following seas depending on test condition) between two and ve boat lengths apart. 3. The data acquisition systems were started. 4. Five seconds later both boats accelerated together to an approximate speed of 15 knots. 5. Both boats continued alongside each other for three minutes. 6. The boats turned around and steps three to ve were repeated for the other wave direction. 7. Steps two to six were repeated three times to capture nine minutes in each direction. 8. The internal pressures of the test vessel were increased to 4 and 4.25 psi in the sponson and keel, respectively. 9. Steps two to seven were then repeated with both boats at dierent internal pressures. 10. Finally, steps one to nine were repeated where the internal pressures of the tests boat were dropped to 2 and 2.25 psi. The body weight of each crew member was measured once during the wave trials, shown in table 3.4. All of the variable boat's crew were the same both days and totalled 270 kg. The helmsman of the reference boat was the same during the two trials but the two crew members changed; the total mass was 244 kg and 238 kg on day 1 and day 2, respectively. The main dierence in the crew's mass is in between the two helmsmen, totalling 25 kg.
3.5.
FULL-SCALE TEST METHOD
67
Reference boat
Variable boat
Helmsman
68
93
Port crew
101
90
Starboard crew
75
87
Total
244
270
(a) Day 1 comparing 3 psi and 4 psi. Reference boat
Variable boat
Helmsman
68
93
Port crew
89
90
Starboard crew
81
87
Total
238
270
(b) Day 2 comparing 2 psi and 3 psi. Table 3.4: Mass of crew members during wave trials (kg).
3.5.7 Calibration Sensor
Trueness
±
Accelerometers
−6 5.6 x 10
Rate gyroscopes
±
Strain gauges String potentiometer Pressure transducer
2 x
10−4
±1 ± 0.1
GPS speed Drop height Trim angle
±1 ± 0.25
Precision
± ± ± ± ±
−6 5.6 x 10 −3 6.6 x 10 2.3 x 2.6 x 9.5 x
10−8 10−3 10−4
± 0.1 ± 10 ± 0.25
Unit g
°/s mm psi knots mm
°
Table 3.5: Uncertainty of sensors.
All the sensors required calibrating to ensure that the measured values were accurate and precise. The majority of the sensors were brand new when they were tted to the D-class; therefore, the manufacturer's calibration values could be used. The manufacturers calibrated the linear scaling coe cients and the zero values at a known voltage (either 12 V or 5 V) so the same known voltage was supplied to the sensors during the experiment. This ensures that the sensors are calibrated but this does not ensure the data acquisition system was accurate. The data acquisition system was checked by placing a constant known voltage across the acquisition system and comparing it to a digital voltage meter (DVM). There was a 0.01 V dierence between the DVM and the data acquisition system. To further ensure that the sensors were accurate, they were checked against other known measurement methods.
The string potentiometers were checked with a ruler to ensure the
measured value was within
±
1 mm. The internal pressure transducers were checked with a
pressure gauge (supplied by the RNLI) and the pressure transducers were within the accuracy of the pressure gauge (± 0.1 psi).
The strain gauges were checked using a tensile testing
machine, the same machine used during the fabric material properties experiment (Instron
68
CHAPTER 3.
HYDROELASTIC METHODOLOGY
E). The strain gauge was bonded to a known composite coupon and extended in the tensile testing machine at quasi-static speeds. The strain gauge readings were within 4 % of the strain reading from the tensile testing machine. The strain gauges on the outboard engine required calibration in a dierent manner because the strain needed to be related to a force on the outboard.
This could have been achieved
by tting the engine horizontally and placing a known mass on the propeller pin; however, there was di culties in bonding the strain gauges to the trim pins.
The water jet coming
o the outboard engine strut constantly impacted the gauges and de-bonded them. Dierent bonding methods were tried but this meant the position of the strain gauge had changed and the original calibration values were wrong. During each of the sub-experiments at least one strain gauge become de-bonded so there is never a full data set and the calibration was not completed for all the strain gauge positions. Nevertheless, the strain on the trim pin can still be used for comparison such as during the at water trials. The strain gauges on the hinge bridges also required calibration between the strain on the bridge and the angle of the hinge. The rst attempt to measure the hinge angle was with an engineer protractor but the shape of the hinge brackets and the components of the D-class (primarily the equipment pod) obstructed the protractor so another method was found. The second method simply involved placing digital bevels (spirit level) on each deck panel. The dierence between the bevels was the hinge angle. The bevels were manufacturer by JingYan, model AS[13/30/60]-[L/LC], and were calibrated at the start of the tests and checked at the end. The calibration of the hinges showed that the aft hinge was more complex than expected. The hinge is formed by gluing and screwing several layers of machine belt across the joint, see gure 2.6. This means that the joint can deform by rotation and shear movement, and this lead to a large scattering in the calibration values. The middle hinge was shown to respond in a uniform, linear manner as anticipated. In the rst calibration experiment no angle change was measured in bow hinge and the strain did not change either; however, later results showed that the strain gauge was not bonded correctly. Unfortunately this means that the bow hinge data is not available and the results assume a two degree hinge angle because that was the mean angle measured during the calibration. The calibration charts for the aft and middle hinge can be seen in gure 3.10.
3.5.8 Post Processing The data was saved to the SD card module on the cRIO in binary format. The data manipulation that was required to convert the binary data into SI units will now be described. The
2
physical quantities that were measured include: acceleration (m/s ), pressure (psi), distance (m), strain and angle (degree). The binary data from the accelerometer module (NI 9234) was converted into voltage using equation 3.2, taken from the NI 9234 operating instructions and speci cations. The mean of one second was used to zero the voltage data. The voltage was converted into acceleration using equation 3.3. The scale factors were taken from the manufacturers calibration valves on the data sheets; seperate valves for each axis of measurement.
3.5.
FULL-SCALE TEST METHOD
69
V oltage = (Binary × 5.2)/8388608
(3.2)
Acceleration = V oltage/ScaleF actor
(3.3)
The binary data from the voltage module (NI 9205) was converted using equation 3.4, taken from NI 9205 operating instructions and speci cations. The voltage measurements required o-setting and the zeroing valves were measured at the during the calibration stage of the experiment. The string potentiometer voltages were converted to meters using equation 3.5, taken from the string potentiometer data sheet. The supply voltage to string potentiometers was measured simultaneously and was used throughout the post processing to ensure any variation in the supply voltage did not aect the measurement accuracy. The voltages measured by the pressure sensors was converted into pounds per square inch using equation 3.6, taken from the pressure sensor data sheet.
V oltage = (Binary × 328)/106
(3.4)
Distance = (1000 × V oltage)/(Sensitivity × V oltageSupply )
(3.5)
P ressure = V oltage × Sensitivity
(3.6)
The manipulation of the rate gyroscope data was more complex and less accurate due to the angular drift with time and the use of integration. The voltage was rst converted into angular velocity using equation 3.7, taken from the rate gyroscope data sheet. Next, the drift with time was removed using a rolling average over one second and, nally, the angular velocity was integrated using cumulative trapezoidal rule to nd the angular displacement.
AngularV elocity = 20 × V oltage
(3.7)
The binary data from the strain module (NI 9236) was converted into voltage using equation 3.8, taken from NI 9236 operating instructions and speci cations. The standard quarterbridge Wheatstone circuit equation was used and is shown in equation 3.9 and 3.10, taken from NI White Paper 4172 (Strain Gauge Con guration Types). The excitation voltage (VEx ) was 3.3 V, gauge factor (GF ) was 2, gauge resistance (RG ) was 350
Ω and the lead wire resistance
(RW ) was measured during the calibration stage using an ohmmeter with an accuracy of
±
1
mΩ.
V oltage = (Binary × 0.09702)/8388607
(3.8)
V r = (VStrained − VU nstrained )/VEx
(3.9)
70
CHAPTER 3.
Strain =
HYDROELASTIC METHODOLOGY
−4V r RW ) × (1 + GF (1 + 2V r) RG
(3.10)
The time signals were transformed into frequency signals using the same fast Fourier's transformation describe in section 3.4.4. The time signals were ltered with the same Butterworth lter discussed in section 3.4.4.
FULL-SCALE TEST METHOD
71
160000
Voltage (µV)
140000 120000 100000
y = 64643x - 9017.5
80000 60000 40000
20000 0
1
1.2
1.4
1.6
1.8
2
2.2
2.4
2.6
Angle (deg)
(a) Aft hinge (B1)
450000 400000
Voltage (µV)
3.5.
350000
y = 60205x + 74315
300000 250000 200000 150000 100000
1
2
3
4
5
Angle (deg)
(b) Middle hinge (B3) Figure 3.10: Hinge angle calibration charts.
6
72
CHAPTER 3.
HYDROELASTIC METHODOLOGY
Chapter 4
Hydroelastic results under vertical loads 4.1 Static tests The static tests were performed on the
29th
of November 2012 in the ILC marina, Cowes.
4.1.1 Eect of internal pressures with no crew on structural deformation The static tests showed how the structure deformed due to only the buoyancy and weight forces of the D-class.
The investigation will start with eects of the internal pressures of
the sponson and keel on the structural deformation while there are no crew on board. The measured internal pressures of the sponson and keel are shown in table 4.1. There were issues with air leaks in the sponson and keel so there are large dierences between the intended internal pressures and the test pressures but there was at least 0.7 psi dierence between the testing conditions. Condition
1
2
3
4
5
6
7
C0 S2.25 K2
2.22
2.17
2.19
2.19
2.25
2.20
2.23
C0 S3.25 K3
3.10
3.07
3.10
3.09
3.15
3.10
3.14
C0 S4.25 K4
3.64
3.68
3.71
3.99
4.04
3.80
3.81
Average 2.207 3.108 3.809
σ 0.0296 0.0286 0.152
(a) Sponson pressure (psi). Condition
1
2
3
C0 S2.25 K2
1.85
1.93
1.96
C0 S3.25 K3
2.87
2.96
2.99
C0 S4.25 K4
4.00
4.08
4.10
Average 1.913 2.937 4.061
σ 0.0566 0.0597 0.0517
(b) Keel pressure (psi). Table 4.1: Internal pressures of the sponson and keel during static tests with no crew.
The sponson rotation due to increasing the internal pressures is shown in table 4.2 and it is clear that the sponson rotation is very small (< 1 mm). It is interesting to note that the sponson initial rotated outwards due to increasing the internal pressures from 2 to 3 psi but 73
74
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
when the pressures were increased from 3 to 4 psi the rotation was inwards. This indicates that there is a non-linear structural deformation, although the deformation is almost negligible at less than one millimetre.
Condition
SR1 - aft (mm) Mean
SR2 - fore (mm)
Dierence
Mean
Dierence
C0 C2.25 K2
66.0
-
65.2
-
C0 S3.25 K3
66.5
-0.4
65.6
-0.4
C0 S4.25 K4
65.7
0.8
64.7
0.9
Table 4.2: Static sponson rotation due to varying the internal pressures with no crew.
The calibration of the deck hinge angles was di cult; so were the static measurements of the hinge angles. Occasionally the aft deck hinge (B1) was not measured correctly, either by the strain gauge or the acquisition system, and the data sets are not complete. The middle deck hinge angle (B3) was measured correctly but without the aft hinge measurement the location of the middle deck hinge could not be found. When this occurred, the aft deck hinge angle was estimated using a linear t and it is represented using a dashed line.
The static
deck hinge angles due to varying the internal pressures with no crew can be seen in gure 4.1 and the red dashed line indicates that the aft hinge angle was estimated using a linear t. The results show that as the internal pressures increase the aft deck hinge decreases but the middle hinge increases. The maximum hinge angle change was 0.58°.
Figure 4.1: Static deck hinge angles due to varying the internal pressures with no crew.
The strains in the deck caused by increasing the internal pressures are shown in gure 4.2. It shows that there is a slight variation in the x-axis strain but the y-axis strain is greater. It reveals that, in the y-axis, the top of all the deck panels are in compression but, in the x-axis, the top of the aft two panels are in tension and the front two are in compression.
It was
originally thought that over in ating the keel would force the centre of the deck upwards and the edge of the deck downwards, resulting in the deck curving around the keel and positive
4.1.
STATIC TESTS
75
Figure 4.2: Static deck micro-strains due to varying the internal pressures from S2.25 K2 to S4.25 K4 with no crew.
strain on the top of the deck; however, the results do not agree with this. They instead suggest the dominant y-axis strains are caused by the tension in the hull which compresses the deck. The hull shapes and the change in the hull shapes are shown gure 4.3 and the sub- gures accumulate so that (a) + (b) = (c) and (c) + (d) = (e). The hull shape is upside down and the origin of the graph is the aft port corner of the deck.
The location of the hull sensors
can be seen in gure 3.8. The largest z-axis deformation is approximately 1800 mm along the x-axis which lines up with the aft hinge between the two aft deck panels, which is 1840 mm from the transom. The two troughs either side are 1650 mm and 2200mm along the x-axis. The z-axis displacements are greater at 4 psi than at 2 psi and this was expected because the keel diameter had expanded. The change in hull shape is greater in between 2 and 3 psi than 3 and 4 psi.
4.1.2 Eect of internal pressures with three crew on structural deformation The eect of varying the internal pressures on the structural deformation with three crew will now be discussed. The internal pressures of the sponson and keel are shown in table 4.3. There were again issues with air leaks in the sponson and keel but there was at least a 0.8 psi dierence between each condition.
The sponson rotation due to increasing the internal
pressures are shown in table 4.4 and the sponson rotation is very small (< 1 mm). The static hinge angles due to increasing the internal pressures with three crew are shown in gure 4.4.
Much of this data is missing due to the problems with the aft deck hinge
measurements and only the 2 psi condition is valid. The strains in the deck caused by increasing the internal pressures with three crew are shown in gure 4.5. The strain pattern caused by increasing the internal pressure with three crew is far more complex than with no crew. It shows that the top of panels 1 and 3 are in compression in both the x-axis and y-axis, whereas the top of panel 2 is in tension. This shows that there is a complex structural interaction with three crew and is most likely caused by the asymmetric loading of the crew. .
76
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
(a) Hull shape: C0 S2.25 K2.
(b) Change in hull shape = C0 S3.25 K3 - C0 S2.25 K2.
(c) Hull shape: C0 S3.25 K3.
(d) Change in hull shape = C0 S4.25 K4 - C0 S3.25 K3.
(e) Hull shape: C0 S4.25 K4. Figure 4.3: Static hull shapes due to varying the internal pressures with no crew.
4.1.
STATIC TESTS
77
Condition
1
2
3
4
5
6
7
C3 S2.25 K2
2.19
2.14
2.17
2.17
2.24
2.18
2.22
C3 S3.25 K3
2.99
2.95
2.97
2.97
3.04
2.98
3.03
C3 S4.25 K4
3.57
3.60
3.63
3.96
3.99
3.75
3.77
Average 2.187 2.991 3.753
σ 0.0321 0.0332 0.171
(a) Sponson pressure (psi). Condition
1
2
3
C3 S2.25 K2
2.05
2.13
2.16
C3 S3.25 K3
2.99
3.07
3.10
C3 S4.25 K4
4.09
4.17
4.20
Average 2.110 3.057 4.155
σ 0.0575 0.0562 0.0550
(b) Keel pressure (psi). Table 4.3: Internal pressures of the sponson and keel during static tests with three crew.
Condition
SR1 - aft (mm) Mean
SR2 - fore (mm)
Dierence
Mean
Dierence
C3 C2.25 K2
65.9
-
65.1
-
C3 S3.25 K3
66.3
-0.4
65.4
-0.3
C3 S4.25 K4
65.6
0.8
64.6
0.8
Table 4.4: Static sponson rotation due to varying the internal pressures with three crew.
Figure 4.4: Static hinge angles due to varying the internal pressures with three crew.
78
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
Figure 4.5: Static deck micro-strains due to varying the internal pressures from S2.25 K2 to S4.25 K4 with three crew.
The eect of the internal pressures on the hull shape, with three crew, is very similar to the change with no crew, see gure 4.6. This contradicts the deck strain results, which were shown to be very dierent to change with no crew. This suggests that the deformation of the hull is not dependent on the deformation of the deck panels.
4.1.3 Eect of keel pressure on structural deformation The investigation will now consider how the structure deforms when only the keel pressure is varied and not the sponson pressure, with no crew. The internal pressures of the sponson and keel are shown in table 4.5. The keel pressures were consistently 1 psi apart; although, the sponson pressures did decrease by 0.18 psi due to leakages. The sponson rotation due to only changing the keel pressure is shown in table 4.6. The values are similar to the previous conditions and are very small (< 1 mm). Condition
1
2
3
4
5
6
7
C0 S3.25 K2
2.96
2.92
2.94
2.95
3.02
2.95
3.00
C0 S3.25 K3
3.10
3.07
3.10
3.09
3.15
3.10
3.18
C0 S3.25 K4
3.14
3.11
3.14
3.13
3.18
3.14
3.18
Average 2.962 3.108 3.144
σ 0.0341 0.0286 0.0276
(a) Sponson pressure (psi). Condition
1
2
3
C0 S3.25 K2
1.88
1.96
1.99
C0 S3.25 K3
2.87
2.96
2.99
C0 S3.25 K4
3.91
3.99
4.02
Average 1.944 2.937 3.972
σ 0.0584 0.0597 0.0564
(b) Keel pressure (psi). Table 4.5: Internal pressures of the sponson and keel during static tests with varying keel pressure.
The eect of changing only the keel pressure on the deck hinge angles can be seen in gure 4.7. The variation in hinge angles due to changing only the keel pressure has the same trend as changing both the sponson and keel pressures; increasing the pressure decreases the aft hinge angle but increases the middle hinge angle. This could be used to suggest the hinge angles are
4.1.
STATIC TESTS
79
(a) Hull shape: C3 S2.25 K2.
(b) Change in hull shape = C3 S3.25 K3 - C3 S2.25 K2.
(c) Hull shape under SOC: C3 S3.25 K3.
(d) Change in hull shape: = C3 S4.25 K4 - C3 S3.25 K3.
(e) Hull shape: C3 S4.25 K4. Figure 4.6: Static hull shapes due to varying the internal pressures with three crew.
80
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
Condition
SR1 - aft (mm) Mean
SR2 - fore (mm)
Dierence
Mean
Dierence
C0 C3.25 K2
65.5
-
64.7
-
C0 S3.25 K3
66.5
- 1.0
65.6
- 0.9
C0 S3.25 K4
66.0
0.5
65.0
0.5
Table 4.6: Static sponson rotation due to varying only the keel pressure with no crew.
Figure 4.7: Static deck hinge angles due to varying only the keel pressures with no crew.
more dependent on the keel pressure that the sponson pressure. The hinge angles are almost identical when the pressure is either S4.25 K4 or S3.25 K4 but there is a greater dierence at lower pressures (i.e. S2.25 K2 or S3.25 K2). This implies that at higher pressures the keel has a dominant eect but as the pressure reduces the sponson becomes more important. The deck strains caused by increasing only the keel pressure are shown in gure 4.8. The pattern is very similar to the pattern caused by varying both the sponson and keel pressures. The hull shapes and the change in hull shapes can be seen in gure 4.9 and again these results are very similar to the shapes caused by varying both the sponson and keel pressures at the same time. This proves that the keel pressure has a more dominant eect on the hull shape and deck strains than the sponson pressure.
Figure 4.8: Static deck micro-strains due to varying only the keel pressure from S3.25 K2 to S3.25 K4 with no crew.
4.1.
STATIC TESTS
81
4.1.4 Eect of crew loading on structural deformation The internal pressures when the crew loading increases from none to three are shown in table 4.7. The internal pressures were always set with one person on board (during the static tests) and not varied when the load was changed. It is expected that as the crew load increases then the internal pressures would increase, especially in the keel, but this is not shown in table 4.7. It is misleading because of the air leaks in the sponson and keel. The no crew condition was measured rst, then there was a wait while the crew assembled (+ 30 minutes). In this time the internal pressures decreased, due to the leakages, to less than 3 psi. When the crew boarded the D-class they caused the pressure to increase back up to 3 psi giving the results a false appearance.
The sponson rotation is shown in table 4.8 and conforms to the other
sponson rotation results. Condition
1
2
3
4
5
6
7
C0 S3.25 K3
3.10
3.07
3.10
3.09
3.15
3.10
3.14
C2 S3.25 K3
2.99
2.95
2.97
2.97
3.04
2.99
3.03
C3 S3.25 K3
2.99
2.95
2.97
2.97
3.04
2.98
3.03
Average 3.108 2.992 2.991
σ 0.0286 0.0334 0.0332
(a) Sponson pressure (psi). Condition
1
2
3
C0 S3.25 K3
2.87
2.96
2.99
C2 S3.25 K3
2.92
3.00
3.03
C3 S3.25 K3
2.99
3.07
3.10
Average 2.937 2.984 3.057
σ 0.0597 0.0579 0.0562
(b) Keel pressure (psi). Table 4.7: Internal pressures of the sponson and keel during static tests with varying crew loading.
Condition
SR1 - aft (mm) Mean
SR2 - fore (mm)
Dierence
Mean
Dierence
C0 S3.25 K3
66.5
-
65.6
-
C2 S3.25 K3
67.0
- 0.5
66.1
- 0.5
C3 S3.25 K3
66.3
0.7
65.4
0.7
Table 4.8: Static sponson rotation due to varying the crew loading.
The eect of crew loading on the deck hinge angles can be seen in gure 4.10. It shows that the hinge angles do vary with crew loading but the direction of variation changes depending on the internal pressures.
For example, increasing the crew loading at 2 psi causes the aft
hinge angle to decrease but at 4 psi the aft hinge angle increases. This emphasises the complex structural interactions found within the D-class. The changes in the deck strains when three crew are added are shown in gure 4.11. It shows that when the D-class is loose and exible (i.e. S2.25 K2) the strains are low and the crew loading does not increase the strains signi cantly. On the other hand, when the D-class is under pre-tensioned stresses and strains due to over in ating the keel (i.e. S4.25 K4) the increase in strain due to crew loading does cause larger, more signi cant variations in strain.
82
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
(a) Hull shape: C0 S3.25 K2.
(b) Change in hull shape = C0 S3.25 K3 - C0 S3.25 K2.
(c) Hull shape: C0 S3.25 K3.
(d) Change in hull shape = C0 S3.25 K4 - C0 S3.25 K3.
(e) Hull shape: C0 S3.25 K4. Figure 4.9: Static hull shapes due to varying only the keel pressure with no crew.
4.1.
STATIC TESTS
83
(a) at S2.25 K2
(b) at S4.25 K4
Figure 4.10: Static deck hinge angles due to varying the crew loading.
(a) at S2.25 K2
(b) at S4.25 K4 Figure 4.11: Static deck micro-strains due to varying the crew loading.
84
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
The hull shape and change in hull shape due to crew loading can be seen in gure 4.12. The no crew condition has a larger z-axis displacement than the three crew condition, which is a result of increased buoyancy and weight forces compressing the keel and hull. Figure 4.12b suggests that the eect of removing one crew has minimal eect on the hull shape; however, the third crew member was removed from the other side of the vessel and the sensors only capture the deformation on one side. The eect of removing the helm can be seen by the large aft deformation in gure 4.12d.
4.2 Quasi-2D drop tests The quasi-2D drop tests were performed in July 2011 at the Chilworth Laboratory, University of Southampton.
4.2.1 Accelerations The frequency spectra from these results were very similar to the previous results obtained using the same test rig (i.e. Lewis et al. (2010)); therefore, the same 250 Hz low pass Butterworth lter was used to remove the unwanted frequencies around 600 Hz. The origin of this 600 Hz peak is unknown. Figure 4.13 shows an example of the spectrum once the data has been passed through a 250 Hz low pass lter. Four examples of the acceleration-time histories of the three stiness conditions (MDF, 1000 N/m, 0 N/m) at the two deadrise angles and two drop heights can be seen in gures 4.14, 4.15, 4.16 and 4.17. Graph (a) shows the acceleration-time history when the data has been processed with a 250 Hz low pass Butterworth lter. Graph (b) shows the same data once it has been passed through a 40 Hz low pass Butterworth lter to smooth the graph purely for viewing purposes. All data has been analysed with a 250 Hz low pass Butterworth lter. Figure 4.15 is the clearest example and the initial impact with the water can be seen after 0.32 s with a peak of nearly 14 g. The accelerations from the MDF condition are very similar to a regular under-damped sinusoidal system, except the second peak is smaller than the third.
This irregularity in the second peak is found in all the hull stiness conditions;
however, the oscillations of the fabric conditions do not represent a regular sinusoidal system. This is clearly shown in the 0 N/m condition where the fourth and sixth peaks are smaller than the fth and seventh. This alternating peak height would imply that there are two outof-phase sinusoidal oscillations occurring but this alternating peak height is not repeated in the other drop conditions. Observations from all the drop conditions show that the typical characteristics are:
The peak accelerations do vary due to the hull stiness and the pre-tensioned stress.
The peak durations do vary due to the hull stiness and the pre-tensioned stress.
The impacts nearly represent an under-damped sinusoidal oscillation but the fabric hull does not undergo constant damping.
4.2.
QUASI-2D DROP TESTS
85
(a) Hull shape under SOC: C3 S3.25 K3.
(b) Change in hull shape: C2 S3.25 K3 - C3 S3.25 K3.
(c) Hull shape: C2 S3.25 K3.
(d) Change in hull shape = C0 S3.25 K3 - C2 S3.25 K3.
(e) Hull shape: C0 S3.25 K3. Figure 4.12: Static hull shapes due to varying the crew loading at S3.35 K3.
86
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
Figure 4.13: Example frequency spectrum of the quasi-2D drop test with a 250 Hz low pass lter (deadrise angle = 5°, drop height = 1 m).
During the rst ve peaks, the fabric hull normally has an higher damping ratio than the MDF hull; however, occasionally the damping ratio in one direction (positive or negative acceleration) of the fabric hull can be equal to that of the MDF hull.
After the rst ve peaks and once the accelerations have reduced dramatically (below 10 %), 50 % of the tests showed that the MDF hull still had a higher damping ratio but in 50 % of the tests the fabric hull had a higher damping ratio.
The mean peak accelerations are shown in gure 4.18.
At 0.5 m drop height it can be
seen that as the structure becomes more exible the mean peak acceleration decreases. These results show that the peak acceleration can be decrease by 18.1 % and 12.4 % at 5° and 15° deadrise angles, respectively (with a drop height of 0.5 m).
On the other hand, this trend
is inverted with a drop height of one metre and a 15° deadrise angle; the peak acceleration increases by 15.1 % with decreasing structural stiness. It is unclear what eect structural stiness has on the peak acceleration when there is a one metre drop height and a ve degree deadrise angle because the MDF hull is unexpectedly high compared to all the other conditions. The fabric hull conditions t the same trend as the other one metre drops. The measured dierence between the mean peak accelerations could be due to uncertainties in the experimental set up and not statistical dierences due to the hull stiness; therefore, the null hypothesis was assumed. To test the null hypothesis a two tailed Student's T-test was performed and the results can be seen in table 4.9. It shows that 5/6 of the drops have a statistical dierence when the deadrise angle is 5° but the hull hypothesis was valid for 4/6 of the drops when the deadrise angle was 15°. Firstly, this implies that hydroelasticity has a greater impact when the deadrise angle is low because there is more statistical dierence at 5° than 15°.
Secondly, it suggests that hydroelasticity has a greater impact when the
4.2.
QUASI-2D DROP TESTS
87
(a) 250 Hz low pass lter.
(b) 40 Hz low pass lter. Figure 4.14: Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 15°, drop height = 0.5 m.
88
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
(a) 250 Hz low pass lter.
(b) 40 Hz low pass lter. Figure 4.15: Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 5°, drop height = 0.5 m.
4.2.
QUASI-2D DROP TESTS
89
(a) 250 Hz low pass lter.
(b) 40 Hz low pass lter. Figure 4.16: Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 15°, drop height = 1 m.
90
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
(a) 250 Hz low pass lter.
(b) 40 Hz low pass lter. Figure 4.17: Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 5°, drop height = 1 m.
4.2.
QUASI-2D DROP TESTS
91
40
35
Peak acceleration (g)
30
25 5° 0.5 m
20
5° 1m 15° 0.5m
15
15° 1m
10
5
0 Rigid
1000 N/m
0 N/m
Hull stiffness condition
Figure 4.18: Mean peak accelerations measured during the 2D drop tests.
drop height is larger because, at 15° deadrise angle, there was no statistical dierence at 0.5 m drop height but at 1 m there was a statistical dierence between 2/3 of the drops. Both of these observations agree with the work of Faltinsen (1999) where the importance of hydroelasticity increases by either decreasing the deadrise angle or increasing the impact velocity, see equation 2.3. Finally, there is always a statistical dierence between the MDF and 0 N/m stiness conditions in all drop conditions with a 90 % certainty meaning that hydroelasticity has de nitely aected the peak acceleration to a 90 % certainty. The peak duration was investigated because it was anticipated that the duration would vary with the internal pressures, similar to the peak acceleration, because each drop height has xed kinetic energy at the point of impact. The integral of the acceleration-time history is proportional to the kinetic energy (Ekinetic
= 21 mv 2 );
thus as the peak acceleration varies,
the peak duration should inversely vary to hold equilibrium. However, this does not consider the damping within the motion.
The problem can simpli ed to damped harmonic motion
for demonstration and the equation for damped harmonic motion is shown in equation 4.1, see Lloyd (1998) page 133. The only variables that the internal pressures could have aected are the damping or spring constants (the aect on mass is negligible). It is probable that the internal pressures aected both the damping and spring constants but it was still expected that the peak duration would have varied inversely (to some degree) with the peak acceleration.
92
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
Table 4.9: Student's T-test of the peak accelerations measured during the 2D drop tests.
m¨ x + cx˙ + kx = 0 This is not the case.
(4.1)
The general trend shows that the MDF conditions had a longer
duration than the fabric conditions; however, the change in duration due to the pre-tension stress in the fabric hull is not consistent. The percentage increase in peak duration from the MDF conditions to the fabric conditions ranges from 7.5 % to 46.1 %. The peak duration is taken as the rst intersection with the x-axis (0 g).
4.2.2 Structural deformation The high speed camera images of the impact from the side through the observation window can be seen in gure 4.19. The wedge is moving down the image and the fabric can be seen wrapped around the wedge. This impact had a deadrise angle of 15°, a drop height of 1 m and a pre-tension stress of 0 N/m. The rst image shows the moment of impact. The second image shows that the fabric behind the wetted edge is pulled tight and ahead of the wetted edge the hull is deformed. The third image shows that the fabric deformation moves with the wetted edge and continues until the wetted edge is outside the window of view.
The forth
image shows an interesting observation that occurs near the end of impact which is a set of ripples in the fabric.
This could be caused by high out-of-plane forces in the centre of the
panel or a re ected structural vibration. In a trial test when the fabric tension was very loose, three of these ripples were observed and in higher tension tests the ripples were occasionally unnoticeable.
4.2.3 Error analysis The MDF stiness condition was chosen to represent an approximately rigid condition but a six millimetre sheet of MDF cannot be assumed rigid during these drop tests. The Young's modulus of MDF is approximately 4 GPa whereas the Young's modulus of hull fabric is 0.28
4.2.
QUASI-2D DROP TESTS
93
Figure 4.19: Transverse images of 2D drop test with a fabric hull under 0 N/m (15 degrees deadrise angle, 1 m drop height).
GPa so it is clear that an MDF panel is considerably more rigid that the fabric hull. The boundary conditions between the MDF and fabric hull were slightly dierent; the fabric hull had a pin joint and the MDF hull had a fully clamped condition. This will aect the panel de ection but the fully clamped boundary conditions of the MDF hull will reduce the panel de ection because there is no rotational de ection. This means the MDF hull can be considered to be more rigid than the fabric hull. Furthermore, a fabric is de ned because it has no out-of-plane bending stiness; therefore, a fully clamped boundary condition will not remove rotational de ection of a fabric. So this change in boundary condition should not aect the response of the fabric. When the wedge impacts the free surface it produces a wave and this wave could potentially be re ected by the end walls of the tank, which may interfere with the impact and the measured accelerations. The speed of a wave in shallow water can be calculated using is the speed of the wave,
g
is gravity and
the speed of the propagating wave is
d
c2 = gd;
where
c
is the water depth. The water depth was 0.5 m so
2.214m.s−1 .
The tank was 5.8 m wide so the wave would
take 2.62 s to re ect and collide with the impacting wedge. This shows that the re ected wave will not aect the measured accelerations. The acceleration-time history graphs of the drop tests show that the free-fall stage is not smooth. This was caused by the old bearings and because the wedge clipped the edge of the tank during descent. The time history graphs do show that the irregularities in the free-fall
94
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
stage were consistent and the impacts were still repeatable.
4.3 Full-scale drop tests The full-scale drop tests were performed on the
6th
and
7th
of March 2013 in the RNLI marina
in Cowes, UK. The mean static internal pressures of the sponson and keel of the D-class can be seen in table 4.10 and it shows that internal pressures were up to 0.49 psi o the intended pressure and this was due to air leakages. Condition
1
2
3
4
5
6
7
S2.25 K2
2.19
2.16
2.18
2.16
2.22
2.19
2.21
S3.25 K3
3.04
3.02
3.04
3.02
3.07
3.06
3.08
S4.25 K4
4.13
4.12
4.14
4.13
4.18
4.15
4.17
Average 2.19 3.05 4.15
σ
Error
0.02
0.06
0.03
0.20
0.02
0.10
(a) Sponson pressure (psi). Condition
1
2
3
S2.25 K2
1.42
1.55
1.55
S3.25 K3
2.89
3.02
30.2
S4.25 K4
3.73
3.85
3.85
Average 1.51 2.97 3.80
σ
Error
0.08
0.49
0.07
0.03
0.08
0.20
(b) Keel pressure (psi). Table 4.10: Mean static internal pressures during the drop tests.
4.3.1 Accelerations This analysis will start by investigating how the internal pressures of the sponson and keel aected the accelerations at the transom, crew position and bow of the D-class. The accelerometers used during the full-scale drop tests did not require any ltering because they had a at frequency response up to 200 Hz and -6 dB above 200 Hz.
The frequency spectrum
showed very few vibrations (< 1 % of peak power) above 200 Hz. The frequency domain and frequency shifts will be discussed in more detail later. The acceleration-time histories of the full-scale drop test from accelerometers in the crew position and on the bow, at both 0.5 and 1 m drop height, can be seen in gures 4.20, 4.21, 4.22 and 4.23. All the graphs show the free-fall stage followed by the major peak acceleration. The overall impacts do follow an under-damped sinusoidal system but there are irregularities causing an inconsistent damping ratio. The typical characteristics of the time history graphs show that:
Changing the internal pressures does aect the peak accelerations at the crew positions and at the bow; although, the aect at the transom was negligibly.
Changing the internal pressures does aect the peak durations; however, there is no consistent trend.
The impacts nearly represent an under-damped sinusoidal oscillations but the damping ratio is not constant.
4.3.
FULL-SCALE DROP TESTS
95
(a) No lter.
(b) 40 Hz low pass lter. Figure 4.20:
Examples of the acceleration-time history from the crew position during the
full-scale drop tests with a drop height of 0.5 m.
96
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
(a) No lter.
(b) 40 Hz low pass lter. Figure 4.21:
Examples of the acceleration-time history from the crew position during the
full-scale drop tests with a drop height of 1 m.
4.3.
FULL-SCALE DROP TESTS
97
(a) No lter.
(b) 40 Hz low pass lter. Figure 4.22: Examples of the acceleration-time history from the bow during the full-scale drop tests with a drop height of 0.5 m.
98
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
(a) No lter.
(b) 40 Hz low pass lter. Figure 4.23: Examples of the acceleration-time history from the bow during the full-scale drop tests with a drop height of 1 m.
4.3.
FULL-SCALE DROP TESTS
99
Mean peak acceleration during the full-scale drop tests 18
Peak acceleration (g)
16
14
Transom 0.5m Transom 1m
Crew 0.5m
12
Crew 1m Bow0.5m
10
Bow 1m
8
6 2 psi
3 psi
4psi
Pressure (psi)
Figure 4.24: Mean peak accelerations measured during the full-scale drop tests.
The mean peak accelerations from all three accelerometers during the full-scale drop tests can be seen in gure 4.24. The peak accelerations measured at the transom of the vessel reveal that the internal pressures had a minimal eect on the accelerations but this was expected because the structure at the transom is considerably stier than the rest of the boat.
The
in atable keel does not extend right up to the transom, which removes the eect of the keel pressure all together. The vertical transom panel will dramatically reduce sponson rotation and this should reduce the eect of the sponson pressure. The null hypothesis was assumed and tested with a two tail Student's T-test, see table 4.11. It showed that with a 95 % certainty there is no statistical dierence between the accelerations at the transom; however, with a 90 % certainty there is some statistical dierence. This suggests that the sponson has a small eect on the peak acceleration but not as signi cant when compared with further forward in the boat. The accelerations measured in the crew position do show a trend of decreasing or increasing due to changes in the internal pressures. At 0.5 m drop height, as the internal pressures increase the peak acceleration also increases, which is the opposite trend to a one metre drop height. The Student's T-test (see table 4.11) con rms that there is a statistical dierence between 5/6 of the drop conditions to a certainty of 95%. Decreasing the internal pressures from 4 psi to 2 psi can lead to a 22.6 % decrease in the peak accelerations experienced by the crew at 1 m drop height; however, at a 0.5 m drop height the peak accelerations can increase by 39.0 %.
100
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
Table 4.11: Student's T-test of the peak accelerations measured during the full-scale drop tests.
The trends in the acceleration at the bow of the D-class are the same as the trends found in the crew position and that is; at 0.5m drop height, an increase in internal pressure causes a decrease in peak acceleration, whereas the trend is inverted at a one metre drop height. The Student's T-test (see table 4.11) again shows that 5/6 of the drop conditions are statistically dierent with a 95 % certainty. Increasing the internal pressures from 2 psi to 4 psi can lead to a 48.9 % decrease, at 0.5 m drop height, and a 8.6 % increase in the peak acceleration at a 1 m drop height. The peak duration was investigated again because it was anticipated that the duration would be proportional to the internal pressure, see section 4.2.1; however, there are less trends than the quasi-2D drops. statistically dierent.
In short, the peak duration results show no trends and few are
Interestingly, Lee and Wilson (2010) ( gure 9) also found that there
was minimal correlation between drop height and impact duration on the hydrodynamic impact of a racing sailboat; although, Lee et al. did nd a strong relationship between drop height and peak pressure.
This indicates that it is not hydroelasticity causing the irregular peak
durations but a hydrodynamic event.
The duration results measured during the full-scale
drop tests agree with the quasi-2D drop tests and the results of Lee and Wilson (2010).
4.3.2 Frequency spectra of acceleration The frequency domain will provide an insight into the accelerations and may help to reveal why the peak accelerations and peak durations change due to hydroelasticity. For the time being the full-scale tests will be discussed because the structural stiness was increased in a linear manner, whereas the stiness of the quasi-2D tests jumped considerably from the fabric
4.3.
FULL-SCALE DROP TESTS
101
hulls to the MDF hull. The mean frequency spectrum for all the full-scale tests can be seen in gure 4.25.
Graphs (a) and (b), from the transom accelerometer, show a clear peak at
1.7 Hz to 2.1 Hz and the frequency of the peak increases with internal pressure and stiness. The rst mode of vibration for a free-free beam is heave and the natural frequency of heave is predicted to be 2.3 Hz using simple harmonic motion and 1.7 Hz to 1.9 Hz using the dry heave method, see Rawson and Tupper (2001) page 477. This indicates that these peaks are caused by heave. There is another set of peaks ranging from 5.3 Hz to 6.4 Hz and this could be the pitch motion. The natural frequency of the pitch motion is 4.1 Hz (at 2.5° static trim angle) predicted using the dry pitch method, see Rawson and Tupper (2001) page 478.. The errors in the predicted heave and pitch natural frequencies are great so it is not anticipated that they will exactly match the measured natural frequencies. The boat was dropped at a trim angle of 4.25° so the transom will impact rst causing the boat to rotate around the transom, resulting in larger heave motions and relatively smaller pitching motions at the transom. The accelerometer tted in the crew position, shown in the middle two graphs, both exhibit the same characteristic peaks at 1.7 Hz to 2.3 Hz and 4.5 Hz to 5.5 Hz for the heave and pitch motion, respectively. to increase.
The increase in internal pressure caused the frequency of the peaks
There is another signi cant peak at 0.6 Hz that is also visible in the transom
frequency spectra and it could be argued that this is heave motion; however, the bow spectra (see the bottom two graphs) do not show any sign of this peak. The bow will heave so this cannot be the heave frequency. The cause of this 0.6 Hz frequency is unclear. A comparison of the transom and crew spectra show that the dominant motion has changed; the transom experienced more heave motion and the crew experienced more pitching motion.
The bow
spectra show the same heave and pitch motions at 2.1 Hz to 2.7 Hz and 4.5 Hz to 5.6 Hz, respectively, and the bow experienced far more pitch motion than heave. There is an interesting super harmonic in the bow spectrum with a one metre drop height at two psi ( gure 4.25f) and it occurs every 2.16 Hz (ranging from 1.98 Hz to 2.29 Hz). This does not appear in any other full-scale drop test condition but was found in all ve repeats under this condition. The peak accelerations under these conditions appears to be higher than expected and do not t the trends of the other drop test conditions, see gure 4.24.
It is
hypothesised that the unexpectedly high peak acceleration is caused by this super harmonic. The quasi-2D drop tests also contained one test condition that repeatedly exhibited higher peak accelerations than expected, which was the MDF wedge with a ve degree deadrise angle at one metre drop height, see gure 4.26. A closer inspection of the frequency spectrum does also reveal super harmonic that occurs every 9.26 Hz (ranging from 8.55 Hz to 9.46 Hz). The magnitude is smaller and less frequent but it is still present.
4.3.3 Structural deformation The internal pressures aected the accelerations of the D-class and now the eect of the internal pressure on the deformation will be demonstrated. The mean peak internal pressures of the sponson and keel are shown in table 4.12. The increase in internal pressures shows that keel pressure increased considerably more than the sponson pressure; the sponson increased by a
102
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
(a) Transom, 0.5 m drop height.
(b) Transom, 1 m drop height.
(c) Crew, 0.5 m drop height.
(d) Crew, 1 m drop height
(e) Bow, 0.5 m drop height.
(f) Bow, 1 m drop height.
Figure 4.25: Mean power spectra of all the full-scale drop tests.
4.3.
FULL-SCALE DROP TESTS
103
Figure 4.26: Mean power spectrum of the quasi-2D drop test with a MDF hull, 1 m drop height and 5° deadrise angle.
maximum of 0.28 psi but the keel increased by up to 3.30 psi. This indicates that the keel has deformed more than the sponson during the impacts. The timing between the peak internal pressures at the various sensors was investigated to explore whether there was a pressure wave travelling through the in atable tubes. It was not possible to detect the instantaneous peak in sponson with enough accuracy to determine whether there was a pressure wave because the ratio between peak pressure and background pressure variation was too low. The ratio was high enough in the keel to nd the instantaneous peak but the peaks occurred at the same time in the three keel pressure sensors. Either the sampling frequency or the spacing between the sensors should be increased to investigate this further. The mean time and frequency domains of the sponson pressures can be seen in gure 4.27. The time history shows that the internal pressures have changed the sponson response, primarily the peak pressure and this is con rmed in the frequency domain because the peak power has increased, although the peak power is also at a slightly lower frequency. The mean time and frequency domains of the keel pressures are shown in gure 4.28 and they show a similar picture to the time and frequency domain of the sponson.
The internal pressures
increased the peak internal pressure and the peak power. The mean peak trim angles are shown in table 4.13 and it reveals that the peak trim angle is lowest at 3 psi. This implies a non-linear relationship. The mean time and frequency domains of the trim angle is shown in gure 4.29. The time histories show a steep increase in trim angle followed by a one second plateau before the trim angle rapidly decreases. The one second plateau is an interesting feature and was not expected; it can also be seen in the video in appendix H.1. The trim angle was measured on the transom. The transom had the engine mounted so this could cause the transom to rotate more than the real trim angle but this will be fairly constant through all the drop tests. Moreover, the trim angle only measures the angle at the transom and thus the aft deck panel as well; however, the bow of the boat could deform and change the trim angle at the bow without it being measured on the transom.
104
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
Sponson pressure (psi)
Condition
Mean
1
2
3
4
5
6
7
H0.5 S2.25 K2
2.38
2.33
2.34
2.32
2.41
2.38
2.41
H0.5 S3.25 K3
3.22
3.17
3.21
3.19
3.24
3.25
3.24
H0.5 S4.25 K4
4.25
4.21
4.24
4.22
4.28
4.25
4.28
Keel pressure (psi)
Condition
1
2
3
H0.5 S2.25 K2
4.19
4.28
4.29
H0.5 S3.25 K3
4.85
4.94
4.93
H0.5 S4.25 K4
5.82
5.90
5.92
Mean 4.25 4.91 5.88
2.37 3.22 4.25
σ
Increase
0.04
0.18
0.10
0.17
0.01
0.10
σ
Increase
0.45
2.74
0.42
1.94
0.07
2.08
(b) Half metre drop height. Sponson pressure (psi)
Condition
Mean
1
2
3
4
5
6
7
H1 S2.25 K2
2.42
2.36
2.39
2.38
2.46
2.42
2.45
H1 S3.25 K3
3.32
3.28
3.30
3.29
3.35
3.35
3.38
H1 S4.25 K4
4.29
4.25
4.29
4.29
4.36
4.30
4.32
Condition
Keel pressure (psi) 1
2
3
H1 S2.25 K2
4.72
7.84
4.86
H1 S3.25 K3
5.75
5.83
5.85
H1 S4.25 K4
6.19
6.27
3.29
Mean 4.81 5.81 6.25
2.41 3.33 4.30
σ
Increase
0.04
0.22
0.06
0.28
0.01
0.15
σ
Increase
0.42
3.30
0.23
2.84
0.17
2.45
(d) One metre drop height Table 4.12: Mean peak internal pressures during the drop tests.
Condition
Maximum
σinter
H0.5 S2.25 K2
10.28
1.83
H0.5 S3.25 K3
5.63
1.34
H0.5 S4.25 K4
7.14
2.02
H1 S2.25 K2
13.11
1.31
H1 S3.25 K3
5.15
0.98
H1 S4.25 K4
10.82
1.16
Table 4.13: Mean peak trim angle during the drop tests.
4.3.
FULL-SCALE DROP TESTS
105
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure 4.27: Mean time and frequency domain of the sponson pressure.
106
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure 4.28: Mean time and frequency domain of the keel pressure.
4.3.
FULL-SCALE DROP TESTS
107
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure 4.29: Mean time and frequency domain of the deck trim angle.
108
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
B1
Condition
B3
Mean
σinter
Mean
σinter
H0.5 S2.25 K2
-8.42
0.15
8.02
0.20
H0.5 S3.25 K3
-7.09
0.33
6.50
0.77
H0.5 S4.25 K4
-6.72
0.15
4.95
0.19
H1 S2.25 K2
-9.14
0.23
8.90
0.30
H1 S3.25 K3
-8.42
0.36
8.78
0.71
H1 S4.25 K4
-8.47
0.39
7.39
1.36
Table 4.14: Peak hinge angles during the drop tests.
The mean peak hinge angles are shown in table 4.14 and it shows that increasing the internal pressures reduces the peak hinge angles. This is to be expected because increasing the internal pressures should increase the stiness of the D-class and result in less deformation. The variation is greater at lower drop heights, which suggests a non-linear relationship. The time and frequency domains of aft (B1) and middle (B3) hinge angles are shown in gures 4.30 and 4.31, respectively. The aft hinge shows that increasing the internal pressures decreased the peak hinge angle and the duration was less; this is matched in the frequency domain where increasing the internal pressures decreased the power of the peak and increased the frequency of the peak. The aft deck hinge has one natural frequency around two hertz. The results for the middle hinge are similar to the aft hinge, expect for the 2 psi with one metre drop height conditions (H10 S2.25 K2), see gure 4.31b. This is the drop that contained a super harmonic and the frequency spectrum is considerably dierent. The super harmonic was only found in the bow so it would not be expected to aect the aft hinge angle. The middle hinge has three natural frequencies at approximately one, three and ve hertz. The mean peak X- and Y-axis micro-strains in the deck can be seen in tables 4.15 and 4.16, respectively. The largest micro-strain was 1872 measured on the aft deck panel (D2x) at 2 psi with a one metre drop height. This could lead to an extension of 3.4 mm over the 1.8 m long, aft deck panel. The maximum strain could be cause by a bending moment and not a compressive force. The surface strain can be related to the radius of curvature using
ε = x/rc ;
where
x
is distance from neutral axis and
rc
is radius of curvature, taken from Gere
and Goodno (2009). It assumes the natural axis is in the centre of the deck panel so
c
= 15
mm. This leads to a radius of curvature of 50 m. The central de ection of the panel can be calculated using the rules of chords
p 2 ; l = 2 rc2 − rcc
where
l
is length of chord and
rcc
is centre
of curvature to centre of chord. This would lead to a peak central de ection of the deck panel of 50.1 mm and this is noticeable; however, the largest change in deck strain due to varying the internal pressures was only 116 micro-strains (between the 2 psi and 4 psi drops with a one metre drop height on the aft deck panel (D2x)). 116 micro-strains could result in a 0.21 mm extension or a 3.1 mm central de ection of the aft deck panel. This means that varying the internal pressures has led to only a 3.1 mm change in the response of the aft deck panel and can be considered negligible; therefore, the mean time and frequency domains of the deck panels will not be explored further but the graphs are included in appendix I.
4.3.
FULL-SCALE DROP TESTS
109
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure 4.30: Mean time and frequency domain of the aft deck hinge angle (B1).
Condition H0.5 S2.25 K2
H0.5 S3.25 K3
H0.5 S4.25 K4
H1 S2.25 K2
H1 S3.25 K3
H1 S4.25 K4
Sensor number
D1x
D2x
D3x
D5x
D6x
D7x
(µ) µσinter Mean (µ) µσinter Mean (µ) µσinter Mean (µ) µσinter Mean (µ) µσinter Mean (µ) µσinter
916
1401
-779
-258
-531
414
24
31
34
8
12
11
955
1291
-750
-218
-480
386
9
38
42
27
36
41
919
1360
-688
-219
-481
434
6
12
12
5
16
5
1226
1872
-1056
-372
-759
669
19
22
16
9
33
41
1348
1572
-965
-420
-866
695
53
80
69
62
73
44
1220
1756
-935
-417
-851
721
55
112
30
50
52
46
Mean
Table 4.15: Peak X-axis micro-strains during the drop tests.
110
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure 4.31: Mean time and frequency domain of the middle deck hinge angle (B3).
Condition H0.5 S2.25 K2
H0.5 S3.25 K3
H0.5 S4.25 K4
H1 S2.25 K2
H1 S3.25 K3
H1 S4.25 K4
Sensor number
D1y
D2y
D3y
D5y
D6y
D7y
(µ) µσinter Mean (µ) µσinter Mean (µ) µσinter Mean (µ) µσinter Mean (µ) µσinter Mean (µ) µσinter
335
262
-
239
208
-
25
27
-
12
13
-
316
320
538
356
345
-
31
51
19
245
58
-
346
348
496
211
209
-
11
22
27
5
18
-
380
499
-
307
314
-
24
66
-
20
14
-
378
554
817
305
-
-
48
149
65
65
-
-
366
542
539
258
359
-
20
22
115
19
80
-
Mean
Table 4.16: Peak Y-axis micro-strains during the drop tests.
4.3.
FULL-SCALE DROP TESTS
Condition H0.5 S2.25 K2
H0.5 S3.25 K3
H0.5 S4.25 K4
H1 S2.25 K2
H1 S3.25 K3
H1 S4.25 K4
111
Sensor number
H9
H11
H14
H15
H17
Mean
49.5
50.3
21.5
91.3
97.0
σinter
3.6
2.6
2.7
6.2
4.1
Mean
62.2
65.5
21.3
102.9
108.3
σinter
1.7
4.0
0.5
2.0
2.4
Mean
69.9
72.9
21.6
110.1
114.5
σinter
0.6
1.0
0.3
0.7
0.9
Mean
44.5
46.1
22.6
83.9
88.7
σinter
3.6
3.2
2.5
5.2
4.9
Mean
56.1
61.5
26.2
88.3
94.3
σinter
3.5
6.6
3.0
1.0
0.5
Mean
62.7
63.0
20.8
96.5
102.1
σinter
2.0
2.3
0.2
1.6
2.8
Table 4.17: Mean minimum-peak hull de ection during the drop tests.
The mean minimum-peak hull de ection is shown in table 4.17. The hull sensors were not waterproof and the circuits would short out if they were submersed (this was expected) so not all of the sensors were still functioning at the end of all the drop conditions. The only ones to be fully functioning during all the drops were H9, H11, H14, H15 and H17 which are shown in table 4.17. These results show that increasing the internal pressures lead to a smaller minimum-peak de ection and less hull deformation, resulting in a 20 mm change in hull de ection.
The mean time and frequency domains of the hull de ection at H9 can be
seen in gure 4.32 and the other hull de ections (H11 to H17) can be found in appendix I. The time domain show that changing the internal pressures changes the peak de ection and the peak duration; however, the frequency domains show very little change due to the internal pressures.
This is unexpected because all the other components have been aected by the
internal pressures. This means the dynamic response of the hull (including the pre-tensioned stresses and the natural frequencies) has not been aected by the internal pressure of the keel; although, the internal pressure will aect the maximum compression of the keel centreline, which in turn aects the deadrise angle. The time and frequency domain of all the measured structural components during the full-scale drop tests have been presented and discussed. All the components show the same trend to increasing the internal pressures (and thus stiness) at both drop heights; the time domain shows a reduction in peak de ection, which is re ected in the frequency domain that shows a decrease in peak frequency. It was anticipated that peak de ections would follow the trends in peak acceleration because double integral of acceleration is displacement; however, this did not occur. Therefore, the trends in the peak de ection data do not help identify the component(s) that may cause the inversion with drop height in the acceleration data.
4.3.4 Error analysis It is inevitable that the full-scale drop tests would be less repeatable than the quasi-2D tests because the set up and environment were less controllable. The drop height was controlled
112
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure 4.32: Mean time and frequency domain of the hull sensor H9.
4.4.
HYDROELASTIC DISCUSSION PART 1
113
by a crane and, although it was surprisingly accurate, it was adjusted when the boat was ve metres away which made it virtually impossible to read one millimetre on the drop height gauge. The water surface was not perfectly smooth because it was outside where the wind and tide could generate small waves, regardless of the settling time. The boat was hung from a very long cable (> 20 m) so the system acted as a pendulum which would also vary the drop height and cause the boat roll angle to not be horizontal.
All of these variables were
controlled and made repeatable through the drop height gauge by ensuring that the variation of the distance between the transom and water's surface was less than
±
10 mm. The drops
were performed in a small mariner so the surface waves generated during the slam did re ect o a nearby wall; however, the waves took 10 seconds to re ect o the wall and impact the boat. The impact took less than 2 seconds so the re ected waves did not aect the test results. The acceleration-time history graphs from the transom and crew, see gures 4.20 and 4.21, show a fairly consistent drop of 1 g; however, the same graphs from bow accelerometer, see gure 4.22 and 4.23, do not exhibit the same characteristics. Instead the bow accelerometer measures very low accelerations until there is a small positive acceleration before the major negative peak.
This implies that the boat is rotating as it drops and that the trim angle
is larger at the point of contact than the trim angle at the time of release but the video recording does not show any signs of rotation. It is unclear why the bow accelerometer does not experience a consistent 1 g drop.
4.4 Hydroelastic discussion Part 1 4.4.1 Whole body vibration of drop tests The quasi-2D and full-scale drop tests actually measured the mechanical shock generated during a slam and not the whole body vibration; therefore, the rst step in quantifying the eect of hydroelasticity on the WBV is to relate mechanical shock to WBV. The primary method to evaluate WBV is via calculation of the RMS of the acceleration; however, the RMS is highly dependent on the time period and the peak acceleration is averaged out. The European directive states that if the crest factor is above six then the VDV value should be used instead; the crest factor of all the drop tests (full and scaled) was well above six, calculated over a 1.5 s time period, so the VDV should be used. The VDV is not designed for individual shocks but it still helps to characterise the slam and weight it depending on the frequency of the vibration. To the author's knowledge there are no standards for evaluating the eect of an individual shock on a human.
The drop tests have been compared to an
under-damped sinusoidal system and this can be characterised using the natural frequency and the damping ratio; however, the damping ratio was shown to be irregular and there are multiple degrees-of-freedom leading to multiple natural frequencies. This is the reason why it was attempted to quantify the shock using the peak acceleration and peak duration but the peak acceleration will now be compared to the VDV. The peak acceleration has been plotted next to the VDV to relate the peak acceleration to a known evaluation method of WBV, see gures 4.33 and 4.34, and the error bars show two
114
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
standard deviations. The crew of the D-class kneel in the boat and there are no frequency weightings for a kneeling person so both seated and standing weightings were used.
It is
anticipated that a kneeling frequency weighting would be in between a seated and standing position because the knees and hips are still able to rotate. The VDV was calculated from the point of impact with the water surface for 1.5 s time period. The VDV from the quasi-2D tests at ve degree deadrise angle does reinforce the conclusions drawn from the peak acceleration and they do show that the human exposure to vibration does change with hull stiness. The VDV increased by 14.3 % when the 0 N/m pre-tensioned hull was changed to a MDF hull. On the other hand, the VDV from the tests with a 15° deadrise angle do not match the trends in the peak accelerations and this is due to the frequency weighting of the VDV. Nonetheless the VDV does show that the stiness and pre-tension of a hull can aect the human exposure to vibration during quasi-2D drop tests. The VDV for the full-scale tests at the transom match the results from the peak accelerations because there is no signi cant eect due to hydroelasticity. The VDV from the crew position and the bow do not always match the trends in the peak accelerations. Increasing the internal pressures tended to increase the VDV in the crew position but the VDV decreased with increasing internal pressures in the bow. The VDV increased by 10.5 % and 16.3 % in the crew position when the internal pressures were increased from 2 psi to 3 psi at 0.5 m and 1 m drop height, respectively.
On the contrary, the VDV decreased by 28.5 % and 13.2 %
on the bow when the internal pressures were increased from 2 psi to 4 psi at 0.5 m and 1 m drop height, respectively. This demonstrates that internal pressures can change the human exposure to vibration during full-scale drop tests. The quasi-2D tests suggest that the VDV could be reduced by 14.3 % by changing the hull from a MDF hull to a 0 N/m pre-tension fabric hull at a 5° deadrise angle; however, when the deadrise angle was 15° the VDV trend was more scattered. This suggests that the hydroelasticity has a greater eect on WBV when the deadrise angle is low, which is backed up by the work of Faltinsen (1999). The full-scale tests suggest that the VDV could be reduced by 9.5 % to 14.0 % in the crew position by reducing the internal pressures from 3 psi to 2 psi at 0.5 m and 1 m drop height, respectively; however, this will lead to an increase in the bow VDV by 2.5 % to 30.1 % but the crew are not exposed to these vibration.
4.4.2 Root cause to hydroelastic slamming The quasi-2D and full scale drop tests showed that hydroelasticity can vary the peak acceleration but what is the root cause of this variation? It is hypothesised by the author that there are two possible reasons for the variation. The rst is that the variation is solely caused by the quasi-static shape of the hull, i.e. the hull deformation is relatively constant and there are no signi cant structural vibrations. Consider rst the 2D problem because only hull stiness was varied. It is well known that the hull shape will aect the slamming characteristics and was brie y reviewed by Lloyd (1998). The structural stiness (which is proportional to the pre-tensioned stresses, see Lewis (2003)) will change the impacting hull shape and this could cause the eect of hydroelasticity; however, increasing the drop height will increase the de-
HYDROELASTIC DISCUSSION PART 1
Peak Acc
115
VDV-Seated
VDV-Standing
45
70
40
60 37.1
50
30 25
28.3
27.2
40
24.3
20
20.7
30
19.9
15
20
VDV (m.s-1.75)
Acceleration (g)
35
10 23.3
5 0
15.4 38.4
13.8 34.5
MDF
13.4
33.6
1000 N/m
0 N/m
58.3
MDF
22.4
21.1 52.6
0.5m drop height
56.1
1000 N/m
10 0
0 N/m
1m drop height
(a) 5 ° deadrise angle
Peak Acc
VDV-Seated
VDV-Standing
25
45 40
22.9
35
20.1
19.9
30
15
25
20
12.1
10
10.8
10.6
15
5 0
22.7
MDF
20.6
8.2
21.3
1000 N/m 0.5m drop height
8.5
0 N/m
38.6
MDF
10
15.5
15.4 9.1
38.8
VDV (m.s-1.75)
20
Acceleration (g)
4.4.
15.1 37.9
1000 N/m
5 0
0 N/m
1m drop height
(b) 15 ° deadrise angle Figure 4.33: Quanti cation of WBV during the quasi-2D drop tests.
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
Peak Acc
VDV-Standing
VDV-Seated
14
35
10.8
10
13.0
30
10.8
25
8
20
6
28.5
27.4
4
19.5
28.1
20.8
20.7
10
2 0
15
VDV (m.s-1.75)
11.3
Acceleration (m.s-2)
13.0
13.0
12
5 7.8
8.3 2 psi
8.3 3 psi
10.9 4psi
11.4 2 psi
11.2 3 psi
0.5m drop height
0 4psi
1m drop height
(a) Transom accelerations
Peak Acc
VDV-Standing
VDV-Seated
18
45
16
Acceleration (m.s-2)
40
16.3
15.5
35
13.3
12
30
10.8
10
25 9.7
8
20 35.7
35.2
30.7
6
26.2
23.7
4
15
25.9 10
2 0
VDV (m.s-1.75)
15.9 14
5 9.5
10.5 2 psi
10.4 3 psi
12.3 4psi
14.3 2 psi
14.1 3 psi
0.5m drop height
0 4psi
1m drop height
(b) Crew accelerations
Peak Acc
VDV-Standing
VDV-Seated 50
20 45 18.5
17.0
40
16.6
15
35 13.7
30 25
10 40.9
9.1
35.5 15
7.0
26.3
5
20
39.9
20.2
VDV (m.s-1.75)
Acceleration (m.s-2)
116
10
18.8
5 0
10.5
8.1 2 psi
7.5 3 psi
16.4 4psi
0.5m drop height
16.0 2 psi
14.2 3 psi
0 4psi
1m drop height
(c) Bow accelerations Figure 4.34: Quanti cation of WBV during the full-scale drop tests.
4.4.
HYDROELASTIC DISCUSSION PART 1
117
formation of the hull and exaggerate the eect of the hull shape but this should not cause the peak acceleration trend to invert with drop height. Increasing the drop height may cause other phenomena to occur such as air cushioning and ow separation, see Faltinsen et al. (2004), and air pockets, see Halswell et al. (2012) but it is not expected that this would cause the trend to invert. It is more likely that the structural vibration of the hull will have a signi cant eect on the rigid body motion and the root cause is a superposition of two (or more) structural vibrations. These structural vibrations will occur at dierent frequencies which will depend on the stiness and boundary conditions of the structure. If these vibrations are in phase at the instantaneous moment of peak acceleration then this will lead to an increase in peak acceleration, whereas, if the two vibrations are 180° out of phase then this will lead to a decrease in peak acceleration. Faltinsen et al. (2004) divided this problem into two time scales; an initial structural-inertia phase and a free vibrations phase. The peak acceleration occurs in the structural inertia phase where large hydrodynamic forces lead to large accelerations on a small structural mass. The frequency spectrum graphs of the accelerations do show a shift in the dominant frequencies. For this reason it is believed that in the 2D tests the inversion of the peak acceleration trend is caused by a phase shift between the structural hull vibration and the rigid body motion, and not the change in shape; although, the reason for the phase shift is not known. Now consider the full-scale drop tests where the internal pressures of the keel and sponson will change the response of the keel, sponson, hull and deck. The root cause to the change in peak acceleration at full-scale could also be due to the change in hull shape or the phase shift between the structural vibrations and the rigid body motion; however, it could also be due to any shape changes or changes in natural frequencies of the keel, sponson or deck. It is probable that the variation in peak acceleration is due to the phase dierence between the rigid body motion and one or more components for the same reason as the 2D tests.
The
inversion of the peak acceleration trend was found in the 2D and the full-scale tests which could imply that it is the structural hull vibration that causes the trend to invert because it was the only parameter changed in the quasi-2D tests. The deformation of the D-class was measured during the full-scale drop tests so this may indicate which components have been signi cantly aected by the internal pressures.
The
deformation of the sponson and keel can be compared using equation 5.3 and the results are shown in table 4.18. This again con rms that the deformation of the keel is considerably larger than the deformation of the sponson; however, equation 5.3 is not valid for the drop tests. It assumes the air is incompressible and under the high impulse forces of a drop test the air compressibility will be signi cant. If air compressibility is considered, then this will allow the in atable cylinder to compress more than without considering air compressibility because the volume of air will decrease as the air is compressed. The full-scale drop tests showed that the deformation of the deck was negligible because the internal pressures changed the peak central de ection of the aft deck panel only two millimetres. The peak hinge angle variation due to internal pressures was up to three degrees in middle hinge (B3) and this may have a noticeable eect on the slamming characteristics;
118
CHAPTER 4.
Condition H0.5 S2.25 K2
H0.5 S3.25 K3
H0.5 S4.25 K4
H1 S2.25 K2
H1 S3.25 K3
H1 S4.25 K4
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
rc
(mm)
rx
Sensor
Percentage Change
(mm)
Sponson (psi)
4.9
254
12.4 58.8
Keel (psi)
100.5
58.5
Sponson (psi)
0.6
254
1.6
Keel (psi)
62.0
58.5
36.2
Sponson (psi)
8.1
254
20.7
Keel (psi)
54.3
58.5
31.8
Sponson (psi)
6.6
254
16.9
Keel (psi)
126.9
58.5
74.2
Sponson (psi)
4.1
254
10.3
Keel (psi)
95.6
58.5
55.9
Sponson (psi)
9.4
254
23.9
Keel (psi)
64.0
58.5
37.5
Table 4.18: Deformation of the sponson and keel related to the change in internal pressure during the drop tests.
however, it is not clear how to investigate the eect of hinge angle because very few boats have been designed with hinged decks. The peak hull de ection was shown to vary by around 20 mm on all the hull string potentiometers but, more importantly, the frequency spectra of the hull was insigni cantly aected by the internal pressures. This is a critical piece of information in regards to the performance of the D-class because it implies that the dynamic response of the hull has not changed during the drop tests, i.e.
the peak hull de ection has changed which is proportional to the peak
de ection of the keel but the natural frequencies have not changed. This also indicates that the pre-tensioned stresses in the hull were not aected by the internal pressure of the keel because, according to Lewis (2003), the pre-tensioned stresses will aect the natural frequencies of a BTM and the natural frequencies of the hull have not been aected by the pressure of the keel. This now leads to confusion between the quasi-2D drop tests and the full-scale drop tests because the trends in the accelerations were similar at both scales; although, only the pre-tensioned stresses were varied in the quasi-2D drop tests but (it now appears that) the pre-tensioned stresses were not varied during the full-scale tests. If the hypothesise about phase dierence between the rigid body motion and one (or more) structural vibrations is correct then this means that dierent structural vibrations have been aected at both scales, but both structural vibrations have led to the same trend in acceleration data. This hypothesis required further validation but this nal observation suggests that if all the structural vibrations were 180° out of phase with the boat motion then a much more signi cant eect on the slamming characteristics could be achieved.
4.5 Summary The full-scale holistic experiment rst measured the static shape of the D-class and it showed that the internal pressures and crew loading can aect the static shape of all the components. It also revealed that the static shape is more dependent on the keel pressure than the sponson
4.5.
SUMMARY
119
pressure. The quasi-2D drop tests proved that the hull stiness and the pre-tensioned stresses can aect the peak acceleration but it depends on the ratio of hydroelastic importance, see equation 2.3. The ratio of hydroelastic importance means that hydroelasticity has a greater aect when the deadrise angle is low or the drop height is large. The trend between the peak acceleration and hydroelasticity was drop height dependent. No trend in the peak duration was found. The full-scale drop tests showed that the internal pressures of the sponson and keel can aect the peak acceleration and, again, the trend was drop height dependent. The frequency spectra revealed that the primary boat motion changed from heave at the transom to pitch at the bow. The spectra also showed a super harmonic in the anomalous result. The magnitudes of deformation of the structural components were analysed and it is believed the dominant component is the keel. The frequency spectra of the hull did not change due to the internal pressures, which suggest the pre-tensioned stress in the hull was not aected by the internal pressures. The WBV generated during the drop tests was investigated using the VDV. The results show that the pre-tensioned stress in the hull or the internal pressures of the sponson and keel can aect the VDV and thus the WBV. The VDV in the crew position was reduced by up to 16 % when the internal pressures were reduced from 3 psi to 2 psi; however, this lead to an increase of 13 % on the bow. Most importantly, the drop tests proved that hydroelasticity can aect the VDV and the WBV.
120
CHAPTER 4.
HYDROELASTIC RESULTS UNDER VERTICAL LOADS
Chapter 5
Hydroelastic results under horizontal loads 5.1 Flat water trials The at water trials were performed on the
12th
of February 2013 on the river Medina, Cowes.
The water conditions are shown in gure 5.1.
Figure 5.1: Water conditions during the at water trials.
5.1.1 Eect of internal pressures on forward speed The primary performance indicator for the at water trials was the quasi-steady forward speed. The eect of the internal pressures on the speed is shown in gure 5.2 and the error bars de ne two standard deviations. This shows that the D-class was 0.44 knots faster when the internal pressures were at 4 psi than 2 psi.
On one run (repeat number 5) at 2 psi a speed of 23
knots was recorded and the helmsman noted that the boat exhibited the pulsing motion; it was regarded as an anomaly and removed from the analysis.
The pulsing motion will be
discussed in more detail in section 5.1.3. The null hypothesis was assumed, tested with a two tailed Student's T-test and the results are shown in table 5.1. It clearly shows that there is a 121
122
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
statistical dierence between the top speeds at the three internal pressures to a 95 % certainty. The internal pressures of the sponson and keel while the boat was still at rest are shown in table 5.2. It shows that the internal pressures were lower than the desired value due to air leakages but they were no more than 8 % o the intended pressure during all the runs.
22.4 22.2
Top Speed Knots
22 21.8 21.6 21.4 21.2 21 20.8 20.6
4psi
3psi
2psi
Figure 5.2: Aect of the internal pressures on the at water speed.
Comparison
Degrees of Freedom
T-values
4 psi to 3 psi
18
2.15
3 psi to 2 psi
17
2.87
4 psi to 2 psi
17
7.11
T-value > T-critical at 95 % certainty
Yes Yes Yes
Table 5.1: Student's T-test of the top speed measured during the at water trials (excluding pulsing run).
The second performance indicator was the thrust from the outboard engine measured using the strain on the trim pin. The mean strain on the trim pin was calculated for the quasi-steady planing at all pressures and Student's T-test was performed. It revealed that the quasi-steady strain did not vary and there was no statistical dierence between the data sets. This means that the internal pressures did not aect the thrust of the outboard engine. The boat was at rest for the rst three seconds and then the helmsmen accelerated at full throttle on to the plane. After the acceleration stage the boat deformed into a quasi-steady shape because the planing forces become quasi-steady, so the investigation will explore how the internal pressures changed the quasi-steady shape.
The mean internal pressures of the
sponson and keel during quasi-steady planing can be seen in table 5.3. The sponson pressures did not increase by more than 0.043 psi, whereas the keel pressure increased by up to 0.734 psi. A change in pressure can be used to infer a change in shape so these results infer that the keel deformed more than the sponson and the keel is likely to have a more dominant eect
5.1.
FLAT WATER TRIALS
123
Condition
1
2
3
4
5
6
7
C3 S2.25 K2
2.28
2.24
2.25
2.23
2.29
2.24
2.27
C3 S3.25 K3
3.21
3.17
3.18
3.17
3.24
3.19
3.22
C3 S4.25 K4
3.84
3.91
3.93
3.92
4.00
3.94
3.95
Average 2.26 3.20 3.93
σ
% error
0.02
0.44
0.03
-1.6
0.05
-7.6
(a) Sponson pressure (psi). Condition
1
2
3
C3 S2.25 K2
2.04
2.15
2.16
C3 S3.25 K3
2.90
3.00
3.02
C3 S4.25 K4
3.73
3.85
3.86
Average 2.12 2.97 3.81
σ
% error
0.07
5.9
0.07
-0.95
0.07
-4.7
(b) Keel pressure (psi). Table 5.2: Mean internal pressures while at rest during the at water trials.
Condition
M eaninitial
C3 S2.25 K2
2.26
C3 S3.25 K3
3.20
C3 S4.25 K4
3.93
M eanplaning
2.30 3.23 3.95
Change
%age Change
σinter
σintra
0.04
1.9
0.023
0.017
0.03
1.0
0.026
0.017
0.02
0.6
0.048
0.017
Change
%age Change
σinter
σintra
0.73
34.6
0.069
0.092
0.76
24.5
0.070
0.087
0.65
16.9
0.070
0.074
(a) Sponson pressure (psi). Condition
M eaninitial
C3 S2.25 K2
2.12
C3 S3.25 K3
2.97
C3 S4.25 K4
3.81
M eanplaning
2.85 3.73 4.46
(b) Keel pressure (psi). Table 5.3: Internal pressures whilst quasi-steady planing during the at water trials.
than the sponson on the planing performance. This is also backed up by the static tests which showed the keel pressure had a greater eect on the static hull shape than sponson pressure. The sponson rotation was also measured and the results are summarised in table 5.4 (position is an inwards rotation). It shows that the sponson did rotate as the boat accelerated on to the plan but only by a few millimetres. Unfortunately one of the sponson rotation sensors was damaged during these trials and it only measured the rotation at 4 psi. This sensor (SR1) was further back where the magnitude of sponson rotation was greater with a mean de ection of 6.79 mm. This shows that the magnitude of rotation does vary along the length of the sponson but it has not been properly quanti ed. There are larger hydrodynamic forces on the aft of the sponson because the bow trims upwards cause sinkage at the aft; however, the transom will reduce the sponson rotation at the aft. Note: the inter-standard deviation is used to de ne the variation between each test condition; whereas, the intra-standard deviation is used to quantify the variation within each repeat of a test condition due to random inputs (such as small waves on the water surface). The intra-standard deviation is actually the mean intra-standard deviation of all the repeats from each test condition. The deck trim angle and hinge angles during quasi-steady planing for the three internal
124
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
Condition
SR1 (aft)
σinter
SR2 (fore)
σinter
σintra
C3 S2.25 K2
-
-
1.01
0.57
0.28
C3 S3.25 K3
-
-
2.11
0.60
0.26
C3 S4.25 K4
6.79
0.76
2.09
0.82
0.70
Table 5.4: Sponson rotation (mm) whilst quasi-steady planing during the at water trials.
Figure 5.3: Hinge angle variation during the at water trials.
pressures are shown in gures 5.3 and 5.4, and are summarised in table 5.5. Figure 5.4 follows the format of (a) + (b) = (c) and (c) + (d) = (e). Firstly, the deck trim angle during quasisteady planing does change due to the internal pressures and a Student's T-test does reveal a statistical dierence between the 2 psi and 4 psi tests, but not between the 2 psi and 3 psi or the 3 psi and 4 psi tests. The trim angle can also be used as a performance indicator because it will aect the hydrodynamic lift generated by planing; however, the trim angle only changed by 0.052° (resulting in a 5 mm raise at the bow) which is probably negligible. The change in the hinge angles due to the internal pressures is considerably larger than the change in deck trim angle, by up to 10 times. As the deck trim angle increases the aft deck hinge (B1) decreases, which suggests that they are coupled due to the displacement of the hull; i.e. as the bow trims upward there are larger weight forces acting on the bow because it is not supported by the buoyancy or hydrodynamic lift forces. The data for the middle hinge (B3) at 2 psi was not captured so a linear t was used to predict it. The validity of the hinge results are unclear due to the multiple problems with the hinge strain gauges. The aft hinge (B1) could have deformed due to shear forces as well as rotational forces and this is backed up by a higher than average inter standard deviation of the aft hinge (B1). The strains generated by the hydrodynamic lifting forces between the static trials with no crew at 2 psi (i.e.
C0 S2.25 K2) and the quasi-steady planing of all internal pressures
5.1.
FLAT WATER TRIALS
125
(a) C3 S2.25 K2.
(b) C3 S3.25 K3 - C3 S2.25 K2.
(c) C3 S3.25 K3.
(d) C3 S4.25 K4 - C3 S3.25 K3.
(e) C3 S4.25 K4. Figure 5.4: Hinge angle variation during the at water trials.
126
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
Condition
Mean (deg)
σinter
σintra
C3 S2.25 K2
4.836
0.053
0.95
C3 S3.25 K3
4.879
0.075
0.71
C3 S4.25 K4
4.888
0.039
0.56
(a) Deck trim angle. Condition
Mean (deg)
σinter
σintra
C3 S2.25 K2
-0.079
0.24
0.28
C3 S3.25 K3
-0.486
0.17
0.21
C3 S4.25 K4
-0.688
0.12
0.13
(b) Hinge angle - B1 (aft). Condition
Mean (deg)
σinter
σintra
C3 S2.25 K2
-
-
-
C3 S3.25 K3
-1.971
0.090
0.13
C3 S4.25 K4
-1.748
0.080
0.080
(c) Hinge angle - B3 (mid). Table 5.5: Aect of the internal pressures on the planing trim angle and hinge angles.
are shown in tables 5.6 and 5.7, along with the inter and intra standard deviations. In the x-axis, table 5.6, it shows that the largest strains were generated on the aft two deck panels which means the majority of the hydrodynamic lifting forces are acting on the aft two deck panels, but the strains are in opposite directions. The top of the aft most deck panel is in tension, whereas, the top of the second aft panel (D3x) is in compression. This could mean that the deck has adopted a slight S-shape in comparison to the static shape.
The largest
strain measured on one panel in the x-axis was 1032 micro-strains (mean of 1114 and 949) which relates to an extension of 1.86 mm over the 1.8 m long, aft deck panel or a 27.9 mm central de ection.
The strain in the front two deck panels could be regarded as negligible.
The y-axis strains, table 5.7, show that increasing the internal pressures increases the y-axis strain in quasi-steady planing. This implies that the crew are forcing the edges of the deck downwards and the keel is forcing the centre upwards resulting in larger bending moments around the keel. The change in strains between the three internal pressure conditions are graphically shown in gure 5.5. The y-axis data is as expected where increasing the internal pressures causes an extension on the top side of the deck due to the more pronounced upwards force of the keel. The x-axis strains show that the aft most deck panel deformed into a S-shape when the internal pressures increased from 2 psi to 3 psi because the strains are in opposite directions. On the other hand, when the internal pressures are changed from 3 psi to 4 psi the curvature is more constant with compression on the top side of the deck panel. The front three deck panels show a consistent extension on the top side of the panels when the internal pressures are increased. The largest strain change measured due to the internal pressures was -140 micro-strains which would lead to a compression of 0.25 mm over a 1.8 m long, aft deck panel, which is very small and negligible. If a bending moment is considered instead of a compressive force, then -140
5.1.
FLAT WATER TRIALS
127
micro-strains could lead to a central de ection of 3.8 mm which would also have a negligible eect on at water performance. Condition
Sensor number Mean
C3 S2.25 K2
C3 S3.25 K3
C3 S4.25 K4
(µ)
µσinter µσintra Mean (µ) µσinter µσintra Mean (µ) µσinter µσintra
D1x
D2x
D3x
D5x
D6x
D7x
1114
949
-840
-25
-66
4
42
157
75
12
11
8
42
129
104
10
22
13
981
972
-806
17
-2
-33
24
47
47
16
14
3
37
130
99
7
15
11
888
832
-743
102
97
-48
35
31
23
30
16
4
34
108
68
5
11
8
Table 5.6: Change in X-axis micro-strains from static C0 S2.25 K2.
Condition
Sensor number
D1y
D2y
D3y
D5y
D6y
D7y
(µ) µσinter µσintra Mean (µ) µσinter µσintra Mean (µ) µσinter µσintra
-146
65
293
109
-50
-
10
12
35
18
17
-
Mean C3 S2.25 K2
C3 S3.25 K3
C3 S4.25 K4
19
27
47
11
8
-
-36
142
385
220
27
-
22
23
29
15
14
-
29
33
50
13
8
-
70
221
459
352
109
-
14
29
23
12
12
-
31
32
39
11
7
-
Table 5.7: Change in Y-axis micro-strains from static C0 S2.25 K2.
The quasi-steady hull shapes and the change in hull shapes are shown in gure 5.6, where (a) + (b) = (c) and (c) + (d) = (e). The quasi-steady hull shapes reveal that there is a concave shape approximately 1000 mm along the x-axis and the peak displacement aligns with the rst deck hinge at 1800 mm. The change in hull shape graphs show that there is a greater change at lower pressures indicating that lower pressures will have a greater eect on hydrodynamic performance. The change in hull shape from 2 psi to 3 psi has a maximum deformation in line with the centre of the second aft deck panel, whereas the change from 3 psi to 4 psi has a maximum deformation aligned with the most aft deck hinge. The change in hull shape has a maximum of approximately 18 mm which would cause a change in peak deadrise angle of 2.18° using Pythagoras' theorem. Interestingly, the change between 2 psi and 3 psi leads to an increase in peak deadrise angle at the point of maximum change but the change between 3 psi and 4 psi leads to a decrease in peak deadrise angle.
The distances measured by the
hull string potentiometers can be seen in table 5.9, along with the inter and intra standard deviations.
The string potentiometer data can be transformed in to mean deadrise angles
using Pythagoras theorem and the mean deadrise angle ranged from 13.05° to 14.39°; this is shown in table 5.8.
The mean deadrise angle increases by 1.34° when the internal pressure
is increased by 2 psi.
This was expected because the keel diameter increases with internal
128
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
(a) C3 S3.25 K3 - C3 S2.25 K3.
(b) C3 S4.25 K4 - C3 S3.25 K3. Figure 5.5: Change in the quasi-steady planing micro-strains in the deck due to the internal pressures.
pressure, which increases the deadrise angle. Internal pressures (psi)
2
3
4
Deadrise angle (°)
13.05
14.02
14.39
Table 5.8: Mean deadrise angles from all the hull sensors during at water trials.
Some of the at water trial results can be validated by comparing them to the results of Dand (2002a,b, 2003b,a,c). Firstly, the top speed of the D-class is reported by the RNLI to be a maximum of 25 knots, which was clearly not achieved in these trials. This shows that the performance of the D-class used in this work had aged and deteriorated over its life and it's regularly report by helmsmen of D-classes that the top speed can vary considerably from boat to boat; a problem that was not complete removed during the last redesign. These speeds are comparable to speeds within the work of Dand et al.; however, the modern 50 HP outboard engine was during these trials but Dand et al. used the older 40 HP outboard engine. Dand et al. measured a running trim angle of approximately 5.6 degrees at 22 knots on the scale model, see Dand (2003b) gure 13. The trim angle on the full-scale boat was not measured at 22 knots but the full-scale trim angle was fairly consistently one degree less than the modelscale trim angle at speeds greater than 10 knots. Therefore, using extrapolation the full-scale
5.1.
FLAT WATER TRIALS
trim angle at 22 knots should be approximately 4.6 degrees.
129
The quasi-steady trim angle
measured during the thesis ranged from 4.84° to 4.89° which shows fairly good agreement and the dierence is most likely due to the position of the helm and crew, which was held constant during this experiment using markers for the knees of the helm and crew. Finally, Dand had estimated the mean deadrise angle to be 13°, which is in agreement with the mean deadrise angles measure during the at water trials of 14.02° at 3 psi.
So far the at water results have shown that all the components of the D-class deform due to the hydrodynamic forces and that the internal pressures change the magnitude of deformation; but which component is likely to have the greatest eect on at water performance? This will be discussed in section 5.3.2 along with the results from the other trials.
5.1.2 Dynamic motion So far the eect of the internal pressures on the deformation of the various components has been shown and it was assumed that the motion was quasi-steady; however, the motion is not quasi-steady. Firstly, there is the dynamic motion as the boat accelerates on the plane and secondly, there is the dynamic motion due to small waves on the water's surface de ned previously using the intra standard deviation. This dynamic motion was investigated using time history plots and videos of moving graphs. Moving-graph videos were used because they show the deformation of the 3D boat through time and the videos can be found in appendix G. The top left graph in the video shows the mean sponson and keel pressures, and the vertical line indicates the run time; similar to gure 5.9. The right graph shows the deck trim angle and the hinge angles (there is one second wait on the deck trim angle due to the use of a rolling average). The bottom left graph shows the deck strains; similar to gure 5.3. The nal plot in the video, bottom right, shows the hull shape in the same manner as gure 5.6.
The rst performance indicator that can be plotted against time is the strain on the outboard engine trim pin and it is shown in gure 5.7 (the three second static stage has been removed). The time history con rms that the internal pressures had a minimal eect on the quasi-steady thrust but it does show that the thrust during the acceleration stage has changed. The acceleration stage could have been aected by either a change in the resistance of the hull or a propeller-hull interaction, but it cannot be con rmed which of these caused the change in thrust. The mean variation in deck trim angle can be seen in gure 5.8 (the three second static stage has been removed) and it shows that the internal pressures did have a minimal eect on the deck trim angle; however, it does reveal that as the boat accelerates it trims bow down before trimming heavily upwards.
130
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
(a) Hull shape: C3 S2.25 K2.
(b) Change in hull shape = C3 S3.25 K3 - C3 S2.25 K2.
(c) Hull shape: C3 S3.25 K3.
(d) Change in hull shape = C3 S4.25 K4 - C3 S3.25 K3.
(e) Hull shape: C3 S4.25 K4. Figure 5.6: Change in quasi-steady planing hull shape during the at water trials.
Pulsing
C3 S4.25 K4
C3 S3.25 K3
C3 S2.25 K2
Condition
0.2 0.2 24.7 16.8
2.4 35.3 16.0
8.8
1.0
0.2
0.2
2.5 52.7
20.0
56.6
4.2
1.4
18.6
3.0
1.0
16.7
2.2
1.2
16.3
H6
18.1
95.4
5.7
1.5
48.1
5.4
1.6
40.4
4.3
4.0
36.3
H7
14.2
69.1
9.1
1.9
89.6
9.9
3.1
84.6
7.6
9.4
88.4
H8
12.8
108.8
8.3
1.7
126.9
9.4
2.8
120.3
7.4
9.1
122.1
H9
14.7
69.8
6.9
2.1
111.1
6.2
2.4
115.6
5.5
1.5
121.5
H10
13.2
114.4
4.2
1.1
153.8
4.0
1.4
155.7
4.0
2.3
160.3
H11
Table 5.9: Distance between the hull and the deck.
17.5
45.6
0.3
0.1
12.1
0.2
0.1
1.0
0.1
8.2
11.6
0.1
1.9 49.7
9.2
H4
0.2
7.8
H2
0.2
1.6
43.5
(mm) σinter σintra Mean (mm) σinter σintra Mean (mm) σinter σintra Mean (mm) σintra
Mean
H1
Sensor number
9.2
61.8
7.4
1.2
71.0
9.9
3.2
75.2
9.7
9.8
75.2
H12
22.3
147.3
17.2
2.8
167.5
23.4
7.1
183.9
22.9
23.1
189.5
H13
4.2
40.2
1.9
2.8
44.6
3.0
1.0
40.9
3.3
1.0
34.0
H14
5.5
133.9
3.9
0.8
142.8
5.3
1.8
134.7
5.6
2.9
118.1
H15
6.7
54.4
2.8
0.6
68.4
4.2
1.4
60.9
4.7
1.5
48.6
H16
6.0
132.9
2.9
0.5
150.9
4.6
1.4
142.0
5.6
2.4
126.5
H17
5.1. FLAT WATER TRIALS 131
132
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
Figure 5.7: Mean strain on the outboard engine trim pin.
Figure 5.8: Mean deck trim angle.
Examples of the time history of the internal pressures and sponson rotation are shown in gure 5.9 (the example is of the rst run of each pressure condition). The time histories show the three seconds while the boat was at rest then it takes a further seven seconds for the internal pressures and sponson rotation to reach their quasi-steady shapes. Once the boat is in quasi-steady planing the internal pressures and sponson rotation do vary, termed the intra variation, but the intra variation is small compared to the overall change from initial to quasi-steady planing. The inter and intra standard deviations are summarised in tables 5.3 and 5.4. The keel pressure initially drops as the D-class accelerates onto the plane and this is shown in gure 5.9b. The drop in pressure is no more than approximately 5 % and the pressure drop
5.1.
FLAT WATER TRIALS
133
(a) The mean percentage change in sponson pressure.
(b) The mean percentage change in keel pressure.
(c) The sponson rotation (position is a inwards rotation). Figure 5.9: Examples of the deformation of the sponson and keel during the at water trials.
134
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
is large at low pressures, summarised in table 5.10. The video reveals that as the keel pressure drops so does the trim angle before both start to increase. It is hypothesised that they both start to increase as the Froude number approaches unity and the boat starts to plane. The video also shows that there is an element of coupling between the keel pressure and the trim angle once the D-class is fully planing; the trim angle varies with the keel pressure (or visa versa). Mean (psi)
σ
S2.25 K2 C3
0.111
0.020
S3.25 K3 C3
0.077
0.006
S4.25 K4 C3
0.080
0.005
Table 5.10: Drop in keel pressure as the D-class accelerates.
The static and quasi-steady hull shapes are shown in gure 5.10 and the change in hull shapes are shown in gure 5.11. The transformation from the static to planing hull shapes can also be seen in the moving-graph videos in appendices G.1 to G.3.
The planing forces
acting on the hull causes a concave shape approximately 1000 mm along the x-axis and a peak displacement aligned with the rst deck hinge at 1800 mm. The pressure distribution on a at planing plate was shown in gure 2.12 and it could be expected that the peak hydrodynamic pressure is aligned with the minimum hull de ection (i.e. centre of the concave section), also suggested by Dand (2003c). Dand (2003c) measured the trim angle and sinkage of the D-class with three crew at 19.4 knots to be 3° and 0.1 m, respectively.
This would mean that the
planing water length is approximately 1.91 m (using Pythagoras' theorem) and almost all the hydrodynamic forces act on the aft most deck panel. On a at plate the peak hydrodynamic pressure due to the planing forces is just behind the front wetted edge so it is expected that the peak pressure is around 1.5 m to 1.8 m along the x-axis. This means that the minimum hull de ection (the concave shape) does not align with the peak pressure and, in fact, it's more likely to align with the maximum hull de ection (1.8 m along x-axis). The videos showed another interesting event where the hull took on another shape for approximately one second. The shape is shown in gure 5.12 and this image was taken from the fourth run at 2 psi. It shows a second hump occurring 1.4 m along the x-axis and at the same time there is a reduction in the primary hump at 1.8 m. This second hump did appear at all pressures. The reason why this second hump occurred is unclear but could simply be due to a wave passing under the boat and this is the shape the hull has adopted; although, the hump does not pass along the length of the hull in the x-axis. The dynamic motion has shown that as the D-class accelerates it initially trims the bow down with a simultaneous drop in keel pressure (and peak in the sponson rotation).
The
thrust of the outboard engine peaks almost instantaneously and as the boat accelerates the thrust decreases as the resistance and propulsion become equal. After approximately seven seconds the D-class deforms into a quasi-steady shape. The transformation of the hull shape from static to quasi-steady is fairly regular but during quasi-steady planing the peak de ection (1.8 m along the x-axis) does vary considerably. A second hump was observed at all pressures.
5.1.
FLAT WATER TRIALS
135
(a) Static hull shape at 2 psi (C3 S2.25 K2).
(b) Quasi-steady planing hull shape at 2 psi (C3 S2.25 K2).
(c) Static hull shape at 3 psi (C3 S3.25 K3).
(d) Quasi-steady planing hull shape at 3 psi (C3 S3.25 K3).
(e) Static hull shape at 4 psi (C3 S4.25 K4).
(f) Quasi-steady planing hull shape at 4 psi (C3 S4.25 K4).
Figure 5.10: Static and quasi-steady hull shape during the at water trials.
136
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
(a) Change in the hull shape at 2 psi.
(b) Change in the hull shape at 3 psi.
(c) Change in the hull shape at 4 psi. Figure 5.11: Change in the hull shape during the at water trials.
5.1.
FLAT WATER TRIALS
137
Figure 5.12: A dynamic shape of the hull during the at water trials at 2 psi (fourth repeat).
The keel pressure increased by approximately 0.7 psi as the boat accelerated onto the plane but the keel pressure was also found to noticeably vary with trim angle.
5.1.3 Pulsing motion The D-class exhibited the pulsing motion during the fth run at 2 psi.
So far it has been
removed from the analysis because it was regarded as an anomaly and the boat did not respond in the same manner as the other run. Moreover, this pulsing run exhibited the highest speed of 23 knots, which was 1.3 knots faster than the non-pulsing runs or a 6.3 % increase. Dand et al. (2008) said that the pulsing motion meant that the D-class was not able to accelerate up to its top speed; however, this run showed that the pulsing motion meant the D-class achieved a higher speed than the normal motion. The strain on the outboard engine trim pin can be seen in gure 5.13. It shows that the acceleration stage has not changed when compared with the mean of the non-pulsing 2 psi runs; however, during the quasi-steady stage the strain is less than the non-pulsing strain. Therefore the speed has increased but the thrust from the engine has decreased.
The nal
performance indicator is the deck trim angle and this is shown in gure 5.14 and the mean and standard deviation can be seen in table 5.11. It shows that the acceleration stage has changed slightly but, more importantly, the intra variation of the running deck trim angle when pulsing is considerably higher than the non-pulsing trim angle. The deformation of the sponson and keel are shown in gure 5.15. The internal pressures of the sponson and keel whilst pulsing are lower than the equivalent internal pressures at 2 psi; although, the intra standard deviation is still greater. It can be seen that the variation of the internal pressures of the sponson and keel are in time and coupled together. The sponson rotation during the pulsing run was shown to be slightly higher than the non-pulsing run at 2 psi.
138
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
Figure 5.13: Strain on the outboard engine trim pin while pulsing.
The deck trim angle and aft hinge angle (B1) are shown in table 5.11.
Unfortunately,
during the 2 psi runs the middle hinge strain gauge (B3) was not functioning. Nonetheless, the aft hinge angle was the lowest of all runs at all internal pressures and the intra standard deviation was similar to the other runs meaning that the aft hinge angle held a fairly constant angle of - 1.63°. The static angle of the aft hinge during the 2 psi runs was - 1.12° (standard deviation = 0.07) so this reveals that the aft hinge angle during the pulsing actually deformed in the opposite direction to the non-pulsing runs.
Condition
Deck trim angle
Hinge - B1 (aft)
Mean (deg)
σintra
Mean (deg)
σintra
C3 S2.25 K2
4.836
0.95
-0.079
0.28
Pulsing
4.770
1.49
-1.625
0.27
Table 5.11: Aect of the internal pressures on the planing trim angle and hinge angles.
The deck strains generated during the pulsing run are shown numerically in tables 5.12 and 5.13, and graphically shown in gure 5.16. The results show that almost all the strains in the deck panels have decreased, especially in x-axis of the after two deck panels (D1x, D2x and D3x).
The intra standard deviation during the pulsing run is fairly close to the intra
standard deviations of the non-pulsing runs, except the aft most strain gauge (D1x).
The
range of the aft deck panel strain (D1x) can be de ned by two standard deviations so the micro-strain ranged from -496 to 604; however, these strains are still less than those generated during the 2 psi non-pulsing runs; i.e. the overall strain during the pulsing run was lower than the non-pulsing runs, even though the variation was greater and this is shown in gure 5.16 (which has the same scaling as gure 5.5). This is also well demonstrated in the moving-graph video of the pulsing run found in appendix G.4.
5.1.
FLAT WATER TRIALS
139
Figure 5.14: Deck trim angle while pulsing.
Condition
Sensor number
D1x
D2x
D3x
D5x
D6x
D7x
(µ) µσinter µσintra Mean (µ) µσintra Mean (µ)
1114
949
-840
-25
-66
4
42
157
75
12
11
8
Mean C3 S2.25 K2
Pulsing Dierence
42
129
104
10
22
13
54
-207
-277
-10
-53
-21
275
137
101
10
23
21
-1060
-1156
564
15
13
-24
Table 5.12: Change in X-axis micro-strains from static C0 S2.25 K2 for the pulsing run.
Figure 5.16: Change in deck micro-strains from static C0 S2.25 K2 to the mean pulsing run.
The hull shape varied considerably during the pulsing run and exhibited a number of dierent shapes including a dynamic response where a wave in the fabric hull passed down the length in a cyclical manner. The intra standard deviation in table 5.9 con rms that the pulsing hull shape varied considerably more than any other run, with a bias of more variation in displacement at the aft of the hull. The quasi-steady hull shapes and the change in hull shapes from the static shape to the mean pulsing shape or quasi-steady non-pulsing shape
140
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
(a) The mean percentage change in sponson pressure.
(b) The mean percentage change in keel pressure.
(c) The sponson rotation (position is a outwards rotation). Figure 5.15: Deformation of the sponson and keel during the pulsing run of the at water trials at 2 psi.
5.1.
FLAT WATER TRIALS
Condition
Sensor number Mean
C3 S2.25 K2
Pulsing Dierence
141
(µ)
µσinter µσintra Mean (µ) µσintra Mean (µ)
D1y
D2y
D3y
D5y
D6y
D7y
-146
65
293
109
-50
-
10
12
35
18
17
-
19
27
47
11
8
-
-106
-87
-10
45
-84
-
35
61
60
18
14
40
-152
-303
-64
-34
Table 5.13: Change in Y-axis micro-strains from static C0 S2.25 K2 for the pulsing run.
can be seen in gure 5.17, where (a) + (b) = (c) and (a) + (d) = (e).
This shows that
the mean pulsing hull shape is considerably dierent from the quasi-steady non-pulsing hull shape. The concave shape at the aft of the hull had disappeared and the hull shape is fairly close to the static hull shape. The author hypothesises that the mean pulsing hull shape has a more uniform and weaker stress distribution than the quasi-steady non-pulsing run because the shape is less distorted; however, this is a dynamic motion so the mean shape is not the best description. The video of moving-graphs in appendix G show this dynamic motion in a way that cannot be achieved using static graphs. The boat accelerates and after six seconds the hull takes on a unique shape with a distinct second hump around 1250 mm along the x-axis and this is shown in gure 5.18. Dand et al. (2008) describes the pulsing motion where initially a pressure wave forms under the bow of the boat before it passes along the length. The author suggests that gure 5.18 is the hull shape when a pressure wave is forming. After 13 s of the run the second hump rolls along the length of the hull and passes under the transom. Dand et al. (2008) then stated that the cycle begins again and a new pressure wave forms. They reported a cycle time of 30 s. This is not the case with the pulsing motion measured here; the cycle time is much quicker than 30 s. The video shows approximately 10 waves pulsing along the hull from 13 s into the run to 22 s so the period of the pulsing motion is approximately 1 s. The pulse only takes around 0.1 s to pass along the hull then there was a 0.9 s wait while another pressure wave built up. Dand et al. did note that the pulsing motion only occurred on very at water and there were very small waves during this experiment. Dand et al. (2008) also reported that by increasing the shear modulus of the hull fabric they removed the pulsing motion. It appears that increasing the shear modulus did not completely remove the pulsing motion and anecdotal evidence from the helmsmen reported that the pulsing motion does occasionally occur.
This implies that increasing the shear modulus considerably decreased
the period of the pulsing motion and made it less noticeable to the helmsmen but did not completely remove it. Therefore, the author suggests that the pulsing motion measured here is distinctly dierent to the pulsing motion measured by Dand et al. (2008).
142
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
(a) Static shape
(b) Changing in pulsing hull shape at S2.25 K2 C3 = mean pulsing - static.
(c) Mean pulsing hull shape
(d) Changing in hull shape at S2.25 K2 C3 = quasi-steady non-pulsing - static.
(e) Quasi-steady hull shape S2.25 K2 C3
Figure 5.17: Mean quasi-steady planing hull shape during the pulsing at water run.
5.2.
WAVE TRIALS
143
Figure 5.18: Hull shape prior to the dynamic pulsing motion.
These results have shone new light on the pulsing motion. Firstly, these results suggest that the pulsing motion increases the speed of the boat compared with non-pulsing. This contradicts the pulsing motion described by Dand et al. (2008); however, the anecdotal evidence from the helmsmen is that pulsing initially decreases the speed and as the pressure wave is released the boat surges forward reaching a higher top speed, before the pressure wave builds up again resulting in a speed reduction. This could suggest that the pulsing motion measured during this experiment was not slowed down by the pressure wave but, instead, allowed to sustain a higher speed by continuously surging forward o the pressure wave. Further work will be required to prove this. These results also revealed that the stresses and strains within the D-class, whilst pulsing, which has never been captured before. In section 2.2.3 it was suggested that the structure of the D-class was statically indeterminate which means that the structure could deform into a number of dierent shapes to still achieve equilibrium. Lewis (2003) also discusses how a BTM can have multiple shapes to achieve equilibrium. It was shown earlier that the deck strains during the pulsing motion were less than those generated during the non-pulsing runs. It was also hypothesised that the mean pulsing hull shape had a more uniform and weaker stress distribution than the non-pulsing quasi-steady hull shape. Therefore, the author hypothesises that the structure has taken on an unstable equilibrium position of minimal potential energy that results in a change in performance. This could suggest that the pulsing motion may only occur with a complex structure like the D-class because all the components achieved a lower potential energy, i.e. a simple fabric planing plate may not exhibit the pulsing motion because the other components are not present.
5.2 Wave trials The wave trials were performed approximately 4 miles due East of Cowes harbour entrance on the
27th
and
28th
of March 2013. The wave data was recorded by Lymington wave buoy
144
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
and ChiMet weather station. The mean wave height, maximum wave height, signi cant wave height, mean wave period and peak wave period are shown in tables 5.14 and 5.15.
Seas
Pressures (psi) Boat 1 vs Boat 2
Head
3 vs 3 psi
Following
2 vs 3 psi
3 vs 3 psi 2 vs 3 psi
Head
3 vs 3 psi
Following
4 vs 3 psi
3 vs 3 psi
4 vs 3 psi
Mean wave
Max wave
Mean wave
Peak wave
height (m)
height (m)
period (s)
period (s)
1
0.05
0.3
2.8
2.1
2
0.05
0.3
2.8
2.1
3
0.05
0.3
2.8
2.1
1
0.03
0.2
2.5
2.7
2
0.03
0.2
2.5
2.7
3
0.03
0.2
4
2.2
1
0.05
0.3
2.8
2.1
2
0.05
0.3
2.8
2.1
3
0.03
0.2
2.5
2.7
1
0.03
0.2
2.5
2.7
1
0
0.1
-
2.2
2
0
0.1
-
2.2
3
0
0.1
-
2.2
1
0.03
0.2
2.5
2.7
2
0
0.1
-
3.1
3
0
0.1
-
3.1
1
0
0.1
-
2.2
2
0
0.1
-
2.2
3
0
0.1
-
2.2
1
0.03
0.2
2.5
2.7
2
0
0.1
-
3.1
3
0
0.1
-
3.1
Run
Table 5.14: Wave data recorded by Lymington Wave Buoy.
The wave trials were set up so that during one session two boats were driven side by side, a reference boat and a variable boat, and 12 runs were performed (6 in head seas and 6 in following seas).
The rst half of the runs were performed with the boats operating at the
same internal pressures. It was not expected that the two boats would perform identically, even at the same internal pressures, because the boats deteriorate with age dierently and the driver's skill has a large impact. So the rst half of the runs were used to gauge the dierence between the two boats at the same internal pressures. The second half of the runs were used to measure the dierence at dierent internal pressures. The dierence in the rst half of the runs was compared to the dierence in the second half of the runs to nd out if the internal pressures had aected the performance. If there was signi cant dierence between the change in the variable boat and the reference boat then the internal pressures had aected the boat motion. Two sessions were performed: the rst session compared 3 psi vs. 4 psi then 3 psi vs. 3 psi, and the second session compared 3 psi vs. 3 psi then 3 psi vs. 2 psi. The boat performance in waves was measured using accelerometers and this can be used to estimate the WBV according to the European directive 2002/44/EC. If the crest factor is above six then the VDV should be used, instead of the RMS, to estimate the WBV. The
5.2.
WAVE TRIALS
Seas
Pressures (psi) Boat 1 vs Boat 2
Head
3 vs 3 psi
Following
2 vs 3 psi
3 vs 3 psi 2 vs 3 psi
Head
3 vs 3 psi
4 vs 3 psi
Following
145
3 vs 3 psi
4 vs 3 psi
Mean wave
Max wave
Signi cant wave
Mean wave
height (m)
height (m)
height (m)
period (s)
1
0.18
0.2
0.23
11.13
2
0.15
0.16
0.21
12.19
3
0.21
0.24
0.21
12.19
1
0.21
0.22
0.19
13.48
2
0.25
0.25
0.19
13.48
3
0.19
0.2
0.19
15.06
1
0.19
0.2
0.23
11.13
2
0.19
0.21
0.21
12.19
3
0.22
0.22
0.22
12.19
1
0.21
0.22
0.19
13.48
1
0.18
0.19
0.2
17.07
2
0.17
0.19
0.2
17.07
3
0.18
0.19
0.26
5.95
1
0.16
0.19
0.16
17.07
2
0.15
0.16
0.16
4.66
3
0.17
0.18
0.16
4.66
1
0.16
0.19
0.2
17.07
2
0.17
0.19
0.26
5.95
3
0.17
0.18
0.26
5.95
1
0.16
0.16
0.16
17.07
2
0.18
0.19
0.16
4.66
3
0.15
0.17
0.17
5.23
Run
Table 5.15: Wave data recorded by ChiMet.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
Pressures (psi)
Run
CHAPTER 5.
Seas
146
Boat 1 vs Boat 2
Head
3 vs 3 psi
Following
2 vs 3 psi
3 vs 3 psi 2 vs 3 psi
Head
3 vs 3 psi
Following
4 vs 3 psi
3 vs 3 psi
4 vs 3 psi
Crest Factor Variable boat
Reference boat
Transom
Crew
Transom
Crew
1
5.4
15.4
17.4
11.7
2
6.2
17.4
7.1
17.2
3
7.1
21.4
15.8
16.2
1
6.1
19.4
15.4
17.0
2
5.8
15.5
12.8
15.9
3
5.9
17.5
11.6
13.5
1
5.4
23.8
4.3
14.3
2
5.5
18.2
5.8
16.7
3
6.4
19.3
5.8
18.9
1
5.5
19.9
7.2
17.5
1
4.6
15.4
5.3
13.0
2
4.5
15.6
4.9
12.7
3
4.3
15.1
7.6
13.9
1
6.0
19.9
5.3
14.7
2
5.6
18.5
6.7
14.0
3
4.3
15.7
13.2
18.0
1
4.8
20.4
3.5
12.9
2
4.2
18.2
3.1
12.4
3
5.0
12.1
6.9
14.1
1
5.3
17.4
4.6
12.2
2
5.3
21.7
3.3
16.5
3
3.9
19.4
3.2
16.9
Table 5.16: Crest factors from the wave trials.
crest factors for all the runs can be seen in table 5.16. It shows that the crest factors were much higher in the crew position than at the transom and this is expected due to the large pitching motion at the bow. The results showed that some runs the crest factors were below six; therefore, the RMS and VDV should be calculated to conform to the European directive. The RMS was calculated rst and the results are shown in table 5.17. The RMS results, as a whole, show that the transom experiences higher levels of vibration than the crew position; however, this is not a fair assessment of the crew position because the crest factors were always above six. Almost all the runs exceeded the Exposure Action Value (EAV) of 0.5
m/s2
in both
2 locations and most of the runs exceed the Exposure Limit Value (ELV) of 1.15 m/s at the transom.
This shows that the levels of vibration in the D-class in waves associated with a
Beaufort force three to four exceed the limits set out by the European directive within three minutes. The RMS values show that there is a variation between the boats and a variation between each run but did the internal pressures statistically aect the RMS values? A Student's T-test was used to compare the dierence (i.e. variable boat - reference boat) between the runs at dierent pressures.
Only one set of runs was shown to be statistically dierent with 95 %
certainty and that was the following seas comparing
3 psi vs. 3 psi and 4 psi vs. 3 psi at
the transom. The RMS is only appropriate for the transom values so the VDV will now be
WAVE TRIALS
147
2
Pressures
Run
Seas
5.2.
(psi) Boat 1 vs Boat 2
Head
3 vs 3 psi
Following
2 vs 3 psi
3 vs 3 psi 2 vs 3 psi
Head
3 vs 3 psi
Following
4 vs 3 psi
3 vs 3 psi
4 vs 3 psi
RMS (m/s ) Variable boat
Reference boat
Variable - reference
Transom
Crew
Transom
Crew
Transom
Crew
1
1.93
0.84
0.83
0.92
1.099
-0.073
2
1.26
0.71
1.07
0.82
0.191
-0.111
3
1.26
0.68
0.91
0.73
0.355
-0.048
1
1.43
0.66
0.70
0.68
0.723
-0.022
2
1.57
0.65
0.90
0.66
0.674
-0.016
3
1.94
0.62
0.65
0.71
1.294
-0.090
1
1.62
0.48
1.00
0.50
0.621
-0.027
2
1.52
0.47
0.53
0.48
0.995
-0.015
3
1.55
0.48
0.50
0.35
1.045
0.129
1
2.01
0.55
1.03
0.58
0.975
-0.029
1
2.02
0.58
1.02
0.69
1.003
-0.108
2
2.28
0.57
1.11
0.67
1.172
-0.099
3
2.41
0.60
0.60
0.71
1.180
-0.144
1
1.59
0.59
1.41
0.66
0.174
-0.069
2
1.76
0.62
1.15
0.69
0.601
-0.075
3
1.84
0.61
0.71
0.73
1.132
-0.113
1
2.11
0.47
0.99
0.50
2
2.38
0.50
1.29
0.57
3
2.25
0.65
0.50
0.48
1
1.96
0.51
1.19
0.59
2
2.02
0.51
1.19
0.62
3
2.10
0.52
1.34
0.57
1.118 1.087 1.744 0.768 0.832 0.769
-0.035 -0.064 0.171 -0.078 -0.103 -0.041
Table 5.17: RMS values from the wave trials.
investigated. The VDV for all the wave trials are shown in table 5.18.
These values have not been
weighted according to the European directive because human weighting might remove parts of the frequency spectrum that the internal pressures have aected. These values cannot be compared to the EAV and ELV of the European directive, and the weighted VDV will be discussed in section 5.3.1. A Student's T-test was used to prove whether or not the internal pressures had statistically aected the VDV. The following seas comparing
3 psi vs. 3 psi
4 psi vs. 3 psi at the transom was statistically dierent to a 95 % certainty; the same as the RMS results. The head seas with the same internal pressures (3 psi vs. 3 psi and 4 psi vs. 3 psi) at the transom were also statistically dierent to a 90 % certainty. These results and
suggest that increasing the internal pressures from 3 psi to 4 psi will reduce the unweighted VDV at the transom; however, this could still be misleading because the variation to cause the statistical dierence came from the reference boat. Also the signi cant wave height measured by ChiMet changed from 0.16 m to 0.22 m. It is more likely that the statistical dierence was caused by the changing sea condition aecting the helmsmen's driving pattern. Another method to assess the eect of hydroelasticity on the boat motion is to use peak acceleration counting, such as Myers et al. (2011).
The top 180 vertical peak magnitude
accelerations with a minimum separation of 0.25 s for each 3 minute run were found using
Pressures (psi)
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
Run
CHAPTER 5.
Seas
148
Boat 1 vs Boat 2
Head
3 vs 3 psi
Following
2 vs 3 psi
3 vs 3 psi 2 vs 3 psi
Head
3 vs 3 psi
Following
4 vs 3 psi
3 vs 3 psi
4 vs 3 psi
Unweighted VDV (m/s Variable boat
Reference boat
−1.75 ) Variable - reference
Transom
Crew
Transom
Crew
Transom
Crew
1
87.8
58.5
46.7
58.9
41.2
-0.3
2
58.3
51.2
50.5
58.1
7.8
-6.9
3
59.4
51.2
50.1
49.9
9.2
1.2
1
68.2
48.4
36.3
45.8
31.9
2.5
2
74.4
43.5
45.9
42.4
28.5
1.0
3
88.3
39.4
33.4
44.3
54.9
-4.9
1
75.1
33.2
46.2
30.1
28.9
3.1
2
70.8
28.6
25.1
28.0
45.6
0.5
3
72.8
30.4
24.1
23.9
48.7
6.5
1
91.7
35.7
48.9
35.8
42.8
0.01
1
91.3
35.6
50.0
42.7
2
101.0
35.3
52.2
40.9
3
106.0
35.7
32.9
42.5
1
73.2
40.1
67.8
43.6
2
79.3
41.9
54.7
43.2
3
83.3
38.7
42.1
47.2
1
95.0
27.8
45.6
28.6
2
106.2
29.7
58.3
33.0
3
100.9
92.4
27.7
27.7
1
91.9
32.2
61.1
36.5
2
91.2
33.8
55.0
37.4
3
94.2
38.7
60.5
33.8
Table 5.18: Unweighted VDV from the wave trials.
41.3 48.7 73.1 5.4 24.6 41.2 49.4 48.0 73.1 30.8 36.3 33.7
-7.1 -5.5 -6.8 -3.5 -1.3 -8.4 -0.8 -3.3 64.7 -4.3 -3.7 4.9
5.2.
WAVE TRIALS
149
ndpeak in Matlab. The magnitude acceleration was found by calculating the 180 top peaks of the positive and negative signals separately, then taking the top 50 % of the positive and negative peaks. This ensures the largest positive and negative peaks are included. 180 peaks were chosen because the wave encounter frequency was observed to be approximately 1 Hz. The acceleration data was un ltered to ensure the frequencies that hydroelasticity may have aected were not removed. The peak acceleration counting histograms are shown in gures 5.19 to 5.24 and each histogram shows the sum of the three repeats run at each internal pressure condition (totalling 9 minutes). The reference boat is shown in blue and the variable boat is in red. The dierence between the reference boat and variable boat ( gure (b) (a)) are shown in gures 5.19c to 5.24c. The peak counting results for head seas comparing 3 psi and 4 psi internal pressures show considerable dierence between the performance at the transom of the variable and reference boats when operating at the same internal pressures, see gure 5.19a. The majority of the peaks acceleration measured in the reference boat are between 2 g and 4 g but the peak accelerations of the variable boat were between 5 g and 7 g.
The performance became more
similar when the two boats operated at dierence pressures, see gure 5.19b. The main peaks of the variable boat reduced to between 4 g and 6 g and the reference between increased to between 2 g and 5 g, shown in gure 5.19c. This was not expected and the same trend was measured at the transom in head and following seas and at dierent internal pressures, see gures 5.21 and 5.23. Clearly, the performance of the reference and variable boats should be more similar than when operate at the same internal pressures than operating at dierent internal pressures. Reasons for the dierence in performance at the same internal pressures include helmsman control, wave spectrum and baseline boat performance (the dynamic response of a D-class to waves at 3 psi internal pressures). The encountered wave spectra were controlled by driving the two boats side-by-side but the wave spectra are never identical. The helmsmen were instructed to hold a straight course and adjust the throttle as little as possible to minimise their eect on boat performance. This leaves the baseline boat performance and it is possible that the two boats have aged and deteriorated dierently. The RNLI have reported that the performance of the D-class does deteriorate with time and use, see Dand et al. (2008). Furthermore, it was shown in section 5.1.1 that the boat used for the at water trials, which was the variable boat for the wave trials, did not reach the expect top speed of 25 knots and it was suggested that the age and deterioration of the boat (probably plastic deformation of components) caused the reduced performance. The author hypothesis that it is the aging of the two boats that has caused the dierence and, in addition, the RNLI have reported that the D-classes do aged dierently. The purpose of the four stage full-scale experiments is to allow knowledge gain through
1
the previous sub-experiments to provide insight into the latter experiments ; mainly using the
1
It was never the intention of the thesis to full analysis the structural deformation data from the wave trials because real seas were used and the wave spectra can only be described from distance weather stations; therefore, the wave energy in equation 5.1 is irregular and poorly quanti ed. It has also been shown that the coe cient of hydroelasticity will most likely vary by an unknown quantity in equation 5.1. This means the structural deformation under investigation is a varying signal disguised in a very noisy background signal.
150
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
(a) Same internal pressures.
(b) Varied internal pressures.
(c) Dierence between same and varied internal pressure; gure (b) - (a). Figure 5.19: Transom acceleration peak counting in head seas comparing the performance at 3 psi and 4 psi internal pressure.
5.2.
WAVE TRIALS
(a) Same internal pressures.
151
(b) Varied internal pressures.
(c) Dierence between same and varied internal pressure, gure (b) - (a). Figure 5.20: Crew acceleration peak counting in head seas comparing the performance at 3 psi and 4 psi internal pressure.
152
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
(a) Same internal pressures.
(b) Varied internal pressures.
(c) Dierence between same and varied internal pressure, gure (b) - (a). Figure 5.21: Transom acceleration peak counting in head seas comparing the performance at 2 psi and 3 psi internal pressure.
5.2.
WAVE TRIALS
(a) Same internal pressures.
153
(b) Varied internal pressures.
(c) Dierence between same and varied internal pressure, gure (b) - (a). Figure 5.22: Crew acceleration peak counting in head seas comparing the performance at 2 psi and 3 psi internal pressure.
154
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
(a) Same internal pressures.
(b) Varied internal pressures.
(c) Dierence between same and varied internal pressure, gure (b) - (a). Figure 5.23: Transom acceleration peak counting in following seas comparing the performance at 3 psi and 4 psi internal pressure.
5.2.
WAVE TRIALS
(a) Same internal pressures.
155
(b) Varied internal pressures.
(c) Dierence between same and varied internal pressure, gure (b) - (a). Figure 5.24: Crew acceleration peak counting in following seas comparing the performance at 3 psi and 4 psi internal pressure.
156
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
at water trials and drop tests to understand the wave trials results. For example, the at water trials showed that the forward speed of the variable boat was lower than the report forward speed, suggesting that age has deteriorated the performance of the variable boat. The quasi-2D and full-scale drop tests showed that the eect of internal pressures of the sponson and keel on peak accelerations was dependent on drop height. The general trend of both drop tests showed that the internal pressures or hull stiness increased the peak accelerations at one drop height but decreased the peak accelerations at the other drop height (i.e. the eect of hydroelasticity inverted with drop height).
This has two implications which will now be
discussed. First, the drop height of the drop tests can be compared to the wave height during the wave trials because the wave height will determine the height at which a hull can loss contact with the water before slamming. Therefore, in a real, random sea the drop height will randomly vary because the wave height randomly varies. This causes the eect of hydroelasticity, internal pressures and/or hull stiness to invert randomly with wave height. For example, if we simplify hydroelasticity and consider that hydroelasticity aects the energy in the boat motion by a simple multiplying coe cient, then we could say;
Eboat = Ewave × HE Where:
Eboat
= boat motion energy,
Ewave
= wave energy and
(5.1)
HE
= coe cient of hy-
droelasticity. In this example the coe cient of hydroelasticity inverts with the wave height; similar to the drop test results that were inverted with drop height.
The coe cient of hy-
droelasticity will be constant if the boat is in regular waves with long time periods and the energy in the boat motion will have a constant oset that could be statistically proven; however, if the waves are irregular then the coe cient of hydroelasticity will irregular invert causing the data to be irregularly scattered and not have a constant oset. The eect of the inversion in term of WBV will be discussed further in section 5.3.3. The second implication of the drop test results is that only the slamming accelerations of the variable boat were measured. It was shown earlier that the forward speed of the variable boat has deteriorated with age and it is likely that the reference boat has also deteriorated but it a dierent degree. Therefore, it would be a false assumption to apply the drop test results to the reference boat because there is no proof that performance of the reference and variable boats are the same.
This deduction is reinforced with the peak counting results where the
performance of reference and variable boat had greater dierence at same internal pressures than at dierent internal pressures.
It would also be an unfair to assume the drop height
that causes the eect of hydroelasticity to invert is the same in both boats. Furthermore, it is unknown how often the eect is inverted with drop height because only two drop heights were tested. The inversion might occur at a number of drop heights due to superposition of the natural frequencies of interacting structural components, see Bishop and Price (1979). All of the peak counting results from the transom show that the performance of the reference and variable boats were more similar when the boats were at dierent internal pressures than the same internal pressures. Another reason for the change in boat motion is the wave
5.2.
WAVE TRIALS
157
pattern and has been assumed the same for both boats during each run because the boats were side-by-side; however, the wave spectrum was not the same when the internal pressures were the same and varied (i.e. gure (a) and (b) were measured over dierent wave conditions). The wave data measured at ChiMet shows that the signi cant wave height during the trials at dierent internal pressure was always below 0.2 m; however, the signi cant wave height was always above 0.2 m when the trials had the same internal pressures, see table 5.15. This presents a dilemma because the eect of hydroelasticity might have inverted due to drop height in between the comparison trials but it is unknown. Therefore, the knowledge gained during the drop tests and at water trials has shown that it is unjusti ed to compare the eect of internal pressures during wave trials because the wave height (and equivalent drop height) has varied, which might cause the eect of hydroelasticity to invert. A systematic series of drop tests with far more drop heights must be performed on both boats to prove whether or not the eect of internal pressures can be compared. It was initially thought that hydroelasticity would have a constant eect on the peak accelerations when the experimental method was developed; therefore, the data would contain a constant oset that could be seen over any chances in the wave spectra. Thus the magnitude of the wave spectrum was not critical to prove the eect of hydroelasticity. It is now known from the drop tests that the eect of hydroelastic on the peak acceleration can invert with drop height, which means the magnitude of the wave spectrum is critical. The drop test results have provided an understanding of the wave trials, which was not known at the beginning of the thesis. The peak counting results discussed so far have all been regarding the accelerations at the transom. The peak counting results for the crew accelerometer are shown in gures 5.20, 5.22 and 5.24, and they shows more consistency between the reference and variable boats than the transom accelerometer. For example, the crew peak counting in head seas comparing 3 psi and 4 psi shows the majority of the peak are between 2 g and 4 g for both the reference and variable boat, see gures 5.20a and 5.20b; plus there is minimal change when either the internal pressures of the variable boat are changed or when the signi cant wave height changed, see gure 5.20c. The head sea results comparing 2 psi and 3 psi, and the following sea results both agree with the previous head sea results. There are some dierence between the reference and variable boat in the crew position but there is more variation in the reference boat than the variable boat. This strengthens the hypothesis that the boats have aged dierently because the baseline boat performance is dierent. The dierences between transom and crew peak counting results are striking; the transom observed large a dierence between the reference and variable boats and a large variation due to wave height, whereas the crew experienced similarities between the reference and variable boats and small but similar variation due to wave height. This indicates that there is an event occurring at the aft of the D-class that does not occur at the crew position and is dierent between the reference and variable boat.
The aging of the two boats has been discussed
and it is possible that the hull and keel at the aft of the boats have plastically deformed over their lifetimes but it is not possible to prove this without knowing the deformation of the
158
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
reference boat. There was another parameter at the aft that was dierent between the reference and variable boats; the mass distribution of the crew members shown in table 3.4.
The
main dierence between the reference and variable boats was body weight of the helmsmen, totalling 25 kg. This lead to a 3.4% dierent in total mass between the reference and variable boats (assuming total mass is 436 kg + 270 kg + 36 kg; lightship minus crew, crew and instrumentation box, respectively). The reported LCG is 1.452 m from the transom (including instrumentation box, see section 3.5.2) so the dierence in masses of the helmsmen would cause the LCG to translate forward 13.5 cm to 1.588 m from the transom because the helmsman's mass of the reference boat was less than the average crew member. Townsend (2008) measured the eect of increasing the ballast in the bow of the RNLI Atlantic 75 and 85 RIB and showed that increasing ballast and moving the LCG forward reduced RMQ bow acceleration and RMQ angular velocity. This could explain the dierence between the reference and variable boats; however, the dierence in helmsmen caused a decrease in ballast (opposite to Townsend (2008)) and moved the LCG forward so the evidence is not conclusive. Newton's second law would suggest that a decrease in mass would lead to an increase in acceleration but the mass also aects the input force to the slam. The author suggests that the dierence in reference and variable boats peak counting at the transom with the same internal pressures is a combination of the dierence mass distribution and aging dierences.
Hydrodynamic planing crafts and the D-class have a non-linear response to waves, see Fridsma (1969) and Dand (2004) respectively, so it could be argued that the opposite response of the reference and variable boats to an increase in wave height at the transom peak counting, see gures 5.19c, 5.21c and 5.23c, is due to the non-linear hydrodynamic response.
Dand
(2004) did show a secondary response to regular wave with a rigid scale-model of the D-class; however, this secondary response was only measured in heave and not pitch.
Therefore, if
the secondary response measured by Dand caused the opposite response to increase in wave height, then the crew peak counting should also show this opposite response but the crew results do not show this. The crew accelerometer was 1.475 m from the transom, see gure 3.5, and the LCG is predicted to be 1.452 m from the transom, thus the crew accelerometer will measure minimal angular acceleration if the D-class rotates about its LCG. Although, a planing craft does not normally rotate about the LCG; instead the centres of resistance and thrust are also important. Fridsma (1969) ( gures 23 and 24) and Begovic et al. (2014) ( gures 6 and 7) both showed that increasing the wave height lead to proportionally similar increases in bow and centre of gravity accelerations using similar scale-models of a prismatic hull with a 16.7° deadrise angle in regular and irregular seas, respectively. This indicates that neither the heave or pitch response (i.e. rigid body motion) has caused the dierence between the transom peak counting results because, if the previous statement was true, the change in response would have been measured in both transom and crew accelerometers. Therefore, a hydroelastic vibration must have aected the boat motion if the evidence indicates it is not a hydrodynamic vibration. Typical hydroelastic modal shapes of ships are shown in Hirdaris and Temarel (2009) and Bishop and Price (1979). It is dierence to predict the modal shape
5.2.
WAVE TRIALS
159
2 but it could be very loosely compared to the open ship shown in gure 3 of
of the D-class
Hirdaris and Temarel (2009). The minimal change in crew peak counting shows that the crew accelerometer is located near a node of the modal shape. The crew accelerometer was located o-centre, in between the crew's knees, suggesting the modal shape was not caused by torsional twisting but, instead, longitudinal vertical bending. It was intended that the drop test results what provide understanding to the wave trials and they have done; however, the drop test results do not agree with the wave trials. The full-scale drop tests show minimal change at the transom but maximum change in the crew, whereas, the wave trials show the opposite. Townsend (2008) highlighted the important of bow up or bow down impacts on slamming accelerations, and the drop tests only explored bow up impacts. The excitation force in the two sub-experiments (drop tests and wave trials) was dierent so that dominant modal shape will be dierent.
Therefore, the hydroelastic modal shapes that caused the change in boat
motion during the wave trials can only be hypothesised. The nal reason that could cause the change in response to an increase in wave height is the response of the helmsmen, such as fatigue and experience. The order of wave trial runs means that fatigue could not cause the change in response (on day one the signi cant wave height increased and on day two the signi cant wave height decreased).
Experience is also
unlikely because the more experienced helmsman was in the variable boat which responded poorly with an increase in peak accelerations to an increase of signi cant wave height but the less experienced helmsman responded well. The dierence in helmsmen response to increase in wave height cannot be ruled out but it is unlikely. The wave trials started by using the RMS and VDV to investigate whether or not hydroelasticity could have aected the boat motion. There were some statistical dierences but the statistical dierences came from the reference boat not the variable boat, leading to inconclusive proof. Peak counting of the vertical accelerations was employed to further explore the eect of hydroelasticity on boat motion.
The peak counting revealed considerable dif-
ferences between the reference and variable boat at the transom and this was linked to the dierence in signi cant wave height. It was hypothesised that the dierences were caused by the dierent mass distributions and aging of the trial boats. Then, it was demonstrated by empirical comparison that a variable in the hydrodynamic performance should be measured at the transom and crew accelerometers, which suggests there dierence in boat motion was caused by a hydroelastic vibration. Finally, it was hypothesised that the hydroelastic modal shape was longitudinal vertical bending with a node orientated near the crew accelerometer. Unfortunately, open-water wave trials provide an uncontrolled environment (which was identi ed in section 3.3.2) making it di cult to rule out certain variables, such as helmsmen control, but the evidence has provided interest hypothesises that justify further work.
2
Analytical models of the D-class structure were investigated at early stages of the project but are not currently capable of solving fabrics structures (BTM) and compression hinges (i.e. hinge stiness depends on friction driven be the compression forces).
160
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
5.3 Hydroelastic discussion Part 2 5.3.1 Whole body vibration of wave trials The VDV calculated during the wave trials so far were not weighted because this may remove part of the frequency spectrum that hydroelasticity might have aected. The crew of the Dclass are kneeling and the European directive 2002/44/EC only provides weightings for humans in a seated or standing position. It is anticipated that a weighting for a kneeling position would be in between the seated and standing weightings because the knees and hips are still able to rotate whilst kneeling. The weighted VDV for seated and standing are shown in tables 5.19 and 5.20, respectively. The rst observation is that the crew position experiences considerably higher levels of vibration than the transom (or helmsmen); in head seas the crew position VDV is on average 2.45 times greater than the transom VDV and in following seas it is 2.06 times greater. This is close to the observations of Dand (2004) which showed that the crew are exposed to twice the accelerations of the helmsmen in head seas. Next, it can be observed that the VDV during the head seas are on average 1.31 times greater than the following seas. Finally, the measured VDV can be compared to the EAV and ELV from the European directive which are 9.1
m/s−1.75
and 21
m/s−1.75 ,
respectively. If the seated weighting is used then, on
average, the EAV is always exceeded by the helmsmen and the ELV is always exceeded by the crew in three minutes in the sea state associated with a Beaufort force three to four. This could be viewed as the worst case scenario because the seated weighting is probably too high for a human in the kneeling position. If the standing position weighting is used then, on average, the crew just exceeds the EAV in three minutes in the sea state associated with a Beaufort force three to four. Either way, this is clear evidence that the levels of WBV experienced by both the helmsmen and crew rapidly exceed the levels set out by the European directive. A Student T-test was performed on the weighted VDV to nd out if the internal pressures had made a statistical dierence; however, no data sets were statistically dierent. This was expected because the weighting will removes part of the frequency domain that the internal pressures have aected and thus it makes data sets more similar. The VDV results can be validated against other VDV measurements in similar craft and the literature review identi ed two similar trials, see Allen et al. (2008) and Myers et al. (2011). There are a few dierences between the trials performed by Allen et al. and Myers et al., and the trials performed here. The trials performed by Allen et al. and Myers et al. were in sea states two to three but here the trials were performed in sea states low two and never reach a sea state three, see tables 5.14 and 5.15. The crafts were also dierent where Allen et al.
and Myers et al.
used RIB but the D-class is an IB. Next, the RIB used by
Allen et al. and Myers et al. have deep-vee hull shapes but the D-class has a relatively low deadrise angle, see table 5.8. The overall length of the D-class is 5.12 m but the Atlantic 75 is 7.5 m (Allen et al. (2008)) and the Artic 28 is 8.7 m (Myers et al. (2011)) overall length. The D-class and Atlantic 75 trials were performed at speed of 15 to 20 knots but the Myer et al. performed their trials at 40 knots. The nal dierence is the duration of the trials. The VDV is dependent on the duration of the trials so the D-class acceleration data was joined
Pressures (psi)
Run
Seas
HYDROELASTIC DISCUSSION PART 2
Boat 1 vs Boat 2
Head
3 vs 3 psi
Following
2 vs 3 psi
3 vs 3 psi 2 vs 3 psi
Head
3 vs 3 psi
4 vs 3 psi
Following
5.3.
3 vs 3 psi
4 vs 3 psi
161
Seated VDV (m/s Variable boat
−1.75 )
Reference boat
Transom
Crew
Transom
Crew
1
16.3
36.4
15.4
36.9
2
12.3
32.4
14.2
37.9
3
11.9
32.6
12.8
31.1
1
12.0
27.8
11.0
29.9
2
12.4
26.3
12.3
25.7
3
13.7
23.3
10.7
27.8
1
11.1
17.6
9.5
18.2
2
10.9
17.2
7.7
17.4
3
10.9
18.0
6.0
14.6
1
13.3
21.2
10.8
21.3
1
13.5
21.8
10.8
25.0
2
14.6
20.5
11.0
22.5
3
15.7
20.8
9.4
24.5
1
11.5
25.8
10.7
26.6
2
12.9
26.2
11.3
26.1
3
13.3
22.9
10.2
28.2
1
13.2
15.9
8.9
17.0
2
14.7
16.4
10.9
19.0
3
13.7
40.9
6.6
16.2
1
13.0
19.5
10.1
20.0
2
12.9
20.5
10.8
21.1
3
13.5
18.8
11.2
19.7
Table 5.19: Seated VDV from the wave trials.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
Pressures (psi)
Run
Seas
CHAPTER 5.
Boat 1 vs Boat 2
Head
3 vs 3 psi
Following
2 vs 3 psi
3 vs 3 psi 2 vs 3 psi
Head
3 vs 3 psi
4 vs 3 psi
Following
162
3 vs 3 psi
4 vs 3 psi
Standing VDV (m/s Variable boat
−1.75 )
Reference boat
Transom
Crew
Transom
Crew
1
6.5
14.6
6.2
14.7
2
4.9
13.0
5.7
15.2
3
4.8
13.0
5.1
12.4
1
4.8
11.1
4.4
12.0
2
4.9
10.5
4.9
10.3
3
5.5
9.3
4.3
11.1
1
4.4
7.0
3.8
7.3
2
4.4
6.9
3.1
6.9
3
4.4
7.2
2.4
5.8
1
5.3
8.5
4.3
8.5
1
5.4
8.7
4.3
10.0
2
5.9
8.2
4.4
9.0
3
6.3
8.3
3.8
9.8
1
4.6
10.3
4.3
10.7
2
5.2
10.5
4.5
10.4
3
5.3
9.1
4.1
11.3
1
5.3
6.3
3.6
6.8
2
5.9
6.6
4.3
7.6
3
5.5
16.4
2.6
6.5
1
5.2
7.8
4.1
8.0
2
5.2
8.2
4.3
8.4
3
5.4
7.5
4.5
7.9
Table 5.20: Standing VDV from the wave trials.
5.3.
HYDROELASTIC DISCUSSION PART 2
Allen et al.
Allen et al.
(trial 1)
(trial 2)
72
70
90
180
2
2
3
3
27.86
27.67
29.29
35.08
61.35
60.79
64.57
75.87
NA
25.94
48.51
57.05
Experimental trial
D-class
Trial duration (min) Sea state VDV of D-class transom using trial duration (m/s
−1.75 )
VDV of D-class crew using trial duration (m/s VDV of literature (m/s
−1.75 )
−1.75 )
163
Myers et al.
Table 5.21: VDV comparison between the D-class and literature.
together using a Hamming window to ensure the disjointed accelerations did not aect the VDV prediction.
A Hamming window was applied to the rst and last 0.5 s of each trial,
using the Matlab hamming function. Only the trials performed at 3 psi were used and both head and following seas were used because both Allen and Myers performed their trials in head and following seas. The joined together D-class acceleration data was either trimmed or reproduced so the duration equalled the duration completed by Allen et al. (2008) or Myers et al. (2011). The comparison of VDV results are summarised in table 5.21. There is 72 minutes of data when all the D-class trails at 3 psi are joined together, including both boats and both head and following seas; this is shown in the second column of table 5.21.
A comparison of the
D-class to trial one of Allen et al., shown in the third column, shows that the D-class produces a higher VDV than the Atlantic 75 at similar speeds and sea states. This is expected because the Atlantic 75 has a deep-vee hull and overall length is greater than the D-class. The second trial by Allen et al. and trials by Myers et al. agree with the D-class results. It shows that the VDV measured at the transom and crew position of the D-class lays either side of the VDV measured by Allen et al. and Myers et al. Therefore, the results show that the VDV and WBV experience by the D-class crew is particularly high but the VDV at the transom is less than the VDV in literature. This validates the VDV measurements by showing they are comparable to literature. The nal VDV comparison can be made in re ection of the wave trial peak counting results, see section 5.2.
It was hypothesised that the dierence in peak counting results
between the transom and crew was caused by a change in the rst mode of vibration.
If
this hypothesis is true, has the rst mode of vibration caused a measureable eect in the weighted VDV results? It was shown that the wave height had a greater eect on the peak counting results than the internal pressures; therefore, the dierent signi cant wave height results will be compared to prove whether or not hydroelasticity aected the VDV results. The mean weighted VDV were calculated for the variable boat and reference boat for the trials performed at two dierent signi cant wave heights from all the head and following seas trials. The results are shown in gure 5.25 and the error bars show the maximum and minimum VDV of each condition. The VDV results agree with the peak counting results; the reference and variable boats' performance was more similar with a signi cant wave height of less than
164
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
18.00
Weighted Standing VDV (m/s-1.75)
16.00 14.00 12.00
13.23
12.85
10.00
10.91
10.27
8.00 6.00 4.00 2.00 0.00 Variable boat
Reference boat
Variable boat
Significant wave height > 0.2 m Same internal pressures
Reference boat
Significant wave height < 0.2 m Varied internal pressures
Figure 5.25: Comparison of weighted VDV measured at the transom during the wave trials comparing dierence signi cant wave heights.
0.2 m than a signi cant wave height over 0.2 m. However, the results with wave height under 0.2 m lay within the maximum and minimum results measured with wave height over 0.2 m so the proof is not conclusive. A student T-test was performed to compare the variable boat at the two signi cant wave heights and to compare the reference boat at the two signi cant wave heights. A statistical dierence was found with a certainty of 90 % in both cases. This does not fully prove that hydroelasticity has aected the WBV experience by the crew of the D-class but it is certainty an indication that hydroelasticity has aected the WBV and justi es further work with the aim to reduce the WBV experience by the crew of high speed planing vessels.
5.3.2 Root cause to hydroelastic planing surfaces The at water trials proved that the internal pressures can aect the at water speed but what is the root cause of this eect and is there a dominant component? So far it has been shown that the sponson pressure varied considerably less than the keel pressure; however, a direct comparison is not fair because the diameter and volume of the sponson is considerably larger than the diameter and volume of the keel. We know;
P1 V 1 = P2 V 2 Where:
P1
pressure and
= undeformed pressure,
V1
V2 = deformed volume (i.e.
= undeformed volume (i.e. a circle),
P2
= deformed
an ellipse). This assumed the material is inextensible,
the air is incompressible and temperature changes are negligible. If the pressure has increased by
α,
where
α
= 1.30 for 30 % increase, then we can say;
5.3.
HYDROELASTIC DISCUSSION PART 2
Condition S2.25 K2
S3.25 K3
S4.25 K4
165
rc
(mm)
rx
Sensor
Percentage Change
(mm)
Sponson (psi)
1.9
254
4.8 20.2
Keel (psi)
34.6
58.5
Sponson (psi)
1.0
254
2.5
Keel (psi)
24.5
58.5
14.3
Sponson (psi)
0.6
254
1.5
Keel (psi)
16.9
58.5
9.9
Table 5.22: Deformation of the sponson and keel related to the change in internal pressure during the at water trials.
V2 = αV1 Where:
V1 = πrc2 L
= volume of a circle and
circumference of the circle (C
= 2πrc )
(5.2)
V2 = πrx ry L q=
and the ellipse (C
= 2π
volume of an ellipse. The
(rx2
+ ry2 )/2)
must be equal if
the material is assumed inextensible. Therefore;
ry = Where:
rc
= radius of a circle,
of an ellipse. So substitute
V1 , V2
rx
√
2rc − rx
= horizontal radius of an ellipse and
and
ry
ry
= vertical radius
into equation 5.2;
rc2 − 2αrc rx + αrx2 = 0
(rc − rx )(rc − αrx ) = 0 The rst solution (rc solution (rc
= αrx )
= rx )
(5.3)
of equation 5.3 is the undeformed solution and the second
is the deformed solution.
This proves that a percentage change in the
internal pressure is related to a percentage change in radius assuming the material is inextensible and the air is incompressible. The maximum diameter of the keel is 117 mm and the diameter of the sponson is 508 mm.
The deformation due to the percentage change in the
internal pressures of the sponson and keel are shown in table 5.22.
This con rms that the
deformation of the keel is considerably larger than the deformation of the sponson irregardless of the dierence in volumes. The static tests also showed that the overall shape of the D-class is more dependent on the keel pressure than the sponson pressure, see section 4.1.3.
So it
shows that the keel is likely to have a more dominant eect on the at water performance than the sponson; however, there are still three more components that could aect the at water performance. The deck has always been considered rigid when compared to the rest of the boat and this has been con rmed.
The central de ection of the aft deck panel changed by 3.8 mm
due to varying the internal pressures. This shows that the response of the deck has hardly changed due to the internal pressures and it is likely to have negligible eect on at water
166
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
performance. The deck hinges were shown to vary due to the internal pressures by no more than 0.6° so it is believed that the deck hinges also have a minor eect on the at water performance.
The nal component of the D-class is the fabric hull and it was shown that
this deformed by a maximum of 37.5 mm due to increasing the internal pressures from 2 psi to 4 psi and this could lead to a 3.6° change in deadrise angle. This change in deadrise angle could aect the at water performance. The eect of altering the deadrise angle due to the change in internal pressures on forward speed can be estimated using the Savitsky prediction, see Savitsky (1964). The Savitsky prediction was never intended for exible craft and its applicability to the D-class is very limited. The Savitsky prediction was developed for constant deadrise angle hulls with hard chines. The D-class has a variable deadrise angle hull with a sponson that could be compared to a round bilge. These are common limitations for most analytical or empirical methods so further work could involve using detailed numerical and CFD methods. Nonetheless, Dand et al. (2008) did compare the D-class to the Savitsky prediction and at approximately 20 knots the Savitsky prediction was very close to the scale model predictions so the Savitsky prediction will now be used to explore the at water trial results. The Savitsky prediction was calculated in MatLab and the script can be found in appendix K. The input values are a combination of measured results from this thesis and data from communication with Ian Dand for aspects that were not measured; these values are summarised in table 5.22. The rst set of input values match the measured at water trials results best as possible using the mean deadrise angle of all the hull sensors, forward speed and trim angle. The predicted drag from the measured input values increases by 34 N as the internal pressures increase; however, this is primarily cause by the increase in speed. If only the speed is varied then there is a 24 N predicted drag variation. The second set of input values in Table 5.23 varies only the deadrise angle and this reveals that the variation in deadrise angle due to internal pressures is predicted to change the drag by 3 N. This is a small quantity but the important detail is that higher internal pressures with larger deadrise angles have higher drag forces, which will reduce the speed. Thus, the Savitsky prediction con rms that large deadrise angles generate larger drag forces and in turn reduces the forward speed. This is opposite to the at water trial results and leads to the conclusion that it was not the change in deadrise angle (due to varying the internal pressures) that changed the forward speed but another factor or parameter of planing. These results indicate that the dominant components are the in atable keel and the fabric hull because they have the largest deformations but they are coupled together through the hydrodynamics of the hull. If the hull is assumed to be rigid and the mean deadrise angles variation, shown in table 5.8, is solely due to keel compression (and not the fabric hull deformation) then the predicted keel compression is 14.3 mm (using Pythagoras theorem). The predicted keel compression in table 5.22 is 10.3 mm (20.2 mm 9.9 mm) which shows that the change in hull shape is caused by both a change in keel compression and a change in the fabric hull deformation. Savitsky (1964) also de ned the pressure distribution on a at planing plate, see gure
5.3.
HYDROELASTIC DISCUSSION PART 2
Parameter (unit)
167
1) Measured
2) Only deadrise
values varied
angle varied
Source
2 psi
3 psi
4 psi
2 psi
3 psi
4 psi
13.05
14.02
14.39
13.05
14.02
14.39
Measured
Speed (knots)
21.49
21.73
21.93
21.73
21.73
21.73
Measured
Trim angle (°)
4.83
4.88
4.89
4.88
4.88
4.88
Measured
Mean deadrise angle (°)
Mass (kg)
655
Dand (2002b)
Beam (m)
1.903
Dand (2002b)
0.11
Dand (2002b)
Sinkage (m)
gure 13 Friction-drag
0.00177
Savitsky (1964)
1.422
I. Dand, private
coe cient Mean wetted length-beam ratio Drag
communication 1103
1124
1137
1122
1124
1125
Predicted
Table 5.23: Savitsky prediction input valves and predicted drag forces.
2.12. This shows that the peak pressure is in line with the stagnation point, just behind the front wetted edge. If the pressure distribution was applied to BTM with uniformed pretension stresses then it is expected that the peak deformation will be in line with the peak pressure. Dand reported (through private communication) that the wetted keel and chine lengths are 4.46 m and 0.96 m, respectively, at 22 knots.
Therefore, it is expected that the peak hull
deformation would form a diagonal line across the hull from the front wetted keel length to the front wetted chine length, if the fabric hull had uniformed pretension stresses. Instead, the hull shape in gure 5.6 shows a peak deformation, the concave shape transversely crossing the hull, at 0.9 m from the transom which is in line with the peak pressure on the chine. This indicates that the peak pressure on the chine is rotating the sponson, thus increasing the pretension stresses across the hull in between the peak pressures on the port and starboard chines (the sponson is glued to the hull and, both the hull and sponson are able to rotation around the edge of the deck), which in turn compresses the keel and forces the concave shape across the hull. Therefore, it is hypothesised the sponson causes the aft concave shape in the hull and that, although the sponson deformation is small, the sponson deformation is critically link to the hull shape. This hypothesis is further supported because the measured sponson rotation varied by only 1 mm due to varying the internal pressures, see table 5.4, and the concave shape in gure 5.6 did not noticeably change due to the internal pressures. In summary, the at water trial results have shown that the forward speed is aected by the internal pressure (and hydroelasticity,) and the deformation of all components is also aected by the internal pressures.
However, when the measured deformations are linked to planing
parameters, such as trim angle and deadrise angle, the predicted change in performance by the Savitsky prediction does not match the measured change in performance. This means that either the measured performance and deformations are incorrect or the parameter that aects the forward speed was not measured or identi ed. The measured performance (speed, trim
168
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
angle and deadrise angle) were validated against the results by Dand et al., which implies that the parameter aecting the forward speed was not measured or identi ed. This is an area for further work.
5.3.3 A hydroelastic planing craft The full-scale holistic experiment and the quasi-2D drop tests have shown that hydroelasticity can aect the performance of the D-class in a number of ways.
The question here is; can
hydroelasticity be used in a real sea to improve the performance of a planing craft?
The
performance will primarily be de ned by the WBV, at water speed and speed in waves. The rst step in being able to use hydroelastic slamming on a planing craft to reduce the WBV is to nd the root cause (i.e. more than just drop height) of the inversion in the peak acceleration trend. With our current knowledge, hydroelastic slamming will decrease the peak acceleration approximately half the time but the other half of the time hydroelastic slamming will increase the peak acceleration for a given hull geometry and stiness. If the root cause is found then it should be possible to optimise the craft to provide an overall reduction in WBV, so that on a few occasions hydroelastic slamming may increase the peak acceleration but overall hydroelasticity may reduce the WBV. The 2D tests provide an insight into the eect of deadrise angle on this interaction. The reduction in peak acceleration from MDF to fabric was larger with a deadrise angle of 5° than 15°; at 5° deadrise angle both drop heights showed a reduction but at 15° only one drop height showed a reduction. Also the percentage reduction was greater at 5° than 15° with a reduction of 18 % and 12 %, respectively. The reason for this is unknown and it could be due to other phenomena such as air cushioning, ow separation and air pockets occurring.
Nonetheless,
it suggests that a hydroelastic hull would be more suitable for crafts with shallow deadrise angles like the D-class or that the structural stiness should be varied with the deadrise angle for an optimal solution. So far hydroelastic slamming has been shown to be an advantage and a disadvantage so much further work is required to optimise it to provide an overall reduction in WBV; however, now compare it to the other solutions available to reduce the WBV on planing craft.
One
major disadvantage for suspension seats and suspension decks is the increase in weight of the craft but a exible hull does not increase the weight. In fact, the US navy have developed a craft where a composite hull was treated as a membrane surface and this allowed them to decrease the overall weight of the hull by 20 %, see Wood (2011). It was stated earlier that the density of the D-class fabric was 57 % lower than aluminium but the ultimate tensile strength was only 27 % lower which also proves that membrane structures could potentially be lighter. Although, the reaction forces on the structure of a exible craft may be considerably higher than found on a rigid craft.
To counter act this, the structure will have to be redesigned
and inevitably lead to a more complex and heavier structural design. This would increase the weight of the scantlings but the US navy still managed an overall weight reduction. A major advantage for a hydroelastic hull is its simplicity; nothing is added to the craft for it to be incorporated. Suspension seats have to be added to the craft which restricts crew
5.4.
SUMMARY
169
movement and raises the vertical centre of gravity, see Townsend et al. (2012a). Suspension decks require a highly complex system of springs and dampers, see Coe et al. (2013). Fins, interceptors and trim tabs all have to be added to the craft which will aect the hydrodynamic performance, increase the complexity of the t out and increase the weight. Hydroelasticity may aect the other performance indicators of a planing vessel including at water speed and the speed in waves. Forward speed is very important for planing craft and the at water trials showed that hydroelasticity could reduce the speed of the vessels by 0.44 knots. On the other hand, the pulsing motion was actually shown to increase the speed by 1.3 knots; although, this contradicts the previous results by Dand et al. (2008). Further work would be required to prove whether or not the pulsing motion could increase the at water speed. Nonetheless, these craft are used in heavy seas where the dominating factor to the speed in waves is actually the WBV experienced by the helmsmen and crew because they cannot cope with the slamming accelerations. So hydroelasticity may reduce the at water speed but the helmsmen may be able to sustain a higher top speed in waves. Advantages
Disadvantages
Potential to reduce the overall
Occasional increased in peak
WBV after optimisation
acceleration after optimisation
Increased speed in waves
Reduction to at water speed
Reduction in weight
Complex structural design
Simplicity
Increase in cost
No change to operational ability Table 5.24: Summary of the advantage and disadvantages of a hydroelastic slamming approach to reducing WBV.
Hydroelastic slamming cannot solve the problem of WBV in planing craft on its own with the current level of knowledge but it does show promise for being part of a combined approach to reducing the WBV. If hydroelastic slamming can reduce peak acceleration and the overall weight then the weight saving can be used to incorporate other WBV reduction strategies. This could lead to an overall reduction in WBV.
5.4 Summary The at water trials proved that the internal pressures of the sponson and keel can aect the forward speed with a loss of 0.44 knots due to reducing the pressures from 4 psi to 2 psi. The dominant components to the at water speed appear to be the keel and the hull. The pulsing motion was actually shown to increase the forward speed which contradicts previous research. The author hypothesises that the D-class has achieved an unstable equilibrium position of minimal potential energy and this is the reason for the pulsing motion. The wave trials were shown to be inconclusive because only one statistical dierence was found due to the internal pressures; however, this can be explained using the drop test results. The trend between the peak acceleration and the internal pressures was drop height dependent. In a real, random sea the drop height will vary causing peak acceleration to sometimes be
170
CHAPTER 5.
HYDROELASTIC RESULTS UNDER HORIZONTAL LOADS
increased by hydroelasticity and other times it will be decreased by hydroelasticity; therefore, the peak accelerations are more scattered but not consistently oset. The WBV during the wave trials were quanti ed using the VDV and this showed that the D-class will exceed the EAV and ELV in less than 10 minutes in the sea states associated with a Beaufort force three to four. This emphasises the need for a WBV reduction strategy. Finally, the possibility of a hydroelastic planing craft with the aim to reduce the WBV was discussed. Hydroelasticity might provide an overall reduction in WBV; although, occasionally hydroelasticity might increase the peak acceleration. It considered whether a membrane hull could be lighter than a conventional hull and this would allow other WBV reduction strategies to be incorporated. This combined solution might provide an overall reduction in WBV.
Chapter 6
Noise literature review Young and Miller (1960) measured the air and water borne noise generated by a 7.5 HP and a 18 HP outboard engine tted to an in atable boat. This is, perhaps, the most relevant and earliest paper relating to this project and is the only paper the author has found that considers both air and water borne noise together. The IB was driven in circles around the microphone and hydrophone to maintain a constant distance of 30m. The microphone was 3.4 m above the water surface and the hydrophone was placed 9 m below the water surface in a water depth of 18 m (ranging from 12 m to 27 m). Young and Miller showed that the water borne noise was about 25 dB greater than the airborne noise, but the dierence was frequency dependent. The measured air and water borne noise spectra for both engines are shown in gure 6.1. It was concluded that an increase in either speed or engine size could increase the air and water borne noise. No transmission loss or propagation model was used, however, the interactions with the seabed and sea surface were considered.
Figure 6.1: Air and water borne noise spectra of two outboard motors measured by Young and Miller (1960) (left - 7.5 HP Johnson AD-12, right - 18 HP Evinrude).
The Sound Pressure Level (SPL) is de ned as the mean square sound pressure divided by a reference acoustic pressure squared expressed in decibels, see Morfey (2001), and is shown in equation 6.1. Nowadays, it is conventional to reference airborne noise to 20 borne noise to 1
μPa;
μPa
and water
however, the work of Young and Miller reference both air and water 171
172
CHAPTER 6.
borne noise to 20
μPa.
NOISE LITERATURE REVIEW
This means that the water borne SPL cannot be directly related to
modern water borne measurements without re-referencing the measurement. Another modern convention is that water borne SPL is reported at one metre, whereas, airborne noise is reported at the measurement distance. This means that a propagation model is required for water borne noise to calculate the frequency dependent transmission loss and predict the SPL at one metre.
SP L = 10log Where;
P RM S
is RMS acoustic pressure and
2 Prms 2 Pref
Pref
! (6.1)
is reference RMS pressure.
6.1 Airborne noise 6.1.1 International standards The rst British Standard on the airborne noise emitted by a vessel on inland water-ways and harbours (BSI 22922:1993) speci es methods and conditions for acquiring reproducible and comparable measurements of the noise levels and spectra emitted by the vessels. It was revised into ISO 2922:2001 in 2001.
Between 2006 and 2009 the standards were updated again, in
accordance to the suggestions of the SoundBoat project, to an international standard called ISO 14509 (Small craft - Airborne sound emitted by powered recreational craft). Small crafts are de ned as vessels less than 24m in length.
ISO 14509 is divided into three parts with
dierent methods for predicting or measuring the noise emissions. These parts are: 1. Pass-by measurement procedures - ISO 14509-1:2008 2. Sound assessment using reference craft - ISO 14509-2:2006 3. Sound assessment using calculation and measurement procedures - ISO 14509-3:2009 Part 1 can be used to measure the airborne noise produced by the D-class from the shore or a second stationary vessel using the pass-by method.
The pass-by method simply involves
driving the source craft in a straight line pass a microphone and is a common method for both air and water borne noise. There are speci c constraints on the location and weather that restrict when and where the test can be performed. The idea of Part 2 is to allow new builds to be compared to a reference craft and if the new build is similar enough then the maximum SPL can be predicted. But it is not applicable for the testing of outboard motors which is currently tted to the D-class. Part 3 uses a combination of methods to predict the SPL at 25m. Firstly, the exhaust noise is measured from on board the vessel and, secondly, the hull-form noise is analytically predicted using equation 6.2. Finally the two measurements are combined to give an overall SPL. Equation 6.2 estimates the hull-form noise generated by vessels with a square transom, no more than two chines and a length greater than 11 m. So the D-class does not t the characteristics needed in Part 3, however, it does state that the industry is still collecting data to verify the procedures and equations.
6.1.
AIRBORNE NOISE
173
LpHF = 30log(vkn ) + 5log(L) + 20 Where:
LpHF
is SPL of hull form,
v kn
(6.2)
is vessel speed (knots) and
L
is Vessel length (m),
as de ned by ISO 8666.
6.1.2 Literature review The International Council of Marine Industry Association (ICOMIA) published several reports from 1993 to 2002 on the study of airborne noise produced by small craft, SoundBoat (2005). Lanpheer (1993) measured the pass-by noise of 16 boat and engine combinations ranging in boat length from 3.4 m to 4.7 m with 13 outboards and 3 stern drives.
Lanpheer (1993)
attempted to study the relationship between the hull noise and the engine noise by performing three pass-by noise measurements of towed vessels (i.e. no engine) and three stationary engine measurements; however, the correlation was poor. Lanpheer and Lassanske (1994) measured the pass-by noise of four boats, from 4.05 m to 5.5 m with one IB, and four outboard engines. The highest level measured was 72 dB(A) by a 5.5 m vessel. The measurements were aimed to validate a standard test method for outboard engines. Towed tests were also performed. Lanpheer (1998) reported the pass-by noise from large twin engine vessels, up to 18 m and 1400 HP. The measured noise ranged from 70 dB(A) to 76 dB(A) at 25 m and the noise from a sailing vessel was also measured at 57 dB(A). One boat showed excessively high noise readings (81 dB(A)) that had an above water exhaust. The results were used to suggest limits on the noise produced by new craft and match those in the current European directive 2003/44/EC. Lanpheer (1999) extended the earlier work of 1994 to included 18 stern drive boats with lengths of 4.8 m to 8.4 m.
The pass-by noise levels ranged from 69 dB(A) to 75 dB(A).
Lanpheer concluded that the hull form and construction had minimal eect on the measured noise providing the exhaust outlet is su ciently submersed. Mattiesen (2002) correlated the pass-by noise with the power/displacement ratio and the Froude number for 62 boats. Gao et al. (2008) used towed models to study the ow noise produced by a surface ship. The study showed that it is possible to measure realistic and reliable results from a towed model although it was stated that this work was only a starting point and further work is still required. The SoundBoat project was set up to develop practical and innovative methods to measure the airborne noise generated by recreational craft with the aim to publish the new methods as an addendum to ISO 14509. Prior to the work of SoundBoat (2005) little was known about the sound propagation over water with surface waves and wash.
The SoundBoat project
critically reviewed the pass-by measurement procedure of ISO 14509 and suggested speci c improvements for: the test sites with shore based microphones, test sites with vessel based microphones, wind speed and wave height. The new method was used to measure 80 boats, which ranged from 7 m to 23.6 m Length Over All (LOA) with a multitude of propulsion devices (single, twin and treble screw, stern drive and surface piercing propellers), and a database was produced containing the boat parameters and sound characteristics.
The new knowledge
174
CHAPTER 6.
NOISE LITERATURE REVIEW
gained regarding sound propagation over surface waves and wash was used to formulate an analytical equation to predict the hull-form noise, which promoted ISO 14509-3:2009. An interesting part of the work performed by SoundBoat (2005) was into the exhaust noise produced and the dierences between above water and below water exhaust outlets. It was found that below water exhaust outlets were considerably quieter than above water outlets so manufacturers may use this as a method to reduce the airborne exhaust noise, however, this is likely to lead to an increase in the water borne noise and, thus possibly, not make an overall reduction to the environmental impact.
6.2 Water borne noise Water borne noise has been given considerably less attention than airborne noise and this is often attributed to the fact that, as humans, we do not regularly hear this noise. First, the international standards will be reviewed and critiqued to show that they are not suitable for HSC in shallow water, and then a literature review will demonstrate the measurements and modelling methods suitable for RIBs and IBs in shallow water.
6.2.1 International standards There are three national and international standards for measuring the water borne noise of a ship;
American National Standards Institute (ANSI) S12.64-2009 - Quantities and procedures for description and measurement of underwater sound from ships - Part 1:
General
requirements.
Det Norske Veritas (DNV) - Rules for classi cation of new buildings, silent class notation, Part 6, Chapter 24, 2010.
ISO/PAS 17208-1:2012 Acoustics - Quantities and procedures for description and measurement of underwater sound from ships - Part 1: General requirements for measurements in deep water.
All three standards use a similar method which is very close to the standard method for measuring the airborne noise (ISO 14509-1:2008). The ship is driven in a straight line passed one or more hydrophones at a set Closest Point of Approach (CPA) and this is called the pass-by method. The arrangement of hydrophones varies depending on the standard and the level of accuracy required. The standards also set requirements for: the speci cation of the hydrophones and data acquisition system, the location, the weather conditions (wind, waves and rain), the background noise and post processing. ANSI S12.64-2009 provides three dierent grades for measuring the underwater noise of a ship with varying degrees of accuracy. There is no limitation to the size of the vessel but they were developed for ships and not small RIBs and IBs. The water depth and CPA are given as the greatest of; 300m or 3 x the ship length and 100m or 1 x the ship length, respectively,
6.2.
WATER BORNE NOISE
for a grade A measurement.
175
There is a 20 knots limitation for the wind speed because it
aects the sea surface, which eects both the background noise, noise propagation and the noise generated by the vessel. The recommended hydrophone geometry can be seen in gure 6.2. The background noise should be at least 10 dB less than the measured noise. The sound propagation model accounts for the spherical spreading of the sound and mitigates the eect of the sea surface sound wave re ections. The seabed re ections are not considered because the water depth is great enough so that they are negligible.
Figure 6.2: Hydrophone geometry suggested by ANSI S12.64-2009 (left - grades A and B, right - grade C).
The DNV silent class notation provides a standard for measuring the underwater noise from ships and they use a unique method involving a sloped seabed.
The geometry of the
hydrophone can be seen in gure 6.3, which uses the slope in the seabed to improve the accuracy by controlling the in uence of the seabed re ections and the standing waves between the seabed and sea surface. This should help to obtain a diused sound eld. It is recommended that the water depth is at least 30m but for HSC the water depth (d) is proportional to the vessel speed (v ), see equation 6.3. The D-class is capable of reaching 25 knots (12.8 m/s) so this would require a water depth over 100 m. The CPA should be 150m to 250m but the background noise should be at least 5 dB less than the measured noise. The sound propagation model accounts for spreading using a
18log.r
relationship, which is slightly dierent to a spherical spreading
model, and a seabed re ection correction of -5 dB is used if the hydrophone is within 0.2 m of the seabed.
d = 0.64v 2
(6.3)
The ISO is currently working with many stakeholders, including BSI, to develop a new international standard for the measurement of water borne noise from commercial ships and it is currently in the development stage of working group TC43. It will not be reviewed here because it is still under development. The standard methods reviewed so far were developed for much larger vessels than the RIBs and IBs considered in this project.
Parameters, such as water depth and CPA, and
176
CHAPTER 6.
NOISE LITERATURE REVIEW
Figure 6.3: Hydrophone geometry suggested by the DNV silent class notation.
considerations, like shallow water eects and wind speed limitations, are no longer applicable to vessels with a length less than 10 m and capabilities of speeds up to 40 knots.
6.2.2 Critique of international standards There are international standards for measuring the water borne noise generated by ships; however, these standards have been drawn up for a dierent class of vessel with considerably dierent boundary conditions. The reasons why the current standards are not applicable to the D-class and other small high small craft will now be discussed. The water depth surrounding the UK is a major obstacle for measuring the water borne noise because the UK waters are relatively shallow when compared to the depths considered within the standards.
ANSI and ISO stipulate a minimum of 100 m water depth which is
impractical around the UK; even the 30 m depth required by the DNV is di cult to nd in the UK and it requires a sloped seabed. To achieve a water depth greater than 30 m in the UK it may require the source craft to be driven for several hours to the deep water locations. In small high speed boats this can be highly impractical due to limitation of the human exposure to vibration, and the health and safety risks of operating a small vessel many miles from the coast. The craft could be transported (by another mode of transport) to a deep water location but this adds considerable cost and time to the experiment. Furthermore, it would be more appropriate to measure the noise in shallow waters because small vessels operate in shallow waters. It is interesting to note the required water depth is often de ned as a ratio to the ship's length. The reason behind this is unclear, however, it is expected that this is a safe guard for when very large vessels are considered, and not an indication that small vessel measurements can be made in shallower water. The sheer scale of ships in the standards is out of proportion to the scale of the vessels considered in this work.
The standards state that there is no limitation to the size of the
vessel but it is anticipated that the standards consider a vessels length from 30 m to over 100 m. Whereas, the RNLI RIBs and IBs range from 3.88 m to 10 m so the scale is out by over a magnitude of 10.
6.2.
WATER BORNE NOISE
177
The DNV silent class notation recommended that for high speed vessels the depth is proportional to the speed,
d ≥ 0.64v 2 .
DNV con rmed (by email) that this relationship is
purely hydrodynamic to remove any shallow water eects on a large ship.
An example of
these shallow water aects is squat which causes the vessel to sink into the water due to the low pressure between the hull and the seabed, see Rawson and Tupper (2001) page 578. This squat eect applies to large displacement vessels but not to planing vessels that are supported by hydrodynamic lift; therefore, it does not apply to small HSC. The limitations for the wind speed are given as 20 knots because it aects the sea surface but this is not suitable for small HSC. If the wind speed is 20 knots, or a Beaufort force ve, this could cause wave heights of two to three metres. This will cause large vertical motions in small HSC which in turn causes the vessel to slam and generate a higher noise signal. The propagation models included in the standards only account for spherical spreading of the noise. It has been shown that a shallow water test location would be more practical but this has dramatic eects on the propagation of the noise due to the re ections from the seabed and sea surface. If the tests are performed in shallow water a new propagation model will be required to handle these sound re ections. ANSI S12.64-2009 has three grades which relate to the accuracy of the measurements. When the accuracy is above 2 % then it is a grade A or B, whereas if it is greater than 5 % it is a grade C. The work within this project should aim for an accuracy greater than 2 % so that it is to the standard of ANSI grade A. This has implications on the accuracy of the measurement method. For example, if a 300 m CPA is measured using a GPS (WASP enabled) with an accuracy of 1.1 m, then the precision is 0.36 %; whereas, if the CPA is reduced to 30 m in proportion to the length of the vessel but still measured with the same GPS, the precision is now only 3.6 %. Therefore, the accuracy of the measurement method needs reconsidering.
6.2.3 Literature review When the water depth is considered shallow then the sound waves re ected from the seabed cannot be ignored because the re ections can amplify or attenuate the received sound, similar to the Lloyd's mirror eect, see Loeser (1999). This has dramatic eects on how the sound can be predicted, measured or propagated. Therefore, in this project it is assumed that the water depth cannot be greater than 30 m and the literature review will not explore deep water measurements in detail. There have only been a few measurements of the water borne noise produced by small crafts and even less have been performed in shallow water. Therefore, this literature review will also explore other methods used for other noise sources such as piling, wind turbines and dredging in shallow water. Erbe (2002) measured the noise emitted by whale watching vessels in Canada using two hydrophones at depths of 5 m and 15 m below the water's surface. A Conductivity, Temperature and Depth (CTD) pro le was taken every 2-3 hours for the sound propagation model and the seabed geometry was known due to a previously performed oceanography experiment. The sound propagation model used in this study was taken from Erbe and Farmer (2000). The model is based on ray theory (Jensen et al. 1994) and includes absorption losses from the
178
CHAPTER 6.
NOISE LITERATURE REVIEW
seabed sediment, and frequency dependent absorption by the ocean water. Rays are traced in two dimensions through an ocean environment described by its bathymetry, sound speed pro les (which could change with range), and bottom sediment.
This method allowed the
predictions of the theoretical zones of noise in uence, discussed in section 6.3.2. The underwater noise produced by a Class 1 powerboat race (top speed of 270 km/h) was measured by Amoser et al., 2004. The hydrophone was located 1.5 m below the water surface and 300 m from the race (due to safety). Both the instantaneous and continuous noise was measured. The noise was reported at the recorded distance so no propagation model was used. Kipple and Gabriele (2003, 2004b) measured the noise produced by a variety of small vessels in Alaska and used a spherical spreading loss equation to relate the measured SPL to the SPL at one yard. No absorption losses, re ections or refractions were considered. Blackwell and Greene (2005) measured the noise from a hovercraft, however, the noise was reported at the recorded distance and again no attempt was made to propagate the noise back to the source level. Nedwell and Edwards (2002) and Nedwell et al. (2003) measured the underwater noise during piling activities for County Wharf, Littlehampton, and the Red Funnel terminal, Southampton, respectively. A multitude of hydrophones were used at distances ranging from 7.5 m to 652 m depending on the received SPL. The hydrophones were taped to rope and lowered to either 1.5 m (Littlehampton) or 2.5 m (Southampton) below the water surface. Nedwell and Edwards (2002) found that a geometrical spreading loss was unsuitable for the results and instead he used an absorption loss model,
SP L = SL − Na (R). N a
is the trans-
mission loss due to absorption and is around 0.07 dB per metre. Nedwell et al. (2003) used the same absorption loss model; however, the absorption loss
Na
was 0.15 dB per meter due
to the water depth. Nedwell et al. (2007) reviewed the models used by Nedwell (2002) and Nedwell et al. (2003) for the application of wind turbines. Nedwell suggested that when the water is deep (> 100 m) and the distance from the source and receiver is large (> 100 m) then geometrical losses dominate the sound propagation and absorption losses can be ignored. However, wind turbines and the D-class operate in considerably shallower water so a dierent propagation model is needed.
The water borne noise generated by a wind turbine was
measured by Betke et al. (2004) to assess the possible environmental impact. A hydrophone was placed 110m from the wind turbine and oated 3m above the seabed with approximately 10m water depth.
No propagation model was used.
Nedwell et al. (2007) found that the
absorption losses in shallow water were very high and this is attributed to the hysteresis losses in the seabed and the sound energy in the shallow water channel. The new model used by Nedwell et al. (2007) includes both geometrical losses and absorption losses and is shown in equation 6.4. Although; this model does not consider the sound re ections from the seabed.
SP LR = SP L1m − N.logR − a.R Where;
SP L1m
is SPL at 1 m,
SP LR
(6.4)
is SPL at R m, R is distance between source and
6.2.
WATER BORNE NOISE
179
receiver, N is Geometric loss factor and a is absorption loss coe cient. Bubliü et al. (2008) measured the noise from research and shing vessels by positioning the hydrophone vertically below the vessel, see gure 6.4. The reason for this was not mentioned but the hydrophone will still receive the re ected sound wave from the seabed.
Gao et al.
(2008) used a towed model in a towing tank to study the noise caused by a surface ship. The use of a towed model means that only the ow noise will be measured and not the propulsion noise. However, the presence of the tank walls will mean that the received noise will contain many re ected noise signals. This has to be carefully accounted for and Gao et al. say further work is needed. Gotz et al. (2009) de nes small crafts as having a length of up to 50m (e.g. recreational crafts, jet skis, speed boats, hover crafts and operational working boats) and describes the typical noise generated by these craft. Small craft normally produce a relatively broadband sound with a SPL of approximately 160 to 175 dB (ref 1
μPa),
although this does depend on
speed and operational conditions. These craft often produce sounds at higher frequencies (> 1 kHz) than larger vessels and often resonate at the vibrational frequencies of the propeller blade, engine or gear box.
Propeller cavitation has signi cant acoustic energy and the frequency
extends above 10 kHz.
Figure 6.4: Noise measurement method from vertically below the vessel, Bubliü et al. (2008).
The majority of the water borne noise measurement methods reviewed so far have not considered the shallow water sound channel which means the measured noise cannot be accurately propagated back to a source level or propagated out to assess the environmental impact. Although, Robinson et al. (2011) measured the underwater noise arising from marine aggregate dredging operations in shallow water and is the most comprehensive underwater noise measurement method found by the author. The method is based on the ANSI standard with extra techniques to compensate for the shallow water sound channel. The rst technique is to use multiple hydrophones at various depths and ranges (50 m to 400 m) to measure the sounds eld (not a point measurement). Robinson et al. (2011) also appreciated that a shallow water channel acts as a low pass lter depending on the water depth. The nal technique was to ac-
180
CHAPTER 6.
NOISE LITERATURE REVIEW
curately model the sound propagation through the shallow water channel. Seven propagation models were compared and it found that the Image source Transmission Loss (ImTL) model was the best for this application. The ImTL model propagates the rst eight sound waves to re ect o the seabed or sea surface, based on Urick (1983). The sound wave re ection o the seabed is based on the work of Brekhovkikh and Lysanov (2003). The sea surface interactions are predicted using; a scattering model by Coates (1988), re ection model by Medwin and Clay (1998) and a wind speed model by Ainslie et al. (1994).
The sound absorption as a
function of frequency is based on the work of Ainslie and McColm (1998). The validation for the ImTL model and more details can be found in pages 29 to 36 and appendix C of Robinson et al. (2011).
6.3 Perception of noise 6.3.1 Perception by humans The human ear transfers the sound vibrations, in either air or water, into an electrical signal in the auditory nerve that the brain interpreters and this is shown in gure 6.5. The sound vibration passes along the auditory canal and causes the ear drum to vibrate. The malleus, incus and stapes transmit the eardrum vibration to the cochlea through the oval window. This is the same mechanism that marine mammals use to hear.
Figure 6.5: Illustration of the human auditory system, Moore (2012).
The evolution of the auditory system means that the human does not perceive all the frequencies within a sound equally. For example; humans can only detect frequencies within the range of 20 Hz to 20 kHz, although this does vary from human to human.
Figure 6.6
shows the contours of equal loudness of the human ear by Robinson and Dadson (1956) and is used in ISO 226:2003 Normal equal-loudness-level contours. It reveals that the human ear is unresponsive to very low frequency but is especially sensitive to frequency around 4 kHz, which corresponds to the frequency of speech, see Everest and Pohlmann (2001) page 51.
6.3.
PERCEPTION OF NOISE
181
Figure 6.6: Equal loudness contours of the human ear, ISO 226:2003.
The frequency weightings have been developed to account for the unequal loudness perception of the human hearing and were published by ANSI S1.4-1971, see Everest and Pohlmann (2001) page 39. These weightings correct the received SPL to account for the non-linear response of the human auditory system and the selection between A, B and C is generally based on the received SPL, as de ned below:
SPL of 20 - 55 dB use weighting A
SPL of 55 - 85 dB use weighting B
SPL of 85 - 140 dB use weighting C
Figure 6.7: A, B and C weighting response characteristics, ANSI S1.4-1971.
The European directive 2003/44/EC sets a maximum allowable airborne SPL emitted by powered recreational craft depending on the engine size, see table 6.1.
This allows an
assessment of the airborne noise to be made and considers the frequency dependence of the human auditory system by using the A weighting. This is su cient to assess the perception of noise for a human in air. This project will not consider the perception of the noise for a human in water because it is rarely ever heard.
182
CHAPTER 6.
NOISE LITERATURE REVIEW
Single Engine Power (HP)
Maximum SPL (dB(A))
P > 13.4
67
13.4 > P > 53.6
72
P > 53.6
75
Table 6.1: Maximum SPL (dB(A)) for an outboard engine depending on engine power.
6.3.2 Perception by wildlife The Oslo and Paris convention for the protection of the marine environment of the North-East Atlantic (OSPAR) convention was set up in 1992 for the protection of the marine environment in the North-East Atlantic but it was not until 2004 that anthropogenic noise was considered a pollutant. In 2009, OSPAR published an Overview of the impact of anthropogenic underwater sound in the marine environment, see Gotz et al. (2009). According to Gotz et al. (2009) this is the most relevant compilation of existing knowledge in relation to the marine environment in the UK and is the basis of the literature within this section. It is worth noting that the dierence in the sea environment, say from the UK to the Mediterranean, will aect the propagation of sound and the types of species found in the region. Most marine organisms use sound for various vital life activities such as; communication, location of mates, searching for prey, avoiding predators and hazards, and for long and short range navigation.
Some sh even use sound to sense the motion of the uid particles, see
Nedwell et al. (2004), but this project will not investigate this use of sound.
The typical
sounds produced by marine mammals and sh compared with the nominal low frequency sounds associated with commercial shipping is shown in gure 6.8. It reveals that commercial shipping clearly has the potential to mask the sounds generated by marine mammals and sh.
Figure 6.8: Typical frequency bands of sounds produced by marine mammals and sh, Gotz et al. (2009).
The possible eects upon wildlife can include masking, behavioural disturbance, hearing loss (temporary or permanent), discomfort, stress, injury or even death. Masking occurs when the noise is strong enough to cover sound signals that the fauna use for activities such as communication, echolocation and passive detection. Behavioural disturbances cause a change
6.3.
PERCEPTION OF NOISE
183
to the activities of the wildlife as a response to the sound. Hearing loss is when the hearing threshold of an animal, usually at a particular frequency, changes due to a high amplitude sound.
It can cause a Temporary Threshold Shift (TTS) or a Permanent Threshold Shift
(PTS). Sound can also cause non-auditory eects by damaging some tissues (such as swimbladder, muscle tissue and brain). These dierent eects can be de ned as theoretical zones of noise in uence depending on the distance between the source and receiver, see gure 6.9, and it will be called the Zones of In uence (ZoI) throughout this paper.
Figure 6.9: Theoretical zone of noise in uence, Richardson et al. (1995).
Each species has evolved dierent hearing thresholds and ranges depending on the frequency and amplitude in a similar manner to contours of equal loudness of the human ear by Robinson and Dadson (1956). The hearing threshold is the average SPL that is just audible to the specie and can be plotted as a function of frequency, this is called an audiogram. Nedwell et al. (2004) provides a summary of available information on the audiograms of marine mammals and sh, including; the hearing mechanisms of marine mammals and sh, detailed discussions about the methods for obtaining audiograms and their quality and an extensive database of audiograms.
Nedwell and Turnpenny (1998) rst proposed the used a generic
weighted frequency scale,
dBht (Species),
to estimate the environmental impact of noise and
it is used in a similar manner to the human frequency weighting scale.
The sound under
investigation is ltered into discrete frequency bands so the SPL in each frequency band can be compared to the hearing threshold of the specie. The eect of the noise depends on the exposure level, which is the dierence between the hearing threshold of the specie and SPL of the noise and this is shown in gure 6.10. Nedwell et al. (2007) proposes that if the exposure level is between 0 to 10 dB then the noise is barely audible, if the levels are over 90 dB it is believed to cause behavioural aects and if the levels are over 130 dB then it is likely to cause a TTS and potentially damage the auditory system. The eect of noise is also dependent on the exposure duration, as well as the exposure level, and if both are exceeded then a TTS may occur and a PTS may occur with repeated exposure. This is summarised in table 6.2, taken from Nedwell et al. (2007). Other authors have used simpler methods to predict the eect of noise by comparing the overall SPL of the anthropogenic noise to a xed value for a
184
CHAPTER 6.
NOISE LITERATURE REVIEW
Figure 6.10: Perceived sound level (dBht ) of sh and marine mammals for an example noise spectrum, Nedwell et al. (2007).
known specie. For example, both Tyack (2009) and Madsen et al. (2006) state that if the SPL is above 190 dB there is a risk of damaging the auditory system of seal; however, this is not frequency dependent and an audiogram shows that the auditory system is highly frequency dependent. Exposure Level (dBht )
Exposure duration
90
8 hours
92
5 hours
99
1 hour
110
Approx 5 minutes
120
Approx 30 seconds
130
Approx 3 seconds
Table 6.2: Comparison of noise exposure level and duration for the same cumulative 90
LEP,D
Noise Dose, Nedwell et al. (2007).
Gotz et al. (2009) says that due to the near shore operation and high frequencies of small vessels (LWL < 50 m) they generally have a geographically limited environmental impact. However, they can dominate the ambient noise in a coastal environment, especially in bays, harbours and estuaries where the wildlife cannot avoid the noise.
These problems can be
exaggerated during vital life stages of the wildlife, such as migration and breeding. This can cause the exposure duration to increase leading to an increased environmental impact. The exposure level can be linked to the ZoI using the
dBht (Species)
scale and it has only
been performed by a few authors, including Erbe and Farmer (2000), Nedwell and Mason (2012), and Nedwell et al. (2012). A water borne noise propagation model is used to predict the area over which the anthropogenic noise may impact a species. An example is shown in gure 6.11. It demonstrates that herring can hear the piling noise at much greater distances than salmon and this may interfere with their spawning area. The discussion above on the perception of noise by wildlife has all been developed for water borne noise, so how can the perception of airborne noise be assessed? The exposure levels and
6.4.
SUMMARY
185
Figure 6.11: Zones of in uence for salmon and herring caused by piling noise, Nedwell et al. (2012).
durations in table 6.2 are based on the perception of noise by the human auditory system in air; therefore, the same exposure levels and durations can be used to predict the ZoI in air as well as in water.
6.4 Summary Measurements of the airborne noise generated by recreational craft, such as RIBs and IBs, started in the 60s and the SoundBoat project proved that the current methods are accurate and repeatable measurements. This has led to an International Standard (ISO 14509) which provides several methods to measure the airborne noise generated by the D-class. The measurement of water borne noise is a dierent story where there are no standards that directly relate to the D-class in shallow water. The majority of measurements in shallow water have not properly considered the shallow water sound channel, except the measurements by Robinson et al. (2011), so the noise cannot be propagated and the received noise cannot be predicted. This means that a water borne measurement method for the D-class must be developed speci cally tailored for small HSC in shallow water but the ANSI standard (S12.642009) and Robinson et al. (2011) can be used as a template. The perception of the airborne noise by humans can be assessed using the European directive 2003/44/EC. There is no directive that apply to the perception by wildlife but it can be assessed using the
dBht (Species) scale
originally proposed by Nedwell and Turnpenny (1998).
This allows the anthropogenic noise to be related to the ZoI for both air and water borne noise. The propagation of water borne noise can be accurately predicted using ImTL model validated by Robinson et al. (2011).
186
CHAPTER 6.
NOISE LITERATURE REVIEW
Chapter 7
Noise methodology 7.1 Water borne noise methodology 7.1.1 Criteria The literature review showed that there are international standards for measuring the water borne noise generated by ships; however, section 6.2.2 revealed that the standards are not directly suitable small HSC. The principle of the pass-by method employed by both the air and water borne noise standards is still suitable but the parameters and boundary conditions need to be carefully considered. The new method is aimed at the D-class but it should also be applicable for other similar vessels; therefore, this method aims to be suitable for all of the RNLI RIBs and IBs. The RNLI RIBs and IBs range in length from 3.8 m to 10 m. All of the RIBs and IBs are capable of planing and this changes some considerations, such as squat, so this method is aimed for planing vessels. The planing speeds associated with the RNLI RIBs and IBs range from 10 knots to 40 knots. The shallow water depth of 20 m has been chosen as a compromise between deep water for high quality measurements and shallow water for practicality.
The headquarters and manufacturing facilities of RNLI are in Poole
and Cowes, respectively, and 20 m is feasible from both locations. Therefore, the new method being developed must meet the following criteria:
Vessel length, ranging from 3 to 10 m.
Planing vessels, mass is supported by the hydrodynamic lifting force.
Vessel speed, ranging from 10 to 40 knots.
Shallow water, approximately 20m.
It is worth considering the stakeholders of this new method because, in the future, it is highly likely that all new-builds will have to meet some form of water borne noise regulation. The main stakeholders here are the boat builders and engine manufacturers so the method should be practical and cost eective. So it would be extremely bene cial for the stakeholders to be able to perform these measurements in shallow water and to measure both the air and water borne noise simultaneously. 187
188
CHAPTER 7.
NOISE METHODOLOGY
7.1.2 Course layout The basic course shape of the pass-by method will stay the same but the dimensions should be scaled according to the vessel length. The CPA stated by the ANSI standard should be at least 100 m or one times the vessels length, whichever is greater.
This is clearly out of
proportion for a ve metre vessel and it is di cult at this stage to know what the optimal distance should be. The airborne noise standard uses a CPA of 25 m and it is intended that both the air and water borne noise is measured simultaneously; therefore, a CPA of 25 m will be used. The next issue relating to the course layout is ensuring that it is constant and does not change during the experiment. This can be very di cult to achieve when dealing with small vessels in strong tides, winds and waves. A trial experiment was performed to test the method under development and a few lessons were learnt. First, the buoys marking the course would not hold their position relative to each. Second, a GPS was used to record the start and the CPA of each test run; however, GPS lost contact with the satellites and resulted in incorrect CPA measurements (sometimes up to 1 km). Finally, a hand-held laser distance-meter was used to measure the CPA but this method was unsuccessful because the target area (the source craft) was too small and the motion of receiver craft were too large to pin point the laser on the source craft. So a new method is required to either ensure the course layout is xed in place or the CPA is measured accurately; if the course layout is xed accurately then the helmsmen can pass the CPA marker within + 1 m and the accuracy of the CPA is
± 0.5 m,
or
the real (driven) CPA is measured accurately and any inaccuracies in the helmsmen or course layout can be accounted for in the post processing because the real CPA is known. Ideally, the eect of the tide could be removed by performing the tests at slack water but this window can be only open for a matter of minutes, then the tide changes direction and the course is no longer accurate. The tide is a particular problem in the UK because the tides are strong and the locations available for testing the RNLI vessels (Poole and Cowes) are near large harbours resulting in even stronger tides. Normally, the course is laid out using anchored buoys; this is appropriate for the airborne noise method where the water is much shallower leading to weaker tides, and it is appropriate for large ships because the real CPA is measured using GPS. A careful balance is required between the tide, wind and waves to ensure an anchored buoy stays within
± 1 m (4 % accuracy
of a 25 m CPA) of its desired position. A more appropriate way to remove the tide is to allow the course to drift with it. This could be referred to as the drifting pass-by method. This also has the bene t that the tide does not aect the hydrophones, which could remove the noise from turbulence and allow the hydrophones to hang vertically. This has removed the eect of tide but a marker is still required to ensure the source vessel pass with a CPA of 25 m. The marker could be tted to a buoyant detectable pole that is 25 m long. This is a commercially available method called the Sonic Boom, see CEProof (2013).
7.1.
WATER BORNE NOISE METHODOLOGY
189
Figure 7.1: Mariner 50 HP outboard motor, Oshoremarine (2013).
7.1.3 Instrumentation The rest of this section will now follow the format of the ANSI standard because it is the most comprehensive and relevant standard current available. This section will show where the new method deviates from the ANSI standard or where the ANSI standard requires re ning. The hydrophones and data acquisition system described in the ANSI standard (section 4.2 and 4.3) are still applicable for the new method. The ANSI grading system from A to C can still be used to judge the accuracy of the hydrophones and data acquisition system. The CPA of the new method has been reduced by a quarter so the CPA measurement method in section 4.4 of the ANSI standard is no longer accurate enough. The ISO standard says that the usual uncertainty of a GPS is
±
1 m. If the CPA was 100 m then a GPS could
measure the distance to an accuracy of 1 %; however, the CPA in this method is 25 m so a GPS would have an accuracy of 4 %. All three standards recommend using GPS to measure the CPA but only the ANSI standard provides an alternative option of an optical or acoustical sensor.
For the results to conform to ANSI grade A the CPA should be measured with an
accuracy of 2 % so GPS is not accurate enough. The accuracy of the GPS could be increased using a Dierential GPS (DGPS) and this would provide su cient accuracy (± 10 cm) but this is an expensive option. Only the ANSI standard considers the location of the acoustic centre. For a grade B and C it is assumed to be half way between the engine room and the propeller and for a grade A measurement it is determined by the user. All the RNLI RIBs and IBs are propelled by outboard motors, see gure 7.1. If the acoustic centre is de ned by ANSI grade B or C then it would lay half way along the main vertical shaft.
Young and Miller (1960) suggest two
acoustic centres; one is the centre of the propeller and the other is the underwater exhaust outlet. In the case of the Mariner 50 HP outboard the exhaust outlet is through the centre of the propeller; therefore, it is assumed that the acoustic centre is the centre of the propeller.
7.1.4 Measurement requirements and procedure The test site requirements outlined by the ANSI standard (section 5.2) are still applicable but the sea bed type and topography requires extra consideration. This is because the sound waves re ecting o the sea bed will signi cantly aect the noise propagation and this can be accoun-
190
CHAPTER 7.
NOISE METHODOLOGY
ted for in a propagation model provided the properties are known. The extra requirements are:
The water depth must be measured with an accuracy of 2 % (±0.5m).
The sea bed must be as at as possible and there should be no large obstacles in the vicinity because this will cause the sound to be re ected back to the hydrophone.
The acoustic properties of the sea bed must to be known.
The acoustic properties of the sea bed should be constant or the variation should be known.
It was shown in section 6.2.2 that the speed-depth relation stimulated by the DNV standard was to remove any shallow water aect, such as squat, on large displacement vessels. These do not apply to small planing vessels. On the other hand, shallow water can aect the planing performance in other ways. The ITTC for resistance measurements of HSC recommends that the water depth should be at least 0.8 times the length of the vessel. The Froude depth number (F rD ) is used to de ne whether the water depth will aect the residuary resistance of a vessel, shown in equation 7.1, see Dand (2002c). The maximum aect occurs when the Froude depth number approaches unity because the wave wash resistance will signi cantly increase. Dand (2002c) also states that solitons can be formed when the Froude depth number ranges from 0.8 to 1.2. Therefore, Froude depth numbers between 0.75 and 1.25 should be avoided. So the water depth for planing vessels should be correctly chosen for both the ITTC recommendation and the Froude depth number.
p F rD = v/ gd
(7.1)
The condition of the sea surface will aect both the background noise and the noise generated by the source vessel, see section 5.3 of ANSI. All the water borne standards state that the measured SPL should be 10 dB higher than the background noise. As long as this requirement is met, then the increase in background noise will not aect the results. Moreover, the sea surface will aect the motion of the source craft and can cause instabilities in the noise source. The ANSI standard places a maximum limit of 20 knots on the wind speed but this is aimed at vessels with a length of up to and beyond 100 m.
It would be more appropriate to use
the limitation set by the airborne standard ISO 14509, which limits the wave height (H) in proportion to the vessel's length (LW L), see equation 7.2. Realistically the behaviour of all HSC depends on the design of the craft (i.e. some are designed for large seas and some are not) so the wave height limitation should be user de ned. Extra attention should be paid to the direction of the tide and wind with small HSC because if the direction of the tide and wind are opposite then this has a tendency to cause steeper waves which will elevate the problem.
H = LW L/50
(7.2)
7.2.
AIR AND WATER BORNE NOISE TRIAL METHOD
191
The hydrophone deployment discussed in section 5.4 of the ANSI standard is still applicable for small HSC. The drifting pass-by method will remove any turbulence around the hydrophones and stop any hydrophone drift (see section 4.4, ANSI). The test course and vessel operation, and the test sequence described in section 5.5 and 5.6 of ANSI 12.64 are still relevant for small HSC.
7.1.5 Post processing The post processing outline in the ANSI standard (sections 6.1 to 6.3) for the Data Window Length (DWL) (equation 7.3), Data Window Period (DWP) (equation 7.4), background noise comparison (equation 7.5), background noise adjustment (equation 7.6), sensitivity adjustment (equation 7.7) and combining hydrophones (equation 7.8) are still suitable for small HSC in shallow water; however, the propagation model is not. This is because the ANSI standard assumes deep water noise propagation (i.e. spherical spreading) whereas this method is developed for shallow water so the re ections from the sea bed and sea surface need to be considered. The literature review revealed that Robinson et al. (2011) compared seven shallow water propagation models and showed that the ImTL model was the most accurate for this water depth and application.
DW L = 2 × CP A × T an(θ)
(7.3)
DW P = DW L/v
(7.4)
4 = Ls+n − Ln = 20log (Ps+n /Pn )
(7.5)
L0p = 10log(10(Ls+n /10) − 10(Ln /10) )
(7.6)
L00p = L0p + ASEN
(7.7)
Ls (r) = 10log{(10Ls (r,h1)/10 + 10Ls (r,h2)/10 )/2}
(7.8)
7.2 Air and water borne noise trial method The noise trials of the RNLI D-class were performed on the 8th of May 2013. The trials were performed in the Solent, UK, approximately 2 to 3 nautical miles due east of Cowes harbour entrance, in between the shore and Ryde Middle. This location was chosen because it provided a water depth of at least 20 m, the shore line provided shelter for the prevailing south westerly winds and it is close to the storage facilities of the D-class. The drifting pass-by method was used in the nal trials and the course layout can be seen in
192
CHAPTER 7.
NOISE METHODOLOGY
gure 7.2. The CPA was controlled using a 24 m buoyant detachable pole system constructed from 12 times two metre sections that were screwed together. The driver was asked to pass the end of the pole with a one metre gap and the anticipated accuracy of the driver was
±
0.5 m; therefore the CPA was 24 m + 1 m. As the acoustic centre of D-class passed the CPA, a time stamp was placed on the data.
The acoustic centre was chosen to be the outboard
engine. The two hydrophones were tted to a weighted rope at depths of 5 and 10 m with an accuracy of
±
20 mm.
The rope was non-metallic and the weights were covered in plastic
to reduce any noise generated by metal on metal contact.
The microphone was tted to a
telescopic pole so that it was 3 m above the water surface. The D-class was set up for standard operation with all fuel and equipment; however, to reduce the sta requirements only two crew were used instead of three.
The D-class was
driven at full speed past the CPA and the speed was measured using a GPS. The structural stiness, and thus the hydroelasticity, of the D-class was varied using the internal pressures of the sponson and keel. Two pressure settings were used; standard operating pressures and plus 1 psi in the sponson and keel. The standard operating pressures are 3 psi in the keel and 3.25 psi in the sponson, and plus one psi relates to 4 psi in the keel and 4.25 psi in the sponson. From now on these conditions will be referred to as 3 psi and 4 psi, rather than stating the 0.25 psi dierence between the keel and sponson. The noise from the D-class was measured 12 times; 3 repeats port side at 3 psi, 3 repeats starboard side at 3 psi, 3 repeats port side at 4 psi and 3 repeats starboard side at 4 psi.
7.2.1 Equipment The airborne noise measurements were made with a ADAU 1/4" omnidirectional eld microphone tted with JIN IN ECM-1OB cartridge ( at frequency response between 50Hz 13kHz) and an ADAU xed gain microphone ampli er. The water borne noise measurements were made with two Neptune D/70 omni-directional hydrophones with a 10 to 100 kHz frequency response (relatively at frequency response between 10Hz - 20kHz). An etec A1001 pre-ampli er was used with 30 dB gain and a 10 Hz low pass lter to remove the eect of surface waves. Both the microphone and hydrophones were fed into a 16 bit Edirol UA-1A ADC (sound card) manufactured by Roland Corporation. This gave an accuracy of
±3.64 × 10−3
±5.22 × 10−3
dB for the hydrophones. Both the microphone and
hydrophones were sampled at 65536 Hz.
A Dell Latitude D630 laptop was used as a data
dB for the microphone and
acquisition system using ADAU Engwaves 2.16.15. The local weather conditions, including wind and temperature, were measured using BrambleMet weather stations. BrambleMet was approximately 4 km away from the test location so the data should be accurate enough. The wave conditions were measured by Lymington wave buoy. The receiver craft had a sh nder on board which was used to measure the water depth with an accuracy of
±
1 m and it was also used to ensure the sea bed was at. A GoPro hero
2 was used to record both videos and pictures of the trials. The speed of the D-class was measured using the on-board GPS with an accuracy of
± 0.1
7.2.
AIR AND WATER BORNE NOISE TRIAL METHOD
(a) Side view.
(b) Top view. Figure 7.2: Drifting pass-by course layout.
193
194
CHAPTER 7.
NOISE METHODOLOGY
knots. The sponson and keel pressures were measured using a pressure gauge with an accuracy of
±
0.1 psi.
7.2.2 Procedures 1. 24 m marker pole is deployed. 2. Background noise is measured for 120 seconds. 3. Environmental data is recorded:
(a) Time. (b) Water depth. (c) GPS coordinates
4. Source craft starts its engine and accelerates to full speed towards the marker pole. 5. Approximately 10 seconds before the source craft passes the CPA the data acquisition system is started. 6. A time stamp is placed when the acoustic centre of the source craft is aligned with the CPA . 7. The source craft continues in a straight line. 8. The data acquisition system is stopped when the source craft is approximately 10 seconds past the CPA. 9. Make note of any vessels in the area (type, position, speed, etc). 10. Steps 3 - 11 are repeated so that each side of the craft is measured at least three times.
7.2.3 Post processing Two post processing methods were used to quantify the noise generated by the D-class; one from the airborne standard (ISO 14509) and one from the water borne standard (ANSI 12.64). The airborne standard quanti es the noise into one single value which is the maximum ASweighted overall SPL and is used to quantify whether or not the airborne noise is within the regulation limits set out by the EU directive 2003/44/EC, plus its simplicity provides an easy comparison between dierent parameters.
The water borne noise standard calculates
the mean one-third-octave SPL and provides an in depth one-third-octave spectrum of the generated noise. To calculate the maximum AS-weighted SPL, rst the data was trimmed to one second to simulate a slow response (i.e. the S-weighting), half a second either side of the CPA. The CPA was de ned using the CPA marker. Then the data was ltered into one-third-octave bands and the RMS value (in dB) was computed using a MatLab code written by C. Couvreur
7.2.
AIR AND WATER BORNE NOISE TRIAL METHOD
based on ANSI S1.11-1986.
195
Next, the signal was compared to the background noise and a
correction value was applied, if applicable. The background noise comparison is identical for both post processing methods and will be discussed shortly. An A-weighting was applied and the one-third-octave bands were summed together to get the overall AS-weighted SPL. The two hydrophones were combined using equation 7.8, taken from ANSI equation 8.
The A-
weighting only applies to the human hearing in air; therefore, the overall S-weighted SPL was also calculated to allow the value to be used for non-human hearing. The one-third-octave spectrum was generated by rst trimming the data down to a two second DWP, one second either side of the CPA, using equations 7.3 and 7.4. The CPA was de ned using the CPA marker. The DWP was then divided into eight segments (each 0.25 s long). Each segment was then ltered into one-third-octave bands and the RMS value (in dB) was computed using a MatLab code written by C. Couvreur based on ANSI S1.11-1986. Then each one-third-octave segment was compared to the background noise level and a correction value was applied, if applicable. The eight segments were averaged together to provide a mean SPL in one-third-octaves for each run. The two hydrophones were combined using equation 7.8, taken from ANSI equation 8.
This could then be plotted as a frequency spectrum in
one-third-octaves. The received air and water borne noise must be compared to the background noise and all the air and water borne standards compare them using equation 7.5. The received signal is then adjusted with a correction value depending on the dierence,
4, and they are summarised
in table 7.1. The airborne standard made no comment on the adjustment value if the dierence was less than six decibels so the results were removed from the analysis, and the same was done with the water borne noise if the dierence was less than three decibels. Airborne noise
4 = Lp+n − Ln 4 ≥ 10 10 > 4 > 6 4≤6
Water borne noise
Correction (dB) 0 -1 blank
4 = Lp+n − Ln 4 ≥ 10 10 > 4 > 3 4≤3
Correction (dB) 0 Equation 7.6 note or discard
Table 7.1: Background noise correction value.
The CPA pole system was used to mark the course for the helmsmen and ensure the boat was 25 m from the microphone and hydrophones; however, the pole deformed due to the wind, waves and tide so the actual CPA was less than 25 m. Photographs were taken of the pole system so that the CPA could be calculated from the deformation of the pole, see gure 7.3. It was assumed that the angle of deformation was equal in between each joining pole, angle between the straight line CPA and the end poles,
α1 and α2 ,
of perspective was removed by averaging these two angles (α1 and
β.
The
were measured. The eect
α2 ).
This showed that the
actual CPA during the measurements was approximately 20 m. Exact values for each run are show in table 8.1. The error in measuring the angle of the pole was maximum CPA error
±
± 0.31 m when α = 90° but when α = 41° (i.e.
± 1°
that relates to a
CPA = 20 m) the error is
0.18 m. It is anticipated that the accuracy of the driver was at least
±
0.5 m and this gives
196
CHAPTER 7.
NOISE METHODOLOGY
Figure 7.3: An example of the deformed CPA pole.
a combined error of
±
0.68 m. When the CPA is 20 m this method provided a percentage
error of 3.4 % which places it in ANSI S12.64 grade C.
7.2.4 Propagation models Two propagation models will be needed; one for air and one for water because the propagation through the mediums are considerably dierent.
Both propagation models will be used to
predict the noise at one metre and to predict the environmental impact and the ZoI. The airborne propagation loss was based on ISO 9613-2 (Acoustics - Attenuation of sound during propagation outdoors) and it considered the geometric spreading and frequency dependent absorption loss. The hemisphere spreading loss was calculated using equation 7.9 and the absorption loss used equation 7.10.
Adiv = 10log
r1 r2
(7.9)
Aatm = α(r1 − r2 )/1000 Where:
Adiv
is attenuation due to geometric spreading (dB),
attenuation due to atmospheric absoption (dB) and
α
(7.10)
r
is radius (m),
Aatm
is
is absoprtion loss coe cient.
The ImTL model will be used to predict the water borne noise propagation.
The de-
tails of the ImTL model and the validation of the model can be found in Robinson et al. (2011) appendix C. The MatLab code was written by M. A. Wood from the ISVR, University of Southampton, and was tailored by the author to predict the ZoI. The ImTL model requires the sediment geoacoustic properties of the seabed including:
density of sediment
7.2.
AIR AND WATER BORNE NOISE TRIAL METHOD
197
3
(kg/m ), compressional sound speed in sediment (m/s), compressional attenuation in sediment (dB/m/kHz ), shear wave speed in sediment (m/s) and shear wave attenuation in sediment (dB/m/kHz ). (1994).
These can be calculated using the equations set out by Bachman
Bachman de nes the sediment using the mean grain size in phi units.
The South
Coast Regional Environmental Characteristics report, see James et al. (2010) page 103, shows that the mean grain size either side of the Solent ranges from two to four millimetres; however, it is extrapolated that the mean grain size inside the Solent is one millimetre because the mean grain size gradient is decreasing as it enters the Solent. The grain size in phi is 0 using grain size (phi) = - log2 [grain size (mm)], see Bachman (1994). The density was calculated using equation 7.11, see Bachman (1994) equation 15. Where;
ρ
3
is density (g/cm ) and
Mz
is grain size (phi). The density is 2.380
g/cm3 .
ρ = 2.38 − 1.725 × 10−1 M z + 6.589 × 10−3 M z 2
(7.11)
The compressional sound speed at the sea oor can be calculated using equations 7.12 and 7.13, see Bachman (1994) equation 1. Where; oor (m/s),
R
V p(0)
is compressional sound speed at the sea
is sound speed ratio and c(0) is sound speed in bottom water (m/s).
V p(0) = R.c(0)
(7.12)
R = 1.296 − 6.01 × 10−2 M z + 2.83 × 10−3 M z 2
(7.13)
The sound speed in bottom water was calculated using the equation of the speed of sound in sea water by Wilson (1960). Solent (2013) stated that the salinity in the Solent can range from 30 to 34.5 mille so the mean salinity was used, 32.25 mille. The mean water temperature
o C . The hydrostatic pressure of 231.7 KPa was calculated at
was taken from table 8.1c, 11.83
23 m depth assuming a seawater density of 1027
kg/m3 .
Therefore, the sound speed in bottom
water, c(0), is 1493.6 m/s and the compressional sound speed at the sea oor is 1935.8 m/s. The compressional sound attenuation was calculated using equation 7.14 when
2.5, see Bachman (1994) equation 19.
0 < Mz 5
Where; kp(0) is compressional sound attenuation at the
sea oor (dB/m/kHz). Therefore, the compressional sound attenuation is 0.230 dB/m/kHz.
kp(0) = 0.230 + 0.026.M z
(7.14)
The shear wave speed was calculated using equation 7.15, see Bachman (1994) equation 18. Where; Vs(0) is shear wave speed (m/s). Therefore the shear wave speed is 61.65 m/s.
V s(0) = V p(0)/31.4
(7.15)
According to Bachman (1994) the shear wave attenuation at the surface is a constant of 13.20 dB/m/kHz, see Bachman (1994) page 7. The input properties used for the ImTL model are summarised in table 7.2. The ImTL model is based on FEM so the mesh density must be investigated. The TL at the
198
CHAPTER 7.
NOISE METHODOLOGY
Properties
Value
Units
Density of sea water
1027
Temperature of sea water
11.83
kg/m3 oC
Salinity
32.25
mille
Speed of sound in sea water
1493.6
m/s
Density of sediment
2380
kg/m3
Compressional sound speed
1935.8
m/s
Compressional sound attenuation
0.230
dB/m/kHz
Shear wave speed
61.65
m/s
Shear wave attenuation
13.20
dB/m/kHz
Wind speed
16
m/s
Table 7.2: Input properties for the ImTL model.
two hydrophones was calculated at frequencies from 100 Hz to 16000 Hz and the mesh density was varied from 1 m to 0.02 m. It was found that the mesh re nement has opposite eects on the TL from the two hydrophones and the trends in the hydrophones were frequency dependent. Therefore, a mesh size was chosen that was in proportion to the error in hydrophone place, which was
±
0.1 m. Matching the mesh size to the error in hydrophone placement results in
averaging the TL over each cell and averages out any errors due to hydrophone placement. The TL calculated over 50 m for each one-third-octave frequency band can be seen in appendix J.
Chapter 8
Noise results The noise trials were performed on the
8th
of May 2013. The noise trials were performed in
the Solent, UK, approximately 2 to 3 nautical miles due east of Cowes harbour entrance and the location is shown in gure 8.1.
The main parameters and environmental conditions are shown in table 8.1. The recorded speed, shown in table 8.1a, shows that the speed varied considerably from run to run, ranging from 20.3 knots to 28.2 knots. This was unexpected because the reported top speed of the D-class was 25 knots.
It is probable that the D-class was riding a wave when it reached a
speed of 28.2 knots. The wave heights measured at Lymington wave buoy are shown in table 8.1b and the maximum wave height was 0.35 m, this is signi cant enough to aect the D-class. The wave direction was quarter following sea on starboard and quarter head sea on port. All the top speeds over 26 knots were measured when the D-class was on the starboard aspect so the high top speeds were caused by the interaction with the waves. This also links with the lowest speed (20.3 knots) being caused by the added resistance in head seas. The CPA varied due to the deformation of the CPA pole; however, the majority of the measurements were at approximately 20 m. The depth also varied by a noticeable amount because the drift method was used and the tides shifted the test area into shallower or deeper waters. The variation of speed and depth meant that on a few occasions the Froude depth number approached unity but there were no signs of solitons being formed and this is not expected to have aected the noise generation.
The wave heights and wind speeds were higher than desired. The wave height did aect the vessel speed and the highest wind speed of 19.5 knots (10 m/s) was higher than the airborne noise standard limit of 7 m/s. The trials were performed in May when historically the winds are weak and probability of rain is low. The date of the experiment was exible in case the weather was poor; however, the experiment had already been delayed due to bad weather and it was decided that these conditions were acceptable. 199
200
CHAPTER 8.
Internal Pressure
Aspect
Time
3 psi
4 psi
Speed
CPA
Depth
(knots)
(m)
(m)
NOISE RESULTS
F rD
S1
1521
28.2
19.9
23
0.97
P1
1522
20.3
19.9
23
0.70
S2
1523
26.5
19.7
23
0.81
P2
1524
25.4
20.6
23
0.87
S3
1525
-
19.8
23
-
P3
1526
-
20.9
23
-
P1
1440
24.3
23
22
0.85
S1
1441
27.5
19.9
22
0.96
P1
1504
25.2
18.5
20
0.93
S2
1503
24.6
18.5
20
0.90
P3
1510
24.6
17.7
25
0.81
S3
1515
25
17.7
27
0.79
(a) Main parameters. Internal Pressure
3 psi
4 psi
Wave Height
Max. Wave Height
Wave Period
(m)
(m)
(s)
S1
0.08
0.3
2
P1
0.08
0.3
2
S2
0.08
0.3
2
P2
0.08
0.3
2
S3
0.08
0.3
2
P3
0.08
0.3
2
P1
0.07
0.35
2.3
S1
0.07
0.35
2.3
P1
0.08
0.3
2
S2
0.08
0.3
2
P3
0.08
0.3
2
S3
0.08
0.3
2
Aspect
(b) Wave conditions from Lymington wave buoy. Internal Pressure
3 psi
4 psi
Wind Speed
Wind Direction
Atmospheric
(knots)
( )
o
(mbar)
S1
16.7
236
1007
11.8
12
P1
16.7
236
1007
11.8
12
S2
16.7
236
1007
11.8
12
P2
16.7
236
1007
11.8
12
Aspect
Sea Temp. (
oC )
Air Temp.
oC )
(
S3
14.6
238
1006.9
11.9
11.9
P3
14.6
238
1006.9
11.9
11.9
P1
15.9
238
1006.9
11.6
12
S1
19.5
238
1006.9
11.6
12
P1
19.2
234
1006.9
11.6
12
S2
19.2
234
1006.9
11.6
12
P3
17.8
234
1007
11.7
11.8
S3
14.7
236
1006.9
11.6
12
(c) Weather conditions from BrambleMet. Table 8.1: Main parameters and environmental conditions measured during the noise trials.
8.1.
OVERALL AIR AND WATER BORNE NOISE
201
Figure 8.1: Navigation chart showing locations of the noise trials and weather stations. Airborne
dBAS 3 psi
4 psi
Port
at ~20 m 77.3
Water borne
dBAS
at 25 m
dBAS
76.7
at ~20 m
134.2
Starboard
77.7
77.1
136.3
Port
76.1
76.1
131.9
Starboard
78.1
77.5
133.9
Table 8.2: Maximum AS-weighted SPL.
8.1 Overall air and water borne noise 8.1.1 Maximum sound pressure level The maximum AS-weighted SPLs are shown in table 8.2. The maximum AS-weighted SPL for the airborne noise at 25 m was 77.1 dB (AS) under standard operating conditions (i.e. at 3 psi). The airborne noise regulation (2003/44/EC) limits the airborne SPL to 72 dB (AS) if the craft has a 50 HP engine. This means that the D-class is 5.1 dB over the limit. Although, it is worth noting that the limit increases to 75 dB (AS) if the engine power is over 53.6 HP and the D-class engine is very close to this limit. Lanpheer (1998, 1999) measured the noise of various craft using the pass-by method and the SPL ranged from 69 dB(A) to 76 dB(A). This shows that the D-class is louder than previous results; although, the reason for this is unknown because the D-class is short and has a smaller engine. It could be due to the engine submersion kit that is retro tted by the RNLI to ensure the engine can be restarted if it gets submerged. The maximum AS-weighted SPL, table 8.2, shows that the starboard side of the D-class is louder. This is the same at 4 psi but the eect of internal pressures on the noise generation will be discussed in more detail in section 8.2.2. The maximum S-weighted SPLs are shown in table 8.3 and they are higher because they have not been A-weighted. The speed of the D-class was shown to vary considerably from run to run and it is expected
202
CHAPTER 8.
Airborne
dBA 3 psi
4 psi
Port
82.2
Port
dBA
83.3
at ~20 m 143.2
82.8
144.5 145.2
81.7
83.9
Starboard
at 25 m 81.4
83.4
Starboard
Water borne
dBA
at ~20 m
NOISE RESULTS
82.5
142.8
Table 8.3: Maximum S-weighted SPL.
Water borne noise at ~20 m 150
88
148
86
146
84
144
82
142
80
140
78
138
76
136
74
134
72
132
70
Maximum S-weighted SPL (dB)
Maximum S-weighted SPL (dB)
Air borne noise at 25 m 90
130 20
21
22
23
24
25
26
27
28
29
Speed (knots)
Figure 8.2: Maximum S-weighted overall SPL against the vessel speed.
that a higher speed would increase the overall SPL. The overall S-weighted SPL allow a simple comparison to be made between the SPL and the speed of vessel, and it is shown in gure 8.2. This shows that the vessel speed was not proportional to the overall SPL and it reinforces the fact that the variation in speed did not aect the measured noise. All the runs were performed at full throttle which means the engine revolutions were relatively constant; therefore, the noise generated by engine would also be relatively constant. The GPS measures the Speed Over Ground (SOG) and not the Speed Over Water (SOW). The trial area was tidal so there could be a SOG even though the D-class was not moving through the water and the noise generation should be proportional to the SOW not the SOG. So this implies that the variation in speed could be misinterpreted and is unlikely to aect the noise generation.
8.2 One-third-octave air and water borne noise 8.2.1 One-third-octave spectra The mean one-third-octave air and water borne noise spectra generated at 3 psi are shown in gure 8.3 and the error bars de ne two standard deviations. The water borne noise shows a
8.2.
ONE-THIRD-OCTAVE AIR AND WATER BORNE NOISE
Table 8.4:
203
Source
Conversion
Frequency (Hz)
Max. Revs
rpm to Hz
93.3
Gear Ratio
13:24
50.5
Propeller Blade
4 blades
202.2
Characteristic frequencies of the Mariner 50 HP outboard engine, see Mariner
website.
fairly consistent broadband noise with no distinguishing peaks or troughs. The SPL is greater at low frequencies than high frequencies by approximately 20 dB. The port and starboard aspects are very similar, especially at frequencies below 500 Hz. Above 500 Hz, the starboard side is slightly louder by 5 dB. The airborne noise, on the other hand, does not show a consistent broadband noise; there is considerably more variation in the spectrum with a 30 dB dierence from low to high frequencies and there is a noticeable peak at 250 Hz (which matches a peak in the background noise). The error in the airborne noise is larger than that of the water borne noise. This means that the airborne noise on the port and starboard side cannot be distinguished at frequencies below 3 kHz but above 3 kHz is appears that the starboard side is slightly louder than the port side.
It was originally thought that the port and starboard aspects would generate almost identical noise spectra because the boat is symmetrical.
The engine is symmetrical with
the exhaust outlet in the centre of the propeller. The transverse weight distribution is symmetrically across the centreline of the boat because the helmsmen was on the starboard and the one crew member was on the port; however, the longitudinally weight distribution was not symmetrical because the helmsmen is next to the transom but the crew is near centre. This will cause the boat to deform torsionally which could cause the dierence between the port and starboard aspects.
There are two main sources for noise on the D-class; the engine (including propeller) and the uid ow over the hull. Gotz et al. (2009) says that small vessels often resonate at the frequency of either the engine revolutions, gear box or blade frequency.
The characteristic
frequencies of the Mariner 50 HP (two stroke, 3 cylinder) outboard engine are shown in table 8.4. The dominant vibrations generated by the outboard engine are low frequencies and do not align with any peaks in the noise spectra. There is a signi cant peak in the airborne noise at 250 Hz but this cannot be caused by the propeller blade frequency because the propeller is under water and the 250 Hz peak is only in the airborne spectrum.
The disadvantage
of a one-third-octave spectrum is the wide band width of each frequency bin. For example, the one-third-octave band that will include the maximum revolution frequency ranges from 89.1 Hz to 112 Hz so it has the eect of averaging out the peak.
A narrow band analysis
was performed and the engine characteristics could be identi ed; however, a thorough narrow band analysis is outside the scope of this thesis and will not be explored further.
204
CHAPTER 8.
NOISE RESULTS
(a) Water borne noise.
(b) Airborne noise. Figure 8.3: Mean one-third-octave noise spectra measured at 3 psi comparing the port and starboard aspects.
8.3.
ERROR ANALYSIS
205
8.2.2 Hydroelastic noise Let us now consider how the internal pressures of the sponson and keel will aect the air and water borne noise generation. The mean of the port and starboard aspects at 3 and 4 psi are shown in gure 8.4 and the error bars de ne two standard deviations. The water borne noise spectra show that above 500 Hz there is a dierence between the internal pressures and the airborne noise spectra agree with this for frequencies above 1000 Hz. The eect of the internal pressures is surprisingly similar to the dierence between the port and starboard spectra. If the dierence between the aspects is due to torsional deformation then it could be assumed that changing the internal pressures has aected the torsional twist in the hull, which gives it a similar appearance to the eect of aspect. It can be seen that the low frequency noise (< 1000 Hz) generated at 4 psi had signi cantly more variation than the noise generated at 3 psi. This is because the 3 psi runs were continuous, i.e. the D-class didn't stop and each run was approximately one minute apart. Whereas, the 4 psi runs were completed in pairs so that there was a 15 minute gap in between the pairs of runs. In these 15 minutes the environmental parameters, such as water depth and CPA, varied which gave the data more variation.
8.2.3 Sound pressure level at one metre The ImTL model was used to calculate the transmission loss of the water borne noise through the shallow water sound channel.
The starboard side (the louder side) spectrum of each
hydrophone measurement was propagated back to one metre then the average of the two hydrophone depths was found and this is shown in gure 8.5. Ideally, the SPL at one metre from both hydrophones should be identical and the dierence can be used to judge the accuracy of the measurements and the ImTL model.
It shows fairly good agreement with frequency
above 50 Hz. This was expected because the shallow water sound channel will act as a low pass lter, see Robinson et al. (2011), so a 20 m depth sound channel will cut-o frequency below approximately 35 Hz. The mean dierence between the two hydrophones above 50 Hz is -0.56 dB with a range of 9.79 dB to -8.65 dB. The SPL at one metre shows that the transmission loss is greater at low frequencies than high frequencies and there is a 60 dB dierence between the low and high frequencies.
8.3 Error analysis There are many sources of error within the noise measurements that will now be discussed and the rst one is the measurement of distance, such as the CPA and water depth. The CPA was estimated using the pole system which gave a combined accuracy of
±
0.68 m but
±
0.5
m was due to random driver error, which cannot be reduced using this method. This accuracy places the measurement in ANSI S12.64 grade C. The uses of a DGPS would increase the CPA accuracy considerably and would also show whether or not the boat was driven in a straight line. The hydrophone placement was very accurate because the drifting pass-by method was
206
CHAPTER 8.
NOISE RESULTS
(a) Water borne noise.
(b) Airborne noise. Figure 8.4: Mean one-third-octave noise spectra comparing the aect of hydroelasticity.
8.4.
PERCEPTION OF AIR AND WATER BORNE NOISE
207
Figure 8.5: Mean one-third-octave noise spectra propagated to one metre.
employed; therefore, the hydrophone line was vertical. A sh nder was used to measure the water depth with an accuracy of
±
2.5 %, which is acceptable but it did con rm that the
seabed was at. The resolution of the microphone and hydrophone measurements were su ciently high but neither have been recently calibrated; therefore, the measurements were very precise but the trueness is unknown. The environmental conditions (wind, waves, temperature, etc) were measured by external parties so the trueness is unknown, but the errors induced by these measurements are minimal. The measured frequency range was 20 Hz to 20 kHz which is very close to ANSI grade B (20 Hz to 25 kHz). The entire airborne noise spectrum was captured because the SPL of the D-class is very close to the background noise at either end of the spectrum, see gure 8.3. This is not true for the water borne noise and all of the high frequencies were not captured. Even an ANSI grade A (up to 50 kHz) may not capture the entire frequency range of the D-class and this should be improved in future work. The shallow water channel cuts o frequencies below 35 Hz (in 20 m water depth) so it is unknown whether the entire low frequency noise was captured but if the D-class always operate in shallow water then this low frequency noise will never propagate.
8.4 Perception of air and water borne noise A case study was performed on one specie to assess the perception of the noise generated by the D-class and relate the frequency dependent noise to the frequency dependent auditory system of that specie. The harbour seal (
Phoca vitulina )
was chosen as a case study species
because James et al. (2010) states that there are seals in the English Channel and an in-depth report on the Solent seals population was performed by Chesworth et al. (2010). The author has also observed seals in Langstone harbour, which is approximately 15 km from the test site.
208
CHAPTER 8.
NOISE RESULTS
The harbour seal is unique for this case study because they inhabit both the land and water, thus allowing both the air and water borne noise of the D-class to be assessed on the same specie. Nedwell et al. (2004) provides four separate water borne audiograms that are within 10 dB agreement of each other and ve airborne audiogram with a 20 dB agreement, shown in gure 8.6. The audiograms were averaged together to give a mean audiogram for the seal that will help remove any intra specie variation or any errors due to the experimental procedure of obtaining an audiogram, as discussed by Nedwell et al. (2004). Anthropogenic noise can have a number of dierent aects from injury to behavioural changes as discussed in the literature review, section 6.3.2. depend on the dierence, of the harbour seal. audible - 0
between the SPL of the D-class and the hearing threshold
Three thresholds of
dBht (Phoca
dBht (Phoca
dBht ,
The aect or perception will
dBht
have been chosen to predict the ZoI; barely
vitulina ), avoidance - 90 dBht (Phoca vitulina ) and risk of injury - 130
vitulina ), see Nedwell et al. (2007).
The author has observed the behaviour of harbour seals in Langstone and made an interesting observation. If a RIB is driven past a beached seal slowly the seal will stay on the beach, whereas, if the RIB is travelling at high speeds then the seal will enter the water. The fact the seals enter the water indicates that water borne noise is not injuring or causing discomfort to the seal, and the seal probably feels more at risk on the shore where it is less mobile. Once the seals are in the water, they often come to the surface to watch the boat pass, which could indicate that the air and water borne noise is perceived dierently or it could mean the seals are able to see the boat more clearly.
8.4.1 Overlay method The rst method to investigate the perception of noise by a specie it to directly overlay the audiogram to the anthropogenic noise under investigation. This was demonstrated by Nedwell et al. (2007) in gure 6.10 and will be called the overlay method. The mean one-third-octave SPL of the D-class at 1 m and 20 m from the starboard (louder) aspect has been overlaid onto the mean audiogram of a harbour seal along with the avoidance (90 (130
dBht )
dBht )
and risk of injury
thresholds and this is shown in gure 8.7.
The overlay of water borne SPL and water borne audiogram, see gure 8.7a, shows that the threshold for injury is much greater than the water borne SPL of the D-class at one metre; therefore, the auditory system of a seal will not be damaged even one metre from the D-class. The SPL at 20 m is greater than the hearing threshold of the seal meaning the seal will de nitely hear the D-class at distances of 20 m or more. The threshold for avoidance is above 10 dB greater than the SPL at one metre suggesting the seal will not avoid the D-class at one meter, although the 90 dB is an estimate so it is expected that the seal will respond in some manner. The overlay of airborne SPL and the airborne audiogram is shown in gure 8.7b. The airborne noise spectrum shows a similar story to the water borne noise spectrum; the noise at one metre is not going to cause damage to the auditory system (either PTS or TTS) and it is unlike to cause a behavioural response. The seal will hear the D-class at 20 m or more but the perception will be mild.
8.4.
PERCEPTION OF AIR AND WATER BORNE NOISE
(a) Airborne audiogram.
(b) Water borne audiogram. Figure 8.6: Audiogram of a harbour seal, see Nedwell et al 2004 page 237 and 238.
209
210
CHAPTER 8.
NOISE RESULTS
(a) Water borne spectrum.
(b) Airborne spectrum. Figure 8.7: SPL of the D-class vs the audiogram of a harbour seal.
8.4.
PERCEPTION OF AIR AND WATER BORNE NOISE
211
Frequency (Hz)
100
400
1000
4000
8000
16000
SPL at 1m (dB)
172.1
154.3
154.7
151.3
153.1
145.9
Audiogram (dB)
96.0
83.9
81.8
72.0
65.0
62.6
Critical TL(dB)
76.1
70.4
72.9
79.2
88.1
83.2
Table 8.5: Water borne barely-audioable threshold of a harbour seal.
8.4.2 Critical transmission loss method A novel method to investigate the perception of noise and to predict the ZoI is to track the vertical and horizontal Transmission Loss (TL) through the shallow water sound channel until a perception threshold (audible, avoidance or injury) is crossed and will be called the critical TL method. The vertical position of the specie in the sound channel can now be investigated. The author only knows of work investigating the horizontal ZoI. The problem is now three dimensional (frequency, horizontal and vertical) so the ZoI are divided into six frequency bands; 100, 400, 1000, 4000, 8000 and 16000 Hz. These bands were chosen because the audiogram frequencies directly overlap with the one-third-octave frequencies and no extrapolation was required to nd the
dBht .
The outlay method showed that the SPL at 1m of the D-class was
less than the avoidance and injury thresholds of the seal at all frequencies so the novel method will only investigate the barely audible threshold. The critical TL that is required so that the D-class is barely audible to a seal is summarised in table 8.5. This demonstrates that at each frequency a dierent TL is required for the D-class to become inaudible to the seal. The ImTL model was used to plot the sound eld up to the critical TL and this is shown in gures 8.8 and 8.9. The colour bar de nes the TL and the white regions de ne where D-class becomes inaudible to a seal. The input values for the ImTL model were the same as the ones used to propagate the SPL of the D-class to one metre. Figure 8.8 shows the sound elds up to 20 km from the D-class and this demonstrates how the propagation of sound through a shallow water channel is highly frequency dependent. The barely audible ZoI at 100 Hz stretches only 2.5 km but at 4000 Hz the D-class could potentially be heard up to 15 km away. At 16 kHz the barely audible ZoI ranges to over 5 km. The sound eld also shows that the re ections of the sound waves can cancel and amplify the sound so the edge of the ZoI is not horizontal through the sound channel. The water's surface release a proportion of the sound pressure which means the ZoI is much shorter near the sea surface and sea bottom. This means that wildlife near the sea surface (marine mammals for breathing) and sea oor (bottom feeders such as skate) are slightly less exposed to the noise.
It was observed in Langstone harbour that the seals
come to the surface to watch a boat pass. It could, in part, be due to the sea surface reducing the SPL. The sound elds up to 1 km are shown in gure 8.9 and provide a more detailed view of the sound eld at levels that are audible to seal. The cancellation and ampli cation of the sound by the re ected sound waves means that a seal could potentially alter its vertical position, without changing it horizontal range, to vary the received anthropogenic noise by up to 30 dB. The critical TL method is based on a few assumptions that will now be discussed to show
212
CHAPTER 8.
NOISE RESULTS
Figure 8.8: Threshold of hearing the D-class by a harbour seal, range 20km.
8.4.
PERCEPTION OF AIR AND WATER BORNE NOISE
Figure 8.9: Threshold of hearing the D-class by a harbour seal, range 1km.
213
214
CHAPTER 8.
NOISE RESULTS
its limitations. Firstly, the bathometry is assumed to be constant and at. This is not true in reality and the depth will have a dramatic eect on the noise propagation. The acoustical sediment properties are also assumed to be constant and over a range of 20 km this is not true; although, the depth variation is likely to have a greater aect than the sediment variation. There are possible errors in the audiograms discussed by Nedwell et al. (2004) that will not be explored in this work. The entire water borne noise spectrum was not captured, especially high frequencies, so the full eect on the seals cannot be predicted. Also the avoidance and injury thresholds are only estimates and the true accuracies of these are unknown. Plus there are errors from the measurement of the D-class discussed in section 8.3. There are also inherent errors in the ImTL model. The nal limitation of this method is the background noise. If the SPL of the D-class drops below the background noise then it becomes inaudible to a specie and this will be discussed in the next paragraph. The critical TL method could be regarded as the worst case scenario because it assumes that there is no background noise and it is the greatest distance that a harbour seal could potentially hear the D-class with these xed environmental conditions.
8.4.3 Background transmission loss method The author suggests another novel method to de ne the environmental impact of anthropogenic noise by comparing to the background noise. The anthropogenic noise becomes unrecognisable to any specie if the TL causes the SPL of the anthropogenic noise to drop below the background noise. It could be said that this de nes the localised region (in time and space) of environmental impact by anthropogenic noise; however, this method is then dependent on the background noise which will vary constantly due to factors such as local shipping, weather condition, water depth, etc. Most importantly, it does not consider what happens when the background noise drops because the region of impact will increase in size in proportion to the decrease in background noise. Nonetheless, this method is very good at de ning the region in which the anthropogenic noise can be recognised over the background noise at the speci c time and location of the measurements. This method is not specie dependent and allows a more general assessment of the environmental impact of the D-class.
This method will be
called the background TL method. The localised region of environmental impact by the anthropogenic noise of the D-class can be seen in gure 8.10 using the background TL method. This produces similar patterns to the critical TL method where the SPL varies with both water depth and range. The primary dierence is the range of the impact region; after 2 km the anthropogenic noise generated by the D-class is not audible above the background noise. The frequency dependence shows that the higher frequencies can propagate over twice the distance of lower frequencies. The background TL method is based on many of the same assumptions as the critical TL method including; at bathometry, constant sediment properties, errors in the measurement of the D-class and background noise, and inherent error of the ImTL model. The systematic errors in the measurement of the D-class and background noise will be identical because they were measured with identical systems; therefore, the systematic errors will not aect these
8.4.
PERCEPTION OF AIR AND WATER BORNE NOISE
Figure 8.10: Audible threshold based on the background noise.
215
216
CHAPTER 8.
NOISE RESULTS
results. This method does assume that the background noise is constant throughout the water channel. There was a dierence in background noise measured at the two hydrophone depths, particularly at low frequencies below 700 Hz with a maximum dierence of 10 dB, which makes this assumption invalid.
8.5 Summary The maximum airborne AS-weighted SPL of the D-class (under standard operating conditions with two crew) was 77.1 dB(AS) at 25 m. This is over the EU regulation and higher than the previous experimental results of larger and faster craft. The maximum water borne S-weighted SPL of the D-class (under standard operating conditions with two crew) was 144.5 dB(S) at 20 m. No trend was found between the air or water borne S-weighted SPL and the SOG. The one-third-octave spectra of the air and water borne noise were analysed. The starboard side was slightly louder at higher frequencies. There were no distinct frequency peaks in either the air or water borne spectra that aligned with the frequencies of the engine revolutions, gear box or blade frequency as suggested by Gotz et al. (2009). It was hypothesised that the exibility of the fabric hull stopped it resonating at the engine frequencies. When the D-class operated at higher internal pressures there was less water borne noise generated above 500 Hz and less airborne noise generated above 1000 Hz. The perception of the air and water borne noise was assessed and a case study was performed on the harbour seal.
First, the overlay method showed that the harbour seal will
not be injured by the air or water borne noise, even at one metre, but the harbour seal will de nitely hear the D-class over 20 m away. The critical TL method showed that in the worst case scenario the D-class could be heard up to a maximum of 15 km away at 4 kHz; although, this assumes that there is no background noise and the range is less at the other frequencies investigated. The background TL method showed that after two kilometres the SPL of the D-class drop below the background noise meaning that the D-class become unrecognisable to any species after that range (at the xed environmental conditions of the test). The new methods showed that the vertical position of a specie in the sound channel could vary the received SPL by up to 30 dB.
Chapter 9
Conclusion This thesis set out to explore the implications of intentionally allowing a planing craft to be exible and studying the eects of the exibility. The study asks; how does hydroelasticity aect the performance, de ned by the boat motion, forward speed and noise, of a planing craft? Furthermore, the perception of the boat motion and noise must also be considered because there can be signi cant dierences between the physical vibration and the perception of the vibration. The work focused on the RNLI D-class in atable lifeboat but the results are not limited to the D-class because the eect of hydroelasticity on the D-class can be compared to other planing craft and HSC. The D-class is a 5 m in atable boat that is capable of achieving 25 knots (in the sea state associated with a Beaufort force 2). The thesis measured the two main types of vibrations found within the D-class, which are boat motion and noise, and to predict the perception of these vibrations. Then, the thesis measured the eect of hydroelasticity on the vibrations, and nally, the thesis predicted the eect of hydroelasticity on the perception of the vibrations. The boat motion of a HSC is an especially important area of research because HSC expose the on-board crew to high levels of vibration. The European directive (2002/44/EC) set the EAV and ELV for the human exposure to vibration. The eects of WBV ranges from chronic and acute, to physiological and psychological, see Ensign et al. (2000). Townsend (2008); Allen et al. (2008); Myers et al. (2011) showed that the levels of vibration measured within RIBs can rapidly exceed both the EAV and ELV. Many reduction strategies are currently being investigated by other researchers, including; suspension seats, suspension decks, active and passive ns, trims tabs, interceptors, gyrostabilisers, exible hulls and elastomer coated hulls, see Townsend et al. (2012a). Although, Coe et al. (2013) and Coats et al. (2003) concluded that a combination of solutions will be required to meet the European directive. The thesis questions whether or not hydroelasticity can be used as a novel method to reduce the WBV and help meet the European directive. The D-class and all surface ships generated noise in two domains: air and water. There are standards (ISO 14509) and regulations (European directive 2003/44/EC) for airborne noise of powered recreational craft but there are no standards or regulations that correctly apply to vessels similar to the D-class. This is a critical area of research because only a few small HSC 217
218
CHAPTER 9.
CONCLUSION
have been measured in the water domain, see Amoser et al. (2004); Kipple and Gabriele (2003, 2004b), and even fewer have attempted to quantify the eect of the water borne noise on the wildlife; therefore, the environmental impact of this type of craft is not understood.
Small
craft (LWL < 10 m) normally operate in shallow waters, which means the water borne noise propagates through a shallow water channel.
A method is required to accurately measure
the water borne noise of small HSC in shallow water, including a method to predict the environmental impact through a shallow water channel. The structural domain was the rst element of hydroelasticity to be investigated.
The
structure of the D-class was divided into four exible components: in atable sponson, hinge deck panels, in atable keel and fabric hull. The eects of hydroelasticity on the performance could only be considered once the exible components were identi ed.
The exibility was
shown to aect the performance in three dimensions and this lead to the three aspects of hydroelasticity: hydroelastic slamming, hydroelastic planning surfaces and global hydroelasticity. Individually, the three aspects of hydroelasticity provide simpli ed and manageable 2D problems that can be experimentally and numerically solved.
Together, the three aspects
give a perspective from which the eect of hydroelasticity on the boat performance can be understood and a framework to study the problem. A multitude of experiments were performed to investigate how hydroelasticity could aect the performance of a planning craft and to answer the research questions. First, the stressstrain relationship of the fabrics used to construct the D-class were measured and the results can be found in appendix D.
Second, a waterline de ection experiment was performed to
give an early assessment of the static deformation and the results can be found in appendix E. Next, the quasi-2D hydroelastic wedge impacting a free surface experiment (drop tests) was completed.
The three aspects of hydroelasticity inspired the third experiment, which
was a four stage full-scale holistic hydroelastic experiment. The sub-experiment was selected to trigger one aspect of hydroelasticity whilst still considering the interaction of the other aspects of hydroelasticity and the coupled structure. The four sub-experiments were: static tests, drop tests, at water trials and wave trials. The nal experiment was the air and water borne noise trials. The main empirical results can be found in chapters 4, 5 and 8 and the following conclusions were drawn from four-stage hydroelastic experiment and the noise trials:
Static tests
The internal pressures of the sponson and keel, and the crew loading did aect the static shape of the D-class.
The keel pressure had a more dominant eect on the static shape than the sponson pressure.
Flat water trials:
The speed of the D-class varied by 0.44 knots due to changing the internal pressure from 2.25 psi and 2 psi to 4.25 psi and 4 psi in the sponson and keel, respectively.
219
The pulsing motion was shown to increase the speed of the D-class by 1.3 knots and this contradicts the results of Dand et al. (2008).
The dominant components that aected the at water performance were the keel and hull.
Drop tests:
Either the pre-tensioned stresses in the hull or the internal pressures of the sponson and keel did aect the peak accelerations.
The pre-tensioned stresses or internal pressures, in turn, did aect the WBV experienced by the crew; although, the frequency weighting of the VDV means the perception is not linearly proportional to the peak acceleration.
The trend between hydroelasticity and the peak acceleration was inverted with drop height.
The eect of hydroelasticity on the peak acceleration was related to the ratio of hydroelastic importance, see Faltinsen (1999).
Super harmonics were shown to cause anomalies in the peak acceleration data.
The dominant components in the full-scale drop tests were the keel and hull, but the dynamic response of the hull was not aected by the keel pressure.
Wave trials:
The internal pressures of the sponson and keel did not statistically aect the accelerations.
The drop tests showed that the trend between hydroelasticity and peak acceleration was inverted with drop height which explains why no statistical dierence was found because the wave height (and thus drop height) varied randomly during the wave trials.
Airborne noise:
The maximum noise generated by the D-class is 77.1 dB (AS), under standard operating conditions with two crew.
This is above the limits set out by the European directive 2003/44/EC.
The D-class is unlikely to cause a PTS or a TTS to a harbour seal at one metre.
A harbour seal will de nitely hear the D-class at 20 m.
The D-class generated less noise at frequencies above 1000 Hz when operating at high pressures.
Water borne noise:
220
CHAPTER 9.
CONCLUSION
The one-third-octave spectra of the noise was predicted at one metre.
The D-class is unlikely to cause a PTS or a TTS to a harbour seal at one metre.
A harbour seal will de nitely hear the D-class at 20 m.
The background TL method revealed that the noise from the D-class will be inaudible to any species after two kilometres (for the xed environmental conditions of the test).
The D-class generated less noise at frequencies above 500 Hz when operating at high pressures.
The rst research question was; how does hydroelasticity aect the boat motion? This was tested using quasi-2D and 3D drop tests, and wave trials. The quasi-2D drop tests showed that the peak acceleration was aected by up to 18 % when the hull stiness was changed on a vee-shaped hull with hard chines; however, the trend was shown to be drop height dependent. Increasing the hull stiness increased the peak acceleration at one drop height but at the other drop height (0.5 m) increasing the hull stiness decreased the peak acceleration. The perception of the acceleration was quanti ed using the VDV and, in contrast to the peak acceleration, the VDV showed that a exible, fabric hull had a lower VDV that a MDF hull in all the quasi-2D drop conditions (up to 14 %) and the trend was not drop height dependent. The quasi-2D drop test results did match the ratio of hydroelastic importance by Bereznitski (2001) and Faltinsen (1999). Townsend et al. (2012a) showed that the hull stiness did not aect the peak acceleration but it is supposed that the hull stiness was not varied su ciently enough to have a measurable eect. As a whole, the 3D drop tests agree with quasi-2D drop tests.
The 3D drop tests demonstrated that the internal pressures of the sponson and keel
of the D-class could aect the peak acceleration by up 39 %; however, again the trend was drop height dependent. The trend in the VDV was shown not to be drop height dependent and low internal pressures always resulted in lower VDV in the crew position (up to 16 %), although the low internal pressures also lead to an increase in VDV on the bow. The drop tests proved that hydroelasticity can aect the peak acceleration and it can reduce the VDV. If the peak acceleration results from the drop tests are applied to a real, random sea, where the wave height and thus drop height will vary random, the eect of hydroelasticity on the peak acceleration will also vary randomly. This was demonstrated using equation 5.1, in section 5.2. Furthermore to the drop tests, the boat motion and the eect of hydroelasticity on the boat motion was also measured during open-water wave trials. There was only one data set that was proven to be statistically dierent to a 95 % certainty and this was expected once the drop test results were considered in the context of a real, random sea. The drop tests demonstrated that hydroelasticity has the potential to be used as a novel method to reduce the WBV and help meet the European directive; however, the wave trials showed that much further work is required to optimise the structural stiness, so that occasionally the peak acceleration would be increased by hydroelasticity but overall hydroelasticity would reduce the VDV. Then this novel reduction strategy could be applied to a real, random sea. The second research question was; how does hydroelasticity aect the forward speed of a
221
planing craft? The at water trials answered this question and showed that a reduction in the internal pressures of the sponson and keel, and thus a reduction in structural stiness, reduced the forward speed of the D-class by 0.44 knots, with a statistical dierence to a 95 % certainty. This is a major disadvantage of hydroelasticity because it reduces forward speed; however, the at water trials also studied a unique phenomenon linked to hydroelastic planing surfaces termed the pulsing motion.
The pulsing motion measured during this study was
actually faster than the non-pulsing runs by 1.3 knots; however, this contradicts the results from previous researchers because Dand et al. (2008) showed that the pulsing motion decreased the forward speed. Further work is required to prove whether or not hydroelasticity (and the pulsing motion) could consistently increase the forward speed. The structure was shown to adopt an unstable equilibrium position of minimum potential energy when pulsing. The nal research question relates to the eect of hydroelasticity on the air and water borne noise. A method for measuring the water borne noise was developed for shallow waters based on the ANSI standard (12.64-2009) and the airborne noise was measured using the ISO standard (14509-1).
The measured airborne noise of the D-class was 77.1 dB (AS), which
was above the airborne noise regulation for powered recreational craft (European Directive 2003/44/EC). The one-third-octave band analysis did reveal that at lower internal pressures (and lower structural stiness) the D-class generated higher noise levels at frequencies above 1000 Hz and 500 Hz for the air and water borne noise, respectively. This highlights another disadvantage for hydroelasticity because the noise results suggest that increasing the exibility can increase in the noise generation. The perception of the air and water borne noise was also considered in the thesis by comparing the measured noise to an audiogram of a harbour seal. This showed that even one metre from the D-class a harbour seal is unlikely to be injured (either PTS or TTS) by the air or water borne noise; however, the seal will de nitely hear the D-class at the measurement distance of 20 m. The thesis has proven that hydroelasticity can aect the peak acceleration and VDV during a drop test, the at water forward speed and, the air and water borne noise, but; can hydroelasticity be used to improve the performance? The drop tests showed that the VDV can be reduced by decreasing the structural stiness but the wave trials reveal that much further work will be required to provide an overall reduction in the VDV in a real, random sea. On the one hand, the at water trials showed that hydroelasticity can reduce the forward speed but, on the other hand, they showed that the hydroelastic pulsing motion can increase the forward speed; although, the pulsing motion requires more research because it contradicts Dand et al. (2008). The noise trials showed that hydroelasticity could increase the noise levels. Another key parameter that hydroelasticity and membrane structures will aect is the mass of the craft because Wood (2011) found that the mass of the hull could be reduced by 20 % if it is treated as a exible membrane structure. This is a massive bene t to hydroelasticity. Suppose that the structural stiness is optimised in the future to provide an overall reduction in WBV and the mass of the hull is reduced because it is a exible membrane structure, then the weight saving can be used to either; reduce the engine size and decrease the noise levels, incorporate more WBV reduction strategies to have a signi cant impact on the VDV or use
222
CHAPTER 9.
CONCLUSION
the weight saving to increase the at water forward speed and/or fuel e ciency. Halswell et al. (2012) hypothesised that the energy in the boat motion is mainly a consequence of the energy from the engine; therefore, by reducing the vertical boat motion at the source, more energy from the engine can be used to increase the forward speed and/or fuel e ciency. The concept of using hydroelasticity to reduce the WBV has political implications because planing craft, RIBs and IBs currently struggle to meet the European directive. Hydroelasticity tackles the problem at the source (unlike the other reduction strategies that are currently available) and still allows other WBV reduction strategies to be incorporated into the design of the craft. This would change the fundamental theories of planing and slamming because the problem switches from hydrodynamic to hydroelastic. Nevertheless, this concept pivots on the ability to optimise the structural stiness so that it provides an overall reduction to the VDV. A program of research will be required to achieve this end goal. The rst step is to extend the quasi-2D drop tests because the measurements made here only scratched the surface. A methodical study into the eect of the hull stiness, deadrise angle and impact velocity (drop height) should be made to systematically characterise the eect of hydroelasticity on the peak acceleration and VDV. Then a prototype hull can be designed where the hull stiness has been optimised according to the deadrise and expected impact velocities to minimise the VDV. The prototype hull should then be extensively tested in a towing tank; initially with regular waves that are proportional to the expected impact velocities. Next, tests in irregular waves should be performed to prove that hydroelasticity will reduce the overall VDV; although, occasionally it might increase the peak acceleration. Finally, at water towing tank tests could be completed to check that forward speed was not dramatically aected by the incorporation of hydroelasticity and that the pulsing motion is not present. The water borne noise measurement method presented in section 7.1 also has political implications because ISO are currently developing a deep water measurement method for ships and wish to develop a shallow water method in the near future. The method developed in this thesis could be used as a framework for future shallow water noise measurements for HSC. The author could not nd any measurements of HSC in shallow water that properly considered the interaction of the sound waves with the seabed and sea surface; therefore, the method developed here is novel because it does considers this interaction. The interaction was modelled using an ImTL model because Robinson et al. (2011) compared it to several other models and they found the ImTL model to be the best. The ImTL model allows the SPL at one metre to be predicted, which is convention for water borne noise. The method developed here measured both the air and water borne noise simultaneously, which has a knock on political impact because they are currently measured and analysed independently. A conclusion of the SoundBoat project was that the airborne noise could be reduced by su ciently submerging the exhaust; however, submerging the exhaust is highly likely to increase the water borne noise. This highlights the importance of assessing the air and water borne noise together. The study into the perception of noise is also unique because only been a few water borne noise studies have considered the full picture, from measurement to perception, including Erbe (2002); Nedwell et al. (2012); Nedwell and Mason (2012); although, none of the studies
223
considered a planing craft. The ImTL model was used to predict the vertical and horizontal TL through a shallow water sound channel. This allowed the measured noise to be compared to the audiogram of a harbour seal as a case study. During the case study, the author developed an interesting noise assessment method that was not species dependent, called the background TL method. Anthropogenic noise becomes inaudible to any species if the TL caused the SPL of the anthropogenic noise to drop below the background noise.
This allows the maximum
distance that a noise is audible to any species to be predicted. It is conventional to assume that the structure of a planing craft is rigid when predicting the performance; however, this thesis has proven that the exibility and hydroelasticity of a planing craft can aect the performance.
Hydroelasticity was shown to aect nearly all
aspects of performance measured during this thesis, except the performance in waves that was explained using the drop test results. This has opened up a new line of research within the area of planing craft because hydroelastic planing craft have rarely been considered in the past. The thesis proposed the use of hydroelasticity to meet the European directive (2002/44/EC) on the human exposure to vibration. The thesis has also provided a framework for measuring and assessing the air and water borne noise generated by HSC in shallow waters. All in all, the thesis has answered the research questions by measuring the performance of a planing craft and the eect of hydroelasticity on the performance, and the thesis has shown the potential advantage and disadvantages of hydroelasticity.
224
CHAPTER 9.
CONCLUSION
Bibliography Ainslie, M. A., Harrison, C. H., and Burns, P. W. (1994). INSIGHT. In
Reverberation modelling with
Proceedings of Institute of Acoustics, volume 16, pages 105112.
Ainslie, M. A. and McColm, J. G. (1998). absorption in sea water.
A simpli ed formula for viscous and chemical
The Journal of the Acoustical Society of America,
103(3):1671
1672. Allen, D. P., Taunton, D. J., and Allen, R. (2008). A study of shock impacts and vibration dose values onboard high- speed marine craft.
International Journal of Maritime Engineering,
150(A3):114. Amoser, S., Wysocki, L. E., and Ladich, F. (2004).
Noise emission during the rst power-
boat race in an Alpine lake and potential impact on sh communities.
Acoustical Society of America, 116(6):37893797.
The Journal of the
Austen, S. and Fogarty, H. F. J. (2004). Design and development of a new in atable lifeboat.
International Journal of Small Craft Technology. Bachman, R. T. (1994). A three-dimensinoal geoacoustic model for the catalina basin. Technical Report 1669, Naval Command, Control and Ocean Surveillance Centre, California. Bagnell, D. G. (1998). The development of a rigid in atable vee-bottom boat. In
Conference on Rigid In atables. RINA.
International
Begovic, E., Bertorello, C., and Pennino, S. (2014). Planing hull seakeeping in irregular head seas.
Transactions of Famena, XXXVIII(3):112.
Bereznitski, A. (2001).
Slamming:
the role of hydroelasticity.
Progress, 4(48):333351.
International Shipbuilding
Betke, K., Glahn, M. S., and Matuschek, R. (2004). Underwater noise emissions from oshore wind turbines. In
CFA/DAGA, pages 12, Oldenburg, Germany.
Bishop, R. E. D. and Price, W. G. (1979).
Hydroelasticity of Ships.
Cambridge University
Press, 1st edition. Blackwell, S. B. and Greene, C. R. (2005). Underwater and in-air sounds from a small hovercraft.
The Journal of the Acoustical Society of America, 118(6):36463652. 225
226
BIBLIOGRAPHY
Bourget, M. (1866). Memoire sur le movement vibratoire des membranes circulaires.
De L'ecole Normale 3. Brekhovkikh, L. M. and Lysanov, Y. (2003).
Fundamentals of ocean acoustics.
Annales
AIP Press
Springer Verlag, Berlin, 3rd edition. Bubliü, I., Skelin, D., and Vukadin, P. (2008). Measurement of vessel underwater noise signature. In
ELMAR 50th International Symposium, volume 2, pages 407410.
Cavallaro, P., Johnson, M., and Sadegh, A. (2003). in ated structures. CEProof
Mechanics of plain-woven fabrics for
Composite Structures, 61(4):375393.
(2013).
Sonic
boom.
Online:
www.ceproof.com/CE-
Marking/EU/Marine/Services/Sound-Testing.html. Accessed: 05/01/2013. Chesworth, J. C., Leggett, V. L., and Rowsell, E. S. (2010).
Solent seal tagging project
summary report. Technical report, Wildlife Trusts South East Marine Programme. Chiu, F. and Fujino, M. (1989). Nonlinear prediction of vertical motions and wave loads of high-speed crafts in head sea.
International Shipbiulding Progress, 36(406):193232.
Choura, S. and Arabia, S. (1997).
Suppression of structural vibrations of an air-in ated
membrane dam by its internal pressure.
Computers and Structures, 65(5):669677.
Chuang, S.-L. (1966). Slamming of rigid wedge-shaped bodies with various deadrise angles. Technical Report DTMB-2268, David Taylor Model Basin, Washington, D.C. Coates, R. (1988).
An empirical formula for computing the Beckmann-Spizzichino surface
re ection loss coe cient.
Control, 35(4):522523.
IEEE Transactions: Ultrasonics, Ferroelectrics and Frequency
Coats, T., Gowing, S., and Shen, Y. (2009). Poruous hulls research - phase 1. Technical report, Naval surface warfare center, Virginia, USA. Coats, T., Haupt, K., and Lewis, J. (2003). Shock mitigation for personnel onboard high-speed combatant craft. In
In Human Systems Integration Symposium, VA.
Coe, T. E., Rutherford, K. T., Dyne, S., and Hirst, J. (2013). Technical solutions for shock mitigation on high speed government craft. In
Craft, pages 112, Poole, UK.
SURV 8 - Surveillance, Search and Rescue
Coe, T. E., Xing, J. T., Shenoi, R. A., and Taunton, D. J. (2009). A simpli ed 3-D human body - seat interaction model and its applications to the vibration isolation design of high-speed marine craft.
Ocean Engineering, 36(9-10):732746.
Comer, R. L. and Levy, S. (1963).
AIAA, 1(7):16521655.
De ection of an in atable cylindrical cantilever beam.
BIBLIOGRAPHY
227
Cooper, M. D. and McCue, L. S. (2011). Experimental study on deformation of exible wedge upon water entry. In
High Speed Marine Vehicles, Naples.
Cripps, R. M., Cain, C. F., Phillips, H. J., Rees, S. J., and Richards, D. (2004). Development of a new crew seat for all weather lifeboats. In
Craft, RINA, pages 6976.
SURV 6 - Surveillance, Pilot and Rescue
Dakshina Moorthy, C. M., Reddy, J. N., and Plaut, R. H. (1995). Three-dimensional vibrations of in atable dams.
Thin-Walled Structures, 21(4):291306.
Dand, I. (2002a). Propeller open water characteristics and strut drag for a Mariner 40 outboard motor. Technical Report C3356.01, BTM SeaTech Ltd. Dand, I. (2002b). Resistance measurements with an RNLI D-class lifeboat. Technical Report C3356.02, BMT SeaTech Ltd. Dand, I. (2003a). Full scale calm water trails on a RNLI D-boat. Technical Report C3356.05, BTM SeaTech Ltd. Dand, I. (2003b).
Resistance experiments with an RNLI D-class model.
Technical Report
C3356.04, BMT SeaTech Ltd. Dand, I. (2003c).
The eect of LGC on the hydrodynamics of an RNLI D-class in atable.
Technical Report C3356.03, BTM SeaTech Ltd. Dand, I. (2004).
RNLI D-class model:
seakeeping measurements in head seas.
Technical
Report C3356.06, BMT SeaTech Limited. Dand, I., Austen, S., and Barnes, J. (2008). The speed of fast in atable lifeboats.
Journal of Small Craft Technology, 150(B2):2332.
International
Dand, I. W. (2002c). The eect of water depth on the performance of high speed craft. In
High Performance Yacht Design Conference, Auckland. Ensign, W., Hodgdon, J. A., Prusaczyk, W. K., Shapiro, D., and Lipton, M. (2000). A survey of self-reported injuries among special boat operators. Technical report, Naval Health Research Centre, San Diego. Erbe, C. (2002).
Underwater noise of whale-watching boats and potential eects on killer
whales (Orcinus Orca), based on an acoustic impact model.
Marine Mammal Science,
18(2):394418. Erbe, C. and Farmer, D. M. (2000). A software model to estimate zones of impact on marine mammals around anthropogenic noise.
The Journal of the Acoustical Society of America,
108(3:1):132731. Everest, F. A. and Pohlmann, K. C. (2001). edition.
Master Handbook of Acoustics.
McGraw-Hill, 5th
228
BIBLIOGRAPHY
Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences,
Faltinsen, O. M. (1997).
The eect of hydroelasticity on ship slamming.
355(1724):575591. Faltinsen, O. M. (1999).
Water entry of a wedge by hydroelastic orthotropic plate theory.
Journal of Ship Research, 43(3):180193. Faltinsen, O. M. (2005).
Hydrodynamics of high-speed marine vehicles.
Cambridge University
Press. Faltinsen, O. M., Kvalsvold, J., and Aarsnes, J. V. (1997). Wave impact on a horizontal elastic plate.
Journal of Marine Science and Technology, 2(2):87100.
Faltinsen, O. M., Landrini, M., and Greco, M. (2004).
Slamming in marine applications.
Journal of Engineering Mathematics, 48(3/4):187217. Faraday, M. (1831). Acoustic streaming over vibrating plates.
the Royal Society, pages 299318.
Philosophical Transactions of
Fichter, W. B. (1966). A theory for in ated thin-wall cylindrical beams. Technical Report D-3466, NASA, Washington, D.C. Fridsma, G. (1969). A systematic study of rough water performance of planing boats: Regular waves. Technical report. Gao, X., Bl, Y., and Miao, G. (2008). Flow noise measurement of surface ship with towed model.
Journal of Hydrodynamics, Ser. B, 20(6):784789.
Gere, J. M. and Goodno, B. J. (2009).
Mechanics of Materials. Cengage Learning, 7th edition.
Gordnier, R. E. (2009). High delity computational simulation of a membrane wing airfoil.
Journal of Fluids and Structures, 25(5):897917. Gotz, T., Hastie, G., Hatch, L. T., Raustein, O., Southall, B. L., Tasker, M., and Thomsen, F. (2009). Overview of the impacts of anthropogenic underwater sound in the marine environment.
OSPAR Biodiversity Series, (441):1133.
Haiping, H. E., Bretl, J. P. E., VanSumeren, H., Savander, B., and Troesch, A. W. (2005). Model tests of a typical rib in waves.
Private Communication.
Halswell, P. K. (2013). Measuring the hydroelasticity of in atable boats to reduce injuries in harsh waters.
National Instruments Case Study.
Halswell, P. K., Wilson, P. A., Taunton, D. J., and Austen, S. (2012). Hydroelastic in atable boats: Relevant literature and new design considerations. 154(Part B1):3950. Hamamoto, T., Hayashi, T., and Fujita, K. (1996).
3D BEM-FEM coupled hydroelastic
analysis of irregular shaped, module-linked large oating structures.
and Polar Engineering Conference, 1:362369.
International Oshore
BIBLIOGRAPHY
229
Hirdaris, S. E. and Temarel, P. (2009).
Hydroelasticity of ships:
recent advances and fu-
Proceedings of the Institution of Mechanical Engineers, Part M: Journal of Engineering for the Maritime Environment, 223(3):305330. ture trends.
Hot Ribs (2011). Rubber coated fabric construction. Online: www.hotribs.com/02articles/059tube-materials/in atable-boat-tube-material.asp. Accessed: 03/09/2011. Hsieh, J., Plaut, R. H., and Yucel, O. (1989). Vibrations of an inextensible cylindrical membrane in ated with liquid.
Journal of Fluids and Structures, 3(2):151163.
Hsieh, J. C. and Plaut, R. H. (1990). Free vibrations of in atable dams.
ACTA Mechanica,
85(3-4):207220. James, J., Pearce, B., Coggan, R., Arnott, S., Clark, R., Plim, J., Pinnion, J., Barrio Frojan, C., Gardiner, J., Morando, A., Baggaley, P., Scott, G., and Bigourdan, N. (2010).
The
south coast regional environmental characterisation. Technical Report OR/09/051, British Geological Survey. Jenkins, C. and Korde, U. (2006). Membrane vibration experiments: A historical review and recent results.
Journal of Sound and Vibration, 295(3-5):602613.
Jenkins, C. H. (1999). Membrane vibrations: a review and new experiment results. In
Joint Applied Mechanics and Material Summer Conference, Blacksburg VA.
ASME
Jenkins, C. H. and Tampi, M. (2000). Local membrane vibrations and in atable space structures. In
Space 2000, Albuquerque NM.
Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 369(1947):2892
Kapsenberg, G. K. (2011). Slamming of ships: where are we now?
2919. Kipple, B. and Gabriele, C. (2003). Glacier Bay watercraft noise. Technical Report NSWCCD71-TR-2003/522, Naval Surface Warfare Center. Kipple, B. and Gabriele, C. (2004a). Glacier Bay Underwater Noise - August 2000 Through August 2002. Technical Report December 2003, Glacier Bay National Park and Preserve. Kipple, B. and Gabriele, C. (2004b).
Glacier Bay watercraft noise - Noise characterization
for tour, charter, private, and government. Technical Report NSWCCDE-71-TR-2004/545, Naval Surface Warfare Center. Korobkin, A. A., Guéret, R., and Malenica, S. (2006). Hydroelastic coupling of beam nite element model with Wagner theory of water impact.
Journal of Fluids and Structures,
22(4):493504. Kvålsvold, J. and Faltinsen, O. M. (1995). Hydroelastic modeling of wet deck slamming on multihull vessels.
Journal of Ship Research, 39(3):225239.
230
BIBLIOGRAPHY
Lanpheer, R. A. (1993). Recreational motorboat sound level test report.
Marine Environment Committee.
A Report to ICOMIA Marine
Lanpheer, R. A. (1998). Merritt island sound level test report.
Environment Committee. Lanpheer, R. A. (1999).
A Report to ICOMIA
Standard boat concept sound level test report - stern drives with
integral exhaust systems - Lake X Florida.
Committee. Lanpheer, R. A. and Lassanske, G. G. (1994).
A Report to ICOMIA Marine Environment Powered recreational craft sound level test
A Report to ICOMIA Marine Environment Committee.
report - Lake X Florida.
Lavro, J., Davis, M., Holloway, D., and Thomas, G. (2007). Experimental analysis of the wet exural mode response of an NPL 6A hydroelastic segmented model.
Mechanical Engineering, 4(2):197.
Australian Journal of
Le Sourne, H., County, N., Besnier, F., Kammerer, C., and Legavre, H. (2003). LS-DYNA applications in shipbuilding. In
4th European LS-DYNA Users Conference, pages 116.
Le Van, A. and Wielgosz, C. (2005). Bending and buckling of in atable beams: Some new theoretical results.
Thin-Walled Structures, 43(8):11661187.
Lee, J. and Wilson, W. A. (2010). Experimental study of the hydro-impact of slamming in a modern racing sailboat.
Journals of Sailboat Technology, SNAME, 1:129.
Leonard, R. W., Brooks, G. W., and McComb, H. G. (1960).
Structural considerations of
in atable reentry vehicles. Technical Report D-457, NASA, Washington D.C. Lewis, S. G., Hudson, D. A., Turnock, S. R., Blake, J. I. R., and Shenoi, R. A. (2006). Predicting motions of high speed RIBs: A comparison of non-linear strip theory with experiments.
Proceedings of the 5th International Conference on High Performance Marine Vehicles (HIPER'06) pp, pages 210224. In
Lewis, S. G., Hudson, D. a., Turnock, S. R., and Taunton, D. J. (2010).
Impact of a free-
falling wedge with water: synchronized visualization, pressure and acceleration measurements.
Fluid Dynamics Research, 42(3):035509.
Lewis, W. J. (2003).
Tension structures: Form and behaviour.
Lloyd, A. R. J. M. (1998).
Thomas Telford.
Seakeeping: Ship behavious in rough weather.
Ellis Horwood, 1
edition. Loeser, H. T. (1999).
Fundamentals of ship acoustics.
SNAME, New Jersey.
Lu, C., He, Y., and Wu, G. (2000). Coupled analysis of nonlinear interaction between uid and structure during impact.
Journal of Fluids and Structures, 14(1):127146.
BIBLIOGRAPHY
231
Madsen, P., Wahlberg, M., Tougaard, J., Lucke, K., and Tyack, P. (2006).
Wind turbine
underwater noise and marine mammals: implications of current knowledge and data needs.
Marine Ecology Progress Series, 309:279295. Main, J. A., Peterson, S. W., and Strauss, A. M. (1994). Load de ection behavior of space based in atable fabric beams.
Journal of Aerospace Engineering, 7(2).
Main, J. A., Peterson, S. W., and Strauss, A. M. (1995). Beam-type bending of space-based in ated membrane structures. Makasyeyev, M. (2010).
Journal of Aerospace Engineering, 8:120.
Two-dimensional unsteady planing elastic plate.
challenges and modeling of hydroelasticity, Edinburgh, UK. Mattiesen, J. (2002).
In
Mathematical
P/D ratio method - identi cation of recreational craft with dierent
sound characteristics using P/D ratio and Frounde number calculation method. Technical report, IMEC/ICOMIA. MCA (2007). The merchant shipping and shing vessels (Control of vibration at work) regulations 2007. Medwin, H. and Clay, C. S. (1998).
Fundamentals of acoustical oceanography academic.
Aca-
demic Press, Boston. Moore, B. C. J. (2012).
An introduction to the psychology of hearing.
Emerald / Brill, 6th
edition. Morfey, C. L. (2001).
The dictionary of acoustics.
Academic Press.
Myers, S. D., Dobbins, T. D., King, S., Hall, B., Ayling, R. M., Holmes, S. R., Gunston, T., and Dyson, R. (2011). Physiological consequences of military high-speed boat transits.
European journal of applied physiology, 111(9):20419. Mysore, G. and Liapis, S. I. (1998). Dynamic analysis of single-anchor in atable dams.
of Sound and Vibration, 215(2):251272.
Journal
Natzijl, P. W. (1998). Bringing a 1.0 metre bouyancy tube to sea on a 18.0 metre rigid hull. In
International conference on rigid in atables, Weymouth, UK.
Nedwell, J. and Edwards, B. (2002). Measurements of underwater noise in the Arun River during piling at County Wharf, Littlehampton. Technical Report 513R0108, Subacoustech Ltd. Nedwell, J., Edwards, B., Turnpenny, A. W. H., and Gordon, J. (2004). mammal audiograms: A summary of available information.
Fish and marine
Technical Report 543R0214,
Subacoustech Ltd. Nedwell, J., Mason, T., Barham, R., and Cheesman, S. (2012). Assessing the environmental impact of underwater noise during oshore windfarm construction and operation.
232
BIBLIOGRAPHY
Nedwell, J., Turnpenny, A. W. H., Langworthy, J., and Edwards, B. (2003). Measurements of underwater noise during piling at the Red Funnel Terminal, Southampton, and observations of its eect on caged sh. Technical Report 558R0207, Subacoustech Ltd. Nedwell, J. R. and Mason, T. I. (2012). Modelling of noise during impact piling operations at the neart na gaoithe oshore wind farm in the rth of forth. Technical report, Subacoustech Environmental Ltd. Nedwell, J. R., Parvin, S. J., Edwards, B., Workman, R., Brooker, A. G., and Kynoch, J. E. (2007). Measurement and interpretation of underwater noise during construction and operation of oshore windfarms in UK waters.
Technical Report 544R0738 to COWRIE
Ltd, Subacoustech Ltd. Nedwell, J. R. and Turnpenny, A. W. (1998). The use of a generic frequency weighting scale in estimating environmental eect. In
Workshop on Seismics and Marine Mammals, London,
UK. Newman, B. (1987).
Aerodynamic theory for membranes and sails.
Sciences, 24(1):127. Newman, J. (1994). Wave eects on deformable bodies.
Progress in Aerospace
Applied Ocean Research, 16(1):4759.
Ochi, M. (1964). Extreme behavior of a ship in rough seas: slamming and shipping of green water.
Society of Naval Architects and Marine Engineers.
Oshoremarine
(2013).
Mariner
50
hp
outboard
engine.
Online:
www.oshoremarine.ie/classi ed/mar50hp2lg.jpg. Accessed: 22/11/2013. Olausson, K. (2012).
Vibration mitigation for high speed craft.
PhD thesis, Royal Institute of
Technology, Stockholm, Sweden. Pike, D. (2003). In atable tubes oer advantages for small craft.
Ship and boat international,
November(D):7882. Plaut, R. H. and Cotton, S. (2005).
Two-dimensional vibrations of air- lled geomembrane
tubes resting on rigid or deformable foundations.
Journal of Sound and Vibration,
282(1-
2):265276. Plaut, R. H. and Fagan, T. D. (1988). Vibrations of an inextensible, air-in atbled, cylindrical membrane.
Journal of applied mechanics, 55:672676.
Plaut, R. H. and Leeuwrik, M. (1988). Non-linear oscillations of an inextensible, air-in ated, cylindrical membrane.
International Journal of Non-Linear Mechanics, 23(516):347353.
Pond, C. (2005). The control of vibration at work regulations. Rawson, K. J. and Tupper, E. C. (2001).
Basic Ship Theory: Volume 2.
BIBLIOGRAPHY
233
Richardson, W. J., Greene Jr, C. R., Malme, C. I., and Thomson, D. H. (1995). Mammals and Noise.
Marine
Academic Press, 1. Conference on Rigid Bottom In atable Craft-
RINA (1981). Full conference proceedings. In
Design, Construction and Handling, London, UK. RINA. RINA (1998). Full conference proceedings. In
International Conference on Rigid In atables,
Weymouth, UK. RINA. RINA (2005). Full conference proceedings. In
International conference on rigid in atables,
Cowes, UK. Robinson, D. W. and Dadson, R. S. (1956). A re-determination of the equal-loudness relations for pure tones.
British Journal of Applied Physics, 7(5):166.
Robinson, S. P., Theobald, P. D., Hayman, G., Wang, L. S., Lepper, P. A., Humphrey, V., and Mumford, S. (2011). Measurement of underwater noise arising from marine aggregate operations. Technical Report MERF Ref no. 09/P108, MALSF. Rojratsirikul, P., Wang, Z., and Gursul, I. (2009). membrane airfoils at low Reynolds numbers.
Unsteady uid-structure interactions of
Experiments in Fluids, 46(5):859872.
Rowe, J. M., Smith, S. W., Simpson, A., and Jacob, J. (2006). Development of a Finite Element Model of Warping In atable Wings.
American Institute of Aeronautics and Astronautics,
pages 119. Santos, F. M., Temarel, P., and Soares, C. G. (2009). On the limitations of two- and three-
Proceedings of the Institution of Mechanical Engineers, Part M: Journal of Engineering for the Maritime Environment, 223(3):457478. dimensional linear hydroelasticity analyses applied to a fast patrol boat.
Saunders, H. E. (1957).
Hydrodynamics in ship design.
Savitsky, D. (1964). Hydrodynamic design of planing hulls.
Marine Technology.
Shuford, C. L. (1954). A review of planing theory and experiment with a theoretical study of pure-planing lift of rectangular at plates. Technical Report TN-3233, NASA, Washington D.C. Solent
(2013).
Report
on
solent
environment.
Online:
www.cefas.defra.gov.uk/media/166904/solent-apr07.pdf. Accessed: 23/12/2013. Song, A., Tian, X., Israeli, E., Galvao, R., Bishop, K., Swartz, S., and Breuer, K. (2008). Aeromechanics of membrane wings with implications for animal ight.
AIAA, 46(8):2096
2106. SoundBoat (2005). Sound evalution and control for recreational marine craft. Technical Report CRAF-1999-71729, BMF.
234
BIBLIOGRAPHY
Stenius, I. (2006).
Finite element modelling of hydroelasticity hull-water impacts.
PhD thesis,
Royal Institute of Technology, KTH, Stockholm, Sweden. Stevens, M. J. (1981). The model testing of in atable craft. In
in atable craft, RINA, page Paper 2. Streeter, M., Hudson, D., and Taunton, D. (2009).
Conference on rigid bottom
Hydroelasticity of a high speed craft.
Master's thesis, University of Southampton. Thomas, J. C. and Wielgosz, C. (2004). De ections of highly in ated fabric tubes.
structures, 42(7):10491066. Townsend, N. C. (2008).
In uencing and in uences of marine vessel motions.
Thin-walled PhD thesis,
University of Southampton. Townsend, N. C., Coe, T. E., Wilson, P. A., and Shenoi, R. A. (2012a). High speed marine craft motion mitigation using exible hull design.
Ocean Engineering, 42:126134.
Townsend, N. C., Coe, T. E., Wilson, P. A., and Shenoi, R. A. (2012b). On the mitigation of human motion exposure on board high speed craft . Strategies and methods , including exible hull design.
Ocean EngineeringEngineering, 42:126134.
Townsend, N. C., Wilson, P. A., and Austen, S. (2008a). model RIB in head seas.
Sea-keeping characteristics of a
International Journals of Maritime Engineering, 151.
Townsend, N. C., Wilson, P. A., and Austen, S. (2008b).
What in uences rigid in atable
Proceedings of the Institution of Mechanical Engineers, Part M: Journal of Engineering for the Maritime Environment, 222(4):207217. boat motions?
Tyack, P. L. (2009). Human-generated sound and marine mammals.
Institute of Physics, pages 3944. Urick, R. J. (1983).
Principles of underwater sound.
Physics Today - American
Peninsula Publishing, New York.
Veldman, S., Bergsma, O., and Beukers, a. (2005). Bending of anisotropic in ated cylindrical beams.
Thin-Walled Structures, 43(3):461475.
von Kármán, T. (1929).
The impact of seaplane oats during landing.
Technical report,
N.A.C.A, Washington, D.C. Wagner, H. (1932). Über Stoss-und Gleitvorgänge an der Ober äche von Flüssigkeiten.
of Applied Mathematics and Mechanics, 12(4):193235. Webber, J. P. H. (1982). beanding and torsion. Wielgosz, C. (2002).
Journal
De ection of an in atable cylindrical cantilever beam subject to
Aeronautical Journal, pages 306312.
De ections of in atable fabric panels at high pressure.
Structures, 40(6):523536.
Thin-Walled
BIBLIOGRAPHY
235
Wilson, W. D. (1960).
Equation for the speed of sound in sea water.
Acoustical Society of America, 32. Wood, K. (2011). Re-inventing the RHIB: Shock mitigation.
The Journal of the
Composite World, pages 14.
Wu, P. and Plaut, R. H. (1996). Analysis of the vibrations of in atable dams under over ow conditions.
Thin-Walled Structures, 26(4):241259.
Yakangler (2011).
Drop stitch technology.
Online: www.yakangler.com/articles/news/new-
products/boats/item/1126-nrs-brings-the-goods. Accessed: 16/08/2011. Young, R. W. and Miller, C. N. (1960). water.
Noise data for two outboard motors in air and in
Noise control, 6(1):2225.
Zhao, R. and Faltinsen, O. M. (1993).
Fluid Mechanics, 246:593612.
Water entry of two-dimensional bodies.
Journal of
236
BIBLIOGRAPHY
Appendix A
Hydroelastic in atable boats: Relevant literature and new design considerations Halswell, P. K., Wilson, P. A., Taunton, D. J., and Austen, S. (2012). Hydroelastic in atable boats:
Relevant literature and new design considerations.
Craft Technology, 154 (Part B1):
39-50.
237
International Journals of Small
Trans RINA, Vol 154, Part B1, Intl J Small Craft Tech, Jan-Jun 2012
TECHNICAL NOTE HYDROELASTIC INFLATABLE BOATS: RELEVANT LITERATURE AND NEW DESIGN CONSIDERATIONS (DOI No: 10.3940/rina.ijsct.2012.b1.125tn) P K Halswell, P A Wilson and D J Taunton, University of Southampton, UK S Austen, RNLI, UK. SUMMARY Inflatable boats are considerably more flexible than conventional metal or composite vessels. The RNLI have developed an inflatable boat, the IB1, with improved performance which has been attributed to its flexibility or hydroelasticity. Current design methodologies for planing vessels predict the performance assuming it is rigid. Designing an entirely hydroelastic boat presents completely new design challenges and will require new design methodologies in the future. This paper considers how to approach an entirely hydroelastic planing vessel and how to divide the boat into practical problems. A design approach taking into account hydroelasticity could potentially improve the performance further by decreasing boat motions, reducing added resistance in waves and minimising the slamming accelerations. This paper reviews the literature relevant to rigid inflatable and inflatable boats and shows the construction of the IB1. The hydroelastic design problem is broken down into three main hydroelastic events: global hydroelasticity, hydroelastic planing surfaces and hydroelastic slamming. Each event is defined, the relevant literature is reviewed and the possible advantages are discussed. A design approach is suggested using a hydroelastic design cycle. The hydrodynamic problem of interacting sponsons is briefly discussed. NOMENCLATURE p P R
Length of cantilever beam (m) Internal pressure (N m-2) Load at tip (N) Cylinder radius (m)
IB IB1 RIB RNLI VDV
Inflatable boat Inshore boat 1 Rigid inflatable boat Royal national lifeboat institution Vibration dosage value (m s-1.75)
1.
minimum of two casualties or one in the prone position. The RNLI use the IB1 in littoral waters where the water can be very shallow and there can be large steep breaking waves caused by the reducing water depth near the shore. The main difference between the IB1 and conventional high speed vessels or RIBs is its flexibility. The main material used within the IB1 is a rubber coated fabric which allows the IB1 to deform considerably. This deformation of the main components, such as the hull and sponsons, affects the fluid flow and this causes a hydroelastic interaction.
INTRODUCTION
This project is supported and partially funded by the Royal National Lifeboat Institution (RNLI). The RNLI is a charity that aims to “save lives at sea” all around the coasts of the UK and Ireland. They design, build, maintain and operate a range of vessels for almost any situation and they own the largest fleet of inflatable boats (IBs) and rigid inflatable boats (RIBs) in the UK. This project will focus on the vessels used in littoral waters, primarily the D class inshore inflatable lifeboat also known as the Inshore Boat 1 (IB1), see figure 1. The IB1 is a five metre inflatable lifeboat which is capable of achieving 25 knots in seas associated with a Beaufort Force 2 and can continue to operate safely up to and beyond seas associated with a Beaufort Force 5. It is powered by a 50 horse power outboard engine and weighs a total of 436 kg (all equipment except crew). It usually has three crew on board and is able to take a
©2012: The Royal Institution of Naval Architects
Figure 1: In the foreground shows the IB1 and in the background shows the Atlantic 85 RIB [1] In 1998 the RNLI performed a feasibility study of the EA16 (the previous version of the D Class) and compared it to seven commercially available vessels that included; RIBs, pure IBs or a combination of both [1]. It
B-39
Trans RINA, Vol 154, Part B1, Intl J Small Craft Tech, Jan-Jun 2012
was found that the EA16 gave the best overall performance and therefore the RNLI have been improving its design and performance through either designers experience or trial and error to achieve the optimum boat. Anecdotal evidence from the feedback of the crew has verified that the flexibility or hydroelasticity within the IB1 improves the performance, especially in waves and surf.
2.
Boat motions and hence human exposure to vibrations Forward speed Added resistance in waves Slamming accelerations Stability when stationary AIMS
Compared with larger boats and ships, there is relatively little scientific understanding about the performance of RIBs and considerably less understanding about the performance of IBs. Their design is usually based on the experience of the designer or trial and error. There has been minimal research into the performance of RIBs and IBs for a number of reasons. One possible reason is that these vessels are primarily used for search and rescue or military purposes so the vessel has no direct profit making abilities. They are also manufactured in low numbers so there is minimal drive to invest capital in research and development.
The first aim of this paper is to provide a review of the current level of knowledge for these types of vessels. A review of the experimental and computational work performed on RIBs and IBs is provided.
The IB1 is unique when compared to almost every other planing vessel due to its highly flexible structure. The longitudinal stiffness is considerably less plus it has specific deck joints to provide control over the longitudinal deformation. The longitudinal bending and torsional twisting is called the global hydroelasticity. The planing surface is constructed from fabric allowing excessive deformation and this is called a hydroelastic planing surface. The fabric hull also causes a hydroelastic slam when a transverse slice of the boat impacts the free surface.
The third aim is to demonstrate to the research/academic community that hydroelastic boats could be designed to change their performance using parameters that are currently not considered in the design process of conventional vessels.
The high flexibility means the importance of hydroelasticity is more pronounced and there is a new area of hydroelasticity which is not commonly considered. This new and novel area is the hydroelastic planing surface. The hydroelastic planing surface links the hydroelastic slamming to the global hydroelasticity, through strip theory. Strip theory uses transverse slices of the vessel to predict the planing performance. So hydroelastic slams are the transverse strips used to predict the hydroelastic planing performance. Then the hydroelastic planing performance is used to predict the global hydroelasticity in waves. This means that all three areas of hydroelasticity need to be designed together. This leads to the question: how do you investigate a planing vessel that is entirely hydroelastic? Currently hydroelasticity is used principally to calculate the stresses and strains in the structure, Price et al. [3] and Hirdaros and Temarel [4], or occasionally to study its effects on boat motion, Hirdaris and Temarel [4], Santos et al. [5] and Senjanovic et al. [6]. The design of the IB1 allows the hydroelasticity within the boat to be adjusted to affect the boat performance in many ways. Once sufficient knowledge is gained the boat can be tuned to optimise the boat performance. Hydroelasticity may affect the boat performance in the following ways:
B-40
The second aim of the paper is to divide an entirely hydroelastic planing vessel into manageable hydroelastic problems. The three main hydroelastic problems (hydroelastic slamming, hydroelastic planing surfaces and global hydroelasticity) are defined and the relevant literature is reviewed. The potential advantages from each hydroelastic event are discussed.
3.
RIB AND IB LITERATURE REVIEW
The first inflatable boat manufacturer was Zodiac and they started in 1936, Williams [7]. The RNLI first introduced the D-class inflatable lifeboat in 1963 after extensive trials. In 1964 at Atlantic College in Wales the first rigid hull was glued to an inflatable boat to form the first RIB, Williams [7]. In 1972 the RNLI launched the Atlantic 21, their first RIB. Although these types of vessels have been around for many years there has still been little research into their performance. In 1981, 1998 and 2005 three international conferences were held in the UK to discuss the design and development of RIBs, [8, 9 and 10]. However most of the evidence was anecdotal and there was little scientific proof using experimental or numerical methods. The topics covered included: history, development, construction techniques, propulsion, problems with model test [57], self-righting issues, example boats, safety, influence of the helmsmen, electronics and equipment. Dand [11 - 13] performed a number of experiments into the performance of the IB1 and measured the resistance and sea keeping performance of the boat at model scale and full scale. Austen and Fogarty [14] documented the development of the IB1 as new materials and construction techniques were being used. When the IB1 was introduced into service it suffered from performance problems due to the fabric floor, ventilation and cavitation so Dand et al. [15] used a careful trial and error process to restore the speed from 20 to 25 knots.
©2012: The Royal Institution of Naval Architects
Trans RINA, Vol 154, Part B1, Intl J Small Craft Tech, Jan-Jun 2012
Haiping et al. [16] undertook experiments into the effect of sponson type on seakeeping performance. It was found that inflatable sponsons had lower response amplitude operators in heave and pitch than foam sponsons in both load conditions. This suggests flexible sponsons improve the ride comfort and seakeeping performance. A computational model of a RIB has been constructed by Lewis et al. [17]. Although the results looked promising the numerical model over-predicted the boat motions when compared to experimental results. Townsend et al. [18, 19] performed a multitude of experiments to characterise the seakeeping performance of a RNLI Atlantic 75 RIB. In [18] they studied the influences of speed, ballast, wave height, encounter frequency, and tube pressure on the boats motions of the Atlantic 75.
forward deck section are bonded to the sponson but the other deck sections are slotted into place.
5.
GLOBAL HYDROELASTICITY
4.
5.1
PROBLEM DEFINITION
DESIGN OF THE IB1
It is important to understand the construction of an IB because it will demonstrate how the craft is able to deform. Figure 2 shows the main components within the IB1. The design of IBs does vary depending on their operational requirements, component materials and construction techniques.
Sponsons
Deck Transom
Inflatable Keel
Fabric Hull
Inflatable keel - this is a tapered inflatable tube that is attached to the centreline of the fabric hull. It is constructed from Hypalon®/Neoprene coated polyester fabrics and is inflated to a pressure of 224 mbar (3.25 psi). Fabric hull - this is a fabric sheet, constructed from two sheets of Hypalon®/Neoprene coated polyester fabrics, that is attached to the sponsons and transom and pulled taught over the keel.
This section investigates the global hydroelasticity of an IB by viewing the boat as a whole and studying the longitudinal bending and torsional twisting vibrations that exist. It has been observed that as the IB1 passes over an oblique wave the deck bends and twists which provides a smoother ride. This dynamic bending and twisting response is similar to the theories presented by Bishop and Price [20] which could be regarded as conventional hydroelastic theories. The flexibility of the boat will affect the wave induced dynamic response of the vessel which in turn affects the boat motion. Figure 3 shows how global hydroelasticity can reduce the vertical motions of a deformable vessel. In conventional vessels there is a coupled interaction between the heave and pitch as a vessel “see-saws” over a wave; however, this interaction will change if the boat is able to bend over the wave. The first advantage of this is reduced boat motions leading to improved ride quality. A reduction of the boat motion means that less energy from the propulsion device is absorbed through vertical motion, which reduces the added resistance in waves. This allows either a higher top speed to be achieved or a smaller, lighter, propulsion device to be fitted. The final advantage is that the boat will be more stable, in pitch and heave, when stationary because the pitching motions will be reduced.
Figure 2: Main components of the IB1 Sponsons - these are the inflatable tubes that surround the boat. They are constructed from Hypalon®/Neoprene coated polyester fabrics and they are inflated to a pressure of 206 mbar (3 psi). Deck - this is the stiffest structural component of the boat made from a composite sandwich panel. The deck is sectioned into four parts (plus the transom) to intentionally allow flexibility and each deck joint has its own stiffness due to the type of joint. The transom and
©2012: The Royal Institution of Naval Architects
Figure 3: Reducing vertical motions through global hydroelasticity
B-41
Trans RINA, Vol 154, Part B1, Intl J Small Craft Tech, Jan-Jun 2012
An inflatable boat has many inter-connected parameters that will affect the global vibrations which include; deck properties (material properties and thickness), deck joints (number, position and stiffness), sponson and keel properties (material properties and internal pressures), fabric hull properties (material properties and pretensioned stresses), mass (centre of gravity and inertia) and construction technique. A static deflection experiment was performed by the authors and it was found that the dominant parameters in the deflection of the boat are the number, position and stiffness of the deck joints. 4.2
which theory should be used for Hypalon®/Neoprene coated polyester fabrics but this is a direction of research for the authors. Leonard, Brooks and McComb [27] derived an equation for the maximum tip loading capabilities of an inflatable cylinder acting as a cantilever, see equation 1. This simply shows that the loading capabilities of inflatable cylinders are proportional to the internal pressure (p). Equation 1:
=
/
5.
HYDROELASTIC PLANING SURFACE
5.1
PROBLEM DEFINITION
LITERATURE REVIEW
4.2 (a) Global Hydroelastic Global hydroelasticity has been studied by many authors starting with the work of Bishop and Price [20]. Bishop and Price developed theories to describe symmetric and anti-symmetric hydroelasticity of ships, but these ships were displacement vessels and not planing vessels. There are numerical models capable of predicting the vertical motions and wave loads on a high speed craft, such as Santos et al. [5] and Chiu and Fujino [21]. Santos et al. [5] modelled a fast patrol boat which had a planing hull form, but it is noted that the approach used was not suitable for planing vessels. They found large differences between the full scale measurements and the numerical model results. To our knowledge no numerical model has yet been validated for a hydroelastic planing vessel.
The planing surface of IBs is normally constructed from fabric which has significantly less out-of-plane bending stiffness than conventional metal or composite hulls. This will allow the planing surface to deform considerably under different loading conditions, see figure 4. The problem is to find the shape of the fabric when it is in steady-state planing and the effect of this deformation on the planing performance. The parameters of a fabric hull are material properties and the pretensioned stresses. These parameters define the out-ofplane bending stiffness of a fabric therefore as they are increased the material becomes stiffer and comparable to a conventional planing surface. A better understanding of a hydroelastic planing surface could lead to an increase in forward speed.
The IB1 has distinct deck joints to allow the boat to hinge in certain points. These deck joints will affect the conventional theories of global hydroelasticity. Newman [22] developed an analytical method to predict the motions of a hinged barge. Hamamoto et al. [23] used a 3D coupled finite element method-boundary element method model to predict the motion of module linked large floating structures. 4.2 (b) Inflatable Cylinders The stiffness properties of inflatable tubes and boundary tensioned membranes (the fabric hull) are not currently considered in hydroelastic models. Early work in the deformation of inflatable cylindrical beams started with Comer and Levy [24] by comparing them to an EulerBernoulli beam. The most recent and relevant work was performed by Wielgosz et al. [25] by using Timoshenko beam theory to account for the shear deformation. A finite element model was made using a stiffness matrix to include internal pressure. Veldman et al. [26] highlighted the importance of using the correct modelling theory; membrane or thin-shell theory. It would be expected that a very thin membrane would correlate better with membrane theory than thin-shell theory. However, [26] found better agreement using thin-shell theory than membrane theory even though the membrane was only 60 nanometres thick. It has not yet been established
B-42
Figure 4: Hull deformation of the IB1 at 19.4 knots from underwater [15] Experiments by Dand [11, 12] were performed on an EA16 D Class at full scale and model scale to measure the resistance, sinkage and trim. The full scale boat was flexible and the fabric hull was able to deform but the scale model was rigid. The comparison of total resistance, see figure 5, showed that the full scale flexible boat had slightly higher resistance than the rigid scaled model. Dand et al. [15] attributed this to the change in trim angle due to the fabric hull deforming and causing a concave camber at the aft end of the hull. They also found an instability when the boat was accelerating on flat water which was described as a “pressure wave” slowly passing under the boat. It caused a “pulsing”
©2012: The Royal Institution of Naval Architects
Trans RINA, Vol 154, Part B1, Intl J Small Craft Tech, Jan-Jun 2012
motion primarily in pitch and heave. Whether the deformation was static or dynamic is unknown.
Figure 5: Comparison of the predicted and measured resistance of the EA16 D Class [15] The first limitation is the “pulsing” motion instability found in the IB1. One hypothesis is that the reduced outof-plane bending stiffness of the hull allowed the concave camber to form. This causes the pre-tensioned stresses in the fabric to change as the camber forms and also results in a change in the hydrodynamic forces on the hull. As the fabric stresses change, the deformation moves aft. The deformation causes a change in hydrodynamics which gives the operator the feeling of this “pressure wave”. It has also been reported that as this “pressure wave” passes under the hull the sponsons can be seen to deflect which indicates high forces and fabric movement. When this deformation reaches the transom the pressure is released and the cycle begins again. This motion is only found on flat water; waves cause the cycle to be broken. So there is a limitation in the minimum outof-plane bending stiffness of the fabric hull to ensure this instability does not occur and this requires quantification. This belief was confirmed through trial and error when the EA16 was developed into the IB1. During the redesign it was found that the fabric had been permanently deformed and low quality control during construction led to a reduction in fabric tension. Once this had been taken into account and the fabric tension was increased the pulsing motion disappeared. 5.2
LITERATURE REVIEW
The most relevant literature to this problem is an analytical model developed by Makasyeyev [28] to describe the planing performance of a 2D planing elastic plate. However this model requires validation and the structural domain deals with conventional materials not membranes or fabrics. No literature directly related to a membrane planing surface has been found. However this fluid structure interaction could be compared with the aeroelasticity of a membrane aerofoil, such as sails and membrane wings. Newman [29] noted skin friction can change the
©2012: The Royal Institution of Naval Architects
membrane tension and in an inviscid flow it is constant. A strong coupling between the frequency of the membrane oscillations and vortex shedding frequency has been shown by Song et al. [30], Rojratsirikul [31] and Gordnier [32]. Gordnier [32] importantly showed that the Reynolds’ Number caused the motion of the membrane aerofoil to change from a standing wave vibration to a dynamic vibration similar to travelling waves. None of the afore-mentioned literature contains a free surface which is vital for the planing fluid forces. It is of interest to note that many new tender-boat designs now employ drop stitch technology for the hull. Dropstitch technology involves two layers of fabric that are sealed together at the edges. Then threads are weaved perpendicular to the layers of fabric to control the shape when inflated, see figure 6. When the two layers are inflated it forms a stiff panel that could be compared to a composite sandwich panel, Bagnell [33].
Figure 6: Drop stitch technology [54] 6.
HYDROELASTIC SLAMMING
6.1
PROBLEM DEFINITION
The problem addressed within this section is regarding the effect of hydroelasticity on the loads and accelerations of a 2D wedge vertically impacting a free surface. An IB has three main flexible components in the vertical direction which are the fabric hull, the inflatable sponsons and the inflatable keel, see figure 7. In reality these three components act together and will affect the response of each other. However, for an initial investigation each can be studied individually. By considering a slamming event as hydroelastic it allows the possibility of changing the impact characteristics. The main characteristics that can be changed, from a boat motion perspective, are the peak acceleration and impact duration. It will also affect the structural loading but this paper will not explore that side of the problem, see Faltinsen [34] for more details. The new parameters for the hull are fabric material properties and pre-tensioned stresses and the new parameters for the inflatable keel and sponsons are material properties and internal pressure. Note that changing the internal pressure is the same as changing the pre-tensioned stresses. The other important variables are impact velocity, deadrise angle and inertia. A simple hull wedge impact was investigated by Townsend et al. [35] to study possible methods of reducing the vertical acceleration on high speed craft. Hull stiffness was reduced from 69 GPa (aluminium) to 6.9 GPa to investigate the effect of
B-43
Trans RINA, Vol 154, Part B1, Intl J Small Craft Tech, Jan-Jun 2012
intentionally reducing the hull stiffness. It was found to have minimal effect on acceleration but it is anticipated that the fabric will have a significantly lower equivalent stiffness which may amplify the effect on acceleration.
importance of including them, especially at low deadrise angles. Air pockets can occur when the structure is very flexible because the fabric hull can deform vertical upwards, as shown in figure 8. Faltinsen et al. [39] noted that the breakdown of air cushions into bubbles requires better understanding and the effect of this is unknown. Flow separation is another consideration and this can be described when there is a hard chine but Faltinsen [41] stated the round bilge flow separation is difficult to handle and here viscosity may need to be included. Finally the membrane behaviour is significantly different from that of conventional solids with nonlinear behaviour due to the interaction of the weave and weft, Lewis [42].
Figure 8: Air pocket formation
Figure 7: The flexible components within a vertically impacting IB It has been proposed but not validated by many authors including Natzijl [36] and Pike [37] that sponsons absorb energy during slamming motions. Townsend [38] did investigate this concept but the internal pressure reduction was shown to have no effect. It is worth noting that the Atlantic 85 investigated by Townsend [38] had a hull shape which caused the sponsons rarely to come into contact with the water which is not the case for the IB1. The experiment proposed for the wedge sections with sponsons will answer this question and allow an investigation into the effect of material properties and internal pressure. Other variables that will affect the amount of energy absorbed by the sponsons include; sponson diameter, sponson overhang and sponson attachment. 6.2
LITERATURE REVIEW
Faltinsen et al. [39] provides a good review of this problem and discusses the challenges within it. Here is a list of particular effects that may require consideration: gravity, viscosity, air cushions, air pockets, air to bubble generation, water compressibility, air compressibility, flow separation and membrane behaviour. Gravity can normally be neglected in this problem, Faltinsen et al. [39]. Viscosity is also commonly neglected but this could affect flow separation when there is not a sharp corner, which will be discussed later, Faltinsen et al. [39]. Air cushions and air compressibility were initially ignored but Bereznitski [40] showed the
B-44
Faltinsen [34] divided this problem into two time scales. The initial time scale is that of the structural inertia phases where the large hydrodynamic forces lead to large accelerations of a small structural mass. This phase is very short compared to the second time scale. The second scale is that of the free vibrations phase and is the highest wetted natural period of the structure. The behaviour is that of the free elastic vibrations of the structure with the initial conditions obtained from the first phase. The maximum stresses occur in the free vibration phase. Faltinsen [43] discusses the importance of hydroelasticity as a ratio between the first period of natural vibration of a wet beam and the duration of the impact. It is quantified in terms of nondimensionalised parameters. Bereznitski [40] uses the same ratio except that it uses the natural vibrations of a dry beam. Bereznitski [40] says that if the ratio is greater than two then hydroelasticity does not play a significant role. Increasing either the material properties or pre-tensioned stresses in the fabric will alter the period of vibration therefore affecting the importance of hydroelasticity. = Cooper et al. [44] were the first to study the deformation of a flexible membrane wedge impacting a free surface. It was found that during the free vibration phase the membrane vibrated at frequencies very near to its natural frequency, which depended on the pre-tensioned stresses. 6.3
CRITIQUE OF MODELLING METHODS
The problem of water entry of 2D bodies started in a purely hydrodynamic sense for a rigid body with the work of Wagner [55] and Von Karman [56] in the 1920s
©2012: The Royal Institution of Naval Architects
Trans RINA, Vol 154, Part B1, Intl J Small Craft Tech, Jan-Jun 2012
and 1930s. This work was advanced by many researchers but it was not until the work of Kvalsvold et al. [45] that the local hydroelastic effects were considered. Using theory alone, Kvalsvold in 1994 studied the slamming-induced local stresses in the wetdeck of a multihull vessel for a doctor of engineering thesis and jointly published the results in Kvalsvold and Faltinsen [45]. The structure was modelled using a 2D Timoshenko beam and the fluid was modelled using Wagner theory. It assumed the fluid to be incompressible and irrotational; and air entrapment and cavitation were not included. This solution was complex and simplified by Faltinsen [34]. Experimental results from Faltinsen et al. [46] and Kvalsvold et al [45] agreed well with both theoretical solutions. Faltinsen [43] used the numerical solution of Kvalsvold and Faltinsen [45] to study the water entry of a wedge including the forward speed of the vessel by solving the coupled non-linear equations by a RungeKutta 4th order scheme. Korobkin et al. [47] demonstrated that it is possible to couple a finite element method for the structural domain directly with Wagner’s theory for the fluid domain. The results were compared with a modal method using a beam model and the results showed very good correlation. Lu et al., [48] used boundary element methods (BEM) for the fluid and finite element method (FEM) for the structure. The non-linear free surface boundary condition was satisfied and the jet was properly treated. Good agreement was found with the results of Zhao and Faltinsen [49].
resistance to spray resistance, Payne [52]. The mechanisms for wave and spray generation are understood for planing vessels with hard chines, Savitsky and Morabito [53]. However, the IB1 and most IBs do not have chines and the mechanisms for generation are not well understood. Figure 5 shows the difference between the measured resistance of the IB1 and the Savitsky prediction. Therefore the problem is to study the wave and spray generation around a vessel with interacting sponsons with speeds from zero to planing and above. Although this problem is not necessarily hydroelastic it is an important stage in predicting the performance of a RIB or an IB. Current theories, such as strip theory and Wagner’s expanding wedge theory, do not consider the effect of a sponson. Therefore this section wishes to define the hydrodynamics around a sponson because the hydroelastic effects of a sponson cannot be explored until the hydrodynamics are understood. By minimising the wave and spray generation it is possible to improve the top speed and acceleration of the craft. In addition, it has the capability to reduce the environmental damage from wave wash, although this may have an adverse effect on the boat motion. The problem can be viewed in 2D transverse slices that allow the effect of the sponsons on the added mass to be investigated; alternatively, the problem can be viewed longitudinally studying the effect of sponsons on the resistance of the craft. 7.2
Bereznitski [40] published an important paper on the role of hydroelasticity in the 2D slamming problem and uses four methods for solving the problem. The first is a Wagner's solution for a rigid body and this can be compared to the work of Faltinsen [34] for an elastic body. Bereznitski also used a self-developed code plus two commercial codes called MSC Dytran and LSDYNA. Bereznitski commented that the most suitable methods were either MSC Dytran or LS-DYNA because they can both deal with the coupled hydroelastic interaction and include air cushion modelling. It is worth noting that MSC Dytran and LS-DYNA are quite similar and the equations for the state of water and air are the same, Bereznitski [40]. LS-DYNA has been used to study this problem by Bereznitski [40], LeSourne et al. [50] and Stenius [51]. Stenius [51] used finite element analysis based on multi-material arbitrary LagrangianEulerian formulation and a penalty contact algorithm and the hydrodynamic loads correlated well with experimental results. 7.
HYDRODYNAMICS
7.1
PROBLEM DEFINITION
As a vessel increases in speed, beyond the hump speed, the main resistance component changes from wave
©2012: The Royal Institution of Naval Architects
LITERATURE REVIEW
Dand [12] performed resistance experiments on the IB1 at full and model scale. No measurements of the wave or spray generation were made but figure 9 shows that the spray is attached to the sponsons until it detaches to form spray sheets. This indicates that surface tension and the coandă effect need to be considered.
Figure 9: Spray generation of an EA16 at 19.4 knots [12] An investigation into the boat motions of RIBs and specifically the RNLI Atlantic 85 were investigated by Townsend et al. [18]. It was found that the sponsons were rarely in contact with the water while planing, resulting in the sponsons having minimal effect on the
B-45
Trans RINA, Vol 154, Part B1, Intl J Small Craft Tech, Jan-Jun 2012
high speed performance. Therefore the sponsons of certain RIBs have negligible effect on the wave or spray generation but this is clearly not the case for the IB1. Waves can be measured using a wave probe but measurement of spray is less common and at present the ITTC do not have any recommended procedures for measuring spray or accounting for spray scaling. The location of the spray sheet separation from the sponsons also needs to be measured. 8.
DISCUSSION
8.1.
HUMAN EXPOSURE TO VIBRATIONS
High speed marine vehicles, such as the IB1, experience non-linear boat motion which results in high and low frequency vibrations with large accelerations. In 2002 a European Directive (2002/44/EC) was proposed to deal with the minimum health and safety requirements regarding the exposure of workers to physical vibrations. The exposure action value for whole-body vibration is 0.5 ms-2 r.m.s (or 9.1 ms-1.75 VDV) and the exposure limit value is 1.15 ms-2 r.m.s (or 21 ms-1.75 VDV). Boat motions and vibrations have been well reviewed in relation to high speed craft by Townsend [35]. Vibrations can not only cause long term injuries to the crew but they can reduce the crew's ability to perform tasks (during and after transit). Possible strategies to reduce human exposure to boat motion have included; suspension seats, suspended decks, active and passive fins, trim tabs, interceptors, gyrostabilisers, flexible hulls and elastomer coated hulls. Townsend et al. [20] showed that the RNLI RIBs exceeded the exposure limit value in a sea with the average of the highest 1/3 significant wave heights equal to 0.4m and average wave period equal to 10.6s. Dand [13] showed that the rigid scale model of the IB1 in regular waves, with a full scale wave height of 0.55m, could be exposed to peak accelerations of up to 4g in the crew's position. Whilst there is considerable debate in the marine community over the validity of applying the European Directive to high speed marine vessel the RNLI are investigating methods to demonstrate how the exposure of their crews and trainers to vibrations can be mitigated. The correct application of global hydroelasticity and hydroelastic slamming may help reduce the boat motions, in terms of vertical acceleration, that cause these high speed vessels to exceed the exposure limit. Hydroelastic slamming has the ability to change the characteristics of a slamming event and reduce human exposure to vibrations. The authors believe that hydroelasticity will reduce the peak slamming accelerations but conversely it will also increase impact duration. At the current stage of understanding about human exposure to vibrations it is unclear which variable (peak acceleration or impact duration) is more important to reduce the harm to the crew. So it is unclear how effective hydroelastic slamming will be at this stage.
B-46
8.2.
DESIGN CYCLE
So far this paper has broken an entirely hydroelastic boat into three main hydroelastic events. The next step is to consider how to design all three events together and a design cycle can be used with a specific order, see figure 10. The first event that requires examination is the hydroelastic slamming. This provides the added mass for the hydroelastic planing surface and the springing and whipping inputs for the global hydroelasticity. Then the hydroelastic planing surface can be studied which provides the calm water planing performance. Finally the global hydroelasticity can be considered to understand the planing performance in waves and the whipping and springing affects. A design cycle is required because all the hydroelastic events are coupled together, as explained in the next subsection.
1. Hydroelastic Slamming
2. Hydroelastic Planing Surface
3. Global Hydroelasticity Figure 10: Hydroelastic design cycle 8.3.
COUPLING OF HYDROELASTIC EVENTS
Global hydroelasticity has the potential to reduce the vertical motion from the coupling of pitch and heave. However this may lead to other issues such as the vessel no longer having the longitudinal stiffness to plane at maximum performance. It appears that a flexible planing surface has a detrimental effect on performance and a rigid surface is more suitable. However, within the design of the IB1 a flexible hull is required to allow the advantages of global hydroelasticity and hydroelastic slamming to emerge. So it is important to quantify the minimum out-of-plane bending stiffness to remove any instabilities so that the maximum flexibility is available for global and slamming hydroelasticity. Hydroelastic slamming may require a low transverse stiffness to improve the slamming characteristics but this may reduce the planing performance. 9.
CONCLUSIONS
The literature that is directly linked to the design and performance of RIBs and IBs has been discussed. This shows how little research has been undertaken in this area. However, research in other fields that is relevant to
©2012: The Royal Institution of Naval Architects
Trans RINA, Vol 154, Part B1, Intl J Small Craft Tech, Jan-Jun 2012
2.
RNLI, Inshore Boat (IB1) – Feasibility Study Report, Private communication, 1998
3.
PRICE, W., SALAS INZUNZA, M., TEMAREL, P., The dynamic behaviour of a mono-hull in oblique waves using two-and three-dimensional fluid structure interaction models, Trans. Royal Institution of Naval Architects 144, pp. 1-26, 2002
4.
HIRDARIS, S. E., TEMAREL, P., Hydroelasticity of ships: recent advances and future trends, Proceedings of the Institution of Mechanical Engineers, Part M: Journal of Engineering for the Maritime Environment 223, (3), pp. 305-330, 2009
5.
SANTOS, F. M., TEMAREL, P., SOARES, C. G., On the limitations of two- and threedimensional linear hydroelasticity analyses applied to a fast patrol boat, Proceedings of the Institution of Mechanical Engineers, Part M: Journal of Engineering for the Maritime Environment 223 (3) pp. 457-478, 2009
6.
SENJANOVIC, I., MALENICA, S., TOMASEVIC, S., Investigation of ship hydroelasticity, Ocean Engineering 35, (5-6), pp. 523-535, 2008
7.
WILLIAMS, S. G., Historical review of 20 years of technical progress of rigid inflatable boats, Int. Conference on rigid bottom inflatable craft, RINA. p. Paper 1, 1981
8.
Conference on Rigid Bottom Inflatable CraftDesign, Construction and Handling, RINA Small Craft Group Conference, November 10 and 11, London, 1981
A hydroelastic design cycle was suggested to analyse the three hydroelastic events in a specific order. The coupling between the three events was examined and showed that a hydroelastic planing surface limits the possible longitudinal and transverse stiffness for global hydroelasticity and hydroelastic slamming.
9.
International Conference on Rigid Inflatables, RINA, May 14 and 15, Weymouth, UK, 1998
10.
International conference on rigid inflatables, RINA, June 2 and 3, Cowes, UK, 2005
10.
11.
DAND, I. W., Resistance measurements with an RNLI D-class Lifeboat, Report by BMT SeaTech; Doc No. 3356.02, 2002
12.
DAND, I. W., Resistance experiments with an RNLI D-class Model, Report by BMT SeaTech; Doc No. C3356.04, 2003
13.
DAND, I. W., RNLI D-class model: Seakeeping measurements in head seas, Report by BMT SeaTech; Doc No. C3356.06, 2004
the approach adopted for this project indicates that hydroelasticity does have the potential to improve boat performance. The construction of the IB1 is described and this shows the areas of flexibility within the design which therefore show where hydroelasticity should be considered in the design of IBs. The optimisation of hydroelasticity may possibly lead to improvements in boat motion (reduced human exposure to vibrations), boat forward speed/acceleration, slamming accelerations, added resistance in waves and stability (pitch and heave) when stationary. Global hydroelasticity was studied first. It may be possible to alter current theories to include the inflatable tubes and deck joints but no current theory has been validated for a hydroelastic planing vessel. Global hydroelasticity has the potential to improve the boat motions and reduce added resistance in waves. The complex problem of a hydroelastic planing surface was then considered. Current results suggest that a flexible surface provides a low quality planing surface. However, a flexible surface is required to allow the other areas of hydroelasticity to function as desired. In the hydroelastic slamming event three different elastic components were described: the hull, sponsons and keel. There are computational models capable of predicting the slamming accelerations and loads with conventional materials. Hydroelastic slamming could alter the slamming characteristics but at the current time the characteristics needed to reduce human exposure to vibrations are unknown. The hydrodynamic problem of interacting sponsons was shown and the error in the current predictions for hull resistance was highlighted.
ACKNOWLEDGMENTS
This project is jointly supported and funded by the RNLI and EPSRC (Engineering and Physical Sciences Research Council). 11.
REFERENCES
1.
Website viewed on 22/08/2011, image of IB1 and Atlantic 85. http://www.rnlisunderland.org/information/lifeb oats/pg18.html
©2012: The Royal Institution of Naval Architects
1.
B-47
Trans RINA, Vol 154, Part B1, Intl J Small Craft Tech, Jan-Jun 2012
14.
15.
16.
17.
AUSTEN, S., FOGARTY, H. F. J., Design and development of a new inflatable lifeboat, SURV VI, RINA, 2004 DAND, I. W., AUSTEN, S., BARNES, J., The speed of fast Inflatable Lifeboats, Int. Journal of Small Craft Technology Vol. 150 Part B2, pp. 23-32, 2008 HAIPING, H. E., BRETL, J. P. E., VANSUMEREN, H., SAVANDER, B., TROESCH, A. W., Model tests of a typical rib in waves, Private Communication, 2005 LEWIS, S. G., HUDSON, D. A., TURNOCK, S. R., BLAKE, J. I. R., SHENOI, R. A., Predicting motions of high speed RIBs: A Comparison of non-linear strip theory with experiments, Proceedings of the 5th Int. Conference on High Performance Marine Vehicles (HIPER'06) pp. 210-224, 2006
26.
VELDMAN, S., BERGSMA, O., BEUKERS, A., Bending of anisotropic inflated cylindrical beams. Thin-Walled Structures 43, (3), pp. 461475, 2005
27.
LEONARD, R. W., BROOKS, G. W., AND McCOMB, H. G., Structural considerations of inflatable re-entry vehicles, NASA TN D-457, 1960
28.
MAKASYEYEV, M., Two-dimensional unsteady planing elastic plate, Workshop on the mathematical challenges and modelling of hydroelasticity, Edinburgh, 21-24 June, 2010
29.
NEWMAN, B., Aerodynamic theory for membranes and sails, Progress in Aerospace Sciences 24, (1) pp. 1-27, 1987
30.
SONG, A., TIAN, X., ISRAELI, E., GALVAO, R., BISHOP, K., SWARTZ, S., BREUER, K., Aeromechanics of membrane wings with implications for animal flight, AIAA Journal 46, (8), pp. 2096-2106, 2008
31.
ROJRATSIRIKUL, P., WANG, Z., GURSUL, I., Unsteady fluid-structure interactions of membrane airfoils at low Reynolds numbers, Experiments in Fluids 46, (5), pp. 859-872, 2009
32.
GORDNIER, R., High fidelity computational simulation of a membrane wing airfoil, Journal of Fluids and Structures 25, (5), pp. 897-917, 2009
18.
TOWNSEND, N. C., WILSON, P. A., AUSTEN, S., Seakeeping characteristics of a model RIB in head seas, Int. Journal of Maritime Engineering 151, (A1), pp. 5-44, 2008
19.
TOWNSEND, N. C., WILSON, P. A., AUSTEN, S., What influences rigid inflatable boat motions? Proceedings of the Institution of Mechanical Engineers, Part M: Journal of Engineering for the Maritime Environment 222, (4), pp. 207-217, 2008
20.
BISHOP, R. E. D., PRICE, W. G., Hydroelasticity of ships, Cambridge University Press, 1979
21.
CHIU, F-C., FUJINO, M., Nonlinear prediction of vertical motions and wave loads of highspeed crafts in head sea, Int. Shipbuilding Progress 36, (406), pp. 193-232, 1988
33.
BAGNELL, D., The development of a rigid inflatable vee-bottom boat, Int. Conference on Rigid Inflatables, RINA, May 14 and 15, Weymouth, UK, Paper 3, 1998
22.
NEWMAN, J. N., Wave effects on deformable bodies, Applied Ocean Research 16, pp. 47-59, 1994
34.
23.
HAMAMOTO, T., HAYASHI, T., FUJITA, K., 3D BEM-FEM coupled hydroelastic analysis of irregular shaped, module-linked large floating structures, Int. Offshore and Polar Engineering Conference, Los Angeles, USA May 26-31, 1996
FALTINSEN, O. M., The effect of hydroelasticity on ship slamming, Phil. Trans. of the Royal Soc. A: Mathematical, Physical and Engineering Sciences 355, (1724), pp. 575-591, 1997
35.
TOWNSEND, N. C., COE, T. E., WILSON, P. A., SHENOI, R. A., On the mitigation of human motion exposure on board high speed craft. Strategies and methods, including flexible hull design, Private Communication (submitted for Journals of Ocean Engineering), 2011
36.
NATZIJL, P. W., Bringing a 1.0 metre buoyancy tube to sea on a 18.0 metre rigid hull, Int. conference on rigid inflatables; 14-15 May, Weymouth, UK, Paper No. 7, 1998
24.
25.
B-48
COMER, R. L., LEVY, S., Deflection of an inflatable cylindrical cantilever beam, AIAA Journals 1, (7), pp. 1652-1655, 1963 WIELGOSZ, C., THOMAS, J. C., LE VAN, A., Mechanics of inflatable fabric beams, ICCES 5, (2), pp. 93-98, 2008
©2012: The Royal Institution of Naval Architects
Trans RINA, Vol 154, Part B1, Intl J Small Craft Tech, Jan-Jun 2012
37.
PIKE, D., (2003) Inflatable tubes offer advantages for small craft. Ship and Boat Int., November, pp. 78-82, 2003
38.
TOWNSEND, N. C., Influencing and influences of marine vessel motions, EngD. Thesis, University of Southampton, UK, 2008
39.
FALTINSEN, O. M., LANDRINI, M., GRECO, M., Slamming in marine applications, Journal of Engineering Mathematics 48, (3/4), pp. 187217, 2004
40.
BEREZNITSKI, A., Slamming: The role of hydroelasticity, Int. shipbuilding progress 4, (48), pp. 333-351, 2001
41.
FALTINSEN, O. M., Hydrodynamics of highspeed marine vehicles, Cambridge University Press, ISBN; 0-521-84568-8, 2005
50.
LE SOURNE, H., COUNTY, N., BESNIER, F., KAMMERER, C., LEGAVRE, H., LS-DYNA Applications in Shipbuilding, In: 4th European LS-DYNA Users Conference pp. 1-16, 2003
51.
STENIUS, I., Finite element modelling of hydroelasticity hull-water impacts, Ph.D. thesis, Royal Institute of Technology, KTH, Stockholm, Sweden, 2006
52.
PAYNE, P. R., Design of High-Speed Boats – Planing, Fishergate Inc, ISBN: 0-942720-06-07, 1988
53.
SAVITSKY, D., MORABITO, M., Origin and characteristics of the spray patterns generated by planing hulls, Davidson Laboratory, Stevens institute of technology, Report No. 2882, 2010
54.
Website viewed on the 07/11/2011, image of drop stitch technology www.yakangler.com/articles/news/newproducts/boats/item/1126-nrs-brings-the-goods
2. 42.
LEWIS, W. J., Tension Structures: Form and Behaviour, Thomas Telford, ISBN: 0-72773236—6, 2003
43.
FALTINSEN, O. M., Water entry of a wedge by hydroelastic orthotropic plate theory, Journal of Ship Research 43, (3), pp. 180-193, 1999
44.
COOPER, M. D., McCue, L.S., Experimental study on deformation of flexible wedge upon water entry, High speed marine vehicles conference, 25-27 May, Naples, 2011.
45.
KVALSVOLD, J., FALTINSEN, O. M., Hydroelastic modelling of wet deck slamming on multihull vessels, Journal of ship research 39, (3), pp. 225-239, 1995
46.
FALTINSEN, O. M., KVÅLSVOLD, J., AARSNES, J. V., Wave impact on a horizontal elastic plate, Journal of Marine Science and Technology 2, (2), pp. 87-100, 1997
47.
KOROBKIN, A., GUÉRET, R., MALENICA, S., Hydroelastic coupling of beam finite element model with Wagner theory of water impact, Journal of Fluids and Structures 22, (4), pp. 493-504, 2006
48.
LU, C., HE, Y., WU, G., Coupled analysis of nonlinear interaction between fluid and structure during impact, Journal of Fluids and Structures 14, (1), pp. 127-146, 2000
49.
ZHAO, R., FALTINSEN, O. M., Water entry of two-dimensional bodies, Journal of Fluid Mechanics, 246, pp. 593-612, 1993
©2012: The Royal Institution of Naval Architects
55.
WAGNER, H., Uber stoss-und Gleitvorgange ander Oberflache von Flussikeiten, Zeitschr. f. Angew. Math. Und. Mech., 12, 4, 193–235, 1932
56.
VON KARMAN, T., The impact of seaplane floats during landing, NACA TN321, 1929
57.
STEVENS, M. J., Model testing of inflatable craft, Conference on rigid bottom inflatable craft, RINA, Paper 2, 1981.
B-49
Appendix B
Measuring the hydroelasticity of in atable boats to reduce injuries in harsh waters Halswell, P. K. (2013). Measuring the hydroelasticity of in atable boats to reduce injuries in harsh waters.
National Instruments Case Study.
249
Measuring the Hydroelasticity of Inflatable Boats to Reduce Injuries in Harsh Waters
"The CompactRIO system worked extremely well as a data logger. It was robust and rugged enough to successfully save the data while under extreme conditions." - Peter Halswell , University of Southampton
The Challenge: Acquiring and analysing data from the hydroelastic response of an RNLI D class vessel with the aim to reduce vibrations experienced by the crew and casualties.
The Solution: The RNLI D-class © RNLI/Martin Fish
Using a CompactRIO system as a stand-alone data logger to acquire data from 52 sensors through 74 channels during various static and dynamic tests on open water.
Author(s): Peter Halswell - University of Southampton The Royal National Lifeboat Institution (RNLI), a charity that aims to “save lives at sea” across the United Kingdom (UK) and Ireland, owns the largest fleet of inflatable boats (IBs) and rigid inflatable boats (RIBs) in the UK. For this project, we studied the RNLI D-class, a five metre IB powered by a 50-horsepower outboard engine capable of achieving 25 knots. Three crew members operate the craft and can rescue at least two casualties. Ocean waves cause almost all high-speed planing vessels, including RIBs and IBs, to vibrate in a nonlinear manner called boat motion. Boat motion exposes the onboard crew and casualties to vibration, which can potentially cause long and short-term injuries with physiological and psychological effects. In 2002, European Union directive 2002/44/EU set limits on maximum human exposure to whole body vibration in the workplace. However, small high-speed planing craft rapidly exceed these limits. We need new strategies to reduce human exposure to these harsh vibrations. A RIB with inflatable sponsons has lower response amplitude operators than a RIB with rigid foam sponsons. We performed a quasi-2D hydroelastic slamming experiment to show that hydroelasticity could change the peak acceleration and impact duration. Our findings indicate that hydroelasticity could reduce boat motion and thus reduce the vibration experienced by the crew. We wanted to study the effects of varying the hydroelasticity to reduce the vibration experienced by the crew and casualties.
Building or Measurement System The best method to study boat motion and hydroelasticity within the D class was to perform four full-scale experiments with each affecting a different aspect of hydroelasticity. The experiments included stationary tests, drop tests, flat water trials and wave trials. Model-scale experiments uncovered issues such as the scaling of fabric properties and atmospheric pressure. Computational models are not capable of predicting the effect of varying the hydroelasticity. The D class had many inflatable and flexible components, including the hull, sponsons, keel, deck, and deck hinges, which led to a very complex hydroelastic interaction. Each component could potentially affect the hydroelasticity and the vibration experienced by the crew in a number of ways. We measured boat motion using 52 sensors through 74 channels attached to various parts of the boat including triaxial accelerometers, rate gyroscopes, strain gages, string potentiometers and pressure transducers. We converted the analogue signals from these sensors into digital signals and saved them during each experiment. We needed a rugged and reliable data logger that was flexible enough to accommodate a huge variety of sensors. The data logger would experience vibrations up to 20 g in 0.1 seconds. It had to be a stand-alone system, meaning no mains power or connection to a laptop, with a minimum sampling rate of 2,500 Hz. Finally, the data logger had to be small enough to fit inside a waterproof case in a restricted space. After careful consideration, we chose the National Instruments (NI) cRIO-9074 embedded controller to meet all these requirements. We reconfigured the flexible CompactRIO system to operate as a data logger. We could swap eight input/output (I/O) C Series modules depending on the requirements of the system. The I/O is controlled and synchronised using the onboard Field Programmable Gate Array (FPGA) chip. The conventional method for using a CompactRIO system as a data logger was to feed the output signal from the FPGA chip into a real-time controller in which the data is saved to either the internal flash memory, a USB stick, or an SD card C Series module. However, the SD card module was accessed through the FPGA chip for the highest possible reading/writing speed of 2 MB/s with the real-time controller removed. We did not find any examples of a data logger application in which the real-time controller was not used, so this method is potentially novel. The benefits include increased reliability, simplified coding and reduced compilation time. With the wide range of C Series modules, we could wire almost any signal into the CompactRIO system. We chose the NI 9236 strain gage module because it matched the 350 ohm quarter-bridge strain gages used on the deck and deck hinges. The NI 9234 dynamic signal acquisition module was suitable for the triaxial accelerometers and the NI 9205 voltage module was flexible enough to support our other sensors (string potentiometers and pressure transducers). We developed the coding for this data logger using NI LabVIEW software, which made programming simple and fast.
Our Findings We successfully saved the data from the sensors during the four experiments and gained valuable information on the performance and deformation of the D class. The results from the static tests measured the static shape of the D class, which had not been done before, especially the deck hinge angle and the hull shape. The drop tests measured the change in peak acceleration due to hydroelasticity and showed that hydroelasticity can potentially affect the human exposure to vibration. The flat water trials measured the effect of hydroelasticity on the top speed of the vessel. The wave trials measured the effect of hydroelasticity on the boat accelerations and human exposure to vibration. The system simultaneously measured performance and deformation. The next step in the analysis is to link the two together to find the origin of the effect of hydroelasticity and isolate the components dominating the effect of hydroelasticity. The CompactRIO system worked extremely well as a data logger. It was robust and rugged enough to successfully save the data while under extreme conditions. We overestimated the power consumption, which meant we could have fit the entire stand-alone system inside a waterproof case half the size of the one we used, making the system even smaller.
Rugged, Reliable, Flexible Testing We used CompactRIO and LabVIEW to successfully test the D class during a range of experiments, from static tests to open-water wave trials. The system provided data on hydroelasticity never before captured. The cRIO-9074 proved a rugged, reliable and flexible data logger that stood up to the challenges we presented. Using this new data, we can improve the D class and other vessels to make them safer for the crew and casualties.
1/7
www.ni.com
Author Information: Peter Halswell University of Southampton University of Southampton, University Road Southampton SO17 1BJ United Kingdom Tel: 07813390513
[email protected]
The RNLI D-class © RNLI/Martin Fish
2/7
www.ni.com
The RNLI D-class © RNLI/Dave Nicoll
3/7
www.ni.com
3D Plot of The Upside-Down Hull Shape During Static Tests (0, 0, 0 is the Starboard Transom Corner)
Deck Hinge Angles During Static Tests (Side View of the Boat)
4/7
www.ni.com
Using The cRIO as a Data Logger With and Without a Real Time Controller
Data Logger Code Utilising Only The FGPA Chip (One of Two Parallel Loops)
5/7
www.ni.com
Acceleration Time History During the Drop Ttests (40 Hz Low Pass Filter)
6/7
www.ni.com
The Stand-Alone Data Logger System With Two Batteries, Voltage Regulator and the cRIO 9074 All Fitted Inside a Waterproof Cse
Legal This case study (this "case study") was developed by a National Instruments ("NI") customer. THIS CASE STUDY IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND AND SUBJECT TO CERTAIN RESTRICTIONS AS MORE SPECIFICALLY SET FORTH IN NI.COM'S TERMS OF USE ( http://ni.com/legal/termsofuse/unitedstates/us/).
7/7
www.ni.com
Appendix C
An experimental investigation into the whole body vibration generated during the hydroelastic slamming of a high speed craft. Halswell, P. K., Wilson, P. A., Taunton, D. J., and Austen, S. (2013).
An experimental
investigation into the whole body vibration generated during the hydroelastic slamming of a high speed craft.
Ocean Engineering, pending.
257
An experimental investigation into the whole body vibration generated during the hydroelastic slamming of a high speed craft P. K. Halswell1, P. A. Wilson1, D. J. Taunton1, and S. Austen2 1
University of Southampton; 2 Royal National Lifeboat Institution.
Corresponding author: P. K. Halswell, email address;
[email protected], Tel; +44(0)238059-7774, postal address; Fluid Structure Interactions Research Group, Engineering and the Environment, Highfield Campus, University of Southampton, Southampton SO17 1BJ.
Abstract High-speed planing craft expose their crew to levels of vibration that regularly exceed the daily exposure limit value set out by the European Directive 2002/44/EU. The human exposure to vibration can cause many effects, from chronic and acute, to physiological and psychological. Many reduction methods are currently being researched, such as suspension seats, active and passive trim tabs and interceptors, and suspension decks, but one method on its own does not appear to solve the problem. Coe et al. (2013); Coats et al. (2003) concluded that a combination of methods will be required to reduce the level sufficiently to meet the legislation. The highest levels of acceleration occur when a high-speed craft slams. This project experimentally investigates whether hydroelasticity can affect the slamming characteristics and Whole Body Vibration (WBV) using quasi-2D drop tests and full-scale drop tests. The quasi-2D impacting wedge was a flexible V-shaped hull with hard chines and this revealed that both the peak acceleration and the Vibration Dosage Value (VDV) can be affected by hydroelasticity. The trend in the peak acceleration was shown to be drop height dependent but the effect on the VDV was more dependent on the deadrise angle. The full-scale tests were performed on a Royal National Lifeboat Institution (RNLI) D-class inflatable lifeboat. The hydroelasticity was controlled using the internal pressures of the sponson and keel where increasing the internal pressures increased the structural stiffness. The full-scale results also showed that the trend in the peak acceleration was drop height dependent. The VDV in the crew position was shown to decrease with decreasing internal pressures and this lead to an increase in the VDV on the bow; however, the crew are not exposed to the bow acceleration. Incorporating an element of hydroelasticity shows the
potential alongside other reduction strategies to alleviate the human exposure to vibration on board high-speed planing craft. Keywords: Whole body vibration, mechanical shock, high speed crafts, rigid inflatable boats, inflatable boat, hydroelastic, slamming, drop tests.
Nomenclatures c
Phase speed of wave (m/s)
cRIO
Compact reprogrammable input output
EAV
Exposure action value
ELV
Exposure limit value
fps
Frames per second
g
Gravity
h
Water depth (m)
HSC
High speed craft
NI
National Instruments
MDF
Medium density fibreboard
RIB
Rigid inflatable boat
RMS
Root mean square
RNLI
Royal National Lifeboat Institution
VDV
Vibration dosage value
WBV
Whole body vibration
1. Introduction In 2002, the European directive 2002/44/EC was passed on the minimum health and safety requirements regarding the exposure of workers to physical vibration. This was included in UK legislation in 2005 through the Control of Vibration at Work Regulations, see Pond (2005), and again in 2007 via the Merchant Shipping and Fishing Vessels (Control of Vibration at Work) Regulations, see MCA (2007). The European directive sets the Exposure Action Value (EAV) for Whole Body Vibration (WBV) to 0.5 ms -2 RMS (or 9.1 ms-1.75 Vibration Dosage Value (VDV)) and the Exposure Limit Value (ELV) to 1.15 ms -2 RMS (or 21 ms-1.75 VDV). The VDV is used instead of the Root Mean Square (RMS) when the crest factor is above six; the crest factor is defined by the peak acceleration divided by the RMS acceleration. High Speed Craft (HSC) in rough water exposes the crew to non-linear vibration that regularly exceed the EAV and ELV. The highest acceleration occurs during a slam and in
the case of HSC this can involve the entire hull losing contact with the water's surface. A slam was defined by Ochi (1964) if the relative motion exceeds the local effective draught and the relative velocity at impact, see Lloyd (1998) page 292. Dand (2004) experimentally tested a rigid scale-model of the Royal National Lifeboat Institution (RNLI) D-class inflatable lifeboat where accelerations of up to 4 g in the crew's position were measured in regular waves, with a full-scale wave height of 0.55 m and full-scale speed of 19.4 knots. Townsend et al. (2008) showed that the RNLI Atlantic 85 Rigid Inflatable Boat (RIB) exceeded the EAV in 30 minutes at 32 knots with approximately 0.4 m significant wave height. Allen et al. (2008) measured the vibrations on the RNLI Atlantic 75 in two trials at speeds of 15 knots to 20 knots. The crest factors were above 6 which meant the VDV should be used instead of the RMS and the z-axis values were 48.51 ms-1.75 and 25.90 ms-1.75 in sea states two and three, respectively. Myers et al. (2011) measured the acceleration on board a military HSC at 40 knots in a sea state of two to three and the VDV for a 3 hour transit was 57.05 ms-1.75 on the deck. This is clear evidence that the vibration within these craft regularly exceeds the ELV and a solution, or combination of solutions, needs to be found. Whilst there is considerable debate in the marine community over the validity of applying the European directive to HSC, the RNLI are investigating methods to demonstrably mitigate the exposure of their crews and trainers to vibration. The human exposure to vibration can have many effects; from chronic and acute, to physiological and psychological, see Townsend et al. (2012). Physiological injuries have been reported by Ensign et al. (2000) to include; spinal and abdominal injuries, damage to internal organs (kidneys), torn ligaments and, broken ankles and legs. Ensign et al. (2000) also reported that the psychological injuries include; annoyance, fatigue, anxiety, loss of visual accuracy and reduced hand-eye coordination (the latter two could be considered a combination of both physiological and psychological effects). Myers et al. (2011) demonstrated that a three hour transit in a 40 knot HSC would reduce the physical performance of the crew (including run distance and vertical jump height). So, by reducing the boat motion and the human exposure to vibration this can reduce the risk of injury, provide a better working environment and increase the crew's effectiveness during and after transit. Researchers have explored many technological solutions to this problem; however, no single solution appears to be completely successful. Coe et al. (2013) concluded that a combination of solutions will be required to reduce the WBV enough to meet the legislation, which was also backed up by Coats et al. (2003). Suspension seats are one solution that is currently been heavily researched, see Coats et al. (2003); Cripps et al. (2004); Coe et al.
(2009); Olausson (2012); Coe et al. (2013); although, Townsend et al. (2012) pointed out that there were many drawbacks. Townsend et al. suggested a number of other strategies to reduce the WBV: suspended decks (also discussed by Coe et al. (2013)), active and passive fins, trim tabs, interceptors, gyrostabilisers, flexible hulls and elastomer coated hulls. Coats et al. (2009) discussed the use of a porous hull to reduce the impact loads and spread the energy over a longer time period, and they showed a significant reduction in impact loads. The RNLI D-class, see figure 1, is a five metre inflatable lifeboat which is capable of achieving 25 knots in the sea states associated with a Beaufort force two and can continue to operate safely up to and beyond the sea states associated with a Beaufort force five. It is powered by a 50 horse-power tiller-steer outboard engine and weighs a total of 655 kg (including all equipment and three crew). It is designed to have a maximum of three crew on board and is able to take a minimum of two casualties or one in the prone position. In 1998 the RNLI performed a feasibility study on the EA16 (the original version of the D-class) and compared it to seven commercially available vessels that included; RIBs, pure inflatable boats or a combination of both. It was found that the EA16 gave the best overall performance. Therefore, the RNLI have been improving its design and performance through either designers experience or trial and error to achieve the optimum boat. Anecdotal evidence from the feedback of the crew has reported that the flexibility within the D-class improves its performance, especially in waves and surf. There are four main structural components in the D-class, see figure 2, which includes; the inflatable sponson (also called the collar or tube), the segmented deck and transom, the inflatable keel and the fabric hull. The hydroelasticity of the D-class has been reviewed by Halswell et al. (2012) and provides the reader with a good overview of the D-class.
Figure 1: The RNLI B-class and D-class. The first component in the D-class is the sponson and it has been proposed by Natzijl (1998) and Pike (2003) (and discussed by others in the maritime community) that the sponson
is able to absorb energy during a slam but so far there is no scientific evidence. Haiping et al. (2005) undertook an experiment into the effect of sponson type on the sea keeping performance. It was found that an inflatable sponson had a lower response amplitude operator in heave and pitch than a foam sponson in both load conditions. This suggests that a flexible sponson can improve the ride comfort and sea keeping performance. Townsend et al. (2008); Townsend et al. (2008) performed experiments into the sea keeping performance of the RNLI Atlantic 75 RIB and they showed that the internal pressure of the sponson had minimal effect on sea keeping. They supposed that the sponson did not contact the water enough to have an effect on the sea keeping performance. So the sponson of the D-class could be responsible for the anecdotal evidence of improved performance because they are fully in contact with the water during operation; however, there are three other flexible components (deck, keel and hull) within the D-class that could also affect the sea keeping performance and the slamming characteristics. Townsend et al. (2012) numerically explored the effect of decreasing the hull stiffness to isolate the humans from the vibration. They reduced the hull stiffness from 69 GPa to 6.9 GPa and found that it had little effect on the response. This would suggest that the hull stiffness has minimal effect on the slamming characteristics; however, table 1 compares the properties of the rubber-coated fabrics, in the transverse direction of the hull (weft), to the properties of aluminium. Firstly, the density of the fabric is 57 % less than aluminium and the Young's modulus is nearly 250 times lower; although, the ultimate tensile strength of the fabric is only 27 % lower than aluminium. This reveals that the fabric used on the D-class hull is 250 times more elastic but is nearly as strong as aluminium. So it is anticipated that this degree of elasticity will affect the slamming characteristics.
Table 1: Fabric properties vs. aluminium properties. Property
Hull fabric (weft)
Aluminium
Density (kg/m3)
1152
2700
Young's modulus (GPa)
0.28
69
Ultimate tensile strength (MPa )
> 80
110
The structure of the RNLI D-class is able to deform significantly more than a conventional planing craft and it is supposed that this may change the slamming characteristics of this hydroelastic vessel. The slamming characteristics, primarily defined by the peak acceleration and peak duration, indicates the magnitude of human exposure to
vibration during a slam; therefore, is it possible to use this hydroelastic interaction to reduce the human exposure to vibration? Thus the aims can be summaries as: 1. Measure the effects of the pre-tensioned stress in a fabric hull on the slamming characteristics of a planing vessel using quasi-2D drop tests. 2. Measure the effects of varying the internal pressures in the sponson and keel of the RNLI D-class on the slamming characteristics during full-scale drop tests. 3. Quantify the change in human exposure to vibration by changing the structural stiffness of a planing vessel.
Figure 2: Main components of the D-class.
2. Experimental Methods 2.1. Quasi-2D Drop Tests It is hypothesised that the pre-tensioned stress in the fabric hull of the RNLI D-class will affect the slamming characteristics, especially the peak acceleration and the peak duration.
2.1.1 Equipment The quasi-2D drop test rig is shown in figure 3 and is the same rig used by Lewis et al. (2010). A pulley system was used to adjust the drop height. The wedge was fitted to two vertical poles via four bearings to ensure the wedge fell vertically. The accelerometer was
fitted to the top of the wedge near the bearings (to avoid submersion). A quick release shackle was used as a release mechanism. The steel framed wedge section was originally designed to test composite panels at a variable deadrise angle but a special fabric panel was wrapped around to form a fabric hull. The deadrise angle could be adjusted to 5°, 15° and 25°. The fabric hull was manufactured by the RNLI from two laminated sheets of Decitex 1100 rubber-coated fabric, the same material and lay up found on the D-class hull. A turnbuckle mechanism was used to adjust the pre-tension stress in the fabric hull. The wedge was dropped into a water tank that was 5.8 m long and 0.75 m wide, and the water height was 0.5 m. In the centre of the tank there was a side window (0.825 m × 0.485 m). A Crossbow CXL-HF accelerometer was used to measure all three axes of acceleration. The sensitivity was ± 10 mV/g and it had a range of ± 100 g. A DaqLab 2000 series data acquisition system was used with a sampling frequency of 5000 Hz and an accuracy of 0.1 mV. This system together gave an accuracy of ± 0.01 g and it was connected to a laptop for storage and processing purposes. A MotionPro X high-speed camera manufactured by Integrated Design Tools Inc. was used to record the event at 2000 frames per second (fps). Lights were required to provide enough contrast.
2.1.2 Parameters The main parameters affecting the slamming characteristics of a quasi-2D drop test are the deadrise angle, drop height and mass. The deadrise angle of the D-class varies from 0° at the transom to a maximum of 15°; therefore, the variable deadrise angle wedge was tested at 5° and 15°. A zero degree deadrise angle was not tested because other phenomena (such as air cushioning) occur at deadrise angles below 4°, see Bereznitski (2001). The maximum drop height of the rig was 1.2 m so 0.5 m and 1 m were tested, which correspond to an impact velocity of 3.13 m/s and 4.43 m/s, respectively. The mass and geometry of the wedge were not varied but they were scaled in relation to the D-class. The mass of the D-class is 655 kg, taken from Dand et al. (2008), and 5 m in length, which is 131 kg/m. The wedge had a longitudinal depth of 0.735 m and a mass of 50.2 kg, which is 68.3 kg/m. The transverse width of the wedge, diagonally from keel to chine, is 0.501 m so at a deadrise angle of 5° and 15° the horizontal width of the wedge (chine to chine) is 0.998 m and 0.968 m, respectively. The D-class has a width of two metres. This means that this wedge was approximately 1/2 scale for both mass and geometry.
Figure 3: Quasi-2D drop test rig.
The main variable of this experiment is the hull stiffness and three stiffness conditions were chosen; approximately rigid and, 1000 N/m and 0 N/m of pre-tensioned stress. The approximately rigid condition was represented by a 6 mm sheet of Medium Density Fibreboard (MDF) (6 mm thickness was the correct panel thickness for the frame) and the Young's modulus of MDF is approximately 4 GPa. MDF was chosen for its light weight and isotropic material properties. Lewis (2003) stated that varying the pre-tensioned stress within a fabric will have the same effect as varying the material properties, so only the pre-tensioned stress was varied. The real pre-tensioned stresses in the D-class' hull are unknown but, in places, the tension is very low because the fabric is very loose; therefore, one stiffness condition was used to represent zero hull tension (0 N/m). It was unclear how much more tension would result in a measurable difference so three times the magnitude was used (1000 N/m). The pre-tensioned stresses were only applied in the transverse direction and were measured using the strain in the fabric. The non-linear material properties of hull fabric showed that a tension of 1000 N/m produced a strain of 0.005. It was found that even under 1000 N/m the fabric deflected perpendicularly to the frame (out-of-plane) due to gravity; therefore, this deflection was also measured to increase the repeatability of the experiment. The central vertical deflection at 1000 N/m was 6 mm and at 0 N/m the deflection was 11
mm. The boundary conditions of the fabric sheet along the chine and centreline were a pin joint but along the two transverse edges the fabric was effectively a roller, only restricting the out-of-plane deflection. The boundary conditions of the MDF were very similar but along the chine and centreline the panel was bolted in place causing a fully clamped condition.
2.1.3 Procedure 1. The deadrise angle and pre-tensioned stress were set. 2. The wedge was raised to the desired drop height. 3. The water was allowed to settle so that the surface movement was less than ± 5 mm. 4. The data acquisition system and high speed camera were started. 5. The wedge was released. 6. The data acquisition system and high speed camera were stopped. 7. The procedure was repeated from step two to acquire three repetitions.
2.2 Full-Scale Drop Tests It has been hypothesised that the internal pressures in the sponson and keel will affect the slamming characteristics of the RNLI D-class, especially the peak acceleration and the peak duration.
2.2.1 Equipment The experimental set up of the full-scale drop tests can be seen in figure 4. The Dclass was lifted with a crane and a simple drop height gauge (a marked length of rope) was fitted to the transom to measure the drop height. The drop height gauge had an accuracy of ± 10 mm. A slip hook was used as a quick release mechanism which could be activated from the shore. The bungee cord was required to stop the heavy shackles (used for trim angle adjustment) from impacting the deck and causing unwanted structural vibration. The trim angle was adjusted using various boat harnesses of different lengths. A GoPro Hero 2 was used to record the drops at 120 fps. Finally a remote trigger was used to start the data acquisition system. Three CFX USCA-TX tri-axial accelerometers were used to measure the accelerations. They had a range of ± 20 g with a DC to 200 Hz flat frequency response. Above 200 Hz the accelerometers had a -6 dB response. Townsend et al. (2012) showed that the physical response of the human body is below 50 Hz, which ensures the accelerometers
capture the entire human response. The accelerometer signals were wired into a National Instruments (NI) 9234 C-series accelerometer module. The module had an input range of ± 5 V, a 24 bit analogue to digital converter and a built in anti-aliasing filter. This combined system has an accuracy of ± 6 µg. The location of the accelerometers is shown in figure 5. The first accelerometer was fitted to the transom next to the helm and the second accelerometer was fitted to the deck between the knees of the crew. These accelerometers measured the vibration of the structure at the point of contact between the helm or crew to provide an estimation of the accelerations experienced by the helm or crew. The final accelerometer was fitted near the bow to investigate how the vibration changes along the length of the D-class. The data logger was a NI 9074 compact Reprogrammable Input Output (cRIO) device. The cRIO was inside a waterproof case with its own battery to form a standalone data acquisition system, which was then strapped to the deck. The cRIO saved the data to two SD cards through a NI 9802 C-series SD card module at a sampling frequency of 2500 Hz.
Figure 4: Full scale drop test set-up.
Figure 5: Location of accelerometers during the full-scale drop tests (metres).
2.2.2 Parameters The main parameters for this experiment are the internal pressures of the sponson and keel; three internal pressure conditions were chosen, the standard operating pressures and ± 1 psi. The standard operating pressures are 3.25 psi in the sponson and 3 psi in the keel; therefore, the three internal pressure conditions are 2.25 psi and 2 psi, 3.25 psi and 3 psi, and 4.25 and 4 psi in the sponson and keel, respectively. The internal pressures of both the sponson and keel will be referred to as 2 psi, 3 psi and 4 psi for simplicity where in reality the internal pressure of the sponson is 0.25 psi higher. It was difficult to judge how much variation would be required to have a distinct effect on the slamming characteristics but anecdotal evidence from experienced helmsmen showed that a 0.25 psi change in the keel pressure would change the performance. This suggested that a variation of ± 1 psi would affect the overall response of the boat. The drop heights of the full-scale drop tests matched the drop heights used in the quasi-2D drop tests; 0.5 m and 1 m. The boat was always released at the same trim angle of 4.25° and this was as close to the running trim angle that could be achieved with the various boat harnesses. Dand et al. (2008) measured a running trim angle of four degrees. The trim angle was measured with a calibrated spirit level that had an accuracy of ± 0.25°.
2.2.3 Procedure 1. The internal pressures and trim angle were set. 2. The boat was raised to the desired drop height. 3. The boat and water surface was allowed time to settle so that the surface variation was less than ± 10 mm (measured using the drop height gauge). 4. The data acquisition system and video recorder were started. 5. The D-class was released.
6. The data acquisition system and video recorder were stopped. 7. The procedure was repeated from step two to six for five repetitions.
3 Results 3.1 Quasi-2D Drop Tests 3.1.1 Accelerations The frequency spectra from these results were very similar to the previous results obtained using the same test rig (i.e. Lewis et al. (2010)); therefore, the same 250 Hz low pass Butterworth filter was used to remove the unwanted frequencies around 600 Hz. The origin of this 600 Hz peak is unknown. Figure 6 shows an example of the spectrum once the data has been passed through a 250 Hz low pass filter.
Figure 6: Example frequency spectrum of the quasi-2D drop test with a 250 Hz low pass filter (deadrise angle = 5°, drop height = 1 m).
Four examples of the acceleration-time histories of the three stiffness conditions (MDF, 1000 N/m, 0 N/m) at the two deadrise angles and two drop heights can be seen in figures 7, 8, 9 and 10. The top graphs show the acceleration-time history when the data has been processed with a 250 Hz low pass Butterworth filter. The bottom graphs show the same data once it has been passed through a 40 Hz low pass Butterworth filter to smooth the graph
purely for viewing purposes. All data has been analysed with a 250 Hz low pass Butterworth filter. Figure 8 is the clearest example and the initial impact with the water can be seen after 0.32 s with a peak of nearly 14 g. The accelerations from the MDF condition are very similar to a regular under-damped sinusoidal system, except the second peak is smaller than the third. This irregularity in the second peak was found in all the hull stiffness conditions; however, the oscillations of the fabric conditions do not represent a regular sinusoidal system. This is clearly shown in the 0 N/m condition (of figure 8) where the fourth and sixth peaks are smaller than the fifth and seventh. The alternating peak height would imply that there are two out-of-phase sinusoidal oscillations occurring but this alternating peak height is not repeated in the other drop conditions. Observations from all the drop conditions show that the typical characteristics are: • The peak accelerations do vary due to the hull stiffness and the pre-tensioned stress. • The peak durations do vary due to the hull stiffness and the pre-tensioned stress. • The impacts nearly represent an under-damped sinusoidal oscillation but the fabric hull does not undergo constant damping. • During the first five peaks, the fabric hull normally has a higher damping ratio than the MDF hull; however, occasionally the damping ratio in one direction (positive or negative acceleration) of the fabric hull can be equal to that of the MDF hull. • After the first five peaks and once the accelerations have reduced dramatically (below 10 %), 50 % of the tests showed that the MDF hull still had a higher damping ratio but in 50 % of the tests the fabric hull had a higher damping ratio.
The mean peak accelerations are shown in figure 11. At 0.5 m drop height it can be seen that as the structure becomes more flexible the mean peak acceleration decreases. These results show that the peak acceleration can be decrease by 18.1 % and 12.4 % at 5° and 15° deadrise angles, respectively (with a 0.5 m drop height). On the other hand, this trend is inverted with a drop height of one metre and a 15° deadrise angle; the peak acceleration increases by 15.1 % with decreasing structural stiffness. It is unclear what effect structural stiffness has on the peak acceleration when there is a one metre drop height and a five degree deadrise angle because the MDF hull is unexpectedly high compared to all the other conditions. The fabric hull results fit the same trend as the other one metre drop heights. The measured difference between the mean peak accelerations could be due to uncertainties in the experimental set up and not statistical differences due to the hull stiffness; therefore, the null hypothesis was assumed. To test the null hypothesis a two tailed Student's
T-test was performed and the results can be seen in table 2. It showed that 5/6 of the test conditions had a statistical difference when the deadrise angle was 5° but only 4/6 of the test conditions had a statistical difference when the deadrise angle was 15°. Firstly, this implies that hydroelasticity has a greater impact when the deadrise angle is low because there is more statistical difference at 5° than 15°. Secondly, this suggests that hydroelasticity has a greater impact when the drop height is larger because, at 15° deadrise angle, there was no statistical difference at 0.5 m drop height but at 1 m drop height there was a statistical difference between 2/3 of the test conditions. Both of these observations agree with the work of Faltinsen (1999) and Bereznitski (2001) where the importance of hydroelasticity increases by either decreasing the deadrise angle or increasing the impact velocity. Finally, there is always a statistical difference between the MDF and 0 N/m stiffness conditions in all test conditions with a 90 % certainty, meaning that hydroelasticity has definitely affected the peak acceleration to a 90 % certainty. The peak duration was investigated because it was anticipated that this would also vary with the internal pressures but this was not the case. The general trend shows that the MDF condition had a longer duration that the fabric conditions; however, the change in duration due to the pre-tension stress in the fabric hull is not consistent. The percentage increase in peak duration from the MDF condition to the fabric conditions ranges from 7.5 % to 46.1 %. The peak duration is taken as the intersection with the x-axis (0 g).
Table 2: Student's T-test of the peak accelerations measured during the 2D drop tests.
(a) 250 Hz low pass filter.
(b) 40 Hz low pass filter. Figure 7: Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 15°, drop height = 0.5 m.
(a) 250 Hz low pass filter.
(b) 40 Hz low pass filter. Figure 8: Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 5°, drop height = 0.5 m.
(a) 250 Hz low pass filter.
(b) 40 Hz low pass filter. Figure 9: Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 15°, drop height = 1 m.
(a) 250 Hz low pass filter.
(b) 40 Hz low pass filter. Figure 10: Examples of the acceleration-time history from the quasi-2D drop test; deadrise angle = 5°, drop height = 1 m.
Figure 11: The mean peak accelerations measured during the 2D drop tests.
3.1.2 Deformations The high speed camera images of the impact through the side observation window can be seen in figure 12. The wedge is moving down the image and the fabric can be seen wrapped around the wedge. This impact had a deadrise angle of 15°, a drop height of 1 m and a pre-tension stress of 0 N/m. The first image shows the moment of impact. The second image shows that the fabric behind the wetted edge was pulled tight and ahead of the wetted edge the fabric was deformed. The third image shows that the fabric deformation moves with the wetted edge and continues until the wetted edge is outside the window of view. The forth image shows an interesting observation that occurs near the end of impact, which was a set of ripples in the fabric. This could be caused by high out-of-plane forces in the centre of the panel or a reflected structural vibration. In a trial test when the fabric tension was very loose, three of these ripples were observed and in higher tension tests the ripples were occasionally unnoticeable.
Figure 12: Transverse images of 2D drop test with a fabric hull under 0 N/m (15 degrees deadrise angle, 1 m drop height). 3.1.3 Error Analysis The MDF stiffness condition was chosen to represent an approximately rigid condition but a 6 mm sheet of MDF cannot be assumed rigid during these drops. The Young's modulus of MDF is approximately 4 GPa whereas the Young's modulus of the hull fabric is 0.28 GPa so it is clear that an MDF panel is considerably more rigid that the fabric hull. The boundary conditions between the MDF and fabric hull were slightly different; where the fabric hull had a pin joint and the MDF hull has a fully clamped condition. This will affect the panel deflection but the fully clamped boundary conditions of the MDF hull will reduce the panel to deflection because there is no rotational deflection. This means the MDF hull can be considered stiffer that the fabric hull. Furthermore, a fabric is defined because it has no out-of-plane bending stiffness; therefore, a fully clamped boundary condition will not remove rotational deflection of a fabric. So this change in boundary condition should not affect the response of the fabric.
When the wedge impacts the free surface it produces a wave and this wave could potentially be reflected by the end walls of the tank, which may interfere with the impact and the measured accelerations. The speed of a wave in shallow water can be calculated using c2=gh; where c is the phase speed of the wave, g is gravity and h is the water depth. The water depth was 0.5 m so the speed of the propagating wave is 2.214m.s-1. The tank was 5.8 m wide so the wave would take 2.62 s to reflect and collide with the impacting wedge. This shows that the reflected wave will not affect the measured accelerations. The acceleration-time history graphs of the drop tests show that the free-fall stage is not smooth. This was caused by the old bearings and because the wedge clipped the edge of the tank during decent. The time history graphs do show that the irregularities in the free-fall stage were consistent and the impacts were still repeatable.
3.2 Full-Scale Drop Tests 3.2.1 Accelerations The accelerometers used during the full-scale drop tests did not require any filtering because they had a flat frequency response up to 200 Hz and -6 dB above 200 Hz. The frequency spectrum showed very few vibrations (< 1 % of peak power) above 200 Hz. The frequency domain and frequency shifts will be discussed in more detail later. The acceleration-time histories of the full-scale drop test from accelerometers in the crew position and on the bow, at both 0.5 and 1 m drop height, can be seen in figures 13, 14, 15 and 16. All the graphs show the free-fall stage followed by the major peak acceleration. The overall impacts do follow an under-damped sinusoidal system but there are irregularities causing an inconsistent damping ratio. The typical characteristics of the time history graphs show that: • Changing the internal pressures does affect the peak accelerations at the crew's position and at the bow; although, the accelerations at the transom are negligibly affected. • Changing the internal pressures does affect the peak durations; however, there is no consistent trend. • The impacts nearly represent an under-damped sinusoidal oscillation but the damping ratio is not constant.
The mean peak accelerations from all three accelerometers during the full-scale drop tests can be seen in figure 17. The peak accelerations measured at the transom reveal that the
internal pressures had a minimal effect on the accelerations but this was expected because the structure at the transom is considerably stiffer than the rest of the boat. The inflatable keel does not extend right up to the transom, which removes the effect of the keel pressure all together. The vertical transom panel will dramatically reduce sponson rotation and this should reduce the effect of the sponson pressure. The null hypothesis was assumed and tested with a two tail Student's T-test, see table 3. It showed that with a 95 % certainty there is no statistical difference between the accelerations at the transom; however, with a 90 % certainty there is some statistical difference. This suggests that the sponson has a small effect on the peak acceleration but not significant compared to further forward. The accelerations measured in the crew position do show a trend of decreasing or increasing due to changing the internal pressures. At a 0.5 m drop height, as the internal pressures increase the peak acceleration also increases, which is the opposite trend to a one metre drop height. The Student's T-test (see table 3) confirms that there is a statistical difference between 5/6 of the drop conditions to a 95% certainty. Decreasing the internal pressures from 4 psi to 2 psi can lead to a 22.6 % decrease in the peak accelerations experienced by the crew at 1 m drop height; however, at a 0.5 m drop height the peak accelerations can increase by 39.0 %. The trends in the acceleration at the bow are the same as the trends found in the crew position and that is; at 0.5 m drop height, an increase in the internal pressure causes a decrease in peak acceleration, whereas, the trend is inverted at a one metre drop height. The Student's T-test (see table 3) again showed that 5/6 of the test conditions are statistically different with a 95 % certainty. Increasing the internal pressures from 2 psi to 4 psi can lead to a 48.9 % decrease, at 0.5 m drop height, and 8.6 % increase in the peak acceleration at a 1 m drop height.
(a) No filter.
(b) 40 Hz low pass filter. Figure 13: Examples of the acceleration-time history from the crew position during the fullscale drop tests with a drop height of 0.5 m.
(a) No filter.
(b) 40 Hz low pass filter. Figure 14: Examples of the acceleration-time history from the crew position during the fullscale drop tests with a drop height of 1 m.
(a) No filter.
(b) 40 Hz low pass filter. Figure 15: Examples of the acceleration-time history from the bow during the full-scale drop tests with a drop height of 0.5 m.
(a) No filter.
(b) 40 Hz low pass filter. Figure 16: Examples of the acceleration-time history from the crew position during the fullscale drop tests with a drop height of 1 m.
Figure 17: The mean peak accelerations measured during the full-scale drop tests.
Table 3: Student's T-test of the peak accelerations measured during the full-scale drop tests.
The peak duration was investigated again because it was anticipated that the duration would be proportional to the internal pressures; however, there are less trends than the quasi2D drops. In short, the peak duration results show no trends and few are statistically differences.
3.2.2 Error Analysis It is inevitable that the full-scale drop tests would be less repeatable than the quasi-2D tests because the set up and environment were less controllable. The drop height was controlled by a crane and, although it was surprisingly accurate, it was adjusted when the boat was five metres away which made it virtually impossible to read one millimetre on the drop height gauge. The water surface was not perfectly smooth because it was outside where the wind and tide could generate small waves, regardless of the settling time. The boat was hung from a very long cable (> 20 m) so the system acted as a pendulum which would also vary the drop height and cause the boat roll angle not to be horizontal. All of these variables were controlled and made repeatable through the drop height gauge by ensuring that the variation of the distance between the transom and water's surface was less than ± 10 mm. The drops were performed in a small marina so the surface waves generated during the slam did reflect off a nearby wall; however, the waves took 10 seconds to reflect off the wall and impact the boat. The impact took less than 2 seconds so the reflected waves did not affect the test results. The acceleration-time history graphs from the transom and crew, see figures 13 and 14, show a fairly consistent drop of 1 g; however, the same graphs from bow accelerometer, see figure 15 and 16, do not exhibit the same characteristics. Instead the bow accelerometer measures very low accelerations until there is a small positive acceleration before the major negative peak. This implies that the boat is rotating as it drops and that the trim angle is larger at the point of contact with the water than the trim angle at the time of release but the video recordings do not show any signs of rotation. It is unclear why the bow accelerometer does not experience a consistent 1 g drop.
4 Discussion 4.1 Frequency Spectrum The frequency domain will provide an insight into the accelerations and may help to reveal why the peak acceleration and peak duration has changed due to hydroelasticity. For the time being the full-scale tests will be discussed because the structural stiffness was increased in a linear manner, whereas the stiffness of the quasi-2D tests jumped from the fabric hulls to the MDF hull. The mean frequency spectra from all the full-scale tests can be seen in figure 18. The top two graphs from the transom accelerometer both show a clear peak at 1.7 to 2.1 Hz and the frequency of the peak increases with internal pressure (and thus structural stiffness). The first mode of vibration for a free-free beam is heave and the natural
frequency of heave is predicted to be 2.3 Hz using simple harmonic motion and 1.7 to 1.9 Hz using the dry heave method, see Rawson and Tupper (2001) page 477. This indicates that these peaks are caused by heave. There are another set of peak ranging from 5.3 to 6.4 Hz and this could be the pitch motion. The natural frequency of the pitch motion is 4.1 Hz (at 2.5° static trim angle) predicted using the dry pitch method, see Rawson and Tupper (2001) page 478. The errors in the predicted heave and pitch natural frequencies are great so it is not anticipated that they will exactly match the measured natural frequencies. The boat was dropped at a trim angle of 4.25° so the transom will impact first causing the boat to rotate around the transom, resulting in larger heave motion and relatively smaller pitch motion. The crew accelerometers, shown in the middle two graphs, both exhibit the same characteristic peaks at 1.7 Hz to 2.3 Hz and 4.5 Hz to 5.5 Hz for the heave and pitch motions, respectively. The frequency of the peaks increased due to increasing the internal pressures. There is another significant peak at 0.6 Hz that is also visible in the transom frequency spectra and it could be argued that this is heave motion; however, the bow spectra (see the bottom two graphs) do not show any sign of this peak. The bow will heave so this cannot be the heave frequency. The cause of this 0.6 Hz frequency is unclear. A comparison of the transom and crew spectra show that the dominant motion has changed; the transom experienced more heave motion and the crew experienced more pitching motion. The bow spectra show the same heave and pitch motions at 2.1 - 2.7 Hz and 4.5 - 5.6 Hz, respectively, and the bow experienced far more pitch motion than heave. There is an interesting super harmonic in the bow spectrum with a one metre drop height at two psi (bottom right graph) and it occurs every 2.16 Hz (ranging from 1.98 Hz to 2.29 Hz). This does not appear in any other full-scale test conditions but was found in all five repeats under this test condition. There is a correlation between this super harmonic and the peak acceleration under the same test condition (i.e. bow accelerations at one metre with two psi). The peak acceleration under this test condition appears to be higher than expected and do not fit the trends of the other test conditions, see figure 17. It is hypothesised that the unexpectedly high peak acceleration is caused by this super harmonic. The quasi-2D drop tests also contained one test condition that repeatedly exhibited higher peak accelerations than expected, which was the MDF wedge with a five degree deadrise angle at one metre drop height, see figure 19. An inspection of the frequency spectrum does also reveal super harmonic that occurs every 9.26 Hz (ranging from 8.55 to 9.46). The magnitude is smaller and less frequent but it is still present.
(a) Transom, 0.5 m drop height.
(b) Transom, 1 m drop height.
(c) Crew, 0.5 m drop height.
(d) Crew, 1 m drop height.
(e) Bow, 0.5 m drop height.
(f) Bow, 1 m drop height.
Figure 18: The mean power spectra of all the full-scale drop tests.
Figure 19: The mean power spectrum of the quasi-2D drop test with a MDF hull, 1 m drop height and 5° deadrise angle.
4.2 The Effect of Hydroelasticity on Whole Body Vibration The quasi-2D and full-scale drop tests actually measured the mechanical shock generated during a slam and not the WBV; therefore, the first step in quantifying the effect of hydroelasticity on the WBV is to relate mechanical shock to WBV. The primary method to evaluate WBV is calculating the RMS of the acceleration; however, the RMS is highly dependent on the time period and the peak acceleration is averaged out. The European directive states that if the crest factor is above six then the VDV should be used instead; the crest factor of all the drop tests (full and scaled) were above eight so the VDV should be used. The VDV is not designed for individual mechanical shocks but it still helps to characterise the slam and weight it depending on the frequency of the vibration. To the authors' knowledge there are no standards for evaluating the effect of a mechanical shock on a human. The drop tests have been compared to an under-damped sinusoidal system and this can be characterised using the natural frequency and the damping ratio; however, the damping ratio was shown to be irregular and there are multiple degrees-of-freedom leading to multiple natural frequencies. This is the reason for attempting to quantify the mechanical shock using the peak acceleration and peak duration but the peak acceleration will now be compared to the VDV. The peak acceleration have been plotted next to the VDV to relate the peak acceleration to a known evaluation method of WBV, see figures 20 and 21, and the error bars show two standard deviations. The crew of the D-class kneel in the boat and there are no frequency weightings for a kneeling person so both seated and standing weightings were used. It is anticipated that a kneeling frequency weighting would be in between a seated and
standing position because the knees and hips are still able to rotate. The VDV was calculated from the point of impact with the water surface for 1.5 s time period. The VDV from the quasi-2D tests at five degree deadrise angle does reinforce the conclusions drawn from the peak acceleration and they do show that the human exposure to vibration does change with hull stiffness. The VDV increased by 14.3 % when the 0 N/m pre-tensioned hull was changed to a MDF hull. On the other hand, the VDV from the tests with a 15° deadrise angle do not match the trends in the peak accelerations and this is due to the frequency weighting of the VDV. Nonetheless, the VDV does show that the stiffness and pre-tensioned stress in a hull can affect the human exposure to vibration during quasi-2D drop tests.
(a) 5° deadrise angle.
(b) 15° deadrise angle.
Figure 20: Quantification of WBV during the quasi-2D drop tests.
(a) Transom accelerations.
(b) Crew accelerations.
(c) Bow accelerations.
Figure 21: Quantification of WBV during the full-scale drop tests.
The VDV for the full-scale tests at the transom match the results from the peak accelerations because there is no significant effect due to hydroelasticity. The VDV from the crew position and the bow do not match the trends in the peak accelerations. Increasing the internal pressures tended to increase the VDV in the crew position but the VDV decreased with increasing internal pressures in the bow. The VDV increased by 10.5 % and 16.3 % in the crew position when the internal pressures were increased from 2 psi to 3 psi at 0.5 m and 1 m drop height, respectively. On the contrary, the VDV decreased by 28.5 % and 13.2 % on the bow when the internal pressures were increased from 2 psi to 4 psi at 0.5 m and 1 m drop height, respectively. This proves that internal pressures can change the human exposure to vibration during full-scale drop tests. The quasi-2D tests suggest that the VDV can be reduced by 14.3 % by changing the hull from a MDF hull to a 0 N/m pre-tensioned fabric hull at a five degree deadrise angle; however, when the deadrise angle was 15° the VDV trend was more scattered. This suggests that the hydroelasticity has a greater effect on WBV when the deadrise angle is low, which is backed up by the work of Faltinsen (1999) and Bereznitski (2001). The full-scale tests suggest that the VDV could be reduced by 9.5 % to 14.0 % in the crew position by reducing the internal pressures from 3 psi to 2 psi at 0.5 m and 1 m drop height, respectively; however, this will lead to an increase in the bow VDV by 2.5 % to 30.1 % but the crew are not exposed to these vibration.
4.3 Root Cause It can be concluded that hydroelasticity can vary the peak acceleration but what is the root cause of this variation? It is hypothesised by the author that there are two possible reasons for the variation. Firstly, that the variation is solely caused by the quasi-static shape of the hull, i.e. the hull deformation is relatively constant and there are no significant structural vibrations. The 2D problem is considered first because only hull stiffness was varied, meaning any measured effect was caused by the hull stiffness and not another variable. It is well known that the hull shape will affect the slamming characteristics and was briefly reviewed by Lloyd (1998). The structural stiffness (which is proportional to the pretensioned stresses, see Lewis (2003)) will change the impacting hull shape and this could cause of the effect of hydroelasticity; although, increasing the drop height will increase the deformation of the hull and exaggerate the effect of hull shape but this should not cause the peak acceleration trend to invert with drop height. Increasing the drop height may cause other
phenomena to occur such as air cushioning and flow separation, see Faltinsen et al. (2004), and air pockets, see Halswell et al. (2012). It is more likely that the structural vibration of the hull will have a significant effect on the rigid body motion and the root cause is a superposition of one (or more) structural vibration and the rigid body motion. The structural vibration will occur at different frequencies which will depend on the stiffness and boundary conditions of the structure. If the structural vibration and rigid body motion are in phase at the instantaneous moment of peak acceleration then this will lead to an increase in peak acceleration, whereas, if the two vibrations are 180° out of phase then this will lead to a decrease in peak acceleration. Faltinsen et al. (2004) divided this problem into two time scales; an initial structural-inertia phase and a free vibrations phase. The peak acceleration occurs in the structural inertia phase where large hydrodynamic forces lead to large accelerations on a small structural mass. The frequency spectrum graphs do show a shift in the dominant frequencies. For this reason it is believed that the inversion of the peak acceleration trend in the 2D tests was caused by a phase shift between the structural hull vibration and the rigid body motion, and not the change in shape; although, the reason for the phase shift is not known. Now consider the full-scale drop tests where the internal pressures of the sponson and keel will change the response of the sponson, keel, hull and deck. The root cause to the change in peak acceleration at full-scale could be due to the change in hull shape or the phase shift between the structural hull vibration and the rigid body motion; however, it could also be due to any shape changes or changes in natural frequencies of the keel, sponson or deck. It is probable that the variation in peak acceleration is due to the phase difference between the rigid body motion and one or more components for the same reason as the 2D tests. The inversion of the peak acceleration trend was found in the 2D and the full-scale tests which could imply that it is the structural hull vibration that caused the trend to invert because it was the only parameter changed in the quasi-2D tests but this cannot be confirmed.
4.4 A Hydroelastic Planing Craft? The first step in being able to use hydroelastic slamming on a planing craft to reduce the WBV is to find the root cause (i.e. more than just drop height) to the inversion of the peak acceleration trend. With our current knowledge, hydroelastic slamming will decrease the peak acceleration approximately half the time but the other half of the time hydroelastic slamming will increase the peak acceleration for a given hull geometry and stiffness. If the root cause is found then it should be possible to optimise the hull design to provide an overall
reduction in WBV, so that on a few occasions hydroelastic slamming may increase the peak acceleration but overall hydroelasticity will reduce the WBV. The 2D tests provide an insight into the effect of deadrise angle on this hydroelastic interaction. The reduction in peak acceleration was larger with a deadrise angle of 5° than 15°; at 5° deadrise angle both drop heights showed a reduction but at 15° only one drop height showed a reduction. Also the percentage reduction was greater at 5° than 15° with a reduction of 18 % and 12 %, respectively. The reason for this is unknown and it could be due to other phenomena such as air cushioning, flow separation and air pockets occurring. Nonetheless, it suggests that a hydroelastic hull would be more suitable for HSC with shallow deadrise angles, like the D-class, or that the structural stiffness should be varied along the length of the vessel with the deadrise angle for an optimal solution. So far hydroelastic slamming has been shown to be an advantage and a disadvantage so much further work is required to optimise the hydroelasticity so that it provides an overall reduction in WBV; however, now compare it to the other solutions available to reduce the WBV on planing craft and the effect hydroelasticity may have on a planing craft. Forward speed is very important for planing craft and it is likely that a hydroelastic hull will slow the craft down. Experienced helmsmen of the D-class have reported that the keel pressure will affect the forward speed. Dand et al. (2008) reported that a flexible planing surface can cause a phenomenon similar to porpoising on flat water but this was removed by increasing the shear modulus of the hull fabric. Hydroelastic planing surfaces were also discussed by Halswell et al. (2012). Although, these craft are used in heavy seas where the dominating factor on the forward speed in waves is actually the WBV experienced by the helm and crew because they cannot cope with the slamming accelerations. So hydroelasticity may reduce the flat water speed but the helms may be able to sustain a higher forward speed in waves. One major disadvantage for suspension seats and suspension decks is the increase in weight of the craft but a flexible hull does not increase the weight. In fact, the US navy have developed a craft where a composite hull was treated as a membrane surface and this allowed them to decrease the overall weight of the hull by 20 %, see Wood (2011). It was stated earlier that the density of the D-class fabric was 57 % lower than aluminium but the ultimate tensile strength was only 27 % lower which also proves that membrane structures could potentially be lighter. Although, the reaction forces on the structure of a flexible craft may be considerably higher than found on a rigid craft. To counter act this, the structure will have to be redesigned and inevitably lead to a more complex and heavier structural design. This
would increase the weight of the scantlings but the US navy still managed an overall weight reduction. A major advantage for a hydroelastic hull is its simplicity; nothing has to be added to the craft for it to be incorporated. Suspension seats have to be added to the craft which restricts crew movement and raises the vertical centre of gravity, see Townsend et al. (2012). Suspension decks require a highly complex system of springs and dampers, see Coe et al. (2013). Fins, interceptors and trim tabs all have to be added to the craft which will affect the hydrodynamics performance, increase the complexity of the fit out and increase the weight.
Table 4: Summary of the advantage and disadvantages of a hydroelastic slamming approach to reducing WBV. Advantages
Disadvantages
Potential to reduce the overall
Occasional increased in peak
WBV after optimisation.
acceleration after optimisation.
Increased speed in waves.
Reduction to flat water speed.
Reduction in weight.
Complex structural design.
Simplicity.
Increase in cost.
No change to operational ability.
Hydroelastic slamming cannot solve the problem of WBV in planing craft on its own with the current level of knowledge but it does show promise for being part of a combined approach to reducing the WBV. If hydroelastic slamming can reduce the WBV and the overall weight then the weight saving can be used to incorporate other WBV reduction strategies. This could lead to a significant overall reduction in WBV.
6. Conclusion The effect of hydroelasticity slamming on the peak acceleration and WBV has been experimentally studied using quasi-2D and full-scale drop tests. The quasi-2D drop tests of a high-speed planing hull with hard chines varied the hull stiffness, deadrise angle and drop height. The full-scale drop tests of a RNLI D-class inflatable lifeboat varied the internal pressures of the sponson and keel (which in turn varied the structural stiffness), and the drop height. The variation of acceleration along the length of the D-class was also studied. The follow conclusions were drawn:
• Hydroelasticity can affect the peak acceleration during 2D and full-scale drop tests. • The drop height can invert the trend of hydroelasticity on peak acceleration. • Hydroelasticity had a greater effect on peak acceleration with a lower deadrise angle or a larger drop height during the quasi-2D drop tests. • Hydroelasticity can affect the WBV during 2D and full-scale drop tests. • The frequency weighting of the VDV means the peak acceleration and WBV does not fit the same trend. • Decreasing the internal pressures on the D-class can reduce the WBV experience by the crew (by 9.5 % to 14.0 %) but this leads to an increase in the bow (by 2.5 % to 30.1 %). • Hydroelasticity had minimal effect on either peak acceleration or WBV at the aft of the D-class whether the structure is considerable more rigid.
There are still many unanswered questions about the effect of hydroelasticity on slamming and WBV, such as how to optimise the hydroelasticity to provide an overall reduction in WBV and why does the super harmonic cause an increase in peak acceleration? Nonetheless, this paper and corresponding work has shown that there is potential to use hydroelasticity to reduce the human exposure to vibration on high-speed planing craft. Membrane structures have been shown by Wood (2011) to reduce the overall weight of a craft whilst this work shows the potential to reduce the human exposure to vibration; therefore, a hydroelastic hull could be combined with other WBV reduction strategies to alleviate the risk of injury to the on-board crew, provide a better working environment and increase the crew's effectiveness during and after transit.
Acknowledgement This work is supported by the RNLI and EPSRC, and a special thanks to the employee at ILC for their support during the full-scale tests.
References Allen, D. P., Taunton, D. J., and Allen, R. (2008). A study of shock impacts and vibration dose values onboard high- speed marine craft. International Journal of Maritime Engineering, 150(A3):1-14. Bereznitski, A. (2001). Slamming: the role of hydroelasticity. International Shipbuild Progress, 4(48):333-351.
Coats, T., Gowing, S., and Shen, Y. (2009). Porous hulls research - phase 1. Technical report, Naval Surface Warfare Center, Virginia, USA. Coats, T., Haupt, K., and Lewis, J. (2003). Shock mitigation for personnel onboard highspeed combatant craft. In Human Systems Integration Symposium, VA. Coe, T. E., Rutherford, K. T., Dyne, S., and Hirst, J. (2013). Technical solutions for shock mitigation on high speed government craft. In SURV 8 - Surveillance, Search and Rescue Craft, pages 1-12, Poole, UK. Coe, T. E., Xing, J. T., Shenoi, R. A., and Taunton, D. J. (2009). A simplified 3-D human body - seat interaction model and its applications to the vibration isolation design of high-speed marine craft. Ocean Engineering, 36(9-10):732-746. Cripps, R. M., Cain, C. F., Phillips, H. J., Rees, S. J., and Richards, D. (2004). Development of a new crew seat for all weather lifeboats. In SURV 6 - Surveillance, Pilot and Rescue Craft, RINA, pages 69-76. Dand, I. (2004). RNLI D-class model: seakeeping measurements in head seas. Technical Report C3356.06, BMT SeaTech Limited. Dand, I., Austen, S., and Barnes, J. (2008). The speed of fast inflatable lifeboats. International Journal of Small Craft Technology, 150(B2):23-32. Ensign, W., Hodgdon, J. A., Prusaczyk, W. K., Shapiro, D., and Lipton, M. (2000). A survey of self- reported injuries among special boat operators. Technical report, Naval Health Research Centre, San Diego, UK. Faltinsen, O. M. (1999). Water entry of a wedge by hydroelastic orthotropic plate theory. Journal of Ship Research, 43(3):180-193. Faltinsen, O. M., Landrini, M., and Greco, M. (2004). Slamming in marine applications. Journal of Engineering Mathematics, 48(3/4):187-217. Haiping, H. E., Bretl, J. P. E., VanSumeren, H., Savander, B., and Troesch, A. W. (2005). Model tests of a typical rib in waves. Private Communication. Halswell, P. K., Wilson, P. A., Taunton, D. J., and Austen, S. (2012). Hydroelastic inflatable boats: Relevant literature and new design considerations. International Journal of Small Craft Technology, 154(Part B1):39-50. Lewis, S. G., Hudson, D. a., Turnock, S. R., and Taunton, D. J. (2010). Impact of a freefalling wedge with water: Synchronized visualization, pressure and acceleration measurements. Fluid Dynamics Research, 42(3):035509. Lewis, W. J. (2003). Tension structures: Form and behaviour. Thomas Telford.
Lloyd, A. R. J. M. (1998). Seakeeping: Ship behaviour in rough weather. Ellis Horwood, 1st edition. MCA (2007). The merchant shipping and fishing vessels (Control of vibration at work) regulations 2007. Myers, S. D., Dobbins, T. D., King, S., Hall, B., Ayling, R. M., Holmes, S. R., Gunston, T., and Dyson, R. (2011). Physiological consequences of military high-speed boat transits. European Journal of Applied Physiology, 111(9):2041-9. Natzijl, P. W. (1998). Bringing a 1.0 metre buoyancy tube to sea on a 18.0 metre rigid hull. In International Conference on Rigid Inflatables, Weymouth, UK. Ochi, M. (1964). Extreme behaviour of a ship in rough seas: slamming and shipping of green water. Society of Naval Architects and Marine Engineers. Olausson, K. (2012). Vibration mitigation for high speed craft. PhD Thesis, Royal Institute of Technology, Stockholm, Sweden. Pike, D. (2003). Inflatable tubes over advantages for small craft. Ship and Boat International, November (D):78-82. Pond, C. (2005). The control of vibration at work regulations. Rawson, K. J. and Tupper, E. C. (2001). Basic Ship Theory: Volume 2. Townsend, N., Coe, T., Wilson, P., and Shenoi, R. (2012). High speed marine craft motion mitigation using flexible hull design. Ocean Engineering, 42:126_134. Townsend, N. C., Wilson, P. A., and Austen, S. (2008a). Sea-keeping characteristics of a model RIB in head seas. International Journals of Maritime Engineering, 151. Townsend, N. C., Wilson, P. A., and Austen, S. (2008b). What influences rigid inflatable boat motions? Proceedings of the Institution of Mechanical Engineers, Part M: Journal of Engineering for the Maritime Environment, 222(4):207-217. Wood, K. (2011). Re-inventing the RHIB: Shock mitigation. Composite World, pages 1-4.
298APPENDIX C. AN EXPERIMENTAL INVESTIGATION INTO THE WHOLE BODY VIBRATION GENERATE
Appendix D
Measuring the stress-strain relationship of rubber-coated fabrics D.1 Aims The fabric material properties experiment aims to further the knowledge regard the behaviour of the rubber-coated fabrics used on the D-class. Currently, only the maximum force at failure is known. This experiment aims to measure the stress-strain relationship of both DTex 1100 and 1670 in the warp and weft direction, and also investigate the eect of thickness and lamination.
D.2 Method The experimental method followed that laid out by ISO 1421:1998.
The experiment was
performed in the Transport Systems Research Laboratories at the University of Southampton on the 8th and 9th of July 2011. The tensile testing machine was an Instron 5569. Only two loads cells were available, 5 kN and 100 kN, and the predicted failure load was up to 8 kN therefore the larger load cell was used.
D.2.1 Test samples The Neoprene coated polyester was supplied by Aerazur and coupons were cut by CNC machine with a circular cutting disk at the Inshore Lifeboat Centre (ILC). Three dierent material types were tested: DTex 1100, DTex 1670 and two ply of DTex 1100 laminated together. Only dry samples were considered during these experiments. The samples were cut to 50 mm wide and were placed in the clamps at 150 mm
±
± 0.5 mm
1 mm apart. Only the usable width of the
fabric was used, not the selvage.
D.2.2 Procedure 1. The top of the coupon was clamped rst (so that it was in the slack condition). 299
300APPENDIX D. MEASURING THE STRESS-STRAIN RELATIONSHIP OF RUBBER-COATED FABRICS
Figure D.1: Tensile test experimental set up
2. The bottom of the coupon was clamped. 3. 10 N± 1 N of pre-tension was applied. 4. The coupon was then extended at a constant rate of 100 mm/min until failure.
5. If the slippage was asymmetric or more than 2 mm then the sample was disregarded and a new sample was tested in replacement. If the sample failed within 5 mm of the clamps than the sample was excluded and a new sample was tested.
6. Each material type was tested in the warp and weft direction, plus each direction was tested 5 times for repetition.
The samples were placed in the clamps with a gap of 150 mm
±
1 mm because of the concern
of slippage. ISO 1421:1998 states to use a gap of 200 mm but if the elongation exceeds 75 % a 100 mm gap should be used. However it was decided that a longer sample would be better so a gap of 150 mm was used. The clamps were tapered which meant they did not grip the fabric right up to the edge of the clamp. It was checked that the top of the coupon did not slip; however, the taper meant that the actual fabric length was longer. The length that the clamp did not grip was measured. The average of 20 samples was 5.525 mm with a range of 4.5 mm to 8 mm.
D.3 Results The average results from the material properties experiment are shown in table D.1. It shows the maximum force and elongation at breakage, plus the standard deviation (σ) and the
D.3.
RESULTS
301
Table D.1: Force and elongation at breakage of DTex 1100, DTex 1670 and two ply DTex 1100.
coe cient of variation (Cv ) as required by ISO 1421:1998. The maximum force and elongation have been converted into stress and strain, and the linear Young's Modulus has been calculated. ISO 1421:1998 states that if the failure is within ve millimetres of the clamp then the result should be excluded because the failure was due to the clamp and a new sample should be tested. 35% of the samples failed due to the clamp using this guide line and in some case there were not enough samples to repeat the tests. After care analysis it was found that all initially excluded results, but one, correlated very well with the other successful samples and often the excluded results had a higher force to breakage than none excluded results. This suggests that the failure was not due to the clamp so the excluded data was used with caution.
D.3.1 DTex 1670 The stress-strain relationship of single ply DTex 1670 at a constant rate of extension of 100 mm/min is shown in gure D.2.
The data sheet reports the minimum tensile strength in
the warp direction as 80.05 MPa (510 kg/50mm) and in the weft direction as 73.77MPa (470 kg/50mm), shown as the data sheet failure lines.
The measured average maximum tensile
strength is above the minimum tensile strength from the data sheet as expected. The data sheet did not contain any information regarding the strain at breakage.
The stress-strain
relationship of the DTex 1670 is clearly non-linear and both the warp and weft directions respond in a similar manner, although the warp direction is slightly more elastic.
If the
Young's modulus is assumed to be linear, shown as the linear t lines, then the measured Young's modulus in the warp and weft directions are 303 MPa and 283 MPa, respectively. Moreover, the material is initially more elastic so the Young's modulus could be as low as 186 MPa and 183 MPa for the warp and weft direction, respectively. This could lead to an error of -38.6 % and -35.5 %, respectively, if the material was at a strain of approximately 0.14.
D.3.2 DTex 1100 The stress-strain relationship of single ply DTex 1170 at a constant rate of extension is shown in gure D.3 and the relationship is clearly non-linear. The minimum tensile strength of this fabric, according to the data sheet, was 70.63 MPa in both directions but all the samples tested exceeded this by at least 20 %. The data sheet, again, did not contain any strain properties. The data sheet reported that the minimum tensile strength of the fabric was the same for both
302APPENDIX D. MEASURING THE STRESS-STRAIN RELATIONSHIP OF RUBBER-COATED FABRICS
Figure D.2: Stress-strain graph of single ply DTex1670.
directions so it was expected that the properties would be the same; however, the properties of the warp and weft are remarkably dierent from each other. The warp direction is considerably stier and initial properties are especially interesting. The measured Young's moduli at failure in the warp and weft directions are 471 MPa and 304 MPa, respectively. The initial non-linear properties in the warp direction mean that the Young's modulus could be as high as 719 MPa or as low as 372 MPa, leading to an error between 52.7 % and -21.0 %, respectively.
The
Young's modulus in the weft direction could drop to 155 MPa with an associated error of -49.0 %.
D.4 Discussion D.4.1 Dierence between DTex 1670 and DTex 1100 It was expected that the stress-strain relationships of the DTex 1100 and the DTex 1670 fabric would be similar and the dierence would be proportional to the thickness or yarn count. Figure D.4 shows a comparison of the stress-strain relationships of both fabrics in the warp and weft directions. It shows that the response of both fabrics in the weft direction are very similar; whereas, the response of DTex 1100 in the warp direction is considerable dierent to the other three cases. These experiments have shown that there is a dramatic dierence and the cause should be found. Before this experiment the construction of the bres inside the Neoprene and Hypalon coating was not considered. It is now clear that the weave construction needs to be considered. The original fabric used on the EA16 had to be changed because the previous manufacturer stopped production.
The RNLI had great di culties nding a replacement fabric because
D.4.
DISCUSSION
303
Figure D.3: Stress-strain graphs of single ply DTex 1100.
none of the fabrics provided a suitable planing surface.
It could be hypothesised that the
dierence between the stress-strain relationship of the warp and weft direction are the reason why this material provides a good planing surface. The warp runs longitudinally down the boat and the weft transversely crosses the boat. The warp is more rigid and this may reduce the longitudinal movement of the fabric therefore providing a better planing surface.
The
weft which crosses the boat may allow the fabric to stretch, which will aect the slamming characteristics.
The anisotropic properties of these fabrics could be used to optimise the
performance of the IB1. This is an example how the aspects of hydroelasticity could be to consider the multi-dimensional eects of hydroelasticity.
D.4.2 Eect of lamination The eect of laminating two sheet of DTex 1100 can be seen in gure D.5, which shows the stress-strain relationship of single and double ply extended at 100 mm/s in the warp and weft directions.
This reveals that lamination causes the fabric to become more elastic and it is
most likely to be due to the addition of the glue used to laminate the sheets together. The glue is considerably more elastic than the fabrics so this will increase the overall elasticity of the laminated sheets, in a similar manner to the composite materials described in Gere and Goodno (2009). Lamination caused a 6 % and 3 % increase in the stress at failure in the warp and weft direction, respectively, and it caused a 16 % and 5 % decrease in the strain at failure in the warp and weft direction, respectively.
304APPENDIX D. MEASURING THE STRESS-STRAIN RELATIONSHIP OF RUBBER-COATED FABRICS
Figure D.4: Comparison of the stress-strain graphs for single ply DTex 1100 and DTex 1670.
Figure D.5: Eect of lamination on the stress-strain relationship of DTex 1100.
D.5.
CONCLUSIONS
305
D.5 Conclusions The fabric material properties experiment has measure the stress-strain relationship of DTex 1100 and 1670 in both the warp and weft direction. It revealed that DTex 1670 has similar stress-strain relationships in the warp and weft direction; however, the warp and weft directions of the DTex 1100 are considerable dierent, especially at low strains. It was shown that the thickness was not proportional to the properties and lamination was shown to increase the elasticity of the fabric.
306APPENDIX D. MEASURING THE STRESS-STRAIN RELATIONSHIP OF RUBBER-COATED FABRICS
Appendix E
Waterline de ection experiment E.1 Introduction The shape and magnitude of the de ection and the origins of the structural stiness of the IB1 are not understood. Therefore a simple experiment was performed to measure the static deformation under various loading conditions in-situ (in water). This experiment was intended to be quick and cheap and to be the rst step in understanding the deformation; this will allow further experiments to be performed if it is justi able.
E.2 Aims This experiment aimed to quickly and cheaply measure the deformation of the IB1 in water under various loading conditions. It measured the shape and magnitude along the waterline as well as the length and width of the vessel. The loading conditions included both realistic crew conditions and central loads that ranged up to realistic slamming conditions. The deformation was checked from symmetry and hysteresis eects. This experiment aimed to check and measure the movement of the fabric around the corner of the deck, however, it was found that the corner was too far under the water to be measured from the shore. No observations could be made. It was also hoped that the sponson and keel pressures could be altered to investigate their eects but time constrictions meant this was not completed.
E.3 Method This experiment was performed on the 16th of November 2010 and took 1 day to complete. It was performed next to the RNLI pontoon at the Inshore Lifeboat Centre (ILC) in Cowes, Isle of Wright. 307
308
APPENDIX E. WATERLINE DEFLECTION EXPERIMENT
Figure E.1: Static de ection experiment; top shows a unloaded IB1 and bottom shows an IB1 loaded with 1.75 tonnes
Figure E.2: Load positions (all dimensions in mm)
E.3.1 Loading conditions Three real crew loading conditions were used: one causality in the prone position.
two crew, three crew and three crew plus
It was assumed that the each crew member weighted
approximately 100 kg to include the weight of the human and their wet equipment (dry suit, helmet, life jacket, etc.).
The weights were placed where the crew/casualty will normally
operate and the exact positions can be seen in gure E.2. The centrally positioned loads were increased in steps of 1/2 a tonne up to the boat's limit which is reported to be 2 tonnes.
E.3.2 Hysteresis eect In atable tubes have been reported by Rowe et al. (2006) to have a hysteresis eect so that the tubes do not return to their original shape after they have been de ected. The hysteresis eect of the sponson could aect the shape of the boat after deformation. This needed to be checked before the rest of the experiment could proceed. To check for a hysteresis eect the boat was measured for an unloaded condition, and then the boat was loaded in steps with measurements taken at each step.
Finally the boat was
unloaded in steps, again, with measurements taken at each step. The loading stages used were 0, 254, 495, 254 and 0kg. Any dierence in the de ection measurements will indicate that the
E.4.
RESULTS
309
hysteresis eect does exist.
E.3.3 Measurement method The black protective rubber strip that surrounds the boat was used as a baseline and measurements were made between this line and the water surface using a ruler. 12 measurements were made along the length of the boat with roughly equal spacing but some measurement points were speci cally positioned on the deck joints to focus the results.
E.4 Results E.4.1 Error analysis This experiment was performed beside a sheltered pontoon but small wavelets were present around the boat and this caused the water level to vary. The water was allowed to settle in between each set of measurements to minimise this error.
Personal judgement was used to
decide the position of the average water level over a short period (30 seconds). This meant that only one measurement was taken per measurement position and loading condition. So the average and standard deviation cannot be calculated. A ruler was used to measure the de ections with an accuracy of
±
0.5 mm. A sprite level
was used to ensure the ruler was vertical. The in atable pressure of the sponson and keel were measured using a pressure gauge with an accuracy of
±
0.1 psi. The loads were not precisely
central as shown in the symmetrical de ection results. The measurements were made on the sponson that is also able to deform. So this means that the measured de ections are not necessarily the de ections of the deck joint or panel. The sponson may have deformed into irregular shapes.
E.5 Analysis and discussion E.5.1 Symmetrical de ection When the vessel was centrally loaded the dierence between the port and starboard de ections had a standard deviation 2.307 showing good signs of symmetric de ection. The mean dierence was -4.691 mm with all results showing a continuous heel to port probably because the load was not precisely central.
E.5.2 Hysteresis eect The vertical and length de ections results showed no signs of the hysteresis eect. The change in the sponson de ection was irregular and no conclusions were made. This is shown in gure E.3.
310
APPENDIX E. WATERLINE DEFLECTION EXPERIMENT
Figure E.3: Change in sponson beam during hysteresis check
E.5.3 Central loading conditions E.5.3.1 Vertical de ection The average de ection of the IB1 can be seen in gure E.4 for the central loading conditions where the change in sinkage and trim have been removed. An outline has been imposed below to show which components are bending and the transverse lines represent the deck joints. Figure E.4 clearly shows that as the load increases the de ection increases. It shows that the main de ection is due to the deck joints, in particular the bow deck joint. It also shows that the deck panel does deform and there is noticeable bending in the main aft deck panel.
140
120
Vertical Position (mm)
100
80
254kg
60
495.3kg 1000 kg
40
1495.3 kg 1749.3 kg
20
0 0
500
1000
1500
2000 2500 Measurement Position (mm)
3000
3500
4000
4500
Figure E.4: Vertical de ection of the IB1 due to centrally placed loads
Figure E.5 shows that as the load is increase the maximum de ection is linear. Point of
E.5.
ANALYSIS AND DISCUSSION
311
2000
1800 1600
Load (kg)
1400
y = 13.84 x
1200 1000 800
600 400 200 0 0
20
40
60
80
100
120
140
Maximum Deflection (mm)
Figure E.5: Linear relationship between maximum vertical de ection and central load
maximum de ection normally lies in between the two middle deck panels although it does move around.
E.5.3.2 Length de ection The results for the change in length showed that the length of the boat decreases as the load increases.
The results are shown in gure E.6 and the relationship was approximated to a
linear correlation of
y = −0.0315x.
The results for the length under a 500 kg load appear to
be an anomaly, this will be discussed later.
Figure E.6: Change in length against load for central loading conditions
E.5.3.3 Width de ection The change in width results again strongly indicated that the sponson width decreases as the load increases. The relative results are shown in gure E.7 and the results closely t the linear relationship of
y = −0.022x.
Figure E.7: Change in width against load for central loading conditions
312
APPENDIX E. WATERLINE DEFLECTION EXPERIMENT
E.5.4 Crew loading condition E.5.4.1 Vertical de ection Figure E.8 shows the vertical de ection of the IB1 under crew loading conditions. The two crew loading condition, which was symmetrical around the longitudinal centreline, showed that the de ections of port and starboard were similar; but conditions with three crew where the loads are not symmetrical around the centreline caused a considerable dierence in de ection from port to starboard. It is interesting to see that the port side under three crew loading condition actually caused an upward de ection.
20
Vertical Deflection (mm)
15
10
2 crew Starboard 2 crew - Port
5
3 crew Starboard 3 crew - Port
0
3 crew+casuality - Starboard 3 crew+casuality - Port 0
500
1000
-5
1500
2000
2500
3000
3500
4000
4500
Measurement Position (mm)
Figure E.8: Vertical de ection of the IB1 due to crew placed loads
E.5.4.2 Length de ection The change in length due to crew loading displayed irregular results. The largest deformation was cause by two crew (200 kg) and the other larger loads (3 crew and 3 crew+casualty) caused smaller deformations. It is not known why this happened. However, this does match the results for the central loading conditions, shown in gure E.9. Small loads cause a large reduction in length then further increase of the loads causes the length increase, before the length starts to decrease because the load is increased again.
E.5.
ANALYSIS AND DISCUSSION
313
60
50
Change in Length (mm)
40
30
Crew Loading Central Loading
20
10
0 0
200
400
600
800
1000
1200
1400
1600
1800
2000
Load (kg)
Figure E.9: Change in length for all loading conditions
45 40
Change in Length (mm)
35 30
25 Crew Loading
20
Central Loading 15 10 5 0 0
200
400
600
800
1000
1200
1400
1600
1800
2000
Load (kg)
Figure E.10: Change in width for all loading conditions
E.5.4.3 Width de ection It was expected that more crew members would cause a larger de ection of the sponson width but this was not shown. All crew loading conditions produced the same sponson de ection. This could be due to the fact that the loading magnitude was too small to cause a signi cant change. It could be hypothesised that a small load, between 200kg and 500kg, causes a stepped de ection of 10 mm. This is because the hysteresis test (loads of 250 and 500kg) and the crew tests (loads of 200 to 400 kg) all caused a de ection of 10 mm. Then there is an activation load level which causes more severe de ections which could be in a linear manner because of the overall de ection test results. In a practical sense this could be explained by saying the stepped de ection is caused by the elastic behaviour of the sponson and the activation level is due to frictional forces between the deck corner and the hull fabric. This hypothesis will need further work to prove that it is correct.
314
APPENDIX E. WATERLINE DEFLECTION EXPERIMENT
E.6 Conclusion It has been shown that the hysteresis eect does not exist in the vertical de ection and in the length of the boat. However, it is unclear whether the sponson width exhibits the hysteresis eect or if the results are invalid. It has been shown that when the vessel is loaded with a central load the de ection is symmetrical. Plus when the vessel is loaded with crew loading the de ection is not symmetrical because the load is not symmetrical. The results show signs that there is complex coupling of the various structural components. But it has also been shown that over a large load range that the response can be approximated to a linear relationship. It was shown that the structural stiness is dominantly aected by the deck joints and deck panels and the point of maximum de ection moves.
E.7 Future work Several methods for making the measurements have been proposed, including using a laser ruler, photographs with a gridded background and extensometers, but the biggest issue is how to support the boat to achieve accurate and repeatable results.
E.7.0.4 Keel and sponson pressure The keel and sponson pressures should be varied during the further work to allow a greater understanding of the coupling of the various components. This is needed to understand the eect of keel and sponson pressure on deck joint stiness.
E.7.0.5 Vertical, length and width de ections Further work could be performed into the vertical, length and width de ection.
A better
measurement method should be found and small increments of load should be used. These changes in the method will hopefully show how the width is distorting and allow the hypothesis stated earlier to be proved valid or invalid.
E.7.0.6 Fabric position The fabric position around the corner of the deck should be inspected.
This will involve
measurement under the waterline and suitable and safe method should be found. Investigating the fabric position will further the understanding of the change in sponson width and this will hopefully allow a greater understanding of the variations in the fabric hull tension.
Appendix F
Compact RIO programming
315
316
APPENDIX F. COMPACT RIO PROGRAMMING
Appendix G
Flat water trial videos This appendix includes the moving graph videos of the at water trials and they can be found on the attached CD. All these videos are at half speed. Press Ctrl, Shift and S at once to slow the video down again using Windows media player.
Top left graph:
Sponson and keel pressure.
Black vertical line indicates time.
Top right graph:
Ti = Trim angle (deg)
B1i = Aft deck hinge angle (deg).
B3i = Middle deck hinge angle (deg).
*
The sensors was not functional at 2 psi.
B4i = Bow deck hinge angle estimated to be 2 degrees.
Bottom left graph:
Deck hinge angles.
The line are indicates and not scaled; although, the scaling is the same in all moving-graph videos (including drop test).
Bottom right graph:
Hull shape.
317
318
APPENDIX G. FLAT WATER TRIAL VIDEOS
G.1 Graphs at 2 psi G.2 Graphs at 3 psi G.3 Graphs at 4 psi G.4 Graphs of the pulsing motion
Appendix H
Drop test videos The drop test videos are on the attached CD. The rst video is a front view at quarter speed and the second video is a side view at one eighth speed. Both drops are at 3 psi and one metre drop height. The moving-graph videos have the same format as the at water videos expect they are at quarter speed.
Press Ctrl, Shift and S at once to slow the video down again using
Windows media player.
H.1 Video of drop H.2 Graphs at 2 psi H.3 Graphs at 3 psi H.4 Graphs at 4 psi
319
320
APPENDIX H. DROP TEST VIDEOS
Appendix I
Full-scale drop tests - additional graphs I.1 Time and frequency domain of hull de ection
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.1: Mean time and frequency domain of the hull sensor H11.
321
322
APPENDIX I. FULL-SCALE DROP TESTS - ADDITIONAL GRAPHS
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.2: Mean time and frequency domain of the hull sensor H14.
I.1.
TIME AND FREQUENCY DOMAIN OF HULL DEFLECTION
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.3: Mean time and frequency domain of the hull sensor H15.
(a) Drop height 0.5 m.
323
324
APPENDIX I. FULL-SCALE DROP TESTS - ADDITIONAL GRAPHS
I.2 Time and frequency domain of deck panel strain
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.5: Mean time and frequency domain of the deck panel strain D1x.
I.2.
TIME AND FREQUENCY DOMAIN OF DECK PANEL STRAIN
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.6: Mean time and frequency domain of the deck panel strain D1y.
325
326
APPENDIX I. FULL-SCALE DROP TESTS - ADDITIONAL GRAPHS
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.7: Mean time and frequency domain of the deck panel strain D2x.
I.2.
TIME AND FREQUENCY DOMAIN OF DECK PANEL STRAIN
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.8: Mean time and frequency domain of the deck panel strain D2y.
327
328
APPENDIX I. FULL-SCALE DROP TESTS - ADDITIONAL GRAPHS
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.9: Mean time and frequency domain of the deck panel strain D4x.
(a) Drop height 0.5 m. Figure I.10: Mean time and frequency domain of the deck panel strain D4y.
I.2.
TIME AND FREQUENCY DOMAIN OF DECK PANEL STRAIN
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.11: Mean time and frequency domain of the deck panel strain D5x.
329
330
APPENDIX I. FULL-SCALE DROP TESTS - ADDITIONAL GRAPHS
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.12: Mean time and frequency domain of the deck panel strain D5y.
I.2.
TIME AND FREQUENCY DOMAIN OF DECK PANEL STRAIN
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.13: Mean time and frequency domain of the deck panel strain D6x.
331
332
APPENDIX I. FULL-SCALE DROP TESTS - ADDITIONAL GRAPHS
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.14: Mean time and frequency domain of the deck panel strain D6y.
I.2.
TIME AND FREQUENCY DOMAIN OF DECK PANEL STRAIN
(a) Drop height 0.5 m.
(b) Drop height 1 m. Figure I.15: Mean time and frequency domain of the deck panel strain D7x.
333
334
APPENDIX I. FULL-SCALE DROP TESTS - ADDITIONAL GRAPHS
Appendix J
Transmission loss predicted by ImTL model
335
336
APPENDIX J. TRANSMISSION LOSS PREDICTED BY IMTL MODEL
337
338
APPENDIX J. TRANSMISSION LOSS PREDICTED BY IMTL MODEL
339
340
APPENDIX J. TRANSMISSION LOSS PREDICTED BY IMTL MODEL
Appendix K
MatLab code for Savitsky prediction % This script will calculate drag on a planing hull using Savitsky 1964. % Author: PKH % Date: 10.12.2014 for run = 1:3; %% Inputs B = [13.0506, 14.0223, 14.3865]; V = [21.49, 21.73, 21.93]; trim = [4.83, 4.88, 4.89];
% Deadrise angle (degrees) % Forward speed (Knots) % Trim angle (degrees)
B = B(run);
% Deadrise angle (degrees)
m = 655; m = m .* 2.20462;
% mass (kg) % mass (lb)
V = V(run); V = V ./ 1.94384449; V = V .* 3.28083989501;
% speed (knots) % forward speed (m/s) % forward speed (fps)
trim = trim(run);
% Measured in the thesis
rho rho b = b =
% % % %
= 1000; = 62.30 ./ 32.2; 1.903; b .* 3.2808399;
d = 0.11; d = d .* 3.2808399; Cf = 0.00177; %% LB pd V1
density of water (kg/m^3) density of water (lb/ft3) beam (m), see Dand et al. 2008 beam (feet)
% sinkage (m), see C3356.04 fig 13 % sinkage (feet)
%
Calculations = 1.422; %Taken from Dands inputs = m./(LB.*b^2.*cosd(trim)); = V.*(1-(2.*pd)./(rho.*V^2))^0.5;
draglb = (m.*tand(trim))+(rho.*V1^2.* Cf.*LB.*b^2)/(2.*cosd(B).*cosd(trim)); dragkg = draglb .* 0.453592; dragN = dragkg .* 9.8066 Drag(run) = dragN; end
341
