Model and full-scale trawl comparison L Fiorentini et al. 10.1046/j.1444-2906.2004.00813.x Original Article349359BEES SGML
FISHERIES SCIENCE
2004; 70: 349–359
Comparison between model testing and full-scale trials of new trawl design for Italian bottom fisheries Loris FIORENTINI,1 Antonello SALA,1* Kurt HANSEN,2 Giulio COSIMI1 AND Vito PALUMBO1 1
Institute of Marine Science – Fisheries Section, National Research Council, Largo fiera della pesca, 60125 Ancona, Italy and 2SINTEF Fisheries and Aquaculture, North Sea Center, DK-9850 Hirtshals, Denmark ABSTRACT: Two large trawl models, at a one-quarter scale, were designed, produced and tested in a flume tank in Hirtshals (Denmark). A traditional trawl, typical for the commercially bottom fisheries in Italy, was selected as a basic design for the development of a new trawl. Characteristics of the new design should include larger meshes in the net areas, where no negative effect on catching power is foreseen, high bosom height, good bottom contact and low towing resistance. The geometry and towing resistance of both the traditional and the experimental trawl models were initially measured in the flume tank for different riggings. Based on the results from detailed flume tank tests, two fullscale trawls were designed and produced. Engineering sea trials have been made on a research vessel to measure the performance of the full-scale trawls and to verify the findings from the flume tank tests. The comparison between sea trials and flume tank tests shows that it is very difficult to accurately model, in the flume tank, trawl sections where the highly elastic Rachel netting made from polyamide is used. KEY WORDS: bottom trawl, experimental fishing, flume tank, gear research, modeling.
INTRODUCTION
MATERIALS AND METHODS
The current paper illustrates a new trawl design for the Italian demersal fisheries, including the use of larger mesh sizes. The new design should combine the features of high bosom height, good bottom contact and low towing resistance. The typical bottom trawl type, which is commonly used in the commercial Italian fisheries, was selected as the reference trawl (traditional trawl). A new experimental trawl design has been developed, based on the results from detailed flume tank tests of the traditional trawl and from an empiric model for the estimation of netting drag. Finally, to verify the results of the flume tank tests, sea trials have been carried out for both traditional and experimental trawls. The purpose of the present paper was to discuss the differences in the results for the Italian bottom trawls from the engineering sea trials and from the flume tank tests.
Trawl specifications
*Corresponding author: Tel: 39-071-207-8828. Fax: 39-071-55-313. Email:
[email protected] Received 5 December 2002. Accepted 8 December 2003.
For the basic design of the Italian traditional trawl, a review of common commercial trawl specifications was made. Some difficulties arose because there are few Italian companies producing ready-made trawls. Normally the nets of a trawler are produced by the oldest member of a fisherman family who no longer goes to sea. The basic design is the result of some modifications introduced by fishermen over the last few years to the old typical Italian trawl. Even if the basic design is always the same, there are some differences from vessel to vessel. Because the full-scale testing at sea would be doneout with the research vessel (RV) S. Lo Bianco (660 HP with a ducted controllable pitch propeller), a traditional trawl for a vessel of 500 HP was selected. Both the full-scale and model trawl designs are, respectively, shown in Figs 1 and 2. The design of the experimental trawl is shown in Fig. 3 and its model in Fig. 4. Additional information on specific materials and methods used
350
FISHERIES SCIENCE
L Fiorentini et al.
Fig. 1 Design of the full-scale Italian traditional trawl.
for data analysis are presented in the following sections. The general features of the new experimental design are as follows. (1) The wings of the new trawl are built from two panels (upper and lower), which have bar cutting along the fishing and float line and in the selvedge opposed to the one-panel wings in the traditional style Italian trawl. This change has been introduced to increase the bosom height of the trawl. The reduction of the netting area in the wing sections by this alteration amounted to approximately 7.5%. (2) The widths of the upper and lower panels were redistributed, compared to the traditional
design. In the new design, the width of the upper belly is only 1.14-fold the width of the lower belly, while in the traditional trawl it is 2.38-fold. (3) A large amount of slack in the bottom panel, which is usual in Italian trawl design, has been incorporated in the new design as well. Trawl model design The design of the models is based on the normal scaling rules.1–3 The linear scale factor l used here is defined as the quantity in the full-scale trawl divided by the corresponding quantity in the model. In general terms, reductions to dimen-
Model and full-scale trawl comparison
FISHERIES SCIENCE
351
Fig. 2 Design of the model of the Italian traditional trawl.
sions of a linear nature are made throughout the model by the amount of the basic factor. Factors concerning drag resistance, which is dependent on surface area for its value, decreases by the square of the basic scale.4,5 Weight and buoyancy forces that rely on volume for their value are reduced by the cube of the basic scale. The fundamental modeling rules may be summarized as follows where f and m refer to full-scale and model, respectively: l=
Lf
Am =
Lm Af l2
(1) (2)
Fm =
F f rm ◊ l3 r f
(3)
Vf l
(4)
Vm =
where L, A, F and V stand for a certain length, area, force and towing speed, respectively. A large model, at a 1/4 scale, of the chosen traditional trawl was first designed and produced. The scaled trawl was entirely constructed from knotted netting materials because knotless model netting does not exist. To compensate for the difference with the full-scale trawl, which uses knotless netting, 15% less netting area of knotted netting was used in the scaled models.6
352
FISHERIES SCIENCE
L Fiorentini et al.
Fig. 3 Design of the full-scale Italian experimental trawl.
The new experimental trawl model was designed following an analysis of the geometry and towing resistance measured on the traditional trawl in the flume tank. During the model tests of this new design, the trawl rigging was altered in order to obtain useful information for the new improved trawl design. In the experimental Italian trawl some of the design ideas normally used in the Danish trawls were introduced in the front part of the trawl. For example, in the traditional Italian trawls the netting in the wing sections is cut alongside knots or never steeper than a 2N2B cut along the sides hung to the headline and fishingline. In the Danish designs an all-bar cut or in more modern designs drop meshes are used along the frame ropes. Moreover, in the Italian design the wings do not have a selvedge dividing it into an upper and a lower wing section. The difference between the two design principles is that in the Italian design all
meshes are closed due to the tension in the bars, whereas the mesh opening is controlled by the hanging ratio when the Danish design principle is used. Flume tank tests The North Sea Center flume tank in Hirtshals (Denmark) has a large test section (21 m long ¥ 8 m wide ¥ 2.7 m deep). Inspection windows are installed throughout the length of the tank test section, which has a moving bottom made from a woven polyester conveyor belt. The purpose of the belt was to generate friction with the footropes in order to represent a hard even sea bed. The tests were conducted by connecting the trawl models directly to the masts and the measurements were taken at three different mast
Model and full-scale trawl comparison
FISHERIES SCIENCE
353
Fig. 4 Design of the model of the Italian experimental trawl.
spreads (corresponding to full-scale wing-end spreads of 12, 15 and 18 m) and at different water speeds (corresponding to the full-scale range of 2.5–4 knots). The parameters measured were: loads ahead of the two wing tips, wing-end spreads and bosom heights in respect to the bottom. Engineering tests at sea Either the traditional or the experimental Italian full-scale trawls were tested in the Adriatic sea, at various depths, using the RV S. Lo Bianco. Polyvalent oval doors were used (1750 mm ¥ 1050 mm). Warp lengths were paid out in relation to the bottom depth and during some hauls their length was altered in order to vary the spread of the net. In each haul, the net was towed on a fixed course and the vessel speed was varied in steps.
In order to determine the effects of the sea current, at least two tows on reciprocal courses were made for each gear arrangement tested. The measured vessel parameters were vessel speed relative to the sea bed, warp loads, engine revolutions, shaft torque, shaft power and fuel consumption of the main vessel engine. The SCANMAR sensors (SCANMAR, Norway) were mounted on the trawls to measure the door spread, the upper wing-end spread and the bosom height above the sea bed. Wire loads were measured ahead of the four wing tips by underwater load cells inserted just in front of the wing-ends. Data were then processed by taking into consideration the effect of the sea currents on the towing speed. To correlate the various parameters with the vessel speed, a dummy variable was added as an independent variable in fullscale regression equations:7 it was given a value of +1 when the data were collected with a counter
FISHERIES SCIENCE
354
current and a value of -1 when the data were collected with a favorable current. For net and door spread, the multiple regression relationship for the full-scale data is: Y = a +b ◊S +c ◊D
(5)
[m ]
where Y(m) is the net or door spread, D(–) is a dummy variable, S(m/s) is the towing speed and a(m), b(s), c(m) are coefficients computed by means of the multiple regression method. After computation of the coefficients, a transformation was possible: Y = a + b ◊ (S + c b ◊ D )
[m ]
(6)
The value c/b represents an estimate of the current speed, to be added or subtracted to the speed in order to obtain the water speed to the gear. Zeroing this term, the relationship between the parameter Y and the speed in absence of any current, was obtained: Y = a +b ◊S
(7)
[m ]
Values of the coefficients for both net and door spread were separately calculated for each gear rigging in the full-scale tests. Two other parameters were considered: the bosom height at the headline center (BH) and the net drag (ND). Because the intention was to compare the behavior of BH(m) and ND(N) in full-scale and model tests according to testing conditions, their relationships, with the other parameters (speed, current, wing-end spread etc.), were first analyzed by means of stepwise regressions.7 The data obtained in the full-scale tests were independently analyzed for each gear arrangement. The analysis showed that, for both parameters, a linear dependence upon speed was reasonably accurate, but a better approximation was achieved by correlating the net drag with the squared speed and the bosom height with the inverse of the speed. The second result of this analysis was that the other independent variable to be considered in the equation was the wingend spread (WS(N)). The use of further variables did not substantially improve the approximation of data, therefore the multiple regression relationships are: ND = a + b ◊ S 2 + c ◊ D + d ◊ WS
[N ]
(8)
[m ]
(9)
for a(N), b(N·s2/m2), c(N) and d(N/m). BH = a + b S + c ◊ D + d ◊ WS for a(m), b(m2/s), c(m) and d(–).
L Fiorentini et al.
A multiple regression analysis was then applied to the data in order to provide coefficient estimates and summary statistics for prediction regression models. For flume tank tests data, the term c · D was not considered. All the statistical procedures were performed using the SPSS for Windows (version 10.0) software package (SPSS, Chicago, IL, USA). RESULTS Tests of Italian trawl models The main test results (bosom height and net drag) and the corresponding regression curves of the two trawl models are summarized in Fig. 5. The bosom height of the traditional trawl, ranging from 1.6 m to 2.8 m, is generally higher than that typical of other Italian bottom trawls, which generally are in the range 0.8–1.5 m.8–10 This indicates that the developments made by the Italian fishermen in the last few years have been effective with regard to increasing the bosom height of traditional trawls. This higher bosom height increases the towing resistance, as shown by the net drag value measured (19 600 N at 3.5 knots and medium spread), which is practically the double of the old commercial Italian trawls. For this trawl, at 3.5 knots and medium spread, 9800 N was measured during the flume tank tests and 12 750–13 730 N in the sea tests.8–10 Initial tests with the experimental trawl showed that, for low wing-end spread and speeds above 3 knots, the footrope had poor bottom contact in the bosom area and along the first part of the wings. Increasing the groundrope weight by 285 N and lengthening the lower bridles by 40 cm solved that problem. Other tests were made to obtain more knowledge of the trawl’s performance when modifications were introduced. The number of floats was reduced to the same amount used in the traditional trawl by eliminating some floats along wings. This reduction caused a reduction in the upper wing height but had no influence on the bosom height. After the foregoing alterations the experimental trawl was tested under the same conditions as the traditional trawl and the results from the tests are shown in Fig. 5. The output results from the multiple regressions analysis are reported in Table 1, where it is possible to observe the estimated coefficients and the fit of the models (R2). The latter indicates that, for all the scaled trawls, more than 90% (i.e. 94–99%) of the variation in net drag and bosom height can be accounted for by the towing speed and net spread.
® Medium spread
Low spread x Low spread
+ Medium spread
• º
ITALIAN EXPERIMENTAL TRAWL MODEL
355
of the experimental trawl and at low wing spread from 2.1–2.8 m to 3.0–4.2 m. Also, the drag of the experimental trawl increased at each wing spread by approximately 1470–4900 N compared to the traditional trawl. The probable reason for the higher resistance is that the mouth area increased, so that more water had to be filtered through the meshes in the belly.
5000
10000
15000
20000
25000
2.3 2.5 2.7 2.9 3.1 3.3 3.5 3.7 3.9 4.1 4.3 Speed [knots]
Upper wing-end spread [m] 30000
High spread
2.3 2.5 2.7 2.9 3.1 3.3 3.5 3.7 3.9 4.1 4.3 Speed [knots] 11
12
13
14
15
16
17
18
19
20
ITALIAN TRADITIONAL TRAWL MODEL
2.3 2.5 2.7 2.9 3.1 3.3 3.5 3.7 3.9 4.1 4.3 Speed [knots] 1.5
2.0
2.5
3.0
3.5
4.0
4.5
Bosom height [m] 35000
High spread
FISHERIES SCIENCE
Model and full-scale trawl comparison
Total force [N]
Fig. 5 Comparison between the Italian trawl models.
The regression model provides useful information to predict the influence on drag from changes in the netting area of the experimental trawl compared to that of the traditional trawl. The geometry of the traditional trawl measured in the flume tank was used as the basis for this estimation. A reduced drag of the new trawl compared with the traditional was expected, when geometry (wing-end spread and bosom height) was kept the same. This can be due to the reduction in the netting area of the wing sections, which will work at higher angles of attack than the belly sections, where the netting area is increased. From Fig. 5 it can be seen that, at high wing spread, the bosom height increased from 1.6–2.3 m of the traditional trawl to 1.8–3.0 m
Full-scale engineering tests of Italian trawls Some problems were encountered during the first hauls, when the traditional trawl was towed at low speed. The net tended to dig and to get stuck in the muddy sea bed, altering the measurements. For this reason the traditional trawl was tested in the speed range of 3.4–4.7 knots. The experimental data from the tests and the corresponding predicted curves are shown in Fig. 6. The wing-end spread was strongly affected by the warp and sweep length (Fig. 6). During the hauls made where the warp length was 150 m and sweep was 203 m, the wing-end spread reached value less than 13 m, in contrast with approximately 18 m measured when the warp paid out was 350 m and sweep was 81 m. The bosom height generally is in the range 1.7–2.4 m. As usual in most of the nets, the increase of the wing-end spread produced a decrease in the bosom height. The experimental trawl presented severe problems, in particular several hauls had to be aborted because the net became stuck in the muddy sea bed. Some alterations to its initial design were introduced during the tests to avoid the problem. After the first tests, the slack in the lower panel was redistributed, but without any success. In order to avoid the digging problem, it was decided to reduce the size of the lower panel bosom. During the hauls carried out with this alteration, no problems with mud have been encountered. The first visual observations showed that the bottom contact of the footrope, especially at the bosom, was not perfect. A chain (total weight 785 N) was attached to the footrope at 50 cm intervals. No digging problems were encountered with these rigging conditions. It was therefore decided to add more weight (195 N of lead) in the bosom area to obtain a better bottom contact. During the subsequent observations the alteration seemed to be effective. The experimental trawl was then tested at sea in three different warp-length conditions. The results from these trials are shown in Fig. 7. At all the rigging configurations (150, 200 and 250 m of warp length), the measured gear parameters showed a greater data dispersion in respect to
-2562.75 515.63 5145.06 82.27 – – 359.66 29.90 (3, 15) 0.996
ASE
ASE
1.96 0.08 2.30 0.09 – – -0.08 0.00 (3, 15) 0.936
Estimate
ASE
393.19 964.54 6168.73 175.20 – – 188.25 53.18 (3, 8) 0.994
Estimate
ASE
3.47 0.19 3.86 0.28 – – -0.19 0.01 (3, 8) 0.991
Estimate
ASE
-9089.41 2095.20 5262.35 267.65 -1527.28 209.37 713.58 85.85 (4, 22) 0.946
Estimate
ND (N) ASE
2.10 0.14 2.19 0.31 0.03 0.01 -0.07 0.01 (4, 22) 0.879
Estimate
BH (m)
Traditional full-scale trawl
5000
10000
15000
20000
25000
30000
35000
3.0
3.4
3.8 4.2 Speed [knots]
4.6
5.0
150
350 250
8
10
12
14
16
18
20
22
3.0
3.4
ASE
-179.47 1153.24 4432.44 115.24 -136.75 106.72 284.37 70.75 (4, 15) 0.991
Estimate
ND (N)
ASE 4.04 0.30 0.17 0.39 0.06 0.03 -0.13 0.01 (4, 33) 0.752
Estimate
BH (m)
Experimental full-scale trawl
4.6
5.0
1.0
1.5
2.0
2.5
3.0
3.5
4.0
3.0
3.4
Fig. 6 Full-scale test results obtained on the Italian traditional trawl. x 150 m warp and 203+48 m sweep
º 250 m warp and 81+48 m sweep
• 350 m warp and 81+48 m sweep
3.8 4.2 Speed [knots]
4.6
5.0
350
250
150
40
45
50
55
60
65
70
3.0
3.4
3.8 4.2 Speed [knots]
4.6
5.0
150
250
350
FISHERIES SCIENCE
ITALIAN TRADITIONAL FULL-SCALE TRAWL
3.8 4.2 Speed [knots]
150
250
350
Bosom height [m]
Upper wing-end spread [m]
Net drag [N]
ASE, astymptotic standard error; ND, net drag; BH, bosom height. Note that ND and BH are not linearly related to towing speed parameter of the model, so non-linear regression is required. To relate ND and BH to the vessel speed, a dummy variable parameter (D) was added as an independent variable in full-scale regression equations.7 For this dummy variable, a value of +1 was used when the data were collected with a counter current and a value of -1 was used with a favorable current.
a b c d DF R2
Estimate
BH (m)
ND (N)
ND (N)
BH (m)
Experimental trawl model
Traditional trawl model
Table 1 Estimated coefficients and ASE of non-linear regression routine applied to the dependent measures ND and BH
Door spread [m]
356 L Fiorentini et al.
FISHERIES SCIENCE
5.0
150
200
40
45
50
55
60
65
3.0
3.4
3.8 4.2 Speed [knots]
4.6
250
70
5.0
x 150 m warp and 203+48 m sweep
º 200 m warp and 81+48 m sweep
5.0 8
10
12
14
16
18
20
22
3.0
3.4
3.8 4.2 Speed [knots]
4.6
150
200
250
• 250 m warp and 81+48 m sweep
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Bosom height [m] 5.0 Speed [knots]
5000
10000
15000
20000
25000
30000
35000
3.0
3.4
3.8
4.2
4.6
200 250
150
Upper wing-end spread [m] Net drag [N]
Although an inductive methodology, such as the regression analysis, can not allow fishing net drags and bosom height to be estimated in detail, for both the experimental and the traditional trawls, multiple regression models were used to compare the main gear parameters (bosom height and net drag) of the models and full-scale trawls. Table 2 shows the comparison between the sea trials and the flume tank tests results for the traditional trawl. Good agreement was found between the model and full-scale trawl tests. Practically no differences were found in drag (
