development of blast resistant steel doors

11th International Conference on Shock & Impact Loads on Structures 14-15 May 2015, Ottawa, Canada

DEVELOPMENT OF BLAST RESISTANT STEEL DOORS Eric Jacques* , Alan Lloyd†, Tim Berry‡, Murat Saatcioglu§, and Jack Shinder** Research Officer and Ph.D. Candidate National Research Council of Canada, 1200 Montreal Road, Ottawa, ON, CANADA e-mail:

Keywords: blast, explosion, shock tube, blast resistance, doors.

Abstract. A new line of blast-resistant steel doors was developed at the University of Ottawa for a local manufacturer through combined experimental and analytical research. The main objective was to develop blast-resistant steel doors meeting performance levels required for the minimum anti-terrorism marketplace. The University of Ottawa’s Shock Tube Testing Facility was used to subject full-scale blast doors to simulated shock waves. The full-scale blast tests were conducted to determine the response of the blast doors relative to ASTM F2927 door and glazing classifications. A number of parameters affecting door response were considered, including door aspect ratio, construction methodology, doorframe construction, as well as anchor size and quantity. The experimental portion of the study was complimented by an analytical investigation to develop door design tools. Finite element analysis was employed to generate door resistance curves. This was followed by single-degree-of-freedom dynamic analyses to predict the various levels of protection for the blast scenarios studied. This paper presents a general overview of the approach adopted for the experimental and analytical investigations. Details of a variable geometry test fixture, capable of accommodating a range of door sizes and anchor configurations, are presented. The experimental test procedure as well as test results are discussed. Analytical work to predict door response, and a summary of key design considerations, are discussed. 1

INTRODUCTION

Over the course of the past decade, much effort has been invested into the development of blast resistant design requirements for critical structural and non-structural building components, such as columns, walls, and windows. This has led to the development of numerous scientific and regulatory publications outlining performance objectives, design requirements, and analysis procedures suitable for blast resistance of these common elements. However, only a limited number of academic literature is available on the response and design of blast resistant doors1,2. The main reason for this lack of information is that it is the responsibility of the door manufacturer to ensure that their product meets the blast rating and protection level required by the engineer or owner. Although speciality door †

Assistant Professor, Department of Civil Engineering, University of New Brunswick, Fredericton, NB, CANADA Senior Design Engineer, AMBICO Ltd., Ottawa, ON, CANADA § Professor, Department of Civil Engineering, the University of Ottawa, Ottawa, ON, CANADA ** President, AMBICO Ltd., Ottawa, ON, CANADA ‡

1

Eric Jacques, Alan Lloyd, Tim Berry, Murat Saatcioglu and Jack Shinder

manufacturers have conducted their own internal research and development programs, this work is generally regarded as trade secret as it was conducted at their own expense. The objective of this paper is to present the preliminary results of a collaborative research and development project completed by AMBICO Ltd., a specialty door manufacturer, and the University of Ottawa. The main goal of the project was to develop a new line of blast resistant doors for the minimum antiterrorism marketplace. The effectiveness of various types of blast resistant door core constructions was evaluated. In addition, the testing provided data to validate a simple and effective analytical procedure based on a combined finite element and single degree of freedom analysis to predict door response to blast loads. 2

EXPERIMENTAL PROGRAM

This section details the design and construction of the AMBICO Ltd. proprietary blast resistant hollow steel door constructions considered in the research program. It also describes the University of Ottawa Shock Tube, used to generate simulated blast pressures, as well as the test fixture and instrumentation. 2.1 Door Specimens Ten single-entry blast resistant door specimens, manufactured by AMBICO Limited, were considered in this study. Each had a height of 2134 mm and width of 914 mm, giving an aspect ratio of 2.33. Overall door thickness was governed by the construction of the internal cores. Doors A1 through D2 were each 45 mm thick while E1 and E2 were 56 mm thick. The main construction material used in door fabrication was 1.99 mm thick 14 ga. galvanneal steel sheet metal with a 120 g/m 2 zinc-iron coating. The sheet metal was used to fabricate the front and back door skins, end channels, as well as the hat stiffeners. As necessary, ASTM A5003 Grade B structural steel tubing was used in doors requiring greater blast resistance. Type 1 rigid insulation meeting CAN/ULC-S7014 was used as a lightweight filler material to maintain the integrity of the front and back door skins. All welds were structural, with a combination of fillet and plug welds used throughout door fabrication.

(a) Stiffened Rigid Insulation

(b) Vertically Stiffened Core

(c) Horizontally Stiffened Core

(d) Structural Rigid Insulation Core

Figure 1: Idealized blast resistant door designs. Four main core constructions, each illustrated in Figure 1, were studied. The most basic core, used for A1, consisted of sheet metal end channels framing a rigid foam layer. The front and back sheet metal skins were fastened to the rigid foam with a construction grade adhesive and then welded to the end channels. An improvement to the basic rigid insulation core, illustrated in Figure 1 (a) and known as the Stiffened Rigid Insulation Core, consisted of additional HSS tubing along the bottom of the door to increase out-of-plane stiffness. Vertically and Horizontally Stiffened Cores, illustrated in Figure 1 (b) and (c), respectively, were each stiffened with sheet metal hat members. An HSS core door, shown in Figure 1 (d), was also manufactured using a steel tube framing system with foam filler material.

2


A test matrix of door specimens is provided is Table 1. Most of the doors were subjected to blast pressure in the seated direction, although B3 was subjected to unseated blast loading to study latch capacity. Schlage L9000 Mortise locks were installed into A1 and B3 at a standard height of 1024 mm above the finished floor. All other doors were tested without latch hardware to focus on the blast resistance of the core construction. Two types of frame construction were considered, each with a height of 2184 mm and width of 1016 mm. A light-duty frame design for low pressure blast resistance, LP, was used with doors A1C2 while a heavy-duty high pressure frame, HP, was used for doors D1-E2. Frame L2 was anchored to the test fixture using 10 ϕ½” Grade 5 bolts. Initially, frame HP was anchored using 15 ϕ¾” Grade 5 bolts, but the frames were field-modified for D2, E1, and E2 to accept ϕ¾” bolts. All frames were constructed with a 23 mm frame stop. Each door was attached to the frame using three 114 mm × 114 mm Heavy Weight Hinges. Door Mass Core Construction Series (kg) A1 Rigid Insulation 66.0 B1 Stiffened Rigid Insulation 69.6 B2 Stiffened Rigid Insulation 69.6 B3 Stiffened Rigid Insulation 81.4 C1 Vertically Stiffened 95.8 C2 Vertically Stiffened 96.4 D1 Horizontally Stiffened 81.8 D2 Horizontally Stiffened 102.9 E1 Structural Insulation 98.7 E2 Structural Insulation 119.1 † – Field-modified to accept ϕ¾” bolts.

Frame Type LP LP LP LP LP LP HP HP † HP HP †

Anchor Seated or Diam. Unseated Test ½” Seated ½” Seated ½” Seated ½” Unseated ½” Seated ½” Seated ½” Seated ¾” Seated ¾” Seated ¾” Seated

Vision Lite 127 mm × 508 mm 127 mm × 508 mm 127 mm × 508 mm 610 mm × 914 mm 610 mm × 914 mm

Table 1: Test Matrix of Door Specimens. Most of the test specimens were constructed without a vision lite (window). However, in some cases potential clients may want a blast-resistant door with suitably blast-resistant window for aesthetic or functional reasons. Accordingly, several doors were constructed with lites to study the effect of window size and placement on door performance. Small, rectangular 127 mm × 508 mm lites were installed in B1, B3 and C2 at eye-level on the latch side. Larger 610 mm × 914 mm lites were installed at eye-level for doors D2 and E2. The lites were framed into the core of the door using bent sheet metal reinforcing channels and fastened using plug welds. The 127 mm × 914 mm lites consisted of two panes of 3 mm annealed glass with a PVB interlayer. The 640mm × 914 mm lites consisted of two panes of 6 mm annealed glass with a PVB interlayer. 2.2 University of Ottawa Shock Tube & Test Fixture The University of Ottawa Shock Tube, illustrated in Figure 2 (a), was used to generate the high pressure blast loading necessary for the door tests. The pneumatically-driven shock tube, seen in Figure 2 (a), consists of three basic components – a variable length driver section, expansion section and end frame. The driver section has a circular diameter of 597 mm and varies in length from 305 mm to 5185 mm. Shock wave parameters are modified by varying driver pressure and driver length. Shock tube firing is accurately and safely controlled by a differential pressure doublediaphragm system located between the driver section and the expansion section. The expansion section is designed to ensure planarity of the shock wave as it propagates from the circular driver section to the 2033 mm x 2033 mm square testing frame opening over a length of 7 m. A shock wave is generated when the aluminium diaphragms are ruptured and the compressed air in the driver section is released into the atmospheric conditions within an expansion section. The shock tube has been shown to produce shock waves which are planar and generate repeated pressureimpulse combinations at the location of the test frame5. A reusable steel test fixture, shown in Figure 2 (b), was constructed. This test fixture was bolted to the shock tube end frame and provided a rigid substrate designed to support the single door tests. The door frames were installed into a support structure with opening dimensions of 2190 mm × 1022 mm. The test fixture was comprised of 150 mm × 150 mm × 6.35 mm hollow steel sections which were welded together. To facilitate anchoring, holes were provided in the fixture to match bolt holes located in the frames. Since the test fixture did not cover the entire shock tube opening, a steel cover was constructed to permit full reflection of the shock wave. A safety bar, placed across the door opening at 3


the latch height, was used to prevent the doors from swinging open excessively during blast testing. Figure 2 (c) shows a typical door prior to testing.

(a) Shock tube driver and expansion section (b) Door test fixture mounted to (c) Door with vision lite shock tube prior to testing Figure 2: Shock tube door test setup. 2.4 Test Procedure Tests were conducted following the requirements of ASTM F2927-126 as closely as possible. Chronologically, doors were manufactured and tested in several batches. This approach allowed the results of the current batch of tests to be used to improve the blast resistance of the cores, as well as find efficiencies in the manufacturing process. This measured research and development approach led to marked improvements in blast-resistance of the doors as the test program progressed. Doors were subjected to incrementally increasing pressure-impulse combinations until either the ultimate capacity of door, or of the shock tube, were achieved. Typically, the first specimen of each type of core construction was subjected to the greatest number of shock tube tests. This was done to ensure that the blast resistance of the doors was determined as accurately as possible, without exceeding the failure level. Subsequent doors of the same core construction were subjected to fewer pressure-impulse combinations, starting the shot generating failure of the previous sample and incrementally increasing as necessary. Door response and hazard level classifications were assigned for each test in accordance with the criterion listed in ASTM F2927. A high-speed data acquisition system recording at 100,000 samples per second was used to capture experimental data. Shock wave pressure time-histories were measured using two piezoelectric pressure sensors located along the bottom and side walls of the shock tube expansion section 50 mm away from the specimens. Door displacements were measured used linear variable displacement transducers (LVDTs) with a stroke of 312 mm, positioned to record the deflected shape of the vertical centerline doors on the unloaded skin. A sawn lumber witness panel was constructed in compliance with ASTM F2927-12 and GSA TS017 for door tests containing vision lites. A sheet of white paper placed over a 25 mm thick rigid-foam insulation, replaced after every test, was used to identify debris impacts. 3

EXPERIMENTAL RESULTS

Experimental results obtained from shock tube testing, including blast loading, hazard classification, and a description of the failure condition of the doors, are summarized in Table 2. The text matrix in Table 1 may be used a reference for core construction and test parameters of the various door series. Figure 3 shows photographs of selected doors after shock tube testing, as well as a selection of common failure modes. Note that the hazard classifications listed in Table 2 are assigned based on the door systems that were tested. Since most doors were tested without latches, the lack of rebound resistance limited door classicisation to relatively low levels of protection despite the fact that doors generally performed well. Prior to failure, doors only exhibited minor distress consisting of small levels of permanent deflection. Addition of suitable latches would be expected to greatly improve door the classification. The most common failure mechanism for doors without large lites was one where the door pushed past the frame stop. Projected shortening of the door width, caused by out-of-plane deflection, resulted in a loss of seating. Generally, this failure was not catastrophic and only resulted in doors becoming wedged in the frame. The following section highlights the general performance and failure modes of the various core constructions.

4


3.1 Rigid Insulation Core As illustrated in Figure 3 (a), the rigid insulation core door A1 failed at the location of least bending resistance along the bottom edge, causing the door to push past the stop. To mitigate this failure, the design of the stiffened rigid insulation core doors was conceived to increase strength by adding an internal HSS tube. Door series B1 and B2, constructed with the improved core construction, demonstrated a modest increase in blast resistance and overall reduction door damage at failure. However, stiffened rigid insulation core doors failed due to door pushing past the frame stop near the latch.

(a) A1 – Rigid Insulation Core

(b) C1 – Vertically Stiffened Core

Stop deformation

(c) Typical stop deformation

(d) Comparison of Lite Performance

Figure 3: Photographs of typical failure modes observed during shock tube tests. 3.2 Vertically Stiffened Core The vertically stiffened core construction used for C1 and C2 showed better blast resistance than the rigid insulation core. In addition to vertical hat stiffeners, both C1 and C2 were constructed using a bottom HSS tube. C1 failed when the door pushed past the stop at the latch location. As illustrated in Figure 3 (b), failure occurred as the vertical hat stiffeners were not sufficiently fastened to the bottom HSS tube. As a result of a lack of integrity between adjacent framing members, the intended two-way response of the door was not activated. Much greater blast resistance was expected by increasing the quantity of weld for door series C2. However, unbeknownst to the engineering staff, C2 was constructed using two smaller lengths of HSS tube tack welded together. As a result, C2 suffered a premature failure when the tack welds fractured and the door pushed past the stop at the bottom of the frame. An investigation found that the manufacturing and quality assurance staff were unfamiliar with the importance of continuity and structural integrity in

5


achieving blast resistance. A suitable training program was implemented to ensure staff were aware of the need for continuity in blast resistant doors.

Door Blast Loading Shot Series (kPa / kPa-ms) A1

ASTM F2927 Classification Door Glazing Response Response Category Category I III III III III IV III H1 IV H1

1 2 3 4 5 6† 1 2†

11.7 / 96.4 24.5 / 195.0 32.6 / 230.6 35.0 / 279.0 39.0 / 267.5 49.0 / 382.9 49.1 / 352.5 51.0 / 319.4

B2

1 2†

50.2 / 379.9 56.4 / 386.9

III IV

-

B3

1†

13.3 / 116.4

-

H1

C1

1 2†

47.2 / 379.5 56.6 / 446.2

III IV

-

C2

1 2†

53.9 / 429.3 77.1 / 637.0

III IV

H1 H1

D1

1 2 3 4 5 6† 1 2†

48.2 / 385.9 58.5 / 460.5 67.2 / 522.5 74.0 / 604.0 78.9 / 635.1 97.1 / 812.5 83.0 / 668.6 103.5 / 941.9

III III III III III III III III

N/A N/A

1 2

89.7 / 743.8 92.3 / 1582.0

III III

-

B1

D2

E1 E2

1 53.1 / 356.4 III 2 72.8 / 542.4 III 3 85.6 / 722.6 III 4† 111.7 / 949.7 III † - denotes a test generating the ultimate capacity of

Description of Failure Condition

Door pushed past the stop at the bottom on the latch side. Permanent deformation, fracture of welds and buckling of skins along unsupported bottom edge. The stop at the failure location was permanently deformed. Door pushed past the stop at the latch location. Significant perm anent deformation and fractured welds between the interior and exterior skins along the bottom HSS. No damage to frame. Window remained in the frame but cracked due to racking of the door. Door pushed past the stop at the latch location. Door permanently deformed, with significant buckling and warping of skins at mid-height. Fractured welds connecting the skins below the latch relief. The frame was undamaged except for minor stop deformation. Door response was elastic and remained undam aged. Minor outward deflection of 4 mm (3/8”) of the skins at the latch location. No damage to frame was observed. Latch hardware failed through rupturing of the bolting mechanism. Egress through the door not possible. Door pushed past the stop at the bottom on the latch side. Lack of continuity between vertical stiffeners and bottom HSS caused permanent deformations of the exterior skins. Fractured welds between the skin seams. W arping of skins around the latch area. The stop at the failure location was permanently deformed. Door pushed past the stop at the bottom on the latch side. Warping of unsupported edge due discontinuous HSS member. Fractured welds and splitting of skins on unsupported latch side. The stop at the failure location was significantly distorted. The frame was visibly deformed at anchor locations and around the lite. The lite was significantly cracked, but no debris. Door remained largely undamaged, and effectively elastic, after testing. Several of the ½” bolts at the bottom of the hinge and latch sides of the door wer e sheared off and others showed significant shear damage after shot 4,5, & 6. Frames HP field-modified to accept ¾” bolts for subsequent doors. Noted some warping of the fram e stop at the bottom of the door. Skin deformation around the internal window frame, with fracture of plug welds connecting frame to skin. Skins separat ed from core on latch side due to failure of welds. Door frame undamaged. Complete lite blowout with >4 kg projectiles travelling >25 km/h. Door largely undamaged. Welds holding skin to HSS core broken on unloaded face. No frame damage. No damage to hinges.

H1 Door largely undamaged with only minor deformation around the lite. No dam age to the door fram e. Lite H1 shattered but film did not separate from frame. Only H4 small amount of glass ejected into the GSA zone 3a. N/A the door systems.

Table 2: Summary of shock tube door test results and associated door response classifications. 3.3 Horizontally Stiffened Core The horizontally stiffened core, D1, constructed incorporating the findings of all previous door tests, performed extremely well during shock tube tests. D1 survived a greater blast pressureimpulse combination than all other doors tested to that point. This door was tested with heavy-duty type HP frame with ½” bolts. The door itself was largely undamaged, and remained effectively

6


elastic throughout testing. During Shot 4 of 7, however, several of the ½” bolts located near the bottom hinge location sheared off, while several adjacent bolts experienced significant shear deformation. Damaged bolts were replaced for the remaining tests but the same anchor locations continued to suffer shear failure. For all subsequent tests, heavy-duty frame HP was field modified to accept ¾” bolts thus eliminating anchor failure. 3.4 Structural Insulation Core The structural insulation core door, E1, with heavy-duty high pressure type HP frame and ¾” bolts, exhibited the highest blast resistance of all specimens. The door, frame, and anchors showed no signs of damage affecting the safety of the door after being subjected to testing at maximum shock tube pressure. The only damage observed was that the skin on the unloaded face had bowed outward as several of the plug welds attaching the sheet metal to the structural core had fractured. 3.5 Latch Performance The latches installed into doors A1 and B3 performed poorly when subjected to blast loading. For seated response, the latch installed on A1 failed on the second shot resulting in the door opening in rebound and rendering the door inoperable. Failure of the latch hardware is illustrated in Figure 3 (a). For unseated response, the latch hardware failed through rupturing of the internal bolting mechanism. The door swung open, struck the safety stop and rebounded back into the frame. The door closed itself and the broken latch jammed the door into the frame preventing egress through the opening. 3.6 Vision Lite Performance Provided the size of the window opening was sufficiently small, it was found that the presence of a vision lite had minimal impact on the blast resistance of a properly designed hollow metal door. A comparison of vision lite performance is illustrated in Figure 4 (d). Although door C2 suffered a premature failure not related to glazing performance, the 127 mm × 508 mm vision lite cracked, but did not fail or release potentially hazardous glass debris into the test cubicle. The PVB interlayer was sufficient to maintain integrity of the glazing unit for the blast pressures considered during testing. The large 610 mm × 914 mm vision lite installed in D2 and E2, however, each suffered a hazardous catastrophic failure. Portions of glazing, weighing over 4 kilograms, were ejected from the frame at speeds in excess of 25 km/h. Lite blowout resulted in severe damage to the witness panel, and garnering a High Hazard glazing rating. 4

ANALYSIS A combined FEM and SDOF analysis procedure was developed to predict the seated response of the blast resistant doors subjected to shock tube tests. This procedure was tailored for use by practicing engineers as a means of quickly and easily evaluating the blast resistance of different core constructions. Accordingly, it incorporates idealized boundary conditions, geometry, and material properties commonly used in an engineering design office. The goal was to provide practicing engineers with the flexibility conduct in-house evaluation of the blast resistance of a particular core design without having to conduct live explosive or simulated blast tests. The following describes the analysis procedure and associated validation. 4.1 Material Properties An elastic-perfect plastic idealization was used to model the isotropic stress-strain response of steel used for the HSS tubing and sheet metal skins. The modulus of elasticity was taken as 200 GPa, with a Poisson's ratio of 0.29. The static yield strength was assumed to be 300 MPa based on typical properties for ASTM 500 Grade B steel. A dynamic increase factor (DIF) of 1.17 was applied to the yield strength in accordance with response of structures in the far design range8. A simplified elastic-perfect plastic idealization was also used for the stress-strain characteristics of the rigid insulation filler material. Although more realistic material relationships developed specifically for foam are available, a simplified material relationship was justified on the basis that the foam does not contribute significantly to door strength. Based on the minimum requirements of CAN/ULC-S701 for Type I rigid insulation, the modulus of elasticity was taken as to be 1.5 GPa, Poisson's ratio was 0.4, and flexural strength and compressive strength were assumed to be 0.2 MPa. 7


4.2 Analysis Methodology The doors were analyzed using finite element software ABAQUS version 6.9-29. Door construction drawings, such as the one illustrated in Figure 4 (a), were used to construct the FEM door models. A typical FEM representation of a vertically stiffened core door is shown in Figure 4 (b). 3D deformable solid elements were used for the rigid insulation core. And the skin feature was used to model the external sheet metal skins. Beam elements with appropriate cross-sectional dimensions were used for the HSS tubes and hat stiffeners. Connections between door framing elements and the sheet metal skins were assumed to be completely rigid to eliminate the complexity of modelling individual welds.

Figure 4: Finite element modelling of vertically stiffened core blast resistant door. Boundary conditions varied depending on whether doors were subjected to seated or unseated loading. For both seated and unseated response, the hinge edge was assumed to be completely fixed against translation but free to rotate, appropriately reflecting the translational fixity provided by the hinges. For the case of the top and latch edges of the doors during seated response, the frame restrained out-of-plane translation, but provided no rotational or in-plane translational restraint. For unseated response, the top edge was unrestrained, while the latch edge was restrained against translation at the latch location only. The bottom of the doors was unrestrained. Typical boundary conditions for seated door response are illustrated in Figure 4 (c). The results of the finite element analysis were used to generate resistance functions for use in SDOF analysis. The total resistance, displacement at the center of the door, and load-mass transformation factors were extracted from the FEM analysis for progressively increasing pressure loading. The nodal displacements of the unloaded door skin, illustrated in Figure 4 (d), were used to compute a load-mass transformation factor, , at each load stage. Although assuming the active degree of freedom at the center of the door did not always correspond to the location of greatest displacement, it facilitated direct comparison against LVDT readings obtained at the same location. The finite element analysis was continued with progressively increasing pressure loads until maximum door displacement exceeded 150 mm, greater than any displacement observed during testing. 4.3 Dynamic Analysis & PI Diagram Development Displacement time-histories and PI diagrams were developed based on SDOF analysis. The equation of motion is given as: ∙

∙ ( ) + ( { }) =

( )∙

(1)

where ( ), and ̈ ( ) are the deflection and acceleration of the active degree of freedom; { ( )} is a piecewise-linear function describing the load-mass transformation factors as a function of deflection; { ( )} is a piecewise-linear function describing the resistance of the member as a function of deflection; is the total mass of the system; is the area impacted by the applied

8


pressure, and; ( ) is the time-varying applied pressure. Functions for and were obtained from FEM analysis, while ( ) was obtained from pressure sensor readings. Software RCBlast10 was used to generate a numerical solution to the SDOF equation of motion listed in Eq. (1). The numerical solution was obtained using the average acceleration method incorporating a predictor-corrector scheme. The predictor-corrector approach was adopted to minimize error arising from the use of resistance and load-mass factors which are a function of the deflection of the current time-step. The doors were weighed prior to testing to obtain the mass. Loaded area was taken as the gross area of the doors, 1.95 m2, and damping was ignored. The SDOF simulation was performed for one-half displacement cycle, up to the onset of rebound, as most doors were tested without latches to provide rebound resistance. 4.4 Results

75

75 Experiment

60

60 45

Pressure

30

30

15

15

0

0

-15

Displacement (mm)

45

Pressure (kPa)

Displacement (mm)

Prediction

-15

-30

-30 0

20

40 60 Time (ms)

80

55 50 45 40 35 30 25 20 15 10 5 0 -5

100

110 100 90 80 70 60 50 40 30 20 10 0 -10

Experiment Prediction Pressure

0

20

40 60 Time (ms)

80

Pressure (kPa)

A comparison of experimental and predicted displacement time-history curves for A1 – Shot 5 and B1 – Shot 6 are presented in Figure 5 (a) and (b), respectively. The results indicate that, for the first half displacement cycle, the experimental mid-height displacement time-history of the doors loaded in the seated direction can be predicted with good accuracy. It is worth noting that the predicted post-peak displacement shows stiffer response than what was observed during blast tests. Better predictions of unloading behaviour would be expected if member hysteresis had been considered.

100

(a) A1 – Shot 5 (b) D1 – Shot 6 Figure 5: Comparison of experimental and predicted displacement time-history curves. The combined FEM and SDOF analysis procedure was used to predict the displacement response of all doors subjected to seated blast loading and constructed without 610 mm × 914 mm vision lites. Based on this analysis, the ratio of predicted to experimental peak displacement was found to be 0.91, with a coefficient of variance (COV) of 24% based on 18 samples. This indicates that the response of blast resistant doors with complex core construction can be approximated using a simple, elastic-plastic FEM/SDOF analysis methodology. Although not presented in this paper, the methodology can also output estimates of reaction forces thus permitting the design of anchor, latch, and hinge hardware. The finite element analysis can also be used to predict other failure modes, such as shortening of the door and flexural failure of door materials. Combined, this gives practicing engineers a new and valuable methodology that can be used during the design, development and evaluation of blast resistant door products subjected to a wide range of pressureimpulse combinations not evaluated experimentally through shock tube or arena blast tests. 5

SUMMARY OF KEY DESIGN CONSIDERATIONS

A number of key design considerations influencing the response of blast resistant hollow steel doors for the minimum anti-terrorism marketplace have been identified as a result of the research & development program. Based on the findings, it is critical that:  Doors have sufficient horizontal stiffness to prevent projected shortening and loss of seating on the frame stop;  Frame stops are designed to resist excess deformations which could result in loss of seating;

9


     

6

Prevent loss of seating by providing excess stop depth greater than the expected level of projected shortening of the door; Sufficient structural integrity of the core be provided to allow for redundant load paths; Splicing of internal hat stiffeners and HSS tubes be avoided; Careful consideration of the magnitude and distribution of forces resisted by anchors, hinges, and latches, whilst ensuring door hardware is designed to meet the target level of protection; Ensure vision lites are positioned in regions of low stress, do not significantly reduce door strength, are properly anchored, and are laminated to prevent fragmentation; Manufacturing staff be trained on the importance of the above list of key design considerations necessary to ensure blast resistance of hollow metal doors.

CONCLUSIONS

An experimental and analytical research and development program was conducted to develop a new line of blast resistant hollow metal doors. Improvements to door construction practices were identified, and implemented in subsequent doors tests. The results of the improvements were compared against the original samples. The response of the doors to simulated blast pressures was evaluated against industry-standard classification protocols. The work highlighted the importance of continuity and integrity of internal stiffening elements in achieving greater blast resistance. Furthermore, analysis demonstrated that the displacement response of the hollow metal doors can be predicted with a reasonable level of accuracy using simple elastic-plastic FEA to generate resistance curves for use in SDOF simulations. 7

ACKOWLEDGEMENTS

This collaborative research project was conducted between the University of Ottawa and AMBICO Ltd. as part of an SME4SME applied research and commercialization initiative. REFERENCES [1]

M.J. Lowak, J.S. Idriss and J.W. Wesevich, Testing and Analytical Evaluation of Doors, Structures Congress 2011, American Society of Civil Engineers (2011). [2] M. Hsieh, J. Hung and D. Chen, Investigation on the blast resistance of a stiffened door structure, Journal of Marine Science and Technology, 16(2), pp. 149-157, (2008). [3] ASTM A500, Standard specification for cold-formed welded and seamless carbon steel structural tubing in rounds and shapes, ASTM International, West Conshohocken, Pennsylvania, (2013). [4] CAN/ULC-S701-11, Standard for Thermal Insulation, Polystyrene, Boards and Pipe Covering, Underwriters Laboratories of Canada, Ottawa, (2005). [5] A. Lloyd, E. Jacques, M. Saatcioglu, D. Palermo, I. Nistor and T. Tikka, Capabilities of a Shock Tube to Simulate Blast Loading on Structures, Behavior of Concrete Structures Subjected to Blast and Impact Loadings, SP-281, Thiagarajan, T., Williamson, E., and Conley, C. (eds.), American Concrete Institute, Farmington Hills, Mich., 2011, pp. 42-61, (2011). [6] ASTM F2927, Standard test method for door systems subject to airblast loadings, ASTM Internation, West Conshohocken, Pennsylvania, USA (2012). [7] United States General Services Administration, Standard Test Method for Glazing and Glazing Systems Subject to Dynamic Overpressure Loadings, GSA TS01-2003,(2003). [8] United States Department of Defense, Structures to resist the effects of accidental explosions, Unified Facilities Code (UFC) 03-340-02, (2008). [9] Dassault Systèmes Simulia Corp., Abaqus Analysis User's Manual, Version 6.9-2, Providence, RI, USA, (2009). [10] E. Jacques, RCBlast (Version 0.5.1), Available for download at http://www.rcblast.ca/ (2015).

10