Sunteți pe pagina 1din 15

Mitigating Risk of Confined Explosion via

Lightweight Sacrificial Cladding


Yasser A. Khalifa 1; Manuel Campidelli 2; Michael J. Tait, M.ASCE 3;
and Wael W. El-Dakhakhni, F.ASCE 4
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

Abstract: A confined explosion inside a critical facility may lead to the total paralysis of pertinent operations. Practical solutions to mitigate
such risk include lightweight sacrificial cladding, which can be deployed as a physical barrier to protect high importance (i.e., critical)
facilities from potential explosions. More specifically, sandwich panels have the potential for high energy dissipation at a relatively low
cost, which makes them suitable for multiple cladding applications against a range of blast-threat scenarios. To investigate the effectiveness
of sandwich panels against such threats, the results of an experimental study carried out by the authors are presented herein. The study
includes 18 cold-formed steel sandwich panels featuring two core configurations selected on the basis of the resistance and modes of failure
of nominally identical panels previously tested under quasistatic loading. The dynamic response of each panel configuration to the detonation
of six different charges of composition C-4 at a constant standoff distance is analyzed in this paper. The test results are presented in terms of
response histories, peak displacements, and observed modes of failure, while the damage levels exhibited by each panel are related to
the response limits specified in design standards currently adopted in North America for the blast protection of buildings. Considerations
of energy absorption and scalability suggest that panels with a unidirectional core configuration may be preferable for blast risk mitigation.
DOI: 10.1061/(ASCE)CF.1943-5509.0001331. © 2019 American Society of Civil Engineers.
Author keywords: Blast risk; Cold-formed steel; Confined explosion; Lightweight structure; Sacrificial cladding; Sandwich panel.

Introduction and the people inside it, thereby increasing the overall level of pro-
tection specified in FEMA 452 (FEMA 2005).
Owing to the confined spaces and the inability to vent pressure, an From a structural perspective, physical barriers made of sacri-
explosion inside a critical facility is likely to cause more damage ficial cladding can be classified into three types: (1) in continuous
than an outdoor explosion of equivalent size and may lead to opera- contact with the main structure, whereby deformations will be in-
tional paralysis. A recent investigation by Geretto et al. (2015) on duced by compression/crushing only; (2) simply supported by the
the damage suffered by steel plates subjected to the detonation of main structure (with free rotations at the supports); and (3) continu-
plastic explosive has shown the effects of blast load confinement. ous across the supports provided by the main structure. The selec-
Tests performed with the same charge mass under fully vented and tion of the type of support should be guided by safety demands
fully confined conditions produced, respectively, impulse values and imposed by the main structure, especially when the latter is con-
permanent deflections 2.7 and 4.1 times as large as those caused by nected to safety-related systems and components. From a materials
unconfined explosions. Hence, it is reasonable to postulate that the perspective, the choice of cold-formed steel (CFS) appears to be
pressure generated by small charges of explosive within a facility, optimal by virtue of its desirable properties—including versatility,
even if tolerated by the main structural system, would damage valu- durability, noncombustibility, and light weight. The use of such a
able nonstructural components. To mitigate the nonstructural dam- construction system has grown significantly over the past decades
age, designers have the option of planning sacrificial spaces that (Yu 2016): CFS is now commonplace in interior, nonload bearing
may provide enhanced protection to both the built environment wall systems as well as exterior, load bearing systems adopted in
midrise commercial buildings, mixed-use buildings, and residential
1
Formerly, Ph.D. Student, Dept. of Civil Engineering, McMaster Univ., constructions; wide-span trusses, hollow-core planks, and floor and
Hamilton, ON, Canada L8S 4L7. Email: khalifya@mcmaster.ca roof systems—made of CFS joists, rafters, bracing elements, studs,
2
Formerly, Research Coordinator, Dept. of Civil Engineering, Institute and a variety of fasteners—are now widespread. Its versatile nature
for Multi-Hazard Systemic Risk Studies, McMaster Univ., Hamilton,
makes CFS the material of choice in many hybrid construction sys-
ON, Canada L8S 4L7 (corresponding author). ORCID: https://orcid.org
/0000-0003-2317-4495. Email: campide@mcmaster.ca; mcampidelli@ tems, where it is combined with and connected to precast concrete,
gmail.com wood, and hot-rolled steel. In the maritime sector, the need for
3
Joe NG/JNE Consulting Chair in Design, Construction, and Manage- weight savings and greater resistance to fire, blast, and penetration
ment in Infrastructure Renewal and Director, Center for Effective Design of has driven the fast development and deployment of laser-welded
Structures, Dept. of Civil Engineering, McMaster Univ., Hamilton, ON, sandwich panels fabricated with CFS (Kujala and Klanac 2005).
Canada L8S 4L7. Email: taitm@mcmaster.ca Laser-welding automation was pioneered by the US Navy in the
4
Director, Institute for Multi-Hazard Systemic Risk Studies, McMaster 1980s to produce the first strength data (Marsico et al. 1993;
Univ., Hamilton, ON, Canada L8S 4L7. Email: eldak@mcmaster.ca
Wiernicki et al. 1991) and prototypes. In the 1990s, several Euro-
Note. This manuscript was submitted on December 20, 2018; approved
on March 12, 2019; published online on October 22, 2019. Discussion pean countries—including the United Kingdom, Germany, and
period open until March 22, 2020; separate discussions must be submitted Finland—took the lead in research and development of laser technol-
for individual papers. This paper is part of the Journal of Performance of ogy (Norris et al. 1989; Tan et al. 1989), the investigation of mechani-
Constructed Facilities, © ASCE, ISSN 0887-3828. cal properties of adhesively bonded panels (Knox et al. 1998),

© ASCE 04019080-1 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


and applications in cruise vessels (Roland et al. 2003). Nowadays, specimens was compared with that of both mild steel plates and
marine applications of CFS technology extend to load bearing air-core panels. For the case of uniform blast loading delivering
structures in naval vessels and leisure yachts and nonload bearing an impulse greater than 20 N · s, results showed the superior resis-
elements on merchant and large cruise ships—and include wing tance of the honeycomb geometry over solid plates. However, for
bulkheads and staircase landings, balcony partitions, stairs and plat- smaller impulse values, honeycomb panels showed higher back
forms in public areas, cabin decks, and shell structures in ice- face deformations than those observed in air-core specimens.
breakers. The production of sandwich panels thrives, and the The use of aluminum foams such as Alporas (Shinko Wire
types and configurations have become so numerous as to require an Company, Amagasaki, Hyōgo Prefecture, Japan) and Cymat
ad hoc classification system (Center of Marine Technologies E.V. (Alusion, Mississauga, Ontario, Canada) in the core was studied
2011). by Theobald et al. (2010). The Alporas cores demonstrated higher
For the purpose of the current study, the following section blast resistance when coupled with thick face sheets, while Cymat
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

provides a brief recapitulation of previous work on sandwich cores did not show any significant improvement over monolithic
panels, with emphasis on testing under extreme dynamic loading, plates. Liu et al. (2013) later conducted a study on the relationship
to gauge the state of technological development of CFS panels and between core foam density and blast attenuation. It was found that a
their potential in the service of sacrificial barriers against blast. lower foam density resulted in higher energy dissipation.
Other studies have focused on panels with asymmetric face
sheets, e.g., Wang et al. (2011), who investigated the effects of five
Previous Work on Metallic Sandwich Panels different thickness ratios (ratios between the front and back sheet
Sandwich panel systems have been investigated for blast protection thicknesses, the front side being the one directly exposed to the blast)
by a number of researchers primarily interested in the panels’ abil- on peak deflection and energy absorption. No clear relationship was
ity to absorb energy and their high stiffness-to-weight ratio. Energy found between thickness ratio and peak deflection; conversely, the
is primarily absorbed by a panel through the deformation of its ability to absorb energy showed a strong inverse correlation with
core; thus the influence of core configuration on a panel’s response said ratio, which seems to favor panels with a thicker back sheet.
has been the focus of several studies. Several researchers have com- Børvik et al. (2008) proposed a cost-effective and lightweight
pared the response of sandwich panels to equivalent monolithic protection system for a 6.1 m (20 ft) standard ISO container, with
solid plates, i.e., plates built of the same material and having the the purpose of using it as a shelter in international operations. The
same mass, to highlight the superiority of sandwich systems. For work was done in six stages and utilized extruded AA6005-T6 alu-
metallic panels, Xue and Hutchinson (2003) numerically investi- minum panels filled with granular material to increase their blast
gated tetragonal truss core geometries and found that they absorb resistance. Although numerical simulations showed significant im-
twice the amount of energy as equivalent monolithic plates. Fleck provement in blast resistance, full-scale test results have yet to be
and Deshpande (2004) analyzed the dynamic behavior of clamped presented to validate this new protection system.
sandwich panels configured in five core topologies—pyramidal, Theobald and Nurick (2007, 2010) conducted a series of tests and
diamond-cell, corrugated, metal foam, and square honeycomb; the numerical simulations to investigate the dynamic response of sacri-
diamond-cell core was found to possess the most desirable blast ficial cladding subjected to uniform blast loads. The panels consisted
resistance against both water and air. Rimoli et al. (2011) investi- of mild steel face sheets and tube cores made of either annealed mild
gated the dynamic response of edge-clamped 6061-T6 aluminum steel or 6063-T6 aluminum alloy. The tests were performed with a
panels with a corrugated core under the impact of explosively driven target impulse of 55 N · s and the panels with the lowest stiffness
wet sand. The panels exhibited a permanent deformation 15%–20% showed the highest deformation and energy absorption.
lower than that of equivalent monolithic plates. McShane et al.
(2010) studied the dynamic response of square-honeycomb and cor-
Explosion Confinement and Energy Output
rugated core configurations in stainless steel sandwich systems sub-
jected to water blasting. Core topology was found to have a strong The current work is focused on small-size blast events originating
influence on the dynamic compressive strength and the degree of from high explosives detonated inside a confined space. Therefore,
core compression, but had a minor effect on the total momentum it is fitting to recall a brief description of the blast phenomena as-
imparted to the sandwich. sociated with such scenarios as well as the engineering principles
Dharmasena et al. (2010) tested five different core topologies commonly adopted to design against it.
of rigidly supported, stainless steel panels that were subjected to The blast load generated by a detonation of high explosive
underwater impulsive loading. The selected topologies were square- within a limited enclosure is determined by a set of variables that
honeycomb, triangular honeycomb, multilayer pyramidal truss, may be grouped into two broad categories: confinement effects and
triangular corrugation, and diamond corrugation. All tested core the energy output of the specific type of explosive in use. Regarding
configurations had a relative density of approximately 5%. The the former, it is well known that an explosion in a confined space—
honeycomb configurations exhibited the highest peak strength and a building, a bus, or a tunnel—can produce complex and unique gas
rapid postpeak softening, while the truss and corrugated cores ex- expansion and blast wave propagation phenomena. The first pos-
hibited the lowest strength and extended postyield plateau. itive pressure phase at the shock wavefront typically presents strong
Cui et al. (2012a, b) tested the dynamic response of metallic similarities with that resulting from outdoor (unconfined) detona-
panels with a tetrahedral lattice core. Results from these studies tions, including a near-instantaneous pressure rise to the peak value
demonstrate the superior performance of tetrahedral lattice geom- followed by exponential decay (Baker et al. 1983; Krauthammer
etry compared to honeycomb. These findings were consistent with 2008); often, however, secondary peaks—unique to indoor events—
those presented by Dharmasena et al. (2011), who investigated the follow, generated by multiple reflections from the surrounding
response of pyramidal lattice cores subjected to a high impulse in surfaces. Another distinctive feature of indoor explosions is a rel-
air through a series of tests and finite-element simulations. atively long-lasting quasistatic pressure, which decays to preceding
Nurick et al. (2009) investigated the behavior of sandwich ambient values as the hot gaseous by-products cool down and are
panels with an aluminum alloy honeycomb core under the effect vented outside the confined enclosure. Depending on the size of
of localized and uniform blast loading. The response of these the explosion relative to the volume of the confining structure,

© ASCE 04019080-2 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


BLAST CHAMBER Tether Potentiometers Outer
Load cells shell
Reflected
pressure
Test frame Pin support
gauges
P1
Specimen
Incident
pressure gauge C-4 charge
P2 Reflected
P3 P4
P5 pressure
gauges
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

Sandwich panel Test frame

(a) (b)

BLAST CHAMBER

Blast
Back face
Test bunker

L = 914.4 mm

L/4
Back face
814 mm
C B A
Incident l/4 l/4

L/4
pressure gauge l/2
Specimen
1.07 m

l=305 mm

(c) (d)

Fig. 1. Test setup and instrumentation: (a) front view; (b) 3D rendering of test frame; (c) 3D rendering of the blast chamber; and (d) panel dimensions
and position of the potentiometers.

the damage experienced in a confined blast can be several-fold configurations were selected on the basis of the results from pre-
that caused by an outdoor event (Geretto et al. 2015; Smith and vious studies conducted by the authors on the ductility and energy
Hetherington 1994); therefore, as a general rule, the consequences absorption of seven different core configurations (Khalifa et al.
of confined explosions inside a critical facility need to be taken into 2017, 2018). Each panel was built using off-the-shelf corrugated
account through detailed risk assessment. sheets commonly available in North America, such as the B and N
The second set of issues relates to the quantification of energy deck profiles. A blast chamber with inner volume equal to 36.4 m3
output generated by a specific type of explosive in a standardized [Figs. 1(a and c)] and Composition C-4 were used to expose the
fashion, to provide the designer with the wavefront metrics needed specimens to confined blast loading; the resulting damage is as-
for engineering computations. The simplest approach that is com- sessed herein to evaluate the potential of this technology for the con-
monly adopted in design practice is that of converting the mass of struction of sacrificial barriers. The consistency of the test data is
explosive in use to an equivalent mass of trinitrotoluene (TNT), in analyzed via semiempirical modeling of the panels’ dynamic behav-
order to match output parameters such as peak pressure and specific ior, whereby dynamic test results are corroborated with numerical
impulse. In the tests reported herein, Composition C-4 was used as simulations based on coarse input parameters such as the panels’
live explosive. C-4 has a density of 1.59 g=cm3 ; detonation veloc- mass, resistance under quasistatic loading, and pressure records.
ity of 8.04 km=s; and TNT equivalency factors of 1.37 and 1.19
for peak pressure and specific impulse, respectively (Krauthammer
2008). Experimental Investigation

Research Scope, Objectives, and Methodology Test Setup and Instrumentation


To better understand the blast-protective properties of CFS The test bunker used for static testing to determine the resistance
cladding, the present investigation focuses on the effects of rela- functions of the test panels (Khalifa et al. 2018) was modified to fit
tively small charges of plastic explosive detonated at small stand- inside the blast chamber. A blast-resistant door with a central open-
off distances from sandwich panels with two different core ing that fits a single panel was installed on the front side of the
configurations—selected on the basis of their potential for high bunker, to allow easy setup of the test panels and access to their
energy dissipation, light weight, low cost, ease of handling and instrumentation. The size of the blast door surrounding each panel
installation, and flexibility to suit multiple applications against a was designed to be sufficient to prevent the reduction in the specific
wide range of threat levels. The specimens evaluated in this study impulse acting on the panel’s surface as a result of rarefaction
are one-way, simply supported, CFS sandwich panels assembled waves. Testing in the blast chamber also required that adequate
either in unidirectional (N-S-core) or in bidirectional (B-T-core) sealing be achieved in the test bunker in order to protect the instru-
configurations and subjected to confined live explosions. These mentation and prevent the blast waves from wrapping around the

© ASCE 04019080-3 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


panels. This objective was accomplished with a tight-fitting design in accordance with the American and Canadian standards for blast
of the blast door—leaving a small gap of a few millimeters between protection (ASCE 2011; CSA 2012). Both panel configurations,
the specimens and the frame of the aperture in which they were including cross-section dimensions and schematics, are rendered
lodged. However, it is reasonable to assume some degree of pres- in Fig. S1.
sure leakage within the bunker during each test, not only as a result All panels were fabricated with cold-formed galvanized steel
of the small openings/fissures in the setup, but because these open- and had width, length, and depth equal to 304.8 mm (12 in.),
ings are expected to increase in size as the specimens deform. It is 914.4 mm (36 in.), and 76.2 mm (3 in.), respectively. Gauge-20
also reasonable to assume a modest effect of the inner pressure [thickness ¼ 0.91 mm (0.036 in.)] sheets were used to fabricate B
buildup on the response of the panels, especially during the quasi- and N corrugated deck profiles. The B profile had a depth of
static pressure phase. Unfortunately, no measurements of internal 38.1 mm (1.5 in.) and was used in the double-layer core of bidi-
pressure were taken; thus its effects remain unquantified. Further- rectional panels, while the N profile had a depth of 76.2 mm (3 in.)
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

more, to prevent damage to the bunker, an outer shell made of and was used in single-layer (unidirectional) panels. Gauge-14
curved steel sheets—shown in Fig. 1(b)—was added to its sides sheets [thickness ¼ 2.03 mm (0.08 in.)] were used for the panels’
and back and the gap between the outer and inner shells was filled front and back sides. The fabrication process made use of metal
with polyurethane foam. inert gas welding—plug and fillet carried out as needed, depending
The reflected pressure was measured using five piezoelectric on the core configuration. The welds were designed to exceed the
pressure gauges mounted on circular galvanized steel plates at- strength of the base metal, as their strength was not an objective of
tached on the door of the test bunker, as shown in Fig. 1(a). Gauges the present investigation; they were equally spaced in the longitu-
P1 to P5 , used to measure the reflected pressure, were Dytran dinal direction at 165 mm (6.5 in.) on the center, while the mini-
model #2300V6 with a sensitivity of 0.5%  20% mV=psi, while mum distance between a panel’s edge and the center of a weld was
the incident pressure (FF) was measured using a PCB gauge— 44.5 mm (1.75 in.); weld spacing in the lateral direction varied,
model #113A21, with a sensitivity of 25%  15% mV=psi. The depending on the dimensions and profile used for the core sheet.
panels were simply supported (pin support at the top and roller Plug welds featured a diameter of 12.5 mm (0.5 in.), while fillet
support at the bottom) and four potentiometers were mounted on welds had a diameter equal to the thickness of the core sheet and
the back side to measure deflections. Three potentiometers (A, B, a length of 51 mm (2.0 in.). Further details regarding welding and
and C) were evenly spaced along the midspan of the specimens, fabrication can be found in the work of Khalifa et al. (2018).
while the fourth potentiometer (D) was connected at the top
quarter-span, midwidth location, as shown in Figs. 1(b and d). The Test Procedure
data acquisition system used in this test program had a sampling
rate of 1 MHz. The blast chamber provides a controlled experimental environment,
where several variables can be manipulated with ease in comparison
to the more difficult conditions typically encountered in the open
Test Matrix
field. Nevertheless, as discussed in later sections, the variability ob-
The tests were conducted on 18 sandwich panels consisting of served in the output of nominally identical tests (reported in Table 3)
uni- and bidirectional core configurations (Khalifa et al. 2018). demonstrates that significant sources of uncertainty related to either
As per Table 1, each core type was tested under three different known test variables or unknown variables (e.g., fabrication details)
scaled distances Z, i.e., three different ratios between the standoff remain unquantified. The sandwich panels were simply supported
distance (SD) and cubic root of the charge mass (W), the latter ex- and held in position by four steel clamps fixed to the frame by
pressed in kg of TNT; all Z-values were selected on the basis of a high strength structural bolts. The bunker door was held close by
preliminary dynamic analysis to achieve different damage levels, high strength bolts as well. Each explosive charge was tethered and

Table 1. Matrix of test specimens


C4 mass Equivalent Scaled distance Scaled distance Specimen Core
Shot (kg) TNT mass (kg) value (m=kg1=3 ) designation designation configuration
1 0.11 0.15 2.82 Z1 N-S-Z1-1 Unidirectional
2 N-S-Z1-2
3 N-S-Z1-3
4 B-T-Z1-1 Bidirectional
5 B-T-Z1-2
6 B-T-Z1-3
7 0.77 1.05 1.48 Z2 B-T-Z2-1 Bidirectional
8 B-T-Z2-2
9 B-T-Z2-3
10 0.94 1.29 1.38 Z3 N-S-Z3-1 Unidirectional
11 N-S-Z3-2
12 N-S-Z3-3
13 B-T-Z3-1 Bidirectional
14 B-T-Z3-2
15 B-T-Z3-3
16 1.39 1.90 1.21 Z4 N-S-Z4-1 Unidirectional
17 1.62 2.22 1.15 Z5 N-S-Z5-2
18 1.92 2.63 1.09 Z6 N-S-Z6-3
Note: Standoff distance ¼ 1.50 m; pressure gauge distance from charge center: gauges P1 , P2 , P3 , P4 , and P5 at 1.59, 1.52, 1.51, 1.51, and 1.52 m,
respectively; and gauge FF at 1.00 m.

© ASCE 04019080-4 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


detonated in front of the specimens’ center at a standoff distance Based on numerical simulations of the deflection curves, elas-
of 1.50 m. The test boundary conditions were selected to simulate tic behavior is apparent in the deflection of the N-S core under
the weakest possible connection between sacrificial cladding and the scaled distance Z1 [Fig. S2(a)], although the phenomenon of
supporting structure. Panels supported on all sides are more likely web crippling is not captured by the model and thus not represented
to be representative of current construction practices and would in the deformed configuration; deformations associated with plastic
probably offer greater performance in terms of blast resistance (Yu behavior, typically characterized by a bilinear shape, are noted in
2016). Therefore, the results discussed in the following sections the B-T core under the scaled distance Z2 [Fig. S2(b)]. The panel
may be regarded as a lower bound of the panels’ performance, es- velocity is calculated over 0.5-ms intervals (0–0.5, 1–1.5, and 2.0–
pecially in the case of B-T cores. 2.5 ms): for example, the N-S core shown in Fig. S2(a) deformed at
a speed varying from 1.1 to 3.9 to 1.2 m=s, respectively.
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

Postblast Observations
After each test, the postblast damage and mode of failure of each Analysis of Test Results
panel were evaluated. Under the first scaled distance (Z1 ), the spec-
imens exhibited elastic behavior, as no permanent deformations
Dynamic versus Quasistatic Panel Failure Modes
were observed either in the core webs or in the face sheets. As ex-
pected, permanent deformations were observed for the more critical Blast-resistant design guidelines, such as those found in the tech-
scenarios, simulated by the scaled distances Z2 –Z4 . No weld failure nical manual UFC 3-340-02 (USDOD 2014), utilize the static re-
was observed in any of the 18 specimens, which demonstrates the sistance function, adjusted for strain rate effects, to build a dynamic
ability of the adopted welding technique to withstand blast loading. model of a structural element under blast load. Hence it is important
The dynamic failure modes experienced by the unidirectional and to determine whether the failure modes of a panel configuration
bidirectional core configurations are discussed in detail in the fol- remain consistent when nominally identical specimens are exposed
lowing sections and compared with failure modes exhibited under to different load regimes. This is accomplished by comparing the
quasistatic loading. dynamic failure modes exhibited by the uni- and bidirectional core
configurations to the quasistatic failure modes reported by Khalifa
Displacement Response History and et al. (2018).
Panel Deflected Shape
Unidirectional Core Sandwich Panels (N-S Core)
Fig. 2 shows the measured midspan displacement histories re- In agreement with preliminary design as well as prior static test
corded under different scaled distances. Fig. 2(a) presents the mid- results, N-S-core panels exhibited web crippling within the support
span displacement history for the B-T-core configuration, where zone, followed by shear failure at high blast load levels. Fig. 3
larger maximum displacements are clearly correlated with lower shows good agreement between dynamic and static modes of fail-
scaled distances. Elastic response of the B-T core is observed as ure, although the unidirectional panels did not show shear-induced
a result of the largest scaled distance Z1 , while the maximum de- general buckling (Khalifa et al. 2018).
formation is produced by the lowest scaled distance Z3 . This behav- The N-S-core panels were tested under five different scaled dis-
ior is consistent with preliminary analysis, whereby different scaled tances, as summarized in Table 1. The first scaled distance (Z1 )
distances were selected to produce different damage states—as produced elastic behavior, while smaller distances instigated differ-
defined by applicable design standards. Similarly, the influence of ent levels of web crippling, as shown in Fig. 3. Panel deformation
the scaled distance on the maximum displacement experienced was initiated by web crippling accompanied by small rotations at
by N-S-core panels can be seen in Fig. 2(b). N-S-core specimens the back corners of the corrugated core. Under lower scaled distan-
exhibit an elastic response under the distance Z1 , and as Z is re- ces (i.e., higher loads), buckling was increasingly noticeable in the
duced, the maximum deformation is found to increase significantly. lower third of the core webs and resulted in the formation of plastic
The residual displacement is also captured in Fig. 2 and follows hinges. Furthermore, the cross-section depth continued to decrease
the same trend observed in the maximum displacement; i.e., it to the point where adjacent inner webs came into contact in the
increases as the scaled distance decreases. form of a triangle, as shown in Fig. 3(g). In summary, the static and

180 80
160 70
Mi d - span di spl acement (mm)
Mid - span displacement (mm)

140 60
120 50 N-S-Z1-3 (W=0.11 kg)
B-T-Z1-3 (W=0.11 kg)
N-S-Z2-2 (W=0.77 kg)
100 B-T-Z2-3 (W=0.77 kg) 40
N-S-Z4 (W=1.39 kg)
B-T-Z3-2 (W=0.94 kg)
80 30 N-S-Z5 (W=1.62 kg)

60 20
40 10
20 0
0 -10
-20 -20
0 10 20 30 40 50 0 10 20 30 40 50
(a) Time (ms) (b) Time (ms)

Fig. 2. Midspan displacement histories: (a) bidirectional panels (B-T core); and (b) unidirectional panels (N-S core).

© ASCE 04019080-5 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


Web
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

crippling

Z1 Z3 Z4
(a) (b) (c) (d)

Z1 Z3 Z4
Deformed Undeformed
(e) (f) (g) (h)

Fig. 3. Dynamic versus static failure modes of N-S-core panels: (a) Z1 ¼ 2.82 m=kg1=3 and (e) low static load; (b) Z3 ¼ 1.38 m=kg1=3 and
(f) intermediate static load; (c) Z4 ¼ 1.21 m=kg1=3 and (g) high static load; and (d and h) failure mode schematics.

dynamic modes of failure observed in N-S-core panels are found to reflected waves and a quasistatic pressure buildup, the latter being
be congruent, which supports the adoption of static resistance func- responsible for a greater overall specific impulse.
tions for the purpose of dynamic analysis. The test records of reflected pressure histories, measured near
the midspan of the test specimen, generally show three positive
Bidirectional Core Sandwich Panels (B-T core) phases [as in Fig. S3(a)] when the charge mass is low—and the
The bidirectional (B-T-core) panels consisted of a two-way cor- scaled distance high, e.g., Z1 ¼ 2.82 m=kg1=3 , which confirms
rugated core made of galvanized B deck profile, which enabled what has been reported by Baker et al. (1983). For lower values
stress redistribution. Fig. 4 shows that the panels’ dynamic response of the scaled distance, from Z2 ¼ 1.48 to Z6 ¼ 1.09 m=kg1=3 ,
remained elastic under the scaled distance Z1 . For more critical the duration of the positive pressure phase of the reflected waves
scenarios, achieved by increasing the charge mass, initiation of is observed to decrease as the scaled distance diminishes. It is
buckling of the face sheet at midspan and buckling of the core postulated that this phenomenon may be the result of the greater
elements were observed. As expected, greater permanent deforma- amount of detonation by-products (hot gases) relative to the volume
tion was observed when the scaled distance was decreased from of the blast chamber (36.4 m3 ), which would counter the propaga-
Z2 to Z3 . Since B-T-core panels experienced the same failure tion of reflected waves in the confined space while contributing to a
modes under both static and dynamic loads, it is reasonable to as- higher quasistatic pressure buildup (gas pressure). Based on the ra-
sume that the static resistance function can be used for dynamic tios of TNT charge mass to chamber volume, ranging from 0.004 to
analysis. 0.07 kg=m3 , the values of quasistatic pressure predicted via empir-
ical charts (Cormie et al. 2009) are expected to monotonically in-
crease with the charge mass within the interval 25–305 kPa; the
Predicted versus Experimental Blast Wave Parameters
associated ratios between quasistatic and reflected peak pressures
As mentioned in the “Introduction” section, a confined blast is a are expected to be within the 7%–10% interval. These numbers
more complex phenomenon than an outdoor blast of equivalent suggest a limited confining effect exerted by the blast chamber.
magnitude. Typically, the damage resulting from a confined blast Given the low level of confinement, the first positive phase
is expected to be significantly greater, owing to the contribution of can be predicted using the same tools commonly adopted when

© ASCE 04019080-6 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


Yielding
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

Z1 Z2 Z3
Undeformed Deformed
(a) (b) (c) (d)

Web
crippling

Z1 Z2 Z3
Undeformed Deformed
(e) (f) (g) (h)

Fig. 4. Dynamic versus static failure modes of B-T-core panels: (a) Z1 ¼ 2.82 m=kg1=3 and (e) low static load; (b) Z2 ¼ 1.48 m=kg1=3 and
(f) intermediate static load; (c) Z3 ¼ 1.38 m=kg1=3 and (g) high static load; and (d and h) failure mode schematics.

dealing with spherical unconfined explosions (Baker et al. 1983; available for calculating the mass of TNT equivalent to that of
Krauthammer 2008). In order to evaluate the parameters describing Composition C-4, generating either the same peak pressure or
the pressure profile at the shock wavefront (e.g., peak pressure the same impulse. For peak pressure and specific impulse values,
and specific impulse), nonlinear regression analysis is used to fit conversion factors of 1.37 and 1.19 are adopted, respectively.
the first positive phase of the pressure records with the modified Accordingly, the ratio of measured (Friedlander fitted) to ConWep
Friedlander equation, expressed as (Baker et al. 1983) predicted peak reflected pressure ranges from 0.59 to 0.83, with
    a minimum error of −41%, while the ratio between measured
t t
PðtÞ ¼ Pmax 1 − exp −α ð1Þ and predicted peak incident pressure values ranges from 0.65 to
td td
1.05, with an associated error between −35% and þ5%. The ratio
where PðtÞ = pressure function; Pmax = peak pressure; t = time; between Friedlander fitted and ConWep predicted reflected impulse
td = positive phase duration; and α = coefficient of exponential values ranges from 0.78 to 1.11, with an associated error between
decay. Key shock wavefront parameters such as the peak pressure, −22% and þ11%. The same ratio calculated for the incident im-
specific impulse, and positive phase duration for a spherical charge pulse ranges from 0.71 to 1.01 and the associated error ranges be-
detonated in free air are predicted using ConWep (Hyde 1991) and tween −29% and þ1%. Given the typical uncertainties associated
systematically compared to the values extracted from test records with blast wavefront metrics (Campidelli et al. 2015), the predic-
via Friedlander fitting, as reported in Table 2 [see also Fig. S3(b)]. tions reported in Table 2 can be regarded as sufficiently accurate for
ConWep predictions are based on conversion factors readily the limited scope of the current investigation.

© ASCE 04019080-7 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


Table 2. Measured versus predicted reflected and incident peak pressure and specific impulse—Spherical air blast
Conversion factor ¼ 1.37 Conversion factor ¼ 1.19
C-4 mass Scaled distance
(kg) (Zi ) Source Pr ðmaxÞ (kPa) Pso ðmaxÞ (kPa) I r a (kPa · ms) I s (kPa · ms)
0.11 Z1 Experimental 335 179 78 40
ConWep 242 224 75 47
Fitted 192 167 76 41
Ratio 0.79 0.75 1.01 0.87
0.77 Z2 Experimental 1,571 1,247 285 183
ConWep 1,523 973 307 165
Fitted 1,028 942 260 166
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

Ratio 0.67 0.97 0.85 1.01


0.94 Z3 Experimental 2,226 968 358 136
ConWep 1,874 1,126 356 186
Fitted 1,560 727 279 132
Ratio 0.83 0.65 0.78 0.71
1.39 Z4 Experimental 3,171 1,207 613 203
ConWep 2,768 1,490 476 228
Fitted 2,153 1,081 428 195
Ratio 0.78 0.73 0.90 0.86
1.62 Z5 Experimental 3,736 1,740 530 248
ConWep 3,232 1,662 495 243
Fitted 2,542 1,746 535 200
Ratio 0.79 1.05 1.08 0.82
1.92 Z6 Experimental 4,084 1,740 779 259
ConWep 3,820 1,877 610 248
Fitted 2,241 1,436 675 241
Ratio 0.59 0.77 1.11 0.97
a
Reflected impulse predictions from Eq. (6): Z ¼ Z1 ; Z2 ; : : : ; Z6 → I r ¼ 98, 380, 438, 579, 646, and 730 kPa · ms.

Further predictions can be made by adapting the formulation ιC ðρÞ ¼ 1 − ab−ρ ρ−c ð4Þ
provided by Geretto et al. (2015) for fully vented (FV) and un-
confined (UC) explosions, wherein the reflected impulse is mod- ιR ðρ; λÞ ¼ ðc00 þ c01 ρÞ þ ðc10 þ c11 ρ þ c12 ρ2 Þλ ð5Þ
eled as a linear function of the charge mass of PE4 (a variant of
Composition C-4) The factor ιC is defined for circular targets as the ratio between
the average specific impulse and the specific impulse at the center
of the target surface, when the charge is detonated in front of it;
Ir ¼ a · W þ b ð2Þ
similarly, ιR is defined for rectangular targets as the ratio between
the average specific impulse calculated over its surface and that
acting on a circular surface of equal area; ρ ¼ SD=r is the standoff
Eq. (2) is unit specific: W is the mass of PE4 explosive in grams,
ratio, that is, the ratio between the standoff distance to the target
I r is the impulse (N · s) delivered by the blast, and a and b are
center and the radius of the circular surface of area equal to that of
constants obtained via linear regression. To adapt this relationship
the target; λ is the aspect ratio of a rectangular surface, that is, the
to the current data set, Eq. (2) needs to be recast in terms of average
ratio of its longest to shortest sides; a ¼ 1.2, b ¼ 2.6, c ¼ 0.1,
specific impulse and scaled range as well as normalized by the cubic
c00 ¼ 1.0, c01 ¼ −1.4 × 10−3 , c10 ¼ −6.2 × 10−2 , c11 ¼ 1.5 × 10−2 ,
root of the charge mass (in kilograms of equivalent TNT). The re-
and c12 ¼ −9.8 × 10−4 are regression constants [see Figs. S4(a
gression constants also need to be adjusted in accordance with
and b)]. These coefficients are found to be suitable for (ρ; λ) values
the particular selection of units. Given a surface area A ¼ 0.04 m2
within the domain ½0.5; 10 × ½1; 10, although the ιC function is
(Geretto et al. 2015), Eq. (2) can be adapted as follows:
tuned to minimize the R2 over the interval ρ ¼ 0.5–2, which is
the part of the domain most sensitive to error. Given Eqs. (4)
  and (5), Eq. (3) can be corrected to account for specific impulse
Ir 250 a
ir ¼ ¼ þb ð3Þ distribution and used to obtain the scaled specific impulse at the
AðW · eÞ1=3 Ze Ze2=3
specimen center:
 
Ir 250 a
where ir = scaled specific impulse in kPa · ms=kg1=3 ; Z ¼ 10SD= ¼ þ b ½ιC ðρÞιR ðρ; λÞ−1 ð6Þ
ðW · eÞ1=3 = scaled range in m=kg1=3 ; e ¼ 1.19 is the impulse- ðW · eÞ1=3 Ze Ze2=3
based factor for the TNT equivalency of PE4; and ða; bÞ ¼
ð1.89; 8.15Þ and (0.62, 2.35) for fully vented and unconfined explo- pffiffiffiffiffiffiffiffi for the test specimens discussed herein, ρ ¼ ð1500=304.8Þ×
where,
sions, respectively. In addition, Eq. (3) can be corrected further to π=3 ¼ 5.04 and λ ¼ 3. The results of Eq. (6) are reported in
account for the nonuniform distribution of specific impulse over the Table 2 and Fig. S5: A comparison with (Friedlander fitted)
specimen surface. To this end, the following correction factors are test results and ConWep output reveals that ConWep gives the
introduced, based on regression analysis of ConWep outputs: least biased, although somewhat more scattered, predictions, as

© ASCE 04019080-8 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


demonstrated by the mean value and standard deviation associated by integrating the deformed shape of a panel, in accordance with
with the test:model ratio, respectively equal to 0.96 and 0.13. The principles of energy equivalency (Biggs 1964; USDOD 2014).
adapted formulation of Geretto et al. (2015) for unconfined explo- 2. Simply-supported boundary conditions, which accurately reflect
sions gives more biased, yet less scattered, predictions, i.e., a test: the test apparatus.
model ratio mean value and standard deviation equal to 0.77 and 3. A load-deflection curve (i.e., resistance function) assumed as
0.10, respectively. Geretto’s formulation for fully vented explo- the average of three test records obtained under quasistatic
sions overpredicts the reflected impulse by a factor of 2 or greater, conditions (see Fig. S6); this assumption implies negligible
which points to the limited confining effect exerted by the blast strain-rate effects. Because the steel strength under high-speed
chamber. loading is underestimated, displacement predictions are ex-
Based on simple formulas provided by Baker et al. (1983), es- pected to be somewhat overestimated (i.e., the model is ex-
timates of the diameter of the fireball produced by the test charges pected to be biased toward peak deflections somewhat greater
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

range from 1.53 to 3.82 m, depending on the charge mass. There- than test measurements).
fore, the pressure gauges, located approximately at 1.5 m from the 4. Negligible damping. The peak displacement typically occurs
explosive, were likely engulfed by the fireball, which raises some during the first cycle of the response, thus the effect of damping
concerns about the effect of heat on pressure records accuracy. is usually negligible (Biggs 1964).
In this respect, it is noted that the operational temperature of the 5. Uniformly distributed pressure over the panels’ exposed surface.
gauges in use is below 135°C and 121°C for the PCB and Dytran Accordingly, the load is determined as the product of the blast
models, respectively, and these upper bounds were likely exceeded overpressure, obtained from test records, and the area of the re-
within the fireball. However, thermal insulation from the environ- flective surface directly exposed to the explosive. Typically, in
ment is provided by the gauge manufacturers; in addition, a second the case of explosions in free air, only the positive phase of the
layer of protection consisting of silicone coating was consistently shock wave is considered. However, as previously mentioned,
applied before every test. the characteristics of a confined blast are different, owing to
the pressure buildup from waves reflected off the walls of the
chamber. Therefore, the full pressure record is used as input in
Dynamic Model the SDOF model to calculate the panels’ dynamic response. The
Several models in the literature, of varying degrees of sophistication, assumption of uniform pressure distribution is, for the most part,
can be adopted to simulate the behavior of sandwich panels under in compliance with current design practice (ASCE 2011), based
explosive loading. Some of these models provide analytical formu- on which the blast overpressure can be considered uniformly
lations to describe the behavior of the entire panel (Cui et al. 2012a; distributed in the far range, that is, for scaled distances greater
Fleck and Deshpande 2004; Kazemahvazi and Zenkert 2009), while than 1.2 m=kg1=3 . As summarized in Table 2, in most cases the
most are based on finite-element approaches (Karagiozova et al. adopted scale distance fell within the far range; two scenarios,
2009; Theobald and Nurick 2007, 2010). A detailed simulation of however, have distances slightly below the far-range threshold
sandwich panel mechanics is beyond the scope of this investigation; (Z5 ¼ 1.15 and Z6 ¼ 1.09 m=kg1=3 ). Therefore, the predictions
thus the adaptation of existing models to the geometry and loading made for specimens N-S-Z5-2 and N-S-Z6-3 may be affected by
conditions of the specimens tested by the authors is not considered. additional error.
Instead, a semiempirical approach is adopted—aimed at predicting 6. Rebound stiffness equal to the inbound stiffness. The inbound
peak displacements and corroborating the congruence of static and and rebound resistance values are defined as the out-of-plane re-
dynamic test results. To this end, single degree of freedom (SDOF) sistance of a component to the positive and negative phases of the
modeling is adopted as a viable approach to predict a component’s pressure wave, respectively. The inbound stiffness is computed
dynamic behavior in the context of blast loaded structures (ASCE as the average slope in the elastic range, on the basis of quasi-
2011; CSA 2012; USDOD 2014). The equation of motion of a static test data (Fig. S6). For the rebound stiffness K r , no data
SDOF system is expressed as (Biggs 1964) are available; thus three different values, defined in terms of the
inbound elastic stiffness K e , are considered, namely, 0.5K e ,
K LM M δ̈ þ RðδÞ ¼ FðtÞ ð7Þ 0.75K e , and K e . These values are assumed to carry out a pre-
liminary assessment of the dynamic response of the test panels
where δ̈ and δ = midspan acceleration and displacement; K LM = and will require further investigation to adequately characterize
load-mass (transformation) factor; M = panel’s mass (12.45 and the panels’ rebound phase. Based on the results yielded by a sub-
15.00 kg for N-S-core and B-T-core panels, respectively); RðδÞ = set of the specimens, a constant rebound stiffness K r ¼ K e is
panel’s average resistance function, determined experimentally via ultimately selected, in compliance with current practice (USACE
static testing, as outlined in the next section; and FðtÞ = load result- 2008a).
ing from the blast pressure, the latter also experimentally determined The model described by the six foregoing assumptions is sim-
via pressure gauge measurements, as described earlier. This semi- ple, fast-running, and the governing differential equation can be
empirical model is based on the following assumptions: integrated exactly via piecewise linear representation of the resis-
1. Full discrete connection (at the weld points) between layers of tance and forcing functions, on the basis of test records. The semi-
each composite panel, designed to exceed the strength of the empirical character of this approach stems from the empirical
base material; this assumption entails negligible slip at the inter- source of the terms RðδÞ and FðtÞ, which does not allow for robust
face, i.e., negligible longitudinal displacement of adjacent layers predictions based on first principles alone; rather, it provides a
relative to one another. The efficacy of such a connection was coarse means of validating static and dynamic data against each
corroborated by postblast inspections, which revealed no altera- other by comparing peak displacement values.
tion of the welds to the naked eye. Therefore, given the loading
and boundary configurations, the specimens’ overall behavior is
Static Resistance Function
assumed to be that of a one-way, flexure-controlled, monolithic
beam, whose cross sections behave in accordance with Euler- The static resistance of the sandwich panels was evaluated by the
Bernoulli theory. The load-mass factor is calculated accordingly, authors in a previous battery of tests by using an airbag to apply a

© ASCE 04019080-9 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


uniform, out-of-plane, quasistatic load (Khalifa et al. 2017, 2018). 5.26 × 10−5
Different corrugated core configurations were investigated, includ- q¼ 2
U4 ð10Þ
1 þ 161.7ðSD
D Þ
ing longitudinal, transverse unidirectional, bidirectional, and X core.
The quasistatic resistance functions were determined in terms of in which U is measured in kelvins. Depending on the mass of C-4,
yield, ultimate forces, and the corresponding displacements. Based Eq. (10) predicts rates of heat transfer ranging from 5.38 × 106 to
on static test results, the unidirectional and bidirectional core con- 3.24 × 107 Wm−2 C−1 at a standoff distance SD ¼ 1.5 m.
figurations were selected for testing with live explosives as they To evaluate the effects of the heat of detonation on the panels’
exhibited desirable ductility and energy absorption. Fig. S6 shows response, a thermal analysis is conducted on the front sheet ex-
the data recorded for three test samples of each selected core con- posed to the fireball, based on the following assumptions:
figuration as well as the mean value of the three data sets: Fig. S6(a) 1. Constant room temperature uR ¼ 20°C within the chamber.
shows the resistance–deflection relationship for the B-T-core panels
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

2. Initial sheet temperature equal to room temperature.


while Fig. S6(b) reports the data obtained from the N-S-core panels. 3. Heat flux density q uniformly distributed over the sheet surface
exposed to the fireball and calculated at the point closest to the
charge (the effect of a variable standoff distance is neglected);
Thermal Effects
timewise, q ¼ qðtÞ is assumed to be constant and equal to the
The effects of the heat generated in each detonation are evaluated value predicted by Eq. (10) for a temperature U ¼ 2,000 K and
by examining the properties of the fireball produced by a spherical a duration T ¼ 20 ms, after which cooling through free convec-
mass of C-4 explosive in free air. Fireball diameter (D) and du- tion is assumed.
ration (T) can be estimated via simple formulations provided by 4. Negligible expansion of the fireball inside the cubicle, as indi-
Baker et al. (1983), wherein diameter, mass, and duration are given cated by postblast observations that revealed no damage to the
in meters, kilograms, and seconds, respectively. High’s formula equipment housed within, including wiring. Accordingly, no
suggests D ¼ 3.86W 0.320 ; for the duration, Hasegawa and Sato’s heat flux is assumed on the sheet back side.
predictions are recommended for charges below 10 kg and show 5. Specimen cooling via free convection: The heat transfer coeffi-
T ¼ 1.07W 0.181 . However, these predictions are not based on tests cient of air (h) is assumed to be independent of temperature and
with C-4 explosive and should be corrected to account for different equal to 100 Wm−2 C−1 .
fireball temperature (U) and specific energy (E). High’s data set 6. Assumed steel properties include its thermal conductivity
is based on propellants, like liquid hydrogen and liquid oxygen, ks ¼ 50.2 Wm−1 C−1 , specific heat cs ¼ 511 Jkg−1 K−1 , and
that produce temperatures approximating 2,500 K (Baker et al. density ρs ¼ 7,850 kg=m3.
1983) and a specific energy output of 13 MJ=kg (Martinez 2018); Based on the foregoing assumptions, for the exposed side of the
Hasegawa and Sato’s data were obtained from propane [for which face sheet a natural (Neumann’s) boundary condition is assumed
U ¼ 2,000 K (Jir-Ming and Jun-Hsien 1996) and E ¼ 46.44 MJ=kg between t ¼ 0 and t ¼ 20 ms, while a convective (Robin’s) boun-
(Elert 2017)], pentane, and octane. For Composition C-4, en- dary condition is assumed afterward; for the back surface of the
ergy density estimates range from 4.87 (Baker et al. 1983) to sheet, a convective boundary condition is assumed for the entire
6.65 MJ=kg (USDA 1990), and a definite value E ¼ 5.38 MJ=kg analysis. The assumption of uniformly distributed heat flux density
is assumed for the purpose of the current investigation on the basis reduces the problem of heat transfer to one spatial dimension.
of TNT specific energy (4.52 MJ=kg) and conversion factor (1.19) By introducing the thermal diffusivity κs ¼ ks =ðcs ρs Þ and the local
as provided by Cormie et al. (2009). The steady-state fireball tem- coordinate x denoting the depth from the sheet exposed side
perature of C-4 charges similar in mass to those described herein (0 ≤ x ≤ G ¼ 2.03 mm), the phenomenon of heat transfer within
was measured to be approximately 2,000 K up to 20 ms from igni- the specimen can be described as follows (Hancock 2006; Lienhard
tion, with peaks of 3,000 to 5,000 K for the first 0.1 ms (Densmore and Lienhard 2001):
et al. 2011).
Based on fireball growth scaling laws (Baker et al. 1983), C-4 ∂u ∂2u
¼ κs 2 ð11Þ
specific scaling factors are calculated relative to the diameter (Do ), ∂t ∂x
duration (T o ), and specific energy (Eo ) of the reference explosive
(or propellant) as follows: with initial and boundary conditions defined as
 1 uðx; 0Þ ¼ uR ; for 0 ≤ x ≤ G ð12Þ
D To E 3
¼ ð8Þ
Do T Eo
∂u qðtÞ
 10  1 ð0; tÞ ¼ − ; for 0 ≤ t ≤ T ð13Þ
T To 3 E 3
∂x ks
¼ ð9Þ
To T Eo
∂u h
ð0; tÞ ¼ ½uð0; tÞ − uR ; for t > T ð14Þ
Eqs. (8) and (9) give scaling factors of 0.80 and 0.49, re- ∂x ks
spectively, which lead to predictions of C-4 fireball diameter and
duration ranging from 1.53 to 3.82 m and from 0.35 to 0.59 s, re- ∂u h
ðG; tÞ ¼ − ½uðG; tÞ − uR ; for t ≥ 0 ð15Þ
spectively, depending on the mass of explosive. However, it should ∂x ks
be noted that the foregoing durations are likely overestimated as
they are based on a steady-state temperature of 2,000 K, which may To verify accuracy, Eq. (11) is solved by using two integration
not reflect the actual temperature variation during fireball expansion. schemes based on finite differences: a forward-time-centered-space
In fact, 2011 records based on different measurement apparati sug- (FTCS) scheme that is explicit and conditionally stable and a
gest that high temperatures can be sustained up to 20 ms (Densmore Crank-Nicolson (CN) scheme that is implicit and unconditionally
et al. 2011) after ignition. In this initial phase, the heat flux density stable (Chapra and Canale 2010; Recktenwald 2011, 2014). Ac-
(q) delivered by the fireball though thermal radiation is calculated cordingly, the boundary conditions from Eqs. (13) and (14) produce
using the following interpolation formula (Baker et al. 1983): the temperature at the first node (integration point) as

© ASCE 04019080-10 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


    sheet—highly susceptible to local thermal stresses—it is therefore
2κs Δt 2hΔt j 2κs Δt
ujþ1
1 ¼ uj1 1− − þ u2 reasonable to conclude that the panels’ overall deformation was not
Δx2 cs ρs Δx Δx2
significantly affected by the thermal shock caused by the fireball. In
2Δt light of this information, a more rigorous three-dimensional analy-
þ ðqj þ huR Þ ð16Þ
cs ρs Δx sis, which would track the thermodynamics of the fireball—its
growth, temperature variation, and emissivity—as well as the non-
    uniform distribution of heat flux density over the specimen’s ex-
1 κ h κ
ujþ1
1 þ s2 þ þ u2jþ1 − s 2 posed surface, is deemed unwarranted.
Δt Δx cs ρs Δx Δx
   
1 κ h κs qj þ qjþ1 þ 2huR
¼ uj1 − s2 − þ uj2 2
þ Panel’s Ductility and Energy Absorption
Δt Δx cs ρs Δx Δx cs ρs Δx
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

ð17Þ The reported experimental deflection is determined using the mean


value of the records from the three potentiometers (A, B, and C)
where uji ¼ uðxi ; tj Þ is the temperature at the ith integration point at mounted at midspan. For simulated deflection histories, the in-
the jth time step; qj ¼ qðtj Þ is the heat flux density at the boundary bound stiffness (K e ) is computed as the slope of the static resistance
at the jth time step; Δt = time step size; and Δx = distance between function through each iteration during the analysis, while the
adjacent nodes. Eqs. (16) and (17) apply to the FTCS and CN rebound stiffness most representative of the specimens’ behavior
schemes, respectively, and give the temperature at the first node is assumed to be K r ¼ K e . Since both American and Canadian de-
induced by Newmann’s and Robin’s boundary conditions by set- sign standards for blast protection classify the damage state of a
ting h ¼ 0 and q ¼ 0, respectively; similarly, boundary conditions structural member according to either its maximum ductility ratio
can be enforced on the last node by replacing the subscripts 1 and 2 or its support rotation, a comparison is performed between the pre-
with those labeling the last and next to last nodes, respectively. dicted and measured maximum ductility and the ratio between the
The results of thermal analysis are conducted for the lowest two is reported in Table 3. A graphical comparison is also included
and highest masses of C-4 reported in Table 1; additional informa- in Fig. S8. It is found that the most accurate prediction of maximum
tion is also provided in Fig. S7. In both cases, a sharp rise in tem- ductility is obtained for specimen N-S-Z3-3, with a ratio of 99%,
perature is noted at the surface exposed to 0.11- and 1.92-kg while the least accurate prediction is obtained for panel B-T-Z3-2,
charges—to peak values of 80°C and 384°C, respectively; as with a ratio of 47%. The variability in prediction accuracy cannot
soon as the fire ball is diminished at t ¼ 20 ms, temperatures at be ascribed to the overly simplistic nature of the adopted model.
the front surface plummet while those at mid-depth and the back Given that nominally identical panels tested under nominally
side keep rising, which is indicative of heat redistribution via con- identical conditions exhibited significantly different responses, un-
duction. After a time approximately equal to fivefold that of the certainties associated with the wavefront parameters, material prop-
fireball duration, the system reaches a steady state of thermal equi- erties, and specimen geometry and fabrication must be invoked to
librium, wherein the entire sheet approximates the same tempera- explain these observations. Future work will be directed toward
ture, which is slowly lowered thereafter by heat exchange with the a rigorous quantification of all these sources of uncertainty.
surrounding air. The temperature gradient induced near the exposed Based on the measured resistance functions, N-S and B-T cores
surface by the detonation of 0.11- and 1.92-kg charges exceeds 600 reveal a capability for energy absorption equal to 2,460 and
and 100°C=mm, respectively, which would induce significant ther- 1,790 N · m, respectively—obtained by numerical integration of
mal stresses in the metal; however, in both instances the gradient the static load-deflection curves up to failure; greater ductility is
rapidly decreases with the distance from the front surface and observed in B-T cores (110-mm versus 60-mm maximum deflec-
it becomes vanishingly small (<1°C=mm) at the back side of the tion). Greater energy absorption through large plastic deformations
front sheet. With the exception of the superficial layers of the front of core elements is generally understood as the most desirable

Table 3. Predicted (SDOF) versus measured maximum displacement, ductility, and support rotation
SDOF analysis predictions Measurements Ratio
Shot C-4 mass Scaled distance Specimen
No. (kg) (m=kg1=3 ) designation δamax (mm) μamax θamax (rad) δ emax (mm) μemax θemax (rad) μamax =μemax
1 0.11 2.82 N-S-Z1-1 2.2 0.31 0.005 3.3 0.47 0.007 0.67
2 N-S-Z1-2 1.7 0.24 0.004 1.29
3 N-S-Z1-3 2.5 0.36 0.005 0.88
4 B-T-Z1-1 3.03 0.45 0.007 3.7 0.55 0.008 0.82
5 B-T-Z1-2 4.5 0.67 0.010 0.67
6 B-T-Z1-3 3.8 0.57 0.008 0.80
7 0.77 1.48 B-T-Z2-1 60.5 9.03 0.132 61.2 9.09 0.129 0.98
8 B-T-Z2-2 44.3 6.61 0.097 1.37
9 B-T-Z2-3 80.1 11.96 0.175 0.76
10 0.94 1.38 N-S-Z3-1 17.3 2.47 0.038 13.1 1.87 0.029 1.32
11 N-S-Z3-2 18.3 2.61 0.040 0.95
12 N-S-Z3-3 17.4 2.49 0.038 0.99
13 B-T-Z3-1 73.8 11.01 0.161 118.3 17.66 0.259 0.62
14 B-T-Z3-2 156.2 23.31 0.342 0.47
15 B-T-Z3-3 150.7 22.49 0.330 0.49
16 1.39 1.21 N-S-Z4 66.2 9.46 0.145 43.9 6.27 0.10 1.51
17 1.62 1.15 N-S-Z5 70.9 10.13 0.155 73.5 10.50 0.161 0.96
18 1.92 1.09 N-S-Z6 77.1 11.01 0.169 N/A N/A N/A N/A

© ASCE 04019080-11 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


Table 4. Predicted (SDOF) versus experimental damage
SDOF analysis prediction Experimental
Shot C-4 mass Scaled distance Specimen
No. (kg) (m=kg1=3 ) designation μamax Damage level μemax Damage level
1 0.11 2.82 N-S-Z1-1 0.31 Superficial 0.47 Superficial
2 N-S-Z1-2 Superficial 0.24 Superficial
3 N-S-Z1-3 Superficial 0.47 Superficial
4 B-T-Z1-1 0.45 Superficial 0.55 Superficial
5 B-T-Z1-2 Superficial 0.67 Superficial
6 B-T-Z1-3 Superficial 0.57 Superficial
7 0.77 1.48 B-T-Z2-1 9.03 Blowout 9.09 Blowout
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

8 B-T-Z2-2 Blowout 6.61 Blowout


9 B-T-Z2-3 Blowout 11.96 Blowout
10 0.94 1.38 N-S-Z3-1 2.47 Heavy 1.87 Heavy
11 N-S-Z3-2 Heavy 2.61 Heavy
12 N-S-Z3-3 Heavy 2.49 Heavy
13 B-T-Z3-1 11.01 Blowout 17.66 Blowout
14 B-T-Z3-2 Blowout 23.31 Blowout
15 B-T-Z3-3 Blowout 22.49 Blowout
16 1.39 1.21 N-S-Z4 9.46 Blowout 6.27 Blowout
17 1.62 1.15 N-S-Z5 10.13 Blowout 10.50 Blowout
18 1.92 1.09 N-S-Z6 11.01 Blowout N/A N/A
Note: μamax = maximum analytical ductility ratio; and μemax = maximum experimental ductility ratio.

feature for sacrificial cladding, because it provides a mechanism to According to qualitative damage descriptions, the damage to the
control and reduce the transfer of pressure to the protected struc- N-S-Z1 (unidirectional core) and B-T-Z1 (bidirectional core) groups
tures (Theobald and Nurick 2010). Therefore, unidirectional panels can be characterized as superficial, as none of the panels in these
appear to offer superior performance for blast-resistant applica- groups exhibited any permanent deformation. The panels in the
tions. It is also worth noting that ductility scales with the same fac- N-S-Z3 group experienced permanent deformations due to web
tor applicable to all linear dimensions, whereas the strain energy crippling and, based on the maximum ductility achieved, their dam-
scales with the cube of that very factor, further reinforcing the age can be classified as heavy. Finally, the panels in the B-T-Z3 and
conclusion that N-S cores are the best option for blast resistance. N-S-Z4 groups achieved the largest ductility and their damage state
Further consideration, however, should be given to the role played can be classified as blowout. It should be noted that the blowout
by the face sheets, especially the ones exposed to the blast. Previous state is below antiterrorism standards and is not permitted in either
investigations on panels featuring tubular core elements revealed the Canadian or the American standards. However, based on their
that sandwich designs capable of distributing the load evenly over static resistance, the N-S-core panels proved to be serviceable in
the core resulted in greater energy absorption (Theobald and Nurick such an extreme state, while providing an ever-greater resistance
2007). Therefore, future work should be directed toward investigat- under large deflections.
ing the optimal tradeoff between face sheet weight and stiffness, in
order to maximize core engagement—and thus energy absorption—
while using a minimum amount of material. Influence of Core Configuration
Sandwich panels with a corrugated core are known to experience a
Experimental versus ASCE 59 and CSA S850 Predicted broad range of failure mechanisms. Those specific to core elements
Damage Levels include face fracture, occurring when the compressive stress in a
core element exceeds the ultimate strength of steel; shear failure,
The test matrix was designed so that each core configuration would
occurring when shear stresses exceed the shear strength of the core
be tested under a minimum of three different scaled distances. The
material; wrinkling (local buckling), caused by compressive stresses
performance of each panel is characterized in terms of damage
greater than the critical buckling stress; and general buckling or
states (response limits), according to the definitions provided in
the two design standards for blast protection available in North shear buckling, wherein the distinction between the two modes is
America for cold-formed, corrugated, one-way panels with a lim- based on the ratio between compressive load and theoretical bifur-
ited tension membrane in flexure (ASCE 2011; CSA 2012). The cation load (Kazemahvazi and Zenkert 2009). In the current data set,
Canadian standard defines four damage levels: superficial, associ- visual observations indicate the recurrence of either general/shear
ated with no visible permanent deformations; moderate, indicating buckling or wrinkling, both classified under the umbrella of web
repairable permanent deformations; heavy, resulting from unrepair- crippling and mostly noted in unidirectional-core panels. Bidirec-
able permanent deformations not leading to failure; and hazardous, tional cores, on the other hand, were dominated by flexure and failed
describing failed components that are expected to cause debris with with a plastic hinge forming near the midspan. Overall, the N-S-core
negligible velocities. In addition, the technical manual PDC-TR panels were found to have higher blast resistance than B-T cores, as
06-08 (USACE 2008b) includes an additional damage state identi- demonstrated by a cross-comparison of panel performance under
fied as blowout, which is associated with components overwhelmed the same scaled distance. For instance, under a scaled distance of
by blast loading and flying debris with significant velocities. 1.38 m=kg1=3 , N-S panels exhibited a maximum midspan deflec-
Table 4 reports the predicted and measured maximum ductility tion ranging from 13.1 to 18.3 mm, whereas B-T panels experi-
ratios for all test specimens; the corresponding damage levels are enced a maximum deflection ranging from 118.3 to 156.2 mm,
determined in accordance with the damage states recapped earlier. as reported in Table 3. An overall comparison is provided in

© ASCE 04019080-12 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


Fig. S9, which demonstrates that N-S cores consistently outperform modes of failure that may arise depending on core and support
B-T cores. geometry.

Conclusions Notation

The performance of lightweight, cold-formed steel, sandwich pan- The following symbols are used in this paper:
els subjected to confined explosions in a blast chamber is examined A = area of exposed surface (m2 );
in this study. On the basis of test results from 18 specimens, the a = regression constant;
following conclusions can be drawn: b = regression constant;
• The proposed panels are potentially viable as a lightweight, c = regression constant;
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

cost-effective solution to the problem of sacrificial cladding c00 = regression constant;


for blast risk mitigation. In particular, the N-S cores demonstrate
c01 = regression constant;
grater energy absorption than B-T cores; hence, under the de-
c10 = regression constant;
scribed experimental conditions, the former appear to be the
better choice for applications in blast protection via sacrificial c11 = regression constant;
cladding. c12 = regression constant;
• Postblast observations demonstrate the congruency between sta- cs = specific heat of steel (m2 s−2 C−1 );
tic and dynamic modes of failure. D = fireball diameter (m);
• No weld failure was observed in any of the test specimens, which Do = fireball diameter of the reference explosive or
suggests the reliability of the proposed welding technique. propellant (m);
• The blast wavefront parameters recorded for the first positive E = specific energy of the explosive charge (J=g);
phase (peak pressure and specific impulse) are found to be in Eo = specific energy of the reference explosive (J=g);
acceptable agreement with the predictions from ConWep and e = equivalent TNT mass factor;
the formulation for reflected impulse adapted from Geretto et al. F = forcing function (N);
(2015). Test:model ratios associated with predictions of specific FF = free-field pressure gauge;
impulse produce mean values and standard deviations respec- G = face sheet thickness (m);
tively equal to 0.96 and 0.13 (ConWep) and 0.77 and 0.10
h = heat transfer coefficient of air (Wm−2 C−1 );
(Geretto’s). Hence, compared to ConWep, Geretto’s adapted
I r = reflected impulse (N · s);
formulation for unconfined explosions gives more biased, yet
less scattered, predictions. I r = reflected specific impulse (Pa · s);
• In terms of maximum ductility, for unidirectional panels the I s = incident specific impulse (Pa · s);
ratio between predictions and measurements ranges from 0.67 ir = scaled reflected specific impulse (Pa · s=g1=3 );
to 1.51; for bidirectional-core panels, the same ratio ranges from K e = inbound elastic stiffness (N=m);
0.47 to 1.37. Given the significant error affecting these predic- K LM = load-mass factor;
tions, future work will be directed at refining the adopted model K r = rebound stiffness (N=m);
to overcome its current shortcomings. ks = thermal conductivity of steel (gms−3 C−1 );
The limitations of this investigation relate to both experimental M = panel mass (g);
conditions and modeling assumptions: P = pressure (Pa);
• The scope of the experimental program is limited to specific P1 = reflected pressure gauge 1;
core configurations and boundary conditions. The test results
P2 = reflected pressure gauge 2;
presented by the authors should not be directly applied to panels
P3 = reflected pressure gauge 3;
with different core configurations and boundary conditions. In
particular, the greater energy absorption observed in N-S cores P4 = reflected pressure gauge 4;
may be partly dependent on the boundary conditions. B-T core P5 = reflected pressure gauge 5;
panels supported on all sides may provide a superior perfor- Pmax = peak pressure (Pa);
mance to that recorded in the current data set. Pr = peak reflected pressure (Pa);
• The current study does not evaluate the effects of face sheet Ps = peak incident pressure (Pa);
thickness. Earlier studies have shown that sandwich configura- q = heat flux density (Wm−2 C−1 );
tions with stiff face sheets, capable of distributing the load evenly qj = heat flux density at time step j (Wm−2 C−1 );
over the core, result in greater energy absorption. Therefore, fu- R = resistance function (N);
ture work should be directed at investigating the optimal tradeoff r = radius of circle of equal area (m);
between face sheet weight and stiffness, in order to maximize SD = standoff distance (m);
core engagement—and thus energy absorption—while using a SDOF = single degree of freedom;
minimum amount of material.
T = fireball duration (s);
• In terms of analysis, the model adopted by the authors does not
TNT = trinitrotoluene;
capture the multiplicity of failure modes known to occur in pa-
nels with corrugated cores, it does not account for strain rate T o = fireball duration of the reference explosive or
effects on the mechanical properties of steel, nor does it quantify propellant (s);
the several sources of uncertainty—associated with wavefront t = time (s);
parameters, material properties, specimen geometry, and fabri- td = positive phase duration (s);
cation defects. Future work will be directed at the computa- U = fireball temperature (°C);
tion of the resistance function from first principles, in order u = steel panels’ temperature (°C);
to address the foregoing limitations and capture the different uji = steel panels’ temperature at node i and time step j (°C);

© ASCE 04019080-13 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


uR = room temperature (°C); Densmore, J., M. Biss, J. Ritter, B. McAndrew, B. Homan, and K.
W = charge mass (g); McNesby. 2011. Time resolved optically based temperature and
pressure measurements of suspended C-4 spheres during and after
x = depth from the sheet surface exposed to the fireball (m); detonation. Technical Rep. No. ARL-TR-5667. Aberdeen, MD. Army
Z = scaled distance (m=g1=3 ); Research Laboratory.
α = coefficient of exponential decay; Dharmasena, K. P., D. T. Queheillalt, H. N. G. Wadley, P. Dudt, Y. Chen,
δ = midspan displacement (m); D. Knight, A. G. Evans, and V. S. Deshpande. 2010. “Dynamic com-
δ̈ = midspan acceleration (m=s2 ); pression of metallic sandwich structures during planar impulsive load-
δ amax = maximum analytical displacement (m); ing in water.” Eur. J. Mech. A. Solids 29 (1): 56–67. https://doi.org/10
.1016/j.euromechsol.2009.05.003.
θamax = maximum analytical support rotation (degree);
Dharmasena, K. P., H. N. G. Wadley, K. Williams, Z. Xue, and J. W.
δ emax = maximum experimental displacement (m); Hutchinson. 2011. “Response of metallic pyramidal lattice core sand-
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

θemax = maximum experimental support rotation (degrees); wich panels to high intensity impulsive loading in air.” Int. J. Impact
Δt = time-step size (s); Eng. 38 (5): 275–289. https://doi.org/10.1016/j.ijimpeng.2010.10.002.
Δx = size of x partition (m); Elert, G. 2017. “Energy density of propane.” The Physics Hypertextbook.
ιC = ratio of average specific impulse to center impulse Accessed September 1, 2019. https://hypertextbook.com/facts/2002
/EricLeung.shtml.
(circular surface);
FEMA. 2005. Risk assessment: A how-to guide to mitigate potential ter-
ιR = ratio of rectangular to circular average specific impulse rorist attacks against buildings. FEMA 452. Washington, DC: FEMA.
(rectangular surface); Fleck, N. A., and V. S. Deshpande. 2004. “The resistance of clamped
κs = thermal diffusivity of steel (m2 =s); sandwich beams to shock loading.” J. Appl. Mech. 71 (3): 386. https://
λ = aspect ratio of a rectangular surface; doi.org/10.1115/1.1629109.
μamax = maximum analytical ductility ratio; Geretto, C., S. Chung Kim Yuen, and G. N. Nurick. 2015. “An experimen-
μemax = maximum experimental ductility ratio; tal study of the effects of degrees of confinement on the response of
square mild steel plates subjected to blast loading.” Int. J. Impact Eng.
ρ = standoff ratio; and
79 (May): 32–44. https://doi.org/10.1016/j.ijimpeng.2014.08.002.
ρs = steel density (g=m3 ). Hancock, M. J. 2006. “The 1-D heat equation.” MIT OpenCourseWare.
Accessed August 31, 2018. https://ocw.mit.edu/courses/mathematics
/18-303-linear-partial-differential-equations-fall-2006/lecture-notes
Supplemental Data /heateqni.pdf.
Hyde, D. W. 1991. Conventional weapons effects program (ConWep)—
Figs. S1–S9 are available online in the ASCE Library (www Application of TM5-855-1. Vicksburg, MS: US Army Engineer Water-
.ascelibrary.org). ways Experiment Station.
Jir-Ming, C., and Y. Jun-Hsien. 1996. “The measurement of open propane
flame temperature using infrared technique.” J. Quant. Spectrosc.
References Radiat. Transfer 56 (1): 133–144. https://doi.org/10.1016/0022-4073
(96)00013-1.
ASCE. 2011. Blast protection of buildings. ASCE/SEI 59. Reston, VA: Karagiozova, D., G. N. Nurick, and G. S. Langdon. 2009. “Behaviour
ASCE. of sandwich panels subject to intense air blasts—Part 2: Numerical
Baker, W. E., P. A. Cox, P. S. Westine, J. J. Kulesz, and R. A. Strehlow. simulation.” Compos. Struct. 91 (4): 442–450. https://doi.org/10.1016
1983. Explosion hazards and evaluation. New York: Elsevier. /j.compstruct.2009.04.010.
Biggs, J. 1964. Introduction to structural dynamics. New York: Kazemahvazi, S., and D. Zenkert. 2009. “Corrugated all-composite sand-
McGraw-Hill. wich structures. Part 1: Modeling.” Compos. Sci. Technol. 69 (7–8):
Børvik, T., A. G. Hanssen, S. Dey, H. Langberg, and M. Langseth. 2008. 913–919. https://doi.org/10.1016/j.compscitech.2008.11.030.
“On the ballistic and blast load response of a 20 ft ISO container pro- Khalifa, Y. A., W. W. El-Dakhakhni, M. Campidelli, and M. J. Tait. 2018.
tected with aluminium panels filled with a local mass—Phase I: Design “Performance assessment of metallic sandwich panels under quasi-
of protective system.” Eng. Struct. 30 (6): 1605–1620. https://doi.org/10 static loading.” Eng. Struct. 158 (Jul): 79–94. https://doi.org/10.1016/j
.1016/j.engstruct.2007.10.010. .engstruct.2017.11.064.
Campidelli, M., W. W. El-Dakhakhni, M. J. Tait, and W. Mekky. 2015.
Khalifa, Y. A., M. J. Tait, and W. W. El-Dakhakhni. 2017. “Out-of-plane
“Blast design-basis threat uncertainty and its effects on probabilistic risk
behavior of lightweight metallic sandwich panels.” J. Perform. Constr.
assessment.” ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ.
Facil. 31 (5): 04017056. https://doi.org/10.1061/(ASCE)CF.1943-5509
Eng. 1 (4): 04015012. https://doi.org/10.1061/AJRUA6.0000823.
.0001018.
Center of Marine Technologies E.V. 2011. Best practice guide for sandwich
Knox, E. M., M. J. Cowling, and I. E. Winkle. 1998. “Adhesively bonded
structures in marine applications. European Commission Contract No.
steel corrugated core sandwich construction for marine applications.”
FP6-506330. Brussels, Belgium: Center of Marine Technologies E.V.
Chapra, S. C., and R. P. Canale. 2010. Numerical methods for engineers. Mar. Struct. 11 (4–5): 185–204. https://doi.org/10.1016/S0951-8339
Singapore: McGraw-Hill. (98)40651-8.
Cormie, D., G. Mays, and P. Smith. 2009. Blast effects on buildings. Krauthammer, T. 2008. Modern protective structures. Boca Raton, FL:
London: Thomas Telford. CRC Press.
CSA (Canadian Standards Association). 2012. Design and assessment Kujala, P., and A. Klanac. 2005. “Steel sandwich panels in marine applica-
of buildings subjected to blast loads. CSA S850. Mississauga, tions.” Brodogradnja: Teorija i praksa brodogradnje i pomorske tehnike
Canada: CSA. 56 (4): 305–314.
Cui, X., L. Zhao, Z. Wang, H. Zhao, and D. Fang. 2012a. “A lattice Lienhard, J. H. I., and J. H. V. Lienhard. 2001. A heat transfer textbook.
deformation based model of metallic lattice sandwich plates subjected Cambridge, MA: John H. Lienhard V.
to impulsive loading.” Int. J. Solids Struct. 49 (19–20): 2854–2862. Liu, H., Z. K. Cao, G. C. Yao, H. J. Luo, and G. Y. Zu. 2013. “Performance
https://doi.org/10.1016/j.ijsolstr.2012.04.025. of aluminum foam-steel panel sandwich composites subjected to blast
Cui, X., L. Zhao, Z. Wang, H. Zhao, and D. Fang. 2012b. “Dynamic loading.” Mater. Des. 47 (May): 483–488. https://doi.org/10.1016/j
response of metallic lattice sandwich structures to impulsive loading.” .matdes.2012.12.003.
Int. J. Impact Eng. 43 (May): 1–5. https://doi.org/10.1016/j.ijimpeng Marsico, T. A., P. E. Denney, and A. Furio. 1993. “Laser welding of light-
.2011.11.004. weight structural steel panels.” In Proc., Int. Congress on Applications

© ASCE 04019080-14 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080


of Lasers & Electro-Optics, 444–452. Orlando, FL: Laser Institute of Theobald, M. D., G. S. Langdon, G. N. Nurick, S. Pillay, A. Heyns, and
America. R. P. Merrett. 2010. “Large inelastic response of unbonded metallic foam
Martinez, I. 2018. “Spacecraft propulsion.” Department of Motopropul- and honeycomb core sandwich panels to blast loading.” Compos. Struct.
sion and Thermofluid Dynamics, Polytechnic University of Madrid. 92 (10): 2465–2475. https://doi.org/10.1016/j.compstruct.2010.03.002.
Accessed August 30, 2018. http://webserver.dmt.upm.es/∼isidoro/bk3 Theobald, M. D., and G. N. Nurick. 2007. “Numerical investigation of the
/c17/Spacecraftpropulsion.pdf. response of sandwich-type panels using thin-walled tubes subject to
McShane, G. J., V. S. Deshpande, and N. A. Fleck. 2010. “Underwater blast loads.” Int. J. Impact Eng. 34 (1): 134–156. https://doi.org/10
blast response of free-standing sandwich plates with metallic lattice .1016/j.ijimpeng.2006.04.003.
cores.” Int. J. Impact Eng. 37 (11): 1138–1149. https://doi.org/10.1016 Theobald, M. D., and G. N. Nurick. 2010. “Experimental and numerical
/j.ijimpeng.2010.05.004. analysis of tube-core claddings under blast loads.” Int. J. Impact Eng.
Norris, C., P. Montague, and K. H. Tan. 1989. “All-steel structural panels to 37 (3): 333–348. https://doi.org/10.1016/j.ijimpeng.2009.10.003.
carry lateral load: Experimental and theoretical behaviour.” Struct. Eng.
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 01/22/20. Copyright ASCE. For personal use only; all rights reserved.

USACE. 2008a. Methodology manual for the single-degree-of- freedom


67 (9): 167–176. blast effects design spreadsheets. Rep. No. PDC TR-06-01 Rev 1.
Nurick, G. N., G. S. Langdon, Y. Chi, and N. Jacob. 2009. “Behaviour of Washington, DC: USACE.
sandwich panels subjected to intense air blast. Part 1: Experiments.” USACE. 2008b. Single degree of freedom structural response limits for
Compos. Struct. 91 (4): 433–441. https://doi.org/10.1016/j.compstruct
antiterrorism design. Rep. No. PDC–TR 06-08 Rev 1. Washington,
.2009.04.009.
DC: USACE.
Recktenwald, G. W. 2011. “Finite-difference approximations to the heat
USDA. 1990. Military explosives: Technical manual TM 9-1300-214.
equation.” KTH Royal Institute of Technology. Accessed August 31,
Washington, DC: USDA.
2018. http://www.nada.kth.se/∼jjalap/numme/FDheat.pdf.
USDOD (US Department of Defense). 2014. Unified facilities criteria:
Recktenwald, G. W. 2014. “Alternative boundary condition implementa-
tions for Crank Nicolson solution to the heat equation.” Portland State Structures to resist the effects of accidental explosions. Document
University. Accessed August 31, 2018. https://web.cecs.pdx.edu/∼gerry No. UFC 3-340-02. Washington, DC: USDOD.
/class/ME448/notes/pdf/Alt_BC_slides.pdf. Wang, T., M. Ma, W. Yu, S. Dong, and Y. Gao. 2011. “Mechanical response
Rimoli, J. J., B. Talamini, J. J. Wetzel, K. P. Dharmasena, R. Radovitzky, of square honeycomb sandwich plate with asymmetric face sheet sub-
and H. N. G. Wadley. 2011. “Wet-sand impulse loading of metallic jected to blast loading.” Procedia Eng. 23: 457–463. https://doi.org/10
plates and corrugated core sandwich panels.” Int. J. Impact Eng. .1016/j.proeng.2011.11.2530.
38 (10): 837–848. https://doi.org/10.1016/j.ijimpeng.2011.05.010. Wiernicki, C. J., F. Liem, G. D. Woods, and A. J. Furio. 1991. “Structural
Roland, F., T. Reinert, and G. Pethan. 2003. “Laser welding in shipbuild- analysis methods for lightweight metallic corrugated core sandwich
ing: An overview of the activities at Meyer Werft.” Weld. Res. Abroad panels subjected to blast loads.” Nav. Eng. J. 103 (3): 192–202.
49 (4): 39–51. https://doi.org/10.1111/j.1559-3584.1991.tb00949.x.
Smith, P. D., and J. G. Hetherington. 1994. Blast and ballistic loading of Xue, Z., and J. W. Hutchinson. 2003. “Preliminary assessment of sandwich
structures. Oxford, UK: Butterworth-Heinemann. plates subject to blast loads.” Int. J. Mech. Sci. 45 (4): 687–705. https://
Tan, K. H., P. Montague, and C. Norris. 1989. “Steel sandwich panels: doi.org/10.1016/S0020-7403(03)00108-5.
Finite element, closed solution, and experimental comparisons, on a Yu, C. ed. 2016. Recent trends in cold-formed steel construction.
6 m × 2.1 m panel.” Struct. Eng. 67 (9): 159–166. Cambridge, MA: Woodhead Publishing.

© ASCE 04019080-15 J. Perform. Constr. Facil.

J. Perform. Constr. Facil., 2020, 34(1): 04019080

S-ar putea să vă placă și