Engineering Failure Analysis 18 (2011) 836–845

Contents lists available at ScienceDirect

Engineering Failure Analysis

journal homepage: www.elsevier .com/locate /engfai lanal

Review

What to look for in the aftermath of an explosion? A review of blastscene damage observables

Andrew Sorensen a,1, William L. McGill b,⇑a The Pennsylvania State University, University Park, PA 16802, United Statesb Information Sciences and Technology, The Pennsylvania State University, University Park, PA 16802, United States

a r t i c l e i n f o

Article history:Received 4 August 2010Received in revised form 18 December 2010Accepted 21 December 2010Available online 30 December 2010

Keywords:Post-blast investigationMaterial failure modesExplosive threatsDamage observablesForensic analysis

1350-6307/$ - see front matter � 2011 Elsevier Ltddoi:10.1016/j.engfailanal.2010.12.010

⇑ Corresponding author. Tel.: +1 814 867 0270.E-mail addresses: asorensen@ist.psu.edu (A. Sore

1 Tel.:+1 814 865 6948.

a b s t r a c t

A review of existing literature dealing with the behavior of often-used building materialssubject to blast loading is presented. Our focus is on the types of structural damage thatmight be observed at a blast scene, and in particular how this evidence can be used to infercharacteristics of the blast (e.g., size and position of the charge). We consider two simplequestions: what does damage look like in the wake of a blast for different structural mate-rials (i.e., ‘‘what damage is seen?’’) and what must the blast pressure have been to causethis damage (i.e., ‘‘what does the damage mean?’’). This review looks at reinforced con-crete, masonry, steel, glass, and timber. We propose the following principle for interpretingobserved damage at a blast scene: if damage is present, a limit state was exceeded. The resultsfrom this review will be of great interest to post-blast investigators that seek to understandthe size and position of a blast.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Unintentional explosions are highly undesired by pretty much everyone. In process industries, steps are frequently takento minimize the causes and consequences of accidental explosions [1]. In a security context, measures are often put in placeto interrupt attack planning and capability development, to deter would-be attackers from following through with theirplans, to prevent those that try to attack from achieving tactical success, and dull the impact of a successful attack [2].In both of these examples, the focus is on what can be done to minimize the risk of harm or loss before an explosionoccurs.

When an explosion happens, attention shifts away from prevention to attribution from the point of view of both causeand effect. The forensic engineer seeks to understand whether any resulting harm to persons or property can be attributedto negligence on the part of those responsible for the design, construction, maintenance or operation of the damaged struc-tural system [3,4]. Following a blast incident, the forensic engineer may be called into assess whether any structures or otherengineering artifacts that suffered damage performed as originally intended. That is, was everything working or functioningas it should at the time of the blast, or can some of the blame be attributed to factors other than the blast itself? If negligencewas not attributed, the findings from a parallel safety investigation may generate guidance for better and safer designs in thefuture. In general, the forensic engineer focuses on the effects of the blast that occurred, and not specifically on the blastitself.

. All rights reserved.

nsen), wmcgill@ist.psu.edu (W.L. McGill).

A. Sorensen, W.L. McGill / Engineering Failure Analysis 18 (2011) 836–845 837

In contrast, post-blast investigators gather evidence at the scene of an explosive incident in order to understand the nat-ure of the blast and ascertain whether it was accidental or deliberate [5,6]. The key questions on the mind of a post-blastinvestigator are:

– What (specifically) exploded?– Why did it explode?– Where did it happen?– Who is to blame?

Post-blast investigators take advantage of a wealth of tangible and testimonial evidence at the blast scene to ascertain thenature of the insult. This includes chemical residue, explosive device fragments, physical injuries in the blast zone and sur-rounding areas and bystander reports. Among the myriad evidence available to the post-blast investigator, the evidence leastoften used in an investigation is observations of structural damage.

Observed structural damage is a useful, but by no means precise, indicator of the strength of an explosion. Given exposureto a blast wave, a material will respond in such a way that results in observable damage or not. It is the observable damagethat is of interest to the post-blast investigator. If damage is observed, the post-blast investigator can say with certainty thatthe force at the location of damage was sufficiently high to damage the material. That is, the response of a material or structureexceeded some damage-inducing limit state. If the extent or pattern of damage is a function of load, then a careful descriptionof the pattern will assist in back-calculating the incident pressure or impulse. If the damage is binary or not easily sized withload, then at least one can resort to saying that the limit state between damage from non-damage was exceeded. It is generallyunderstood that different explosive types produce different pressure–time histories; however the majority of observable blastdamage to materials that can be useful for this purpose lie outside of the immediate impact area or initial time history.

The purpose of this paper is to present a summary of what is known about how structural materials break when exposedto a blast. This paper is oriented toward the needs of the post-blast investigation community; its aim is to provide post-blastinvestigators with insights on how to turn observations of damage into useful evidence. Specifically, the paper considers thefollowing two-part question: What does material damage look like in response to a blast and to what extent can observations ofdamage help us back-calculate properties of the actual blast? Post-blast investigators must be able to recognize and describestructural damage in order to make the best use of this evidence in their investigations. In future work, we intend to dem-onstrate a general framework that uses these and other observables to ascertain the center of the explosion, charge size anddirectional properties (if any). We believe that this review is a prerequisite for developing comprehensive guidance, proce-dures and technologies for evidence collection supporting future post-blast investigations.

2. Methodology

In this review, our focus is on the different types of damage sustained by traditional engineered materials following expo-sure to blast loading. To this end, we sought articles that describe how structural materials fail in response to a blast, thatprovide qualitative or quantitative descriptions of the damage caused by explosions, or that present analytical tools or limitstate formulations relating observed damage to experienced pressure. Materials include reinforced concrete, masonry, struc-tural steel, non-structural glass, and timber. We also considered the structural purpose for each material, such as whether itwas used as a slab, beam or column.

This review did not focus on the performance of systems of comprised of multiple structural components, though we rec-ognize that such concerns have been the topic of a number of recent engineering analyses of high-profile blast events [7–15].Furthermore, papers that utilized only numerical modeling without corresponding analytical calculations were also ex-cluded from this review with the exception of those papers that discuss the physical/observable behavior or failure modesof structural materials.

Our search was limited to scholarly articles (e.g., journal papers, conference proceedings), whether published or in-press,retrieved through a systematic search on Google Scholar, academic databases (e.g., INSPEC) and publisher websites (e.g.,Elsevier, Science Direct, ASCE publications, ACI publications). When feasible, references were traced from child to parentin order to broaden and deepen the extent of our search. The cutoff date for publications considered in this review was15 August 2010.

The organization of our synthesis of the literature is as follows. For each material considered, the types of observablesfollowing blast exposure are discussed in terms of (a) what is seen and (b) what it means. We discuss what is observed,how to describe it, how it happens, and what it implies. Again, our interest in not so much in material performance as itis on what can be gleaned about the blast from observed damage; the review is focused on the needs of the post-blast inves-tigator, not of the forensic engineer.

3. Material response to blast loading

The literature on the performance of five traditional building materials is reviewed in terms of observed damage follow-ing exposure to a blast load. The materials we considered are reinforced concrete, masonry, steel, non-structural glass, and

838 A. Sorensen, W.L. McGill / Engineering Failure Analysis 18 (2011) 836–845

timber. For each material, we consider different types of blast-induced damage and associated limit state formulations (ifavailable). In general, a limit state defines the boundary between failure and not-failure. Limit state functions are n-ary,where n can be 2 (binary), 3 (trinary) and so on. The response of some materials may progress through multiple damagestates, thus allowing for a more constrained estimate on the pressures needed to cause damage. For example, as discussedin a later section, damage to concrete and masonry proceeds through multiple distinct damage states where the transitionbetween states occurs at progressively higher levels of incident pressure. In such cases, the ability to discriminate betweenmultiple states of damage enhances our ability to make precise statement of what the load needed to be to cause such dam-age. Other situations are limited by the binary nature of failure. For example, ductile materials such as steel behave non-lin-early in the plastic region and suffer permanent deformation when the loads exceed the elastic limit (i.e., the yield stress).When this happens, the observed deformation is often not enough to (practically) produce accurate estimates of the causalload. However, evidence of permanent deformation is enough to say that the elastic limit has been exceeded. Similarly, ifpermanent deformation is not observed, then we have enough information to say that the yield strength was not exceeded.Thus, at the very least, characterizing a material as failed or not failed with respect to a particular type of damage providessome constraint on what the pressures can and cannot be. The basic idea behind this review is embodied by the followingstatement: if damage is present, a limit state was exceeded. In what follows we present a summary of the literature in this areaas it pertains to reinforced concrete, masonry, steel, glass, and timber.

3.1. Reinforced concrete (RC)

One phenomenon that occurs with RC slabs and plates is the development of spallation on the opposite side of the blastexposed panel (Fig. 1). Spallation, or spall, is a type of damage that results from the combination of pressure and heat result-ing from the blast load. However, engineers are often cautioned to not confuse spall caused by a resulting fire after a blastand spall caused by the blast itself [16]. Spall due to a resulting fire is characterized by char around the location of spall, aswell as the whether the spall is located on the side of the slab exposed to fire or on the opposite side. Concrete spall can alsooccur due to impact from a projectile such as shrapnel accompanying a blast [17].

In one of the first publications on the matter, McVay divides the level of spall damage into the following three categories[18]:

i. No damage (or light cracking): from initial state to a few barely visible cracks.ii. Threshold for spall: from a few cracks and a hollow sound to a large bulge in the concrete with a few small pieces of

spall on the surface.iii. Medium spall: from a very shallow spall to spall penetration up to one-third of the plate thickness.

Using these descriptors and building upon a previous work by Nash and others [16], Xu and Lu [19] use numerical sim-ulations to develop a set of best fit curve equations that relate the level of observed spall damage with charge weight (up to100 kg) at a specific standoff distance for square plates of varying thickness.

RC panels and slabs that are exposed to pressures resulting from blast loads, but not the heat that causes spall, can beanalyzed using yield line theory. In the development of yield line theory, it was observed that when RC plates are loadeduntil failure, the crack lines develop in the same locations for specimens with the same boundary conditions, aspect ratio,and loading type. Using these restraints, equations were developed to calculate the ultimate flexural capacity of RC slabs.An often quoted resource for calculating yield lines under dynamic loads can be found in the book Structural Dynamics by

Fig. 1. Concrete spall due to blast exposure.

A. Sorensen, W.L. McGill / Engineering Failure Analysis 18 (2011) 836–845 839

John Biggs published in 1964 [20]. This text includes diagrams of the different yield lines than can develop for differentaspect ratios and boundary conditions. Using these diagrams, the boundary conditions of a slab can usually be determinedbased on the location of the yield lines. Additionally, if more information is known about the slab such as concretecompressive strength and reinforcing steel size and spacing, then the approximate experienced pressure can also becalculated.

Recently, research has been carried out to determine the percentage of strength that is lost when openings in a RC panelare included [21]. The results of this study show that the resistance of a panel decreases by up to 60% for openings that rep-resent 20% of the total panel area. Figure 3 of this study also shows yield line configurations for a variety of opening locations.Additionally, a table (Table 1 of the study) is included in the study that presents modification factors for panels with differentaspect ratios and the ratio of the area the openings to the area of the panel. These factors can be used to modify the yield linetheory resistance calculations specified in Biggs work.

For cases where the blast charge was placed directly on a reinforced concrete slab, the development of a crater may occur.A study in 2004 on the behavior of concrete pavement slabs under blast loads uses limit analysis of a truncated cone geom-etry to represent the resulting blast crater. The authors developed and validated a simplified equation for calculating thepressure on slab using the resulting crater diameter and depth [22].

Another useful RC structure useful for post-blast analysis is the shear wall. A study of the behavior under blast loading ofload bearing RC shear walls was carried out by Naito and Wheaton in 2006 [23]. Using a single degree of freedom model, theauthors identified four stages of inelastic behavior that the walls experience before failure as follows:

Stage 1: The member remains fixed at the supports with a concave curvature beginning a short distance away from thesupports. The resulting curve has two inflection points.

Stage 2: The supports at one end of the member remains fixed and the end forms a hinge at the support. The resultingcurve has one inflection point near the fixed support.

Stage 3: The supports at both ends of the member form a hinge resulting in a concave curve with no inflection points.Stage 4: Both supports are hinged and the center of the member cracked resulting in the creation of two straight mem-

bers of equal length.

Figure 9 of Ref. [23] shows a diagram representing this behavior. Using Biggs as a reference, the authors show that resis-tance functions for the four stages of inelastic behavior (Table 1) can be used to match the results of a pressure–impulse dia-gram obtained by experiment. The observed failure shape of a load bearing RC shear wall can therefore be correlated to theproper resistance function and a corresponding blast pressure calculated using the methods prescribed by Biggs.

Another topic of interest for RC slabs is the resulting cracking/localized failure due to localized impact loads that oftenresult of blast debris. This topic is amply covered in the doctoral dissertation of Zineddin which discusses the different failuremodes of RC slabs subject to localized impact loads [17,24]. The dissertation states that if a particular building is subjected toa blast load, the floor slab will be exposed to a large transient dynamic load. As a result two possible modes of failure canoccur: (1) slab localized failure and (2) global failure where the slab diaphragms that transfer the lateral loads may be weak-ened to such an extent that the whole building may become laterally unstable. In both cases, these types of failures couldincrease the risk of an ensuing progressive collapse. Pictorial depictions of these failures can be found in Figure 2.9 of Ref.[24]. Furthermore, it is important to identify the modes of failure that can occur in order to model the response correctly.Impact can lead to different types of global or localized damage, including flexure penetration and scabbing, spalling, per-foration and punching shear failure.

The failure modes of slabs are dependent on the rate the load is applied. Static or quasi-static loading favors a global bend-ing mode. If the same load is applied during a soft impact, the supports have higher stresses and a direct shear failure at thesupport could replace the bending mode. If the impact becomes harder, time is too short for the stress wave propagation, andthe local response under the load could dominate the slab behavior. The probability of punching or direct shear increases andthere may be a combination of failure modes. According to the main orientation of major cracks, the following modes aregenerally distinguished during impact events: bending or flexure resulting in cracks around the impact location, shear alongthe supports, punching shear resulting in a hole at the point of impact, and direct shear which results in a hole at the point ofimpact as well as shear around the hole. Pictorial depictions of these failures can be found in Figure 2.10 of Ref. [24]. Spallingand scabbing generally take place during bending failure, but we do not see them in full penetration since the whole areafails. It is nevertheless suggested to distinguish between the projection of concrete debris out from the side opposite tothe load and the punching of the concrete cover from the side directly hit by the blast.

Table 1Load bearing RC shear wall component transformation factors (reproduced from [22]).

Stage no. Description Resistance Flexural stiffness (k)

1 Elastic 12MyB/L 384EI/L3

2 Elastic–Plastic I (8MyA + 4MyB)/L 185EI/L3

3 Elastic–Plastic II (12MyA + 4MyB)/L 384EI/5L3

4 Plastic (12MyA + 4MyB)/L 0

840 A. Sorensen, W.L. McGill / Engineering Failure Analysis 18 (2011) 836–845

3.2. Masonry

The analysis of masonry can be quite complex due to wide variations in materials and boundary conditions. One of themost often utilized masonry applications in structures is the masonry wall. In 2005, Wesevich and Oswald compiled the re-sults of 236 open-air and shock tube tests on conventional masonry walls of various spans, thicknesses, support, reinforce-ment configurations, and damage levels [25]. The results of their compilation were input into the Department of Defense’sConcrete Masonry Unit Database Software. In the study the authors observed that the level of damage to the concrete ma-sonry unit (CMU) walls could be separated into four categories: level 1 – reuse, Level 2 – replace, level 3 – collapse, and level4 – blowout. The given qualitative descriptions for the each category are given in Table 2 and pictures depicting each failuretype can be found in Figure 1 of Ref. [25] (see Fig. 2).

In 2008, Moradi and others prepared a current state of analysis paper on the resistance of membrane retrofitted CMUwalls to lateral pressure [26]. This paper also includes a review of the behavior of unreinforced CMU walls under such load-ing. The study considered unreinforced CMU walls supported along their top and bottom, and spanning vertically betweenthese supports. As flexural stresses increase, cracks develop along a course of the bed joints. The weight and surcharge (P) ofthe wall produces uniform compressive forces which initially counteracts any tensile forces that result from bending in thewall due to lateral forces (p). As the lateral forces increase, the impact of the compressive forces on the tension side of thewall (R) decrease and tensile forces begin to develop and increase. Once these tensile forces develop, cracks occur on two orthree units of the wall and eventually extend the full length of the wall. As the crack widths increase, compressive forcesbegin to increase on the compressive face of the wall. This behavior is known as arching action. The resulting compressionstress block increases up to the compressive strength of the masonry units and as it increases the center of the stress blockscontinues to move towards the compressive face of the wall. The wall will fail when the line of action of the compressionstress block moves outside of the acting centerline of the vertical loads caused by the wall self weight and surcharge. Figure 3provided in Ref. [26] illustrates this behavior. The paper also presents the derivation of an equation for determining the walldeflection based upon the lateral load. However, for the purpose of this study, the equation can be rearranged to determinethe lateral load when the resulting deflection is measured.

Similarly, Garbin and others discuss a similar equation presented in the Eurocode [27] for determining the out of planeload resistance for a reinforced concrete frame with a non-load bearing masonry-infill wall undergoing arching action [28].

Table 2CMU wall panel damage level summary (reproduced from [25]).

Damagelevel

Unreinforced without arching Unreinforced with arching Reinforced Qualitativeresponse

1 Hairline cracks, no noticeablepermanent deflections

Permanent deflections not exceeding0.15 h

Peak dynamic and residual supportrotations not exceeding 7 and 1degrees, respectively

Reusable

2 Wall is still standing, significantcracking with noticeable permanentdeflections

Wall is still standing, permanentdeflections exceed 0.15 h

Peak dynamic and residual supportrotations exceed 7 and 1 degrees,respectively

Replace

3 A major portion of the wall has fallen tothe ground either immediately in frontor behind the wall’s originalconfiguration

A major portion of the wall has fallen tothe ground either immediately in frontor behind the wall’s originalconfiguration

A major portion of the wall has fallen tothe ground either immediately in frontor behind the wall’s originalconfiguration

Collapse

4 Significant wall debris velocities aregenerated

Significant wall debris velocities aregenerated

Significant wall debris velocities aregenerated

Blowout

Fig. 2. Masonry cracking representative of level 2 – replace.

Fig. 3. Example of lateral/torsional buckling of steel members.

A. Sorensen, W.L. McGill / Engineering Failure Analysis 18 (2011) 836–845 841

The equation utilizes the compressive strength of the masonry and as such, an idea of the compressive strength must beknown. The authors also present failure modes for the infill walls based on experimental and analytic knowledge. The dif-ferent in plane failure modes of masonry-infill walls can be categorized into three distinct modes, namely [29]:

(1) Sliding shear mode: Represents horizontal sliding shear failure through bed joints of a masonry infill. This failure modeis associated with infill built with weak mortar joints and frame with strong members and joints. The occurrence ofthis failure mode causes what is known as the knee brace effect on the frame.

(2) Diagonal cracking mode: In the form of a crack connecting the two loaded corners. This failure mode is associated withframes with weak joints and strong members, and infill with strong blocks and mortar joints.

(3) Corner crushing mode: Represents crushing of the infill in at least one of its loaded corners. This failure mode is usuallyassociated with infill having weak masonry blocks surrounded by a frame with weak joints and strong members.

Diagrams of these three failure modes are shown in Figure 9-1 of Ref. [28].

3.3. Structural steel

There are three main failure modes associated with steel structures: buckling, necking, and shear tearing. It is possible tosee combinations of these three modes on the same specimen in what is called combined failure modes such as the casewhere a column first buckles which causes shear tear at bolted connections. There is also the possibility that the failuremodes can happen in more than one plane such as in the case of lateral/torsional buckling as shown in Fig. 3.

Lateral buckling is commonly seen in steel columns used to transfer loads vertically down to the foundation. Structuralloads are transferred to columns using beams and floor slabs. Because the loads are carried vertically steel columns are usu-ally considered to be compression members. One benefit of using steel columns over concrete columns is the ability to createmoment frames where under certain loading conditions, some column members would be exposed to tension forces. Underblast loading, tensile loading of columns is common as floor systems lift under the blast load. The failure mode for a columnloaded only in axial compression is buckling. Buckling is a lateral movement in the beam that shortens the length of the col-umn. The buckled shape of the column is a function of the support conditions and follows the Eigen-value shapes. Columnbuckling behavior is analyzed using the Euler buckling equation which calculates the applied force necessary to buckle a col-umn and which depends on the geometric and material properties of the column [30].

A steel column exposed to a tension load will fail when the applied tension load causes an internal stress in the memberthat is greater than the tensile strength of the material. The applied stress is calculated by dividing the axial tension force bythe cross sectional area of the member. The failure location in the member is determined by slight differences of the materialstrength at the micro level of the member. However, if a uniform material strength is assumed, the steel member will beginto elongate and the cross sectional area of the steel member will begin to diminish or neck until the column is completelypulled apart into two pieces.

In an article in Structure Magazine, three practitioners discuss the combined effects of axial load and load distribution forair-blast loaded steel columns [31]. The authors determine that in addition to axial load effects, column response can beinfluenced by the assumed shape of the air-blast loading. The distribution of air pressure along the height as a uniform loadwith the formation of a plastic hinge at mid-height of the column is appropriate for far-field effects; however, this simpli-fication is inaccurate for columns located in close proximity to the explosion (see Figure 2 of Ref. [31]). The authors continueby discussing that near-field effects occur when columns are subjected to a close-in explosion causing a non-uniform pres-sure distribution along the height of the column which typically results in a plastic hinge located below mid-height of thecolumn (see Figure 3 of Ref. [31]). The combined effects of load distribution and axial load were also studied by examiningcolumn response to close-in explosions using a non-uniform load distribution where the expected plastic hinge location was

842 A. Sorensen, W.L. McGill / Engineering Failure Analysis 18 (2011) 836–845

determined by analysis (Figures 4a and b of the study). Comparison of damaged steel columns to figures given in the articlecan be utilized to determine whether the damage was caused by close-in explosion which aids in the determination of theblast center. After the column damage is categorized as close in or normal, the resulting rotation due to the blast of the col-umns can be measured and compared to figures given in the article to estimate the blast load.

The behavior of steel beams under blast loading has been the subject of several studies. Menkes and Opat were one of thefirst to explore the behavior of explosive loads on rigid end aluminum beams [32]. Although their research utilized alumi-num instead of steel, their results have successfully been used by others to model steel behavior [33–35]. The purpose oftheir study was an attempt to replicate the forces that may be experienced by a space vehicle re-entering the earth’s atmo-sphere. The authors’ experimental procedure involved the application of sheet explosives to aluminum beams of three dif-ferent thicknesses and at two different lengths. The beams were buffered with a neoprene pad to prevent spall and its effects.The results of testing showed that the failure modes of the beams varied as the moment of inertia of the beams increased.The authors grouped the failures into three categories:

I. Large inelastic deformation.II. Tearing (tensile failure) in outer fibers, at or over the support.

III. Transverse shear failure at the support.

Category I beams have the smallest moment of inertia and Category III had the largest. The authors also attempt to verifytheir results with early forms of finite element codes and Timoshenko beam theory. From the experiments and modelingresults the authors develop a descriptive evaluation for each failure category:

(a) For Mode I, the severity of the damage may be described by the residual central deflection. This depends on theimpulse intensity, the density and constitutive relations of the materials, the manner in which the impulse is parti-tioned (determining the initial velocity in the substrate), and the beam thicknesses and length.

(b) For Mode II, the threshold depends on the same variables cited in (a), with the exception of the length. It is charac-terized by a small tear in the top fiber.

(c) For Mode III, the threshold depends on the same variables as for Mode II, occurring at higher values of the impulseintensity. It is characterized by no appreciable plastic deformation. The initial velocity appears to be critical.

A diagram depicting these three failure modes is shown in Figure 1 of Ref. [36].In their 1993 study of blast loaded square plates, Olson and others determined that an additional failure mode occurred

beyond to the three failure modes mentioned above [36]. Through experimental testing of circular and rectangular plates,the authors confirmed that the plates behaved as predicted for Mode I. However, during the initiation of Mode II, when tear-ing begins along a side, the following was observed (see Figure 10 of Ref. [36]):

(i) Tearing occurred along one side, two sides or four sides before any of the corners are torn out. These cases are desig-nated as II-1, II-2 and II-4 respectively.

(ii) In the cases where some of the corners were torn out, the specimen rotated about the other corners.

Nurick and Shave later detail the relationship between the type of failure experienced in Mode II and the impulsereceived by the steel plates [37]. The authors introduced an additional Mode II behavior labeled Mode II�. The first phase

Fig. 4. Representative glass cracks due to blast loading.

A. Sorensen, W.L. McGill / Engineering Failure Analysis 18 (2011) 836–845 843

of Mode II� occurs when there is only partial tearing of the plate in addition to shear lift. Tearing begins at the center ofone side and moves towards the corners, while the other three sides exhibit shear lift. As the impulse increases, thenumber of torn sides increases, progressing from one side of tearing to two sides of tearing to three sides of tearing untilall four sides have torn, but not completely. The experimental results show that the tearing will begin simultaneously atthe center of each side, and for higher impulses the tearing will occur simultaneously along the central region of eachside. In most cases, the plate sides tore all the way into the corner, while in some cases, the plate sides tore across thecorner.

3.4. Non-structural glass

In his 1981 book, Explosion Investigation, Yallop notes that the breakage of glass windows presents a special case becausealmost every blast scene will have some sort of glass damage and most sites will have multiple if not hundreds of observ-ables [38]. The main parameters which determine the pressure at which a pane of glass will break are the area, the ratio ofheight to width, and the thickness. As previously discussed, once a glass window is shattered there is no way to determinethe exact pressure the window experienced. However, if a glass pane is only cracked, then a series of graphical diagramspresented by Main stone can be utilized by measuring the previously mentioned parameters and utilizing the diagrams[39]. Fig. 4 shows an example of cracked glass resulting from blast. Yallop also described a method to determine blast shapecharacteristics by identifying locations where glass panes have and have not shattered or cracked and graphing their loca-tions. By shading areas where cracked and shattered glass was observed, some idea of the pressure distribution causing theobserved damage can be inferred.

Loughran points out that care must be taken when identifying the cause of failure in glass as it possible that the failure isdue to thermal changes [40]. A thermal break in glass can be identified as the crack occurs at a right angle (90�) to the surfaceof the glass. This distinction is important because equations relating to the pressure at which glass breaks should not be ap-plied to breaks caused by thermal expansion.

Norville and Conrath present a series of papers on the utilization of ASTM codes F 2248-03 [41], and E 1300-04 [42], aswell as the Department of Defense (DOD) code UFC 4-010-01 [43,44]. Included in these papers and codes are diagrams thatrelate the static design load, standoff distance, and TNT load for glass during a 60 s and a 3 s duration (Figure 1 of Ref. [43]and Figure 7 of Ref. [44] respectively). These figures can be utilized in instances when an approximation of the distance fromthe center of the blast insult and knowledge of the flexural strength of the glazing panes are known. Furthermore, careshould be taken to insure that the panes are far enough away from the blast center so as not to have been affected bythe negative pressure phase as the negative pressure phase may have some affect on the glass pane depending on the sizeof the blast load and the distance [45]. The flexural strength of the glazing panes can be determined using charts taken fromASTM E 1300-02, ‘‘Standard Practice for Determining Load Resistance of Glass in Buildings’’, [46]. To utilize these charts, theglass thickness, support conditions, and unsupported lengths must be known.

3.5. Timber and other materials

In an US Army report of debris hazard from blast loaded plywood sheet closures, the author presents experimental resultsof blast loading of plywood sheet closures suitable for basement shelters [47]. The report presents pressure–time histories ofthe blast loading and the panel deflection that occurs as result of the loading forces. While timber products are not normallythought to be blast resistant, the author discovered that the ultimate strength under the blast load (load to break through ofpanels) was found to be about eight times the calculated allowable static load values. Bursting loads were found to vary fromabout 85.5 kPa (12.4 psi) for 1.27 cm (1/2 in.) thickness to 338 kPa (49 psi) for the 2.54 cm (1 in.) panels. The author notesthat almost no damage occurred to a test panel for loads just below the bursting pressure and that for pressures abovethe burst level, the panel would nearly always blow out. At loads much higher than the burst pressure it was almost impos-sible to find recognizable remnants of the panel. The report also presents extensive load–deflection curves and high speedcamera measurements of the test specimens, which can be used as references for investigators that observe blast damage toplywood panels.

The analysis of other timber products exposed to blast loadings for the purpose of identifying exposed pressures is lim-ited. This is due to the brittle nature of timber under this type of loading. However two results of this occurrence can be uti-lized: the location and direction of timber fragments, and the analysis of burn markings. In the previously mentioned text byYallop, the author discusses utilizing the burn markings on timber products to determine the blast and pressure directions[38]. However, the author notes that care must be taken to insure that the markings are a direct result of the blast wave andnot from subsequent fires.

One glaring omission from the review of existing literature is the behavior of non-load bearing engineered materials otherthan glass. These materials include heating and air conditioning duct work and equipment, appliances, large machinery, andother furniture. Yallop does specify some investigative measures for utilizing these items such as measuring the distancethat these objects may move during the blast event, or measuring the indention of a refrigerator door [38]; however, wewere unable to find any articles that present a detailed discussion on the behavior or utilization of these materials thatcan be used for investigative purposes.

Table 3Summary of failure modes and methods of analysis.

Column Beam Slab/plate Wall

Concrete Development of crack lines orformation of a crater/yield lineanalysis or equations based oncrater dimensions

Development of crack lines/yield lineanalysis

Masonry X X Bending of wall, formation of cracks,or crushing of units/measuredeflections, yield line analysis,utilization of equations

Steel Buckling of column/AISCbuckling equations;compare buckling shape toloading diagrams

Buckling of beams/AISCbuckling equations;compare buckling shape toloading diagrams

Buckling or tearing of plates/compare buckling shape to loadingdiagrams X

Glass X X Formation of cracks/comparison ofcracks to diagrams or utilization ofstandoff design charts X

Timber X Panel blow out and/or burst intounrecognizable fragments/staticstress analysis

844 A. Sorensen, W.L. McGill / Engineering Failure Analysis 18 (2011) 836–845

4. Summary and conclusions

A summary of the failure modes for each material and their structural use along with a summary of the analysis toolsdescribed in this review are presented in Table 3.

All observables discussed in the preceding section, if actually observed at a blast scene, constitutes a portion of the corpusof evidence available to post-blast investigators as they seek to understand and ascertain the details of an explosion event.We observe after this literature review that several materials/configurations are under-addressed in the literature, yet arejust as likely to be exposed to explosive environments as the other materials. These include RC and masonry columns, typicalsteel connections, other engineered non-load bearing materials such as mechanical duct work and appliances.

Other issues that are of importance to the post-blast investigator, yet not addressed in this paper, include techniques fordetermining whether observed damage was the result of the blast in question or whether it was pre-existing or due to othercauses (e.g., responder actions). Future work will develop methods for using observed structural damage at a blast scene toarrive at some conclusion of the size and center of the blast.

Acknowledgement

Funding for this research is provided by the Department of National Intelligence’s (DNI) Intelligence Community Postdoc-toral Fellowship Program.

References

[1] Hattwig M, Steen H. Handbook of explosion prevention and protection. Hoboken, NJ: Wiley; 2004.[2] Talbot J, Jakeman M. Security risk management body of knowledge. Hoboken, NJ: Wiley; 2009.[3] Carper K. In: Forensic engineering, editor. 2nd ed. Boston, MA: CRC Press; 2000.[4] Noon R. In: Forensic engineering investigation. Boston, MA: CRC Press; 2000.[5] Beveridge A. In: Forensic investigation of explosives. Boston, MA: CRC Press; 1998.[6] Thurman J. In: Practical bomb scene investigation. Boston, MA: CRC Press; 2006.[7] Ramabhushanam E, Lynch M. Structural assessment of bomb damage for world trade center. ASCE J Perform Construct Facil 1994;8(4):229–42.[8] Corley W, Mlakar P, Sozen M, Thornton C. The Oklahoma City bombing: summary and recommendations for multihazard mitigation. ASCE J Perform

Construct Facil 1998;12(3):100–12.[9] Mlakar P, Corley W, Sozen M, Thornton C. The Oklahoma City bombing: analysis of blast damage to the Murrah building. ASCE J Perform Construct Facil

1998;12(3):113–9.[10] Sozen M, Thornton C, Corley W, Mlakar P. The Oklahoma City bombing: structure and mechanisms of the Murrah building. ASCE J Perform Construct

Facil 1998;12(3):120–36.[11] Vaidogas E. Explosive damage to industrial buildings: assessment by resampling limited experimental data on blast loading. J Civ Eng Manage

2005;11(4):251–66.[12] Ambrosini D, Luccioni B, Jacinto A, Danesi R. Location and mass of explosive from structural damage. Eng Struct 2005;27:167–76.[13] Tagel-Din H, Rahman N. Simulation of the Alfred P. Murrah federal building collapse due to blast loads. In: Proceedings of the 2006 AEI conference,

Omaha, Nebrask, vol. 190(40798); 2006. p. 32–47.[14] Osteraas J. Murrah building bombing revisited: a qualitative assessment of blast damage and collapse patterns. ASCE J Perform Construct Facil

2006;20(4):330–5.[15] Byfield M. Behavior and design of commercial multistory buildings subjected to blast. ASCE J Perform Construct Facil 2006;20(4):324–9.

A. Sorensen, W.L. McGill / Engineering Failure Analysis 18 (2011) 836–845 845

[16] Nash P, Vallabhan C, Knight T. Spall damage to concrete walls from close-in cased and uncased explosions in air. ACI Struct J 1995;92(6):680–7.[17] Zineddin M, Krauthammer T. Dynamic response and behavior of reinforced concrete slabs under impact loading. Int J Impact Eng 2007;34:1517–34.[18] McVay M. Spall damage of concrete structures. Technical Report SL 88-22, US Army Corps of Engineers Waterways Experiment Station; 1988.[19] Xu K, Lu Y. Numerical simulation study of spallation in reinforced concrete plates subject to blast loading. Comput Struct 2006;84:431–8.[20] Biggs JM. Introduction to structural dynamics. New York, NY: McGraw-Hill; 1964.[21] Mays G, Hetherington J, Rose T. Resistance-deflection functions for concrete wall panels with openings. J Struct Eng 1996;124(5):579–87.[22] Luccioni B, Luege M. Concrete pavement slab under blast loads. Int J Impact Eng 2004;32:1248–66.[23] Naito C, Wheaton K. Blast assessment of load-bearing reinforced concrete shear walls. ASCE Pract Period Struct Des Construct 2006;11(2):112–21.[24] Zineddin M. Behavior of structural concrete slabs under localized impact. Doctoral Dissertation, University Park, PA: The Pennsylvania State

University; 2002.[25] Wesevich J, Oswald C. Empirical based concrete masonry-pressure impulse diagrams for varying degrees of damage. In: Metropolis & beyond,

proceedings of the 2005 structures congress and the 2005 forensic engineering symposium; 2005.[26] Moradi L, Davidson J, Dinan R. Resistance of membrane retrofit concrete masonry walls to lateral pressure. ASCE J Perform Construct Facil

2008;22(3):131–42.[27] Eurocode 6. Design of masonry structures – part 1-1: general rules for reinforced and unreinforced masonry structures. CEN European committee for

standardization, Ref. No. EN 1996-1-1; November 2005.[28] Garbin E, Galati N, Nanni A. Design guidelines for the strengthening of unreinforced masonry structures using glass grid reinforced polymers (GGRP)

systems. Technical Report Prepared for Bondo Inc. & TechFab LLC., University of Missouri-Rolla, Rolla, Missouri; 2005.[29] ACI 440M. Guide for the design and construction of externally bonded FRP system for strengthening unreinforced masonry structures. American

Concrete Institute, ACI Committee 440, Farmington Hills, MI; 2004.[30] AISC. In: Steel construction manual. Chicago, IL: American Institute of Steel Construction; 2005.[31] Godinho J, Montalva A, Gallant S. Analysis of steel columns for air-blast loads. Structure magazine. A Joint Publication of NCSEA, CASE, and SEI; 2007. p.

13–4.[32] Menkes S, Opat H. Tearing and shear failures in explosively loaded clamped beams. Exp Mech 1973;13:480–6.[33] Nurick G, Martin J. Deformation of thin plates subjected to impulsive loading – a review: part I – theoretical considerations, part II – experimental

studies. Int J Impact Eng 1989;8(2):159–86.[34] Nurick G, Shave G. The deformation and tearing of thin square plates subjected to impulsive loads – an experimental study. Int J Impact Eng

1996;18(1):99–116.[35] Rudrapatna N, Vaziri R, Olson M. Deformation and failure of blast-loaded square plates. Int J Impact Eng 1999;22:449–67.[36] Olson M, Nurick G, Fagnan J. Deformation and rupture of blast loaded square plates – predictions and experiments. Int J Impact Eng

1993;13(2):279–91.[37] Nurick G, Shave G. The deformation and tearing of thin square plates subjected to impulsive loads – an experimental study. Int J Impact Eng

1996;18(1):99–116.[38] Yallop HG. In: Explosion investigation. Edinburgh, Scotland: Forensic Science Society and Scottish Academic Press; 1980.[39] Mainstone R. The response of buildings to accidental explosions. Garston, Eng. Building Research Establishment. Current paper – Building Research

Establishment; 1976 [CP 24/76].[40] Loughran P. Falling glass. Berlin, Germany: Birkhäuser; 2003.[41] ASTM. Standard practice for specifying an equivalent 3-s durations design loading for blast resistant glazing fabricated with laminated glass. F 2248-

03, West Conshohocken, PA; 2003.[42] ASTM. Standard practice for determining the load resistance of glass in buildings. E 1300-04, West Conshohocken, PA; 2004.[43] Norville H, Conrat E. Considerations for blast resistant glazing design. ASCE J Archit Eng 2001;7(3):80–6.[44] Norville H, Conrath E. Blast-resistant glazing design. ASCE J Perform Construct Facil 2006;12(3):129–36.[45] Krauthammer T, Altenberg A. Negative phase blast effects on glass. Int J Impact Eng 2000;24:1–17.[46] ASTM. Standard practice for determining load resistance of glass in buildings. E 1300-02, West Conshohocken, PA; 2003.[47] Coulter, G. Debris hazard from blast loaded plywood sheet closures. Memorandum Report ARBRL-MR-02917, US Army Armament Research and

Development Command. Aberdeen Proving Ground, Maryland: Ballistic Research Laboratory; 1979.