Blast-Resistant Highway Bridges: Design and Detailing Guidelines (2010)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

82.1 Overview The purpose of conducting the research described in this report is to increase the security of the U.S. transportation in- frastructure by providing the maximum possible contribu- tion to the design and analysis of blast-resistant bridges given the available resources. Although the issues associated with various threats and the blast-resistant design of different bridge components will require many years of study, this work makes an important first step in the security of highway bridges by focusing on the areas of greatest need. The final work plan carefully considers information found in the literature, exten- sive past experience with blast testing and explosive effects on structures, and practical considerations concerning the im- portance of certain design issues and individual bridge com- ponents. This chapter describes the state of knowledge prior to the start of the current work and outlines the basis for the selected research plan. 2.2 Literature Review While the field of blast-resistant bridges is relatively new, information regarding the design of blast-resistant build- ings does exist in the literature, and a few recent and current studies do focus on bridges subjected to blast loads. Because sound research typically advances the state of knowledge by building on the foundation of the past, the literature review conducted for this project provides comprehensive coverage of basic blast-resistant design principles found in the litera- ture and recent research related to both blast-resistant build- ings and bridges, including an overview of seismic design and detailing provisions as they may apply to blast-loaded bridges. The review includes information regarding the analysis, design, and retrofit of structures to resist blast loads, along with infor- mation on the topics of bridge security, risk management, and vulnerability assessment. This chapter provides a brief summary of the most relevant information on a variety of topics relative to such a threat. As with any structural design, proper determination of the load and the resulting response is essential. To that end, the following section focuses prima- rily on airblast phenomenology and blast-wave propagation as they interact with structures, followed by current blast- resistant design guidelines and practices. Emphasis is given to reinforced concrete structures to be consistent with the focus of the test program. 2.2.1 Blast Loads and Shock Phenomena Three principal effects of an explosion are important for structural design: the total impulse, the peak pressure, and fragments (velocity, distribution, and mass). While the first two aspects of blast loading can usually be computed if the explosive type, explosive weight, explosive shape, and standoff distance are known, the load imparted to a structure by frag- ments is often difficult to quantify because fragments are typ- ically irregular in nature (Conrath et al., 1999). For terrorist threat scenarios involving vehicle-delivered bombs, however, fragment loads are negligible for structural design, though they still pose a human injury risk, because typical casings for terror- ist weapons (i.e., car or truck parts, sheet metal, and plastic) are not massive (especially relative to cased military weapons) (Conrath et al., 1999). Because the focus of this research is on protecting bridge components subjected to blast effects asso- ciated with terrorist events, this review focuses primarily on airblast loading from uncased charges with no fragments. The detonation of a high explosive is a high-rate chemical reaction producing a localized sudden release of energy that dissipates violently through a shock wave, which is a region of highly compressed air that radiates spherically away from an explosive source. This region of compressed air creates an over- pressure and a dynamic pressure as it passes by a given point in space. Idealized curves for overpressure and dynamic pressure are shown in Figure 1. The overpressure is the pressure result- ing from the explosion in excess of the ambient pressure, and C H A P T E R 2 Research Background

9the dynamic pressure is the pressure of the resulting air flow. At the arrival of the shock front, the overpressure rises nearly instantaneously to its peak before decreasing to zero, at which time a small negative overpressure (i.e., suction) occurs. The dynamic pressure increases nearly instantaneously with the ar- rival of the shock front and consists of a strong wind away from the explosion, then a weak wind toward the explosion, and then a very weak or feeble wind away from the explosion. Un- like the overpressure, which is a measure of air pressure rela- tive to the atmospheric pressure, the dynamic pressure always remains positive. There are two reasons for this phenomenon. The first is that the dynamic pressure is determined using the square of the wind velocity, making it positive regardless of the direction of the wind. The second is that the dynamic pressure is a measure of kinetic energy (i.e., “energy of motion”), which is a pressure without reference to another pressure. The incident wave is the term used to describe the shock wave that radiates spherically from an explosion. This incident wave will reflect off any surface in its path, and the term reflected wave is used to describe the wave that reflects off a surface. The reflected wave travels at a higher velocity than the incident wave because it travels through air that has already been heated and compressed. As a result, waves reflecting off rigid surfaces (e.g., the ground) can potentially catch up to and merge with the in- cident wave to create a single wave front called a Mach front (Figure 2). At any point prior to the joining of the incident and reflected waves, two independent pressure peaks will occur, each of which is smaller than that of the single Mach front pressure amplitude. The point at which the incident and re- flected waves merge is known as the triple point. An example of the two independent pressure peaks prior to the formation of the Mach front is shown in Figure 3. Figure 3a shows a W IN D R EV ER SA L W IN D R EV ER SA L A IR F LO W C EA SE S DYNAMIC PRESSURE POSITIVE PHASE NEGATIVE PHASE TIME SH O CK F RO N T A RR IV ES OVERPRESSURE POSITIVE PHASE NEGATIVE PHASE COMPRESSION TIME ATMOSPHERIC PRESSURE SUCTION A M BI EN T PR ES SU RE R ES TO RE D D Y N A M IC P RE SS U RE O V ER PR ES SU RE + 0 - STRONG WIND AWAY FROM EXPLOSION (DECREASING TO AERO) WEAK WIND TOWARD EXPLOSION FEEBLE WIND AWAY FROM EXPLOSION (tq+) (tp+) Figure 1. Overpressure and dynamic pressure variation with time (Glasstone and Dolan, 1997).

10 pressure–time history of a shock wave after the reflected wave and incident wave merge, forming a single incident wave, and Figure 3b shows a pressure–time history before the two waves merge, in which the separate incident and reflected waves are visible. The degree to which a wave reflects depends on the ter- rain over which the blast wave travels. Hilly land masses can in- crease blast effects in some areas but decrease them in others. All pressure–time histories, except those very close to the detonation, have the same general assumed form shown in Figure 4, while the exact values that define the curve vary de- pending on the size of the explosive charge and the location of interest. For example, the peak pressure, Pso, decreases signifi- cantly with standoff distance, while the positive phase dura- tion, to, increases with standoff. By definition, the impulse is equal to the area under the pressure–time history curve. Different charge weights and standoff distances scale to cre- ate similar shock waves. “Self-similar blast waves are produced at identical scaled distances when two explosive charges of the same explosive material with similar geometry but of different weights are detonated in the same atmosphere” (Conrath et al., 1999). Scaling equations relate the parameters needed to define the curve in Figure 4, and the most common scaling relationship is the Hopkinson-Cranz or “Cube-Root” scaling law (Conrath et al., 1999), which is shown in Equation 1. where: Z = scaled standoff (ft/lb1/3) R = standoff, distance between center of blast source and target (ft) WTNT = charge weight of explosive (lb equivalent TNT) The charge weight, WTNT, is in terms of a TNT-equivalent charge weight. TNT equivalencies relate the energy output Z R WTNT = 1 3 1( ) Slant Distance, R Angle of Incidence, α H Ground Zero Path of Triple Pont Incident Wave Reflected Wave Mach Front Shelter HT RG Ground Surface Figure 2. Unconfined air burst showing formation of Mach front (Department of the Army, 1990). PSO (INCIDENT OVERPRESSURE) PS (INCIDENT OVERPRESSURE AT MACH FRONT) O V ER PR ES SU RE O V ER PR ES SU RE Pr (REFLECTED OVERPRESSURE) (a) POINT AT MACH FRONT (b) POINT ABOVE MACH FRONT Figure 3. Two graphs showing difference in overpressure before and after formation of Mach front (Department of the Army, 1990).

of common explosives to that of TNT. In reality, a high ex- plosive’s TNT equivalency varies as a function of standoff, explosive geometry, target orientation, and atmospheric conditions. For the purposes of design, however, a given explosive’s TNT equivalency is treated as a constant, which is the standard to which much of the available data have been reported. The Hopkinson-Cranz scaling law presented above has been verified experimentally using TNT equivalencies for scaled distances greater than 0.4 ft/lb1/3 and may not be valid below this value. The Hopkinson-Cranz scaling law enables the prediction of blast-load parameters for full-scale explosions using data from smaller-scale tests, and an engineer can create an ideal- ized pressure–time curve for a free-air explosion, shown in Figure 4, with the TNT equivalent charge weight and stand- off distance. A hemispherical burst on a perfect reflecting sur- face will double the effective charge weight, while a reflection factor of 1.8 is more realistic when significant ground cratering is present (Conrath et al., 1999). Commonly used “standard” airblast curves, often referred to as “spaghetti charts,” exist to predict the parameters required to define an idealized blast wave; Figure 5 shows one of these charts. A vast collection of empirical data and theoretical predictions provides the basis for these charts, and curves exist for both hemispherical and spherical “free-field” bursts. The transmission of a shock front through a fluid (i.e., air) is a nonlinear process, and the interaction of a blast wave with a structure is a complex problem leading to significantly mag- nified pressures and impulses. Figure 6 illustrates a blast wave reflecting off a structure. The magnification of the reflected pressure and impulse will vary depending on the magnitude of the peak incident overpressure and the orientation and location of the structure relative to the explosion source. Military design manuals (Department of the Army, 1986; Department of the Army, 1990) contain empirically derived Pso Pso- Po Pressure Time After Explosion to to- Positive Specific Impulse Negative Specific Impulse tA Figure 4. Idealized pressure–time curve for free-air explosion (Department of the Army, 1990). Notes: Pso = Peak positive incident pressure (psi) Pr = Peak positive normal reflected pressure (psi) is/W 1/3 = Scaled unit positive incident impulse (psi-ms/lb1/3) ir/W 1/3 = Scaled unit positive normal reflected impulse (psi-ms/lb1/3) tA/W 1/3 = Scaled time of arrival of blast wave (ms/lb1/3) to/W 1/3 = Scaled positive duration of positive phase (ms/lb1/3) U = Shock front velocity (ft/ms) W = Charge weight Lw/w 1/3 = Scaled wavelength of positive phase (ft/lb1/3) Scaled standoff Z = R/w1/3 L w /W 1/ 3 , P r , P s o , ir/ W 1/ 3 , i s/ W 1/ 3 , U t A /W 1/ 3 , t o /W 1/ 3 Figure 5. Positive phase airblast parameters for hemispherical surface TNT detonation at sea level (Department of the Army, 1990). 11

12 curves that provide the magnified values of the reflected pres- sure and reflected impulse as functions of the peak incident overpressure, the angle of incidence (i.e., the angle between a shock wave’s direction of travel and a structure’s surface normal vector), and charge weight. Clearing effects on the front surface of a structure decrease the reflected pressure near free edges. When clearing occurs, the reflected pressure seeks relief toward the lower pressure regions at the free edges, forming a rarefaction (or relief) wave that propagates from the low-pressure region at the free edges to the high-pressure region at the middle of the surface. Therefore, the pressure differential between the free edge and front face causes the pressure at point B in Figure 6 to dissi- pate faster than at point A. According to TM 5-855-1 (Depart- ment of the Army, 1986), the clearing time at a given point on a reflective surface is the time required for the reflected pressure to dissipate from that point, and that time is given by Equation 2 (which is also known as the three-transits-to- edge-rule). where: tc = clearing time(s) Sx = distance from nearest free edge to point of interest (ft) Us = shock front velocity (ft/s) The type of blast load a structure must resist depends on the design threat and the type of structure. When a charge is detonated extremely close to a structure, it imposes a highly impulsive, high-intensity pressure load in a localized region of the structure. When a charge is detonated farther away, it produces a lower-intensity, longer-duration uniform pres- sure distribution over the entire structure. As the standoff increases, the pressure distribution over the surface becomes t S U c x s = 3 2( ) more uniform (Department of the Army, 1990). Because the shape and intensity of the loading can vary depending on the charge weight and standoff distance, three blast-loading cat- egories exist: contact, close-in, and plane-wave. Figure 7 illus- trates these categories (Department of the Army, 1990). A contact blast load consists of a high-intensity, non-uniform load where breaching is a typical response. Breach is defined as the complete loss of concrete through the depth of a cross- section. A close-in blast load is the result of a spherical shock wave striking a structure to produce a non-uniform load and StructureFree Edge Free Edge SB SA B A Reflected Pressure Incident Pressure Point A Point BPr Po Pr es su re Time tc = 3SA/US Pso Pr Po Pr es su re Time tc= 3SB/US Pso Reflected Pressure Dynamic + Incident Pressure Reflected Pressure for Case of Infinite Reflecting Plane Figure 6. Blast wave reflecting off structure (U.S. Army Engineer Research and Development Center, 2003). Structure Structure Structure (a) (b) (c) Figure 7. Blast-loading categories: a) contact, b) close-in, c) plane-wave (Departments of the Army, Air Force, and Navy and the Defense Special Weapons Agency, 2002).

an impulse-dominated response. A plane-wave blast load is a far-field explosion that produces essentially planar waves and a uniform load on each surface when it reaches a structure. Additional load categories include unconfined free air burst (no immediate wave amplification), unconfined air burst (amplification from ground reflections), unconfined surface burst (amplification from ground reflections), confined fully vented explosion (one or more surfaces open to atmosphere), partially confined explosion (limited openings), and fully con- fined explosions. Most bridge loads will be either an uncon- fined air burst, unconfined surface burst, or a confined fully vented explosion (e.g., under the deck between girders). 2.2.2 Structural Response to Blast Loads While the response of bridges to terrorist explosive threats is a relatively new topic for the structural engineering com- munity, several observations from past incidents involving the performance of buildings during terrorist attacks and the response of bridges to cased military weapons can be made regarding the general expected response of blast-loaded bridges. Explosions located beneath a bridge deck will cause large uplift forces, and pressure buildup between girders and near the abutments can greatly amplify the applied load, as shown in Figure 8. Detonations directly between girders or at the abutments can generate extremely severe loads on the girders and deck, and underwater explosions can result in large water plumes that cause damage to structural compo- nents of bridges crossing waterways. When designing for any explosion beneath the deck, it is best to sacrifice the deck and focus on saving the girders and columns. Longer girders typically are more resilient than shorter girders because of their greater mass, strength, and flexibility. Two cost-effective ways to strengthen bridges for blast loads are to provide continuous reinforcement in the tops of concrete girders for uplift resistance and to provide stiffeners for steel girders to prevent local buckling. In addi- tion, using hinge restrainers or extended column seats can possibly prevent girders from falling off of the piers in the event of large deformations resulting from blast loads. Columns that are rigidly connected to the superstructure can experience tensile forces due to the uplift of girders dur- ing below-deck explosions. During above-deck explosions, columns can experience increased axial loads due to blast forces in addition to gravity loads. Although recent attacks in Iraq may suggest otherwise (ABC News, 2007), prior obser- vations from military operations indicate that most sub- structures are generally large enough to withstand anticipated above-deck explosions (Winget, 2003). Columns will experience significant shear forces from close-in charges for below-deck scenarios, and proper de- sign requires enough transverse reinforcement to prevent shear failure and force a flexural failure. “When an explo- sion occurs below the deck of a bridge, bents and piers will be subjected to large lateral forces, possibly resulting in large deformations, shear, or flexural failures. Additionally, con- crete cratering and spalling from the blast-wave impact may lead to significant losses of concrete, especially if the stand- off distance is small” (Winget et al., 2004). Columns should be designed for lateral blast loads and resulting deflections in addition to the axial loads present due to gravity. Most bridge columns have much greater axial capacity than axial demand, however, and they experience service axial loads that are below their balance point. Therefore, ignoring the axial load is usually a conservative assumption for typical bridge columns subjected to lateral blast loads because the inclusion of axial loads will typically increase both shear and flexural resistance and improve performance. An exception to this guideline would be for cases in which the axial load is in excess of the balance point load and/or P-Δ (i.e., second- order) effects are significant, as they can be in tall, slender columns. Very close-in or contact blasts create high-intensity blast pressures that often cause breaching (i.e., section loss) of a column, spalling of concrete cover from high-intensity blast pressures, back-face scabbing of concrete cover due to large deflections, and post-failure fragments. While blasts directly aimed at bridge structural members may be the first choice for a terrorist attack, military records indicate that large explo- sions near the columns create ground craters that can cause foundation instability, column failure, and bridge collapse. As a result, these military records state that it may be easier to “shake down the columns” than to “shoot up the superstruc- ture” (Bulson, 1997). 2.2.3 Dynamic Material Strength and Strain-Rate Effects The materials used in reinforced concrete construction have response characteristics that depend on the rate of load- ing. Most of the data supporting this topic are experimental; therefore, the physical cause is not completely understood. Incident wave Reflected wave Figure 8. Blast-wave propagation during below-deck explosion (Winget et al., 2005). 13

14 The widely held belief is that crack propagation occurs at a limiting velocity and cannot crack fast enough to keep up with a high rate of loading, and thus concrete exhibits a strain-rate threshold where concrete strength increases significantly above that value (Tedesco, 1999). Figure 9 shows the relationship between strain rate and concrete strength for tension and compression. The strain-rate threshold for concrete response in tension is approximately 1 to 10 per second, and the strain- rate threshold for compression is approximately 50 to 80 per second. Above these threshold values, concrete demonstrates significantly higher strengths than under static loading con- ditions. The elastic modulus is not as strain-rate sensitive as strength, and as a result, the strain rate can easily be converted to a stress rate (Tedesco, 1999). High rates of loading also affect the mechanical properties of ductile metals, as they also cannot deform fast enough to keep up with extreme loads. Therefore, ductile metals, like steel, have a limiting deformation velocity that results in material strength increases with increasing strain rates. This limiting velocity occurs at a much higher strain rate than concrete, and various relationships between strain rate and strength for unspecified metals are shown in Figure 10 (Tedesco, 1999). The yield strength and ultimate tensile strength increase substantially, while the modulus and the elongation at rupture largely do not change (Tedesco, 1999). Techniques exist to calculate the strength increase for both concrete and steel at a given strain rate. Dynamic increase fac- tors (DIFs), which are defined as the ratio of dynamic material strength to static material strength, are used to account for the high strain rates present in blast events. Table 1 lists DIFs for reinforced concrete design, which are dependent on the Figure 9. Concrete strain-rate effects on strength (Tedesco, 1999). Figure 10. Effects of strain rate on yield stress for various metals (Tedesco, 1999). type of stress (i.e., flexure or shear). These dynamic increase factors are used to increase material strengths over those that are measured under standard static testing protocols to account for the actual strength expected to be present under various blast scenarios. These values have been derived empir- ically from blast-test experiments (Department of the Army,

1990). For design, DIFs are typically assumed to be constant, even though they vary with strain rate. Strength and age increase factors are used to determine realistic material properties under dynamic loads if the actual material strengths are unknown. Strength increase factors (SIFs) are used to account for actual material (con- crete and steel) strengths in excess of the specified design values, as shown in Table 2. In addition, age increase factors are used with concrete to account for strength gains beyond 28 days. From empirical data collected by the Department of the Army (1986), an age increase factor of 1.1 is specified for concrete less than six months old, and a value of 1.15 is specified otherwise. 2.2.4 Application of Seismic Design to Blast-Loaded Bridges Seismic loads are similar to blast loads in that both produce dynamic structural response and often lead to large deforma- tions with inelastic material response. The subsections below compare blast and seismic loads and response characteris- tics to determine relevant design concepts. A summary of current seismic design practice and research to investigate whether or not seismic design principles can provide ade- quate structural resistance for blast-loaded structures is also provided. 2.2.4.1 Comparison of Blast and Seismic Designs Seismic and blast loads are both time dependent, and both induce a dynamic structural response that generally results in inelastic behavior. While the allowable damage to both structural and nonstructural components depends on the structure’s purpose for both blast and seismic loads, life safety is usually more important than the prevention of structural damage for most designs. Designs for both seismic and blast loads allow large inelastic deformations to help dissipate en- ergy, and improving strength, ductility, redundancy, and con- nection capacity in any structure can improve the performance under both seismic and blast loads. Blast and seismic loads, however, have fundamental dif- ferences that prohibit the direct application of all seismic detailing and design requirements to blast-loaded struc- tures. Blast loads have a higher amplitude and shorter dura- tion than seismic events. Blast-load duration is measured in milliseconds, approximately one thousand times shorter than that of an earthquake (Conrath et al., 1999). Due to the uncertainty associated with selecting an appropriate design threat, the magnitude of a blast load is more difficult to predict than seismic loads and is independent of geograph- ical location, unlike earthquakes, as shown in Figure 11. A blast load radiates in all directions from the source, creating a complex pressure–time history that varies according to the location of a structure, while lateral load effects will domi- nate a structure experiencing a seismic event. Also, an earth- quake is a widespread event, while blast effects typically remain local. Therefore, seismic design and detailing should not be assumed to provide adequate protection for blast- loaded structures. According to the National Research Council report ISC Security Design Criteria for New Federal Office Buildings and Major Renovations (2003), “Although design for seismic resistance and design for blast resistance share some common principles, the two types of design must not be mistakenly viewed as redundant.” The report also states, “Attempts to link seismic and blast design re- quirements by simply comparing the lateral or shear forces on a structure produced by these events (an equivalent seis- Concrete Concrete fdu/fufdu/fufdy /fy fdy /fy f'dc/f'cf'dc/f'c Flexure 1.17 1.05 1.19 1.23 1.05 1.25 Diagonal Tension 1.00 1.00 1.00 1.10 1.00 1.00 Direct Shear 1.10 1.00 1.10 1.10 1.00 1.10 Bond 1.17 1.05 1.00 1.23 1.05 1.00 Compression 1.10 1.00 1.12 1.13 1.00 1.16 *Far Design Range: Z 2.5 ft/lb1/3 †Close-in Design Range: Z < 1.0 ft/lb1/3 Source: Department of the Army, 1990 Stress Type Far Design Range* Close-in Design Range† Reinforcing Bars Reinforcing Bars Table 1. Dynamic increase factors. Material SIF Structural (f y 50 ksi) 1.10 Reinforcing Steel (f y 60 ksi) 1.10 Cold-Formed Steel 1.21 Concrete* 1.00 * The results of compression tests are usually well above the specified concrete strengths and may be used in lieu of the above factor. Some conservatism may be warranted because concrete strengths have more influence on shear design than bending capacity. TM 5-1300 specifies a SIF of 1.10. Source: American Society of Civil Engineers, 1997 Table 2. Strength increase factors. 15

16 mic base shear) perpetuate the erroneous impression that seismic design is an umbrella for blast resistance.” 2.2.4.2 Seismic Design Guidelines The AASHTO LRFD Bridge Design Specifications (AASHTO, 2007) specify seismic design and detailing provisions for bridges. The Caltrans Seismic Design Criteria (Caltrans, 2003) and Bridge Design Specifications (Caltrans, 2006) are addi- tional resources for design and detailing requirements. The essential requirement for a reinforced concrete column sub- jected to a strong ground motion is that it retain a substantial portion of its strength as it experiences severe loading rever- sals into the nonlinear range of response (Pujol et al., 2000). Current seismic design procedures specify a “performance- based” or “limit-state” design. In general, structural per- formance criteria are available for not only the traditional life-safety level but also for more restrictive design levels, such as serviceability and damage control. Quantitatively, “service- ability” implies that a structure will not need repair after an earthquake, while “damage control” implies that only re- pairable damage occurs (Kowalsky, 2000). In general, “it is uneconomical to design structures to withstand lateral forces corresponding to full elastic response to design-level earth- quakes. The alternative, and widely accepted approach, is to design for a lower force level and detail structure for ductil- ity” to ensure that it can sustain the inelastic displacements associated with seismic loads without significant strength degradation (Priestley and Park, 1987). Recognizing the difficulty in determining the actual maxi- mum shear that a critical column may experience during an earthquake, AASHTO LRFD recommends a plastic hinge analysis for seismic design. A plastic hinge analysis considers all potential plastic hinge locations to determine the maxi- mum possible shear demand. A typical seismic column with fixed supports and a displacement at one end due to a later- ally applied force will form two plastic hinges, as shown in Figure 12. This concept, which ensures that a member will have sufficient shear capacity to allow a ductile flexural failure mech- anism to form, is one that applies equally well to structures sub- jected to blast loads. Therefore, a plastic hinge analysis using a blast-load distribution is an ideal method to determine the maximum shear demand in blast-resistant design. 2.2.4.3 Seismic Detailing Guidelines The ACI Building Code (2005) and AASHTO LRFD Spec- ifications (2007) provide extensive detailing guidelines for seismically loaded concrete members in buildings and bridges. According to Sezen and Moehle (2006), “surveys of reinforced concrete building collapses in past earthquakes identify column failures as the primary cause. Such failures are com- monly attributed to widely spaced and poorly anchored trans- verse reinforcement.” Therefore, the amount and anchorage of transverse reinforcement in seismically loaded columns requires further research to determine their applicability to blast-resistant design. “Earthquakes and laboratory experience show that columns with inadequate transverse reinforcement are vulnerable to damage including shear and axial load failure” (Sezen and Moehle, 2006). Sezen tested four full-scale reinforced concrete columns with light transverse reinforcement to collapse under a simulated seismic loading. Figure 13 shows the crack pattern of one of the specimens, which illustrates the forma- tion of plastic hinges at the top and bottom of the column as load increased. After the flexural strength was reached, deterioration of the cross-section due to a lack of sufficient transverse re- %g contour labels %g contours %g intervals > 40 20-40 10-20 7-10 3-7 2-3 1-2 0-1 States Legend Figure 11. Seismic hazard map: peak ground acceleration (USGS, 2008).

inforcement triggered a shear failure, as shown in Figure 14. Equations 3, 4, and 5, from the AASHTO LRFD Specifications (2007) seismic provisions, specify a minimum volumetric reinforcement ratio and minimum area of transverse re- inforcement to help provide adequate core confinement for circular and rectangular columns, respectively. A sh f f sh c c y ≥ ′0 12 4. ( ) ρs c y f f ≥ ′0 12 3. ( ) where: f ′c = specified compressive strength of concrete at 28 days (psi) fy = yield strength of reinforcing bars (psi) s = vertical spacing of hoops, not exceeding 4 in. (in.) hc = core dimension of column in the direction under con- sideration (in.) Ag = gross cross-sectional area (in.2) Ac = area of concrete core (in.2) A sh f f A A sh c c y g c ≥ ′ −⎛⎝⎜ ⎞ ⎠⎟0 30 1 5. ( ) (a) (b) (c) INFLECTION POINT PLASTIC HINGE PLASTIC HINGE L ˜ 0. 5L MP MP DISPLACEMENT DUE TO SEISMIC LOADS DISPLACEMENT DUE TO SEISMIC LOADS Figure 12. Plastic hinge analysis for seismic column: a) deflected shape, b) plastic hinge locations, c) plastic moment. Figure 13. Crack pattern for specimen 1 (Sezen and Moehle, 2006). 17

18 The seismic provisions require more transverse reinforce- ment than a typical gravity-loaded column “to ensure that the axial load carried by the column after spalling of the con- crete cover will at least equal the load carried before spalling and to ensure that buckling of the longitudinal reinforcement is prevented” (AASHTO, 2007). Thus, the spacing of trans- verse reinforcement is important for shear resistance and core confinement in the plastic hinge regions of a seismically loaded column. Likewise, blast-loaded columns may require increased transverse reinforcement to ensure ductile behavior. Recent work by Bae and Bayrak (2008) on the seismic per- formance of full-scale, reinforced concrete columns is the first to demonstrate the opening of seismic discrete ties using hooks with a 135°-bend, plus an extension of 8.0 db, as shown in Figure 15. AASHTO LRFD Section 5.10.2.2 defines seismic hooks as a “135°-bend, plus an extension of not less than the larger of 6.0 db or 3 in.” According to the researchers, “unlike the full-scale concrete columns, the hooked anchorages often reach close to the center of the core concrete in scaled column specimens.” The remaining specimens in the research study used a minimum hook length of 15.0 db to prevent the open- ing of hoops. The larger “hook length proved to be very effec- tive, and opening of the 135° hooked anchorages of the ties was not observed in the other tests.” To avoid anchorage pull- outs and to improve the performance of seismically loaded (and blast-loaded) columns with discrete hoops or ties, longer hook lengths than currently specified should be used. 2.2.5 Design Issues The structural engineering community has gained much ground in the field of blast-resistant structural design. Despite this progress, however, the bridge engineering community still lacks design guidelines specific to blast-resistant highway bridges. The probability for a specific type of attack against a specific bridge is usually very low; however, “a low probabil- (a) (b) Figure 14. Damage after failure: a) specimen 1, b) specimen 3 (Sezen and Moehle, 2006). Figure 15. Opening of discrete ties (Bae and Bayrak, 2008).

ity of occurrence does not justify minimizing our efforts to manage the potential adverse effects of a catastrophic event” (Williamson and Winget, 2005). Several papers have been writ- ten regarding the risk management, analysis, and design of crit- ical bridges subjected to blast loads. The findings from these papers have been summarized in this section and can provide guidance for those officials considering the blast-resistant design of highway bridges. An important aspect of designing bridges for security in an economically feasible way is to have in place plans for evalu- ating the criticality of any one structure on the transportation network. Thus, in deciding how to allocate resources, bridges considered more essential to the transportation infrastructure, or those thought to be at higher risk for a terrorist attack, should be given priority in the implementation of protective measures over other, less critical bridges. The references contained in this section of the review describe methods of carrying out threat and vulnerability analyses and risk assess- ments. Once the risks to a given bridge have been assessed, measures may need to be taken to mitigate these risks if they are deemed unacceptable. These measures generally attempt to deter an attack by increasing surveillance or limiting access, but they can also include actions to limit the effects of blast loads or procedures to aid in rescue and recovery. Usually, deterrence and prevention measures will provide the least expensive solution to mitigate risk initially. Therefore, a risk manager should consider implementing these measures for short-term risks before hardening a structure is specified. Deterrence and prevention, however, may not always provide the most cost-effective solution for long-term risks when con- sidering lifetime costs, such as maintenance, replacement, personnel, and surveillance costs. 2.2.5.1 Risk Assessment Designing or retrofitting all bridges to resist extreme loads is cost prohibitive, and engineers need a method to identify credible threats, prioritize assets, and manage risk. Although blast-resistant bridge design is a relatively new topic in the field of structural engineering, many state agencies and re- search organizations already have strategies for risk assess- ment and management. While each approach differs slightly in implementation, all sources provide the same general guid- ance. This section outlines a risk assessment strategy based on a compilation of relevant sources, and the following section provides a risk management method based on a comprehen- sive literature review. Publications by Abramson et al. (1999), Rummel et al. (2002), and SAIC (2002) provide valuable information regarding risk assessment and management. A report by Williamson and Winget (2005) outlines a comprehensive approach to risk assessment, and the method they propose is a compilation of the best practices found in the literature. Step one of the risk assessment process is to identify all criti- cal assets within a jurisdiction and determine the criticality of these assets based on function, average daily traffic, access to populated areas, access to emergency and medical facilities, military importance (Strategic Highway Network), impor- tance to commerce, international border access, symbolic importance, availability of detours, presence of utility lines, and estimated repair time. The risk assessor should weight each criterion separately to reflect its importance within the jurisdiction, while not double counting for related criteria (e.g., average daily traffic may relate to major trade routes or availability of detours). This criticality score represents the importance of a bridge and indirectly captures the conse- quences of a potential attack (i.e., the consequences of an attack on a very important bridge likely will be great), and it is important to note that it should not include vulnerabilities to an attack or specific structural weaknesses because a later step considers these issues. Additionally, certain criteria may warrant a bridge’s placement in a higher importance category despite its overall criticality score. Examples include signature bridges whose failure may cause great socio-economic harm and bridges with significant military importance that have no detour capable of carrying the required traffic. Step two consists of identifying all possible internal and external threats to the critical bridges identified in step one. Internal and external threats can be a variety of actions such as the threat of earthquakes, high winds, or fire; however, the proposed risk assessment process only considers terrorist actions. Examples of potential attacks include, but are not limited to, a vehicle-delivered bomb on a superstructure, a vehicle- or maritime vessel-delivered impact and bomb against a column, hand-placed explosives between girders or inside box girders, or a series of timed events incorporating some or all of the above. As previously mentioned, designing a bridge to resist all possible threats is not feasible, and the risk manager should identify the most likely threat scenarios. Although terrorist activity is uncertain, a threat-point-of- view analysis can provide insight on the most likely terrorist threats. This analysis considers factors such as the terrorists’ potential objectives, available resources, availability of targets, and the impact of a successful attack. Once all possible terror- ist actions have been determined assuming that a terrorist has no limitations, each action should receive a score that indicates the relative probability of that action occurring compared to other actions. A multiplication decision matrix is best for this process, and the risk manager should conservatively assume that terrorists are experts in demolition, have structural engi- neering experience, and will encounter no resistance. In step three, the risk manager formulates the potential scenarios by pairing critical assets identified in step one with potential threats identified in step two. Once scenarios are 19

20 formulated, the process includes determining the probability of each event and assessing the vulnerability of assets with each scenario. Step four consists of assessing the consequences of each attack scenario, and the risk manager should assume the worst-case consequences of an attack not considering poten- tial mitigation measures. Potential consequences include, but are not limited to, loss of life, severe injuries, loss of bridge function due to structural damage, and financial losses. Once each scenario (i.e., combination of threat and asset) is identi- fied, a risk manager should categorize each scenario according to the probability of successful occurrence and the severity of impact. The probability of occurrence is subjective and comes from the threat-point-of-view analysis, and the criticality score is the basis for the severity of impact. All information is then combined in tabular format to determine which scenarios have the greatest risk and therefore require the most attention. An example of this table can be seen in Figure 16. Each scenario can be placed in one of the boxes based on severity of impact and probability of successful occurrence. Those bridges that fall in the severe and high range will receive the most attention. 2.2.5.2 Risk Management Once the risks are assessed, measures must be taken through a risk-management process to mitigate the risks to a level that is appropriate and economically feasible. The first step of risk management is to identify potential countermeasures avail- able to mitigate the risks previously identified. Such counter- measures as deterrence, detection, or defense can reduce the probability of occurrence, while others can lessen the sever- ity of the consequences through methods such as structural hardening, warning devices that indicate failure, or emergency operations planning. Additional considerations for selecting countermeasures include resource availability, implementa- tion difficulty, level of inconvenience, adverse environmental effects, adverse effects on serviceability, or usefulness. The second step of the risk-management phase is to deter- mine the costs for each countermeasure considered. Cost considerations should include initial purchase, installation, maintenance, replacement, and service life. Step three consists of a cost–benefit analysis to determine which countermeasures would be the most effective and efficient. Williamson and Winget (2005) recommend that the benefits be in terms of risk mitigation achieved and that a countermeasure summary sheet be used. An example of a countermeasure summary sheet can be seen in Figure 17. Because some countermeasures may also reduce other risks [e.g., fiber reinforced polymer (FRP) wrapping can reduce the risk of failure due to both seismic and blast events], the countermeasure benefits should be con- sidered during the design process for all risks associated with the bridge under consideration in order to get a complete picture. The goal of step three is to ensure the maximum pro- Severity of Impact Threat Scenario Categories Catastrophic (Criticality > 75) Very Serious (Criticality 51 - 75) Moderately Serious (Criticality 26 - 50) Not Serious (Criticality < 25) Highly Probable Severe Severe High Moderate Moderately Probable Severe High Moderate Low Slightly Probable High Moderate Low Low Pr ob ab ilit y of S uc ce ss fu l O cc ur re nc e Improbable Moderate Low Low Low Figure 16. Threat scenario categories (Williamson and Winget, 2005).

tection for all assets or the asset under consideration given the available resources. Prioritizing bridge importance may assist in allocating scarce resources among bridges. Step four consists of implementing the countermeasures and reassessing the risk with the countermeasures in place. If the countermeasures do not reduce the risk to an acceptable level, the scenario may re- quire additional countermeasures, or senior managers will need to accept the risk to an asset until additional resources are available. It is important to note that no level of mitigation will completely eliminate all risk, and officials will need to deter- mine the amount of risk they are willing to accept. The fifth and final step of risk management is monitoring the effectiveness of the countermeasure(s) for future decisions and using this information to guide future risk-management decisions. 2.2.5.3 Characterization of Analysis Methods for Blast-Loaded Structures Blast prediction techniques are often separated into load determination and response determination methods, and both of these categories can be further divided into two groups: first-principle and semi-empirical methods (National Research Council, 1995). First-principle methods solve sys- tems of equations based on the basic laws of physics. Accurate predictions of blast load and response can be obtained with these methods if the equations are solved correctly. Although first-principle programs use fundamental laws of physics and constitutive laws of materials, they have several limitations that are difficult to overcome without the use of empirical models. Blast propagation in real scenarios can be compli- cated by such things as atmospheric conditions, boundary effects, explosive material inhomogeneities and rates of re- action, as well as many other parameters, and first-principle methods cannot easily account for these factors (National Research Council, 1995). Additionally, the calculation of changes in blast pressure due to large structural deformations and localized failures can be quite problematic because accu- rate constitutive equations for materials responding in this range are not readily available. Moreover, because of the highly nonlinear nature of structural response to blast loads, an ana- lyst using first-principle methods to compute behavior should validate any predictions with actual experimental results to ensure that the methods are being implemented correctly. It can be very difficult, however, to find validated first-principle models due to a lack of experimental data available in the pub- lic domain, and any validation applies only to the specific sce- narios that were experimentally considered (National Research Council, 1995). Despite these limitations, response predic- tions based on first-principle results can be developed when a lack of applicable data exist, but interpretation of the results requires engineering judgment and experience. Semi-empirical models, in contrast to first-principle mod- els, utilize extensive data from past experiments. As a result, they require less computational effort and are generally preferable over first-principle programs. However, because a lack of experimental results for responses to blast loads exists in the public domain and because semi-empirical pro- grams are only slowly becoming available to the general engi- neering community, structural engineers must often rely on first-principle methods and good engineering judgment to determine blast effects (National Research Council, 1995). In addition, semi-empirical models are often valid only for Function/Effectiveness Costs Per Year Countermeasure D et er re nc e D et ec t D ef en d R ed uc e Im pa ct Ca pi ta l O pe ra tin g M ai nt en an ce Countermeasure 1 M L L $ $ $ Countermeasure 2 M H $ $ $ Countermeasure 3 H $ $ $ Countermeasure 4 L H $ $ $ L = Low Effectiveness M = Medium Effectiveness H = High Effectiveness Source: Modified from SAIC A Guide to Highway Vulnerability Assessment for Critical Asset Identification and Protection. Figure 17. Countermeasure summary sheet (Williamson and Winget, 2005). 21

22 the structural members and scenarios considered during the formulation of the model. Weighing the relative advantages and disadvantages of each modeling approach, semi-empirical approaches are always preferable because the models include validated empirical data. In fact, because of their efficiency over first-principle models for such cases, semi-empirical models are much better for design. For those cases in which the scenario in question relies on extrapolation of test data, or for cases where data are not available, first-principle models should be used by themselves, preferably after validation against experimental data or semi-empirical models for known cases, to predict blast loading and response. While the suggestions presented throughout this document address both first-principle and semi-empirical analysis methods, semi-empirical methods should be used whenever possible. Some methods utilize both first-principle and semi- empirical procedures (Winget, 2003). Equations first cal- culate blast-wave propagation and structural response, and the results are then compared to, and corrected by, empiri- cal data from similar scenarios. These methods have wider ranging applicability than semi-empirical methods, and they require less computational effort and provide more accuracy than first-principle methods. Therefore, methods that utilize both first-principle relationships and semi-empirical data are very practical for design use. Although empirical data do not widely exist in the public domain, a few of these programs are available. The origin, accuracy, and applicability of the data used in these methods, however, may be difficult to verify. Given that possibility, the guidelines presented in this doc- ument emphasize pure first-principle structural analysis methods over combined procedures, but an analyst should understand that legitimate techniques based on, or corrected by, legitimate empirical data are preferable over pure first- principle methods at all times. Most blast-analysis programs separate the calculation of blast-wave propagation effects from the determination of structural response. Thus, loads resulting from the chosen blast source are first calculated, and then they are applied to the structure using a separate response analysis method. Such separated methods are considered “uncoupled,” and they typically provide conservative predictions of loads acting on structural components. Because the analysis typically assumes the structure is rigid during load calculations, structural de- flections and localized member failures, which can vent and redistribute pressure, are neglected, and the analysis typically overestimates blast pressures and forces in unfailing members. Accordingly, use of uncoupled methods often provides con- servative load values for designing structural members. Coupled analysis methods, unlike uncoupled analyses, con- sider blast-wave propagation and structural response together as they interact over time. Thus, a structure can vent pressure through localized failure, and the forces resulting in many members will be smaller and more realistic than those pre- dicted by uncoupled analysis approaches. For scenarios in which local failure or large deformations result, coupled analysis techniques may be necessary. Although coupled pro- grams are expected to provide more accurate results than uncoupled ones, they do so at considerable costs due to the number of input parameters required, the time and experi- ence needed to create a model and interpret the output, and the computational resources and time required to compute results. Because uncoupled analysis methods usually provide conservative blast propagation and structural response predic- tions, most design cases do not require the increased costs associated with coupled analysis methods. Uncoupled and coupled analytical programs belong in two further subdivided groups based on the characteristics of the analytical methods employed. Uncoupled analysis methods have two categories: static analyses and dynamic analyses. Within each of those categories are single-degree-of-freedom (SDOF) models and multiple-degree-of-freedom (MDOF) models. Figure 18 and the following sections describe these divisions. It is important to note that the level of accuracy, computational time and cost, and complexity of analysis increase when moving from left to right in the figure. The next section describes further details, applications, and limitations of these methods. 2.2.5.3.1 Uncoupled Static Analyses. A static analysis for a blast scenario consists of an “equivalent wind design” (ASCE, 1997; Bounds, 1998), which is similar to the equiva- lent static procedure used for seismic design. Such an analysis can compute response for both single- and multiple-degree- of-freedom systems. The approximated blast pressure under consideration is a static force applied to the structure being analyzed, and the analysis does not account for inertial effects. Because this method is very general, no program specifically exists for equivalent static blast analyses; however, those pro- grams currently used for ordinary structural analysis can be used for this purpose. Although this method is relatively simple, its main weakness is accuracy. Unlike seismic events, vehicle impact incidents, or vessel impact scenarios, blast- loading characteristics cannot be easily defined based on historical data. The loads acting on a structure for a given blast event can vary greatly depending on the type of explo- sive, the location of explosive, the surrounding reflecting geometry, and the geometric and material properties of the structure being investigated. Accordingly, a static blast design requires the introduction of many approximations. In addi- tion, no general equation exists to determine a conserva- tive static load (Bounds, 1998), which makes determining an appropriate load for design difficult (ASCE, 1997). Thus, accuracy is limited (ASCE, 1997; Bounds, 1998), and bridge

designers should not use an equivalent static design for any purpose. 2.2.5.3.2 Uncoupled Dynamic Analyses. Dynamic un- coupled analyses vary from simple SDOF systems to more com- plex MDOF systems. SDOF dynamic analysis methods are relatively simple, and design engineers commonly use them to determine individual member response. The mathematical procedure required to derive the properties of the equivalent SDOF system is similar to that of a modal analysis used for seismic-resistant design. A separate load determination method can calculate the time-varying blast loading under consideration, and the SDOF analysis assumes a deflected shape for the response of the member being analyzed, often using a static loading response shape that approximates the dynamic response shape (Biggs, 1964; Department of the Army 1990). This deflected shape is then integrated along the length of the member with the actual mass and force to de- termine an equivalent mass and force for the dynamic system, and a simple spring-mass-damper system is then assumed and analyzed. The resistance used for the spring corre- sponds to the pattern of deformation for the member being analyzed. With this approach, the analysis of the SDOF model includes inelastic material behavior by noting the formation of plastic deformation mechanisms that correspond to the assumed displaced shape. For example, in a fixed–fixed beam under uniform loading, the bending moment acting at the supports will reach the plastic moment or section capacity as the magnitude of the load is increased. The analysis can in- clude the plastic hinges that occur at the ends of a member by modifying the assumed deflected shape to account for the presence of the hinges, as shown in Figure 19. Such analyses can be solved in closed-form, but they often employ numer- ical solutions to allow for a wide range of loading histories and nonlinear material behavior. Uncoupled dynamic analyses of MDOF systems can range from simple, dynamic 2-D frame analyses to very sophisticated 3-D finite element analyses. Models of the structural systems under consideration are constructed in commonly used analy- sis software, and the time-varying load for the analysis comes from a separate load determination method. Because a cate- gory containing uncoupled dynamic MDOF analyses can rep- resent a wide range of methods with varying capabilities, this document considers MDOF frame analyses and detailed finite element analyses separately. First Principle and Empirical Models Uncoupled Analysis Coupled Analysis Dynamic Analysis Dynamic Analysis Static Analysis SDOF MDOF SDOF MDOF SDOF MDOF Increasing accuracy, complexity, and cost Figure 18. Flowchart of possible analysis methods (Winget, 2003). Elastic Elastic-Plastic Plastic Plastic hinge Plastic hinges Figure 19. Stages of beam response (Biggs, 1964). 23

24 2.2.5.3.3 Coupled Dynamic Analyses. Coupled analy- ses are intrinsically dynamic because they “couple” blast pres- sures with response to consider how the loading and structure interact over time. Sophisticated software currently avail- able can model such complex fluid–solid interaction. When modeling MDOF systems, these programs allow the engineer to investigate global changes in response due to failure or large deformations of individual components. Although these methods can provide significant increases in accuracy over uncoupled analyses, they require a considerable amount of time to input the many variables needed to define such com- plex systems and perform the analyses. Furthermore, these analyses require a very experienced engineer to interpret the results. In addition, many codes claim to be coupled, but only a limited number of codes truly have the capability to couple blast pressures with structural response. For the vast majority of design scenarios involving blast loads acting on bridges, this level of accuracy is usually not necessary due to uncer- tainties that exist with blast loadings, and simpler analytical methods can often provide conservative and reasonably accu- rate results at a fraction of the cost (Winget, 2003). Although conducting coupled SDOF analyses may be tech- nically possible, doing so would not be practical because the results would not be very useful. If a model of a structural sys- tem and blast scenario requires a coupled analysis to account for expected load changes due to events such as venting from localized failures or large deformations, more than one degree of freedom would be necessary to investigate the change in response of one component due to the behavior of another component. Therefore, for all practical purposes, coupled analyses are useful only for MDOF models. 2.3 Research Needs Only a limited amount of information regarding blast- resistant design of bridges is available in the open literature. While it is likely that the military has extensive classified re- search on the topic, nearly all publicly available references on the blast response of structures focus on buildings. Bridge engineers can access design guidelines for military and petro- chemical building structures when considering blast-resistant bridges, although the applicability of the principles developed for buildings to bridges is uncertain. Thus, there is a need to develop an understanding of the principles of blast-wave propagation, the potential effects of blast loads on bridges, and the resulting response and potential failure mechanisms of bridge members. The research presented in this report focuses primarily on columns because they are integral parts of most bridges. Loss of a single column could result in the collapse of multiple spans or the loss of multiple bridges in the case of multi-level interstate exchanges. In addition, bridge columns are especially vulnerable because military records show that it can be easier to destroy a bridge by shaking down the columns with large, nearby explosions than shooting up the superstructure (Bulson, 1997). While past research and currently available guidelines may not directly apply to the case of bridges, they do provide a valuable foundation for ad- vancing the field of blast-resistant bridge design. Therefore, the following sections describe the information currently available in the literature. 2.3.1 Research Needs and Focus of Current Study Several publications on weapons effects and structural re- sponse to blasts are available, and information included in these references provides useful information for the design of bridge components to resist blast loads. One such military document is TM 5-1300, Structures to Resist the Effects of Ac- cidental Explosions (Department of the Army, 1990), which is the technical manual for design against accidental explosions. According to the introduction, “The purpose of TM 5-1300 is to present methods of design for protective construction used in facilities for development, testing, production, storage, maintenance, modification, inspection, demilitarization, and disposal of explosive materials” (Department of the Army, 1990). The manual establishes blast-resistant design proce- dures and construction techniques to provide protection for personnel and valuable equipment. For the purposes of the current study, TM 5-1300 provides allowable design response limits for structural elements in terms of support rotations. The manual also requires the uniform distribution of shear reinforcement throughout a member, and it provides dynamic increase factors and strength increase factors for different materials. This manual recommends Grade 60 reinforcement and a minimum of 4,000 psi concrete to provide sufficient reinforcement ductility and concrete strength for structures that need blast resistance, and it does not permit high-strength concrete unless laboratory testing can demonstrate that it has sufficient toughness. Fundamentals of Protective Design for Conventional Weapons provides engineers with useful information regarding the design of hardened structures. Conventional weapons can range from an airblast alone to direct hits from precision- guided cased bombs, and a hardened structure’s primary task is full functionality after a wartime attack. As it relates to the current study, this manual provides response limit criteria based on support rotations. UFC 3-340-01 (Departments of the Army, Air Force, Navy, and the Defense Special Weapons Agency, 2002) supersedes TM 5-855-1 and carries the title Design and Analysis of Hardened Structures to Conventional Weapon Effects. Both of these manuals are available only to federal contractors and are not approved for general pub- lic release.

Structural Design for Physical Security: State of Practice (Conrath et al., 1999) is a commonly used non-governmental blast-resistant design guideline for buildings. The purpose of this reference is to “provide methods, guidance, and refer- ences for structural engineers challenged with a physical secu- rity problem” for a civilian facility. This book introduces general concepts of structural response and behavior under the effects of severe short-duration dynamic loads and a design philosophy that can be adopted to enhance safety. This design philosophy is “simpler is better,” recommending the use of simple geometries with minimal ornamentation, which can become airborne during explosions. Ductility and redun- dancy are two important considerations when trying to pro- vide designs that localize and isolate failures. Specific infor- mation that may be relevant for the current research is the specified ductility limits based on typical structural members’ observed level of damage under blast loads for a given defor- mation, as shown in Table 3. In flexure, the formation of a plastic hinge requires that the hinge size be approximately equal to a member’s depth. Detailing plays an important role in achieving these levels of ductility, and the book recom- mends design approaches that improve the bending and shear strength of columns under blast loads. For reinforced concrete, this philosophy includes using adequate confining steel to ensure ductile behavior and sufficient development of reinforcement. Design of Blast Resistant Buildings in Petrochemical Facili- ties (ASCE, 1997) contains civilian blast-resistant design guidelines primarily intended for petrochemical facilities. The book focuses on the structural aspects of designing build- ings for blast resistance, and it details equations for several parameters needed to define the blast wave shown in Fig- ure 4. Additionally, the text also recommends modifications to the current concrete and steel design codes (ACI, 2005 and AISC, 2006, respectively) to increase structural blast resis- tance. Designs for structures needing full functionality after a blast event should provide for elastic response under the pre- dicted loads. In most cases, however, inelastic structural re- sponse is allowed and provides a means to dissipate the energy of a blast. This document also provides a design approach based on evaluating the ductility and hinge rotations of each member. Much of the information in this book is relevant to the current research. 2.3.2 Blast-Resistant Design: Buildings versus Bridges The design approach and guidelines presented in the pre- vious section are primarily for the blast-resistant design of buildings. There are several main differences between the de- signs of buildings and bridges to resist blast loads. Figure 20 illustrates a typical building and a highway overpass in Austin, Texas. Buildings, such as the UT Performing Arts Center, are able to create a large standoff to structural mem- bers through landscaping and site layout. In contrast, a high- way overpass is a bridge that crosses over another road or railway, and these structures commonly have extensive access below the deck and near columns via parking areas, traffic lanes, sidewalks, or other general unobstructed areas. There- fore, gaining access to structural members is much easier for bridges than buildings, creating the possibility of a design threat with a significantly smaller standoff distance. Also, structural members in buildings, such as columns and beams, are usually behind a façade and are only indirectly loaded by a blast wave, while bridge supports, such as columns, are di- rectly exposed to blast waves. Additionally, blast loads on flat walls and their resulting response have been experimentally verified and are well understood. However, very little is known about shock-wave interaction with slender structural members. Therefore, additional research should determine if current design guidelines for buildings are directly applicable to bridges. Light Moderate Severe Global Bending/ Membrane Response /L 4 8 15 Shear v 1 2 3 Bending/Membrane /L 4 8 15 Shear v 1 2 3 Columns Compression Shortening/Height 1 2 4 Load-Bearing Walls Compression Shortening/Height 1 2 4 Shear Walls Shear v 1 2 3 * /L = Ratio of Centerline Deflection to Span † v = Average Shear Strain Across Section ‡ For Reinforced Concrete with ρ > 0.5% per face Slabs Element Type‡ Type of Failure Criteria*† Damage Level (%) Beams Source: Conrath, 1999 Table 3. Typical failure criteria for structural elements (Conrath et al., 1999). 25

26 2.3.3 Current and Past Research on Highway Bridges Although the field of bridges subjected to blast loads is relatively new, some past and current research focuses on the development of design guidelines and analytical models for various bridge components subjected to these types of loads. While this effort within the open structural engineer- ing community is in its infancy, these studies provide a good foundation on which to base the work presented in this report. This section summarizes the findings from these studies to il- lustrate the progress made in the field of blast-resistant bridge design. Researchers at Florida State University analyzed the re- sponse of typical AASHTO girder bridges using STAAD.Pro, a commonly used finite element program. The researchers applied the peak pressures computed from the defined threat to the bridge as a static load. The model bridge failed under the applied loads above and below the bridge deck. The au- thors specifically claim that the AASHTO girders, pier caps, and columns were not resistant to typical blast loads (Islam, 2005). While these results may be quite alarming, they are extremely conservative and unrepresentative of the expected behavior of a bridge to dynamic blast loads. The study uti- lized a static analysis, and analysis of a structure subjected to blast loads should always be conducted dynamically. Peak pressures produced by a shock wave are often very high, but they dissipate within milliseconds. As a result, the impulse governs the response of most structures for a typical blast load. Additionally, this study did not use material increase factors in the analysis, which is not representative of real blast scenarios. Researchers at the University of California at Berkeley studied the response of single-cell and multi-cell steel and composite bridge columns loaded with “simulated effects of car bomb explosions” using finite element analyses (Rutner et al., 2005). While the conclusions included general obser- vations about response mechanisms of these columns based on analytical results, they did not provide any design guide- lines. Researchers at the New York State Department of Transportation and City College of New York plan to de- velop high-precision analytical models and design guidelines for blast effects on highway bridge components based on ex- isting seismic guidelines (Agrawal, 2007). This study is cur- rently ongoing, and limited information is available in the literature. Winget et al. (2004) carried out parameter studies with SDOF analyses to “evaluate the effectiveness of structural retrofits, refine the performance-based standards, and develop general blast-resistant guidelines specifically for bridges.” The researchers analyzed a reinforced concrete bridge with three- column bents, changing the pier diameter, pier shape, and concrete strength. The study considered two terrorist threat scenarios: a vehicle bomb below the deck and hand-placed charges in contact with the pier. “It was expected that the vehicle bomb possesses the potential to produce large lateral blast pressures, resulting in localized breaching damage of the concrete and causing a flexural failure of the pier” (Winget et al., 2004). The contact charge was thought to “possess the potential to breach enough of the pier to render it incapable of supporting the dead loads” (Winget et al., 2004). The researchers used BlastX version 4.2.3.0 (SAIC, 2001) to generate loads for all cases considered, and ConWep V. 2.0.6.0 (U.S. Army Corps of Engineers, 2001) provided predictions courtesy of http://www.utpac.org courtesy of Kim Talley (a) (b) Figure 20. Differences between blast-resistant building and bridge design: a) UT Performing Arts Center, b) I-35 bridge near Town Lake.

for the reduced area of the piers due to breaching from local blast damage. Winget et al. (2004) state that “the pressure dis- tribution varied significantly along the height of the pier at any given time. This phenomenon was due to the reflected pres- sures off the ground and the reflected pressure buildup be- tween the girders under the deck.” Calculations for loads on the curved column surface included a reduction of the re- flected pressure based on the changing angle of incidence. Additionally, recent parameter studies conducted by Winget et al. (2005) assumed blast loads to be directed laterally for col- umn design, and that work provided a preliminary approach for estimating blast loads on slender square and circular mem- bers; however, the design loads for these scenarios are still unclear, and additional work is needed to develop a method to predict accurate blast loads on bridge columns. Winget et al. (2004) utilized SPAn32 (U.S. Army Corps of Engineers, 2002) to calculate the flexural response of the piers subjected to the vehicle blast loads. A plastic analysis yielded the ultimate resistance, which was adjusted during the re- sponse predictions using dynamic increase factors to modify material strengths based on the instantaneously calculated strain rate. The analyses modeled the pier as an SDOF flex- ural member fixed at both ends and considered the effects of nonlinearity due to material behavior. Winget et al. (2004) reported that, for simplicity, “when determining the flexural response of the piers due to the blast pressure and reduced cross-sectional area from local damage, it was conservatively assumed that the cross-sectional area along the entire height of the pier was reduced to its minimum predicted diameter at the location of maximum breaching.” Winget et al. (2004) then predicted damage levels using deformation-based em- pirical data derived primarily from building members for concrete beam elements in flexure. For the piers, the maxi- mum support rotation corresponded to 1.3 degrees for slight to moderate damage and 2 degrees for moderate to heavy damage. The analysis assumed that the pier lost structural integrity at 3 degrees of rotation. Significant for the current study, Winget et al. (2004) also noted that the support rotation values used in this study may require adjustment based on future experimental data specifically for bridge piers. Research completed by Bruneau et al. (2006) at the Univer- sity of Buffalo aimed to develop “a multi-hazard bridge pier concept capable of providing adequate protection against col- lapse under both seismic and blast loading.” Quarter-scale concrete-filled circular steel columns (CFCSC) linked by a fiber-reinforced concrete cap beam and foundation beam were subjected to blast loads. The assumed threat was a small vehicle bomb below the deck at a close standoff distance. “The CFCSC exhibited a ductile behavior under blast loads and no significant damage was suffered by the fiber reinforced concrete cap-beam” (Bruneau et al., 2006). Observations included permanent deformations halfway up the height of the column and shear at the column base, including fracture of the steel tube halfway around the perimeter from the front, as illustrated in Figure 21. Current experimental research on blast-resistant bridges includes an FHWA and state pool-funded project consisting of analytical studies and large-scale experimental blast tests on steel bridge towers subjected to blast loads (Ray, 2006). The University of Washington is also performing experimen- tal blast tests on two full-scale prestressed girders and two full-scale girder and deck structures. These studies are cur- (a) (b) Figure 21. Concrete-filled steel tube subjected to blast loads: a) ductile response, b) shear response (Bruneau et al., 2007). 27

28 rently under way, and limited or no information is available in the literature. 2.3.4 Summary This literature review provides an overview of the princi- ples of shock propagation in free-field and reflected condi- tions, the computation of blast loads, general observations of blast effects on structures, and basic analytical procedures for predicting the response of structures to dynamic loads. Although this review does not provide a comprehensive dis- cussion of every available source related to explosive effects on bridges, it does provide the basic design knowledge a bridge engineer needs to predict blast loads and the resulting response of structures, and it outlines the background for the current research. The information discussed above, along with more in-depth coverage, can be found in numerous additional sources in the open literature that exists on the topics of munitions and blast effects on structures. 2.4 Focus of Current Work Recent increases in the frequency and intensity of terrorist attacks on transportation infrastructure highlight the need for blast-resistant design guidelines for bridges. Most currently available blast-resistant design and detailing requirements focus only on buildings, and additional research should deter- mine if these methods and requirements, as well as seismic guidelines, are applicable to blast-resistant bridge design. “Signature” bridges (suspension bridges, cable-stay bridges, tied arches, etc.) sometimes receive priority over other bridge types for blast-resistant research and design for two reasons. First, many in the bridge engineering community believe that an attack against a signature bridge is more likely than an attack against a regular highway bridge, and second, an attack against a high-profile bridge target seems to carry a greater perceived socio-economic impact than an attack on a regular highway bridge. However, past experience does not necessar- ily support either of these beliefs as historical data show that terrorists desire to attack ordinary bridges even in industrial- ized nations that have high-profile signature bridges, and recent bridge collapses in the U.S. have shown that the failure of a typical highway structure can have a devastating socio- economic impact. A report from the Mineta Transportation Institute (Jenkins and Gersten, 2001) includes the analysis of 53 terrorist attacks that specifically targeted bridges between 1980 and 2006, and the report notes that 58% of bridges tar- geted worldwide and 35% of bridges targeted in industrial- ized nations during that time were highway bridges other than signature bridges. Considering that 60% of attacks on all transportation targets during that same time were bombings, a bombing of an ordinary highway bridge is a realistic scenario. Furthermore, the recent accidental collapses of the Oklahoma Weber Falls I-40 bridge (Blue Ribbon Panel, 2003), the Queen Isabella Causeway in Texas (Texas Office of the Governor, 2001), and the I-35 interstate bridge in Minnesota (Minnesota Department of Transportation, 2007) show the great socio- economic harm and fear that can result from the failure of one of these typical highway bridges. Moreover, an Al-Qaeda training manual specifically reinforces the threat to ordinary bridges by stating that a main goal of terrorism is “blasting and destroying bridges leading into and out of the city” to “strike terror into the enemies” (“Military Studies in the Jihad against the Tyrants,” 1995), and regular attacks on ordinary bridges during the war in Iraq illustrate the success of these goals. Therefore, a bombing of a regular highway bridge in the U.S. is in fact a realistic scenario, and bridge engineers need guidelines for designing blast-resistant “ordinary” high- way bridge structures. Focusing the current research on typical highway bridge structures provides the most effective use of available proj- ect funds. Although understanding the effects of blast loads on signature bridges undoubtedly is an important research need, these structures are often complex, requiring special- ized design provisions even for typical gravity and wind loads, and research on such bridges likely will provide results appli- cable only to the specific bridges considered. Additionally, while a significant number of existing critical bridges in the U.S. may require retrofits to resist blast loads, each of these bridges is different, with various types of structural systems in various states of disrepair. In-depth research on existing crit- ical bridges would produce results applicable only to a lim- ited number of scenarios. Focusing on the future design of typical highway bridge structures, however, will maximize the return on allocated research funds, as the results can guide designs for future typical highway bridges and spawn addi- tional research on other types of bridge structures and retro- fits of existing highway bridges. Therefore, the focus of the current research is on the blast-resistant design of new, typi- cal highway bridges, and the primary objective is to develop blast-resistant design guidelines that can be easily incorpo- rated into the AASHTO bridge design specifications. While highway bridges have several structural components with blast resistance worthy of investigation, bridge columns are arguably the most important, and they appear to be the most vulnerable to direct attack. Most bridges intrinsically have open access to substructure members for either hand-placed or vehicle-delivered explosives in the form of traffic lanes, side- walks, waterways, or other unrestricted areas. Although super- structure traffic also has unrestricted access to the deck, the type of superstructure can vary significantly from one bridge to another, while concrete bridge columns are integral to an overwhelming percentage of bridges. Thus, focusing on a particular superstructure type would not provide the same

comprehensive benefit as focusing on bridge columns. More- over, while a successful attack against a superstructure likely will mean the loss of only one span, the failure of a critical bridge column likely will mean the collapse of at least two spans, and failure of a bridge column in an extensive inter- state highway interchange may cause the progressive col- lapse of numerous spans as superstructures collapse onto one another. Two aspects are necessary for any sound design: a thor- ough understanding of the load, and a good prediction of the expected response. As stated previously, most past studies of blast loads acting on structures involve flat surfaces (i.e., walls and building façades), and little is known about the interaction of a shock wave with a slender square or circular member (i.e., column). Furthermore, experts in the field of blast-resistant design are uncertain as to whether the observa- tions regarding the response of building components to blast loads also apply to bridges. Therefore, this study includes two parts: Phase I aims to characterize the structural loads acting on square and circular bridge columns due to airblast, and Phase II investigates the response of half-scale bridge columns to blast loads, specifically focusing on various design param- eters that may increase blast resistance. The following chap- ters provide details of both parts of this research program. 29