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.

60 5.1 Overview Observations from analytical research and experimental test programs are presented in this chapter. The experimental research was divided into two phases. The first phase focused on determining the blast-load variation as a function of time and position for scaled round and square non-responding columns. In the second phase of experimental testing, half- scale reinforced concrete columns were tested under close-in blast loads. Both Phase I and Phase II test programs are described below. 5.1.1 Phase I Tests The observations from each blast test on the small-scale, non-responding columns tested in Phase I of this experimental research program are summarized in this section. The entire Phase I test program and all data processing were completed in December 2006 at the U.S. Army Corps of Engineers’ Engineer- ing Research and Development Center in Vicksburg, MS. The Phase I test program included eight small-scale blast tests at four sets of standoff distances. The same scaled stand- off was used for all tests. The objective of the small-scale tests was to characterize the structural loads on square and round bridge columns due to blast pressures. The experimen- tal observations indicate how cross-sectional shape, standoff, and geometry between the charge and column positions influence blast pressures on the front, side, and back faces of bridge columns. The data gathered from these tests also allow the assessment of the accuracy of classical methods used to predict blast loads on slender structural components such as bridge columns. Prior to these tests, the majority of the data used to develop empirical blast-load models came from free- field blast tests as well as blasts against large, flat panels. Thus, the focus of the Phase I tests was to study blast-load variation as a function of time and position on slender components for which there were limited data available. 5.1.1.1 Characterization of Blast Loads on Columns ERDC recorded 160 channels of pressure–time history data during the Phase I blast tests. Figure 43 shows an example of the pressure–time and impulse–time history for one of the test series. For each test, the results from a free-field pressure gauge on each side of the charge were compared to ensure that a similar amount of blast energy was directed toward each specimen. The percent difference in free-field impulse ranged from 6% to 35% and were very reasonable for tests at such a small scale. For any given test, the data show that the pres- sures and impulses at the bottom of a column are significantly higher and arrive much earlier in time than those at the top of the same column. Figure 44 shows the difference in pres- sure and arrival time along the height of a column. This dif- ference becomes significantly more prominent as the physical standoff decreases, and this finding suggests that shear will likely be the dominating mode of response for blast scenarios with similar scaled standoffs and heights-of-burst as those considered in the Phase I tests. Furthermore, given that pres- sures and impulses at the bottom of a column are greater and arrive sooner than those at the top of the column, shock waves reflecting off the deck will not likely control the response of a typical bridge column because a typical column will reach its peak response very early in time for the most serious design threats. Scenarios with smaller standoff distances tend to produce lower pressures and impulses at the top of columns than do scenarios of the same scaled standoff but with larger physical standoff distances. For a series of blast tests having the same scaled standoff, cases with smaller actual standoff distances have larger scaled distances to elevated gauges on the column than do cases with larger standoff distances. Additionally, for a given scaled standoff, the angle at which the shock front strikes elevated gauges on a column increases as the standoff decreases. For the geometries considered in the Phase I test program, the reflected pressure decreases as the angle at which C H A P T E R 5 Observations and Research Findings

the shock front strikes the column increases, resulting in sig- nificantly decreased pressure and impulse near the top of the column as the standoff distance decreases. Figure 45 shows the difference in geometry for two different standoff distances with the same scaled standoff, and one can see that the ratio R′2/R′1 is significantly greater than the ratio R′2/R′1. Bridge Explosive Loading (BEL) (U.S. Army Corps of Engineers, 2000) and BlastX (SAIC, 2001) are two programs that can predict blast effects on flat surfaces. Comparisons between computed values and experimental data show that these programs routinely over-predict pressures and impulses, especially at locations near the top of a column. Figures 46 and 47 show comparisons between the predictions obtained using BEL and experimental pressures and impulses for the front gauges of the square columns for a representative test in the Phase I program. The BEL predictions overestimate the 61 Figure 43. Example plots of pressure and impulse for the middle front gauge. Figure 44. Front gauge pressure–time histories. ab12s7.grfTime, msec Pr es su re , P si Im pu ls e, P si -m se c 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -50 -15 0 0 50 15 100 30 150 45 200 60 250 75 300 90 350 105 400 120 450 135 500 150 550 165 600 180 650 195 ab2r7.grfTime, msec Pr es su re , P si Im pu ls e, P si -m se c 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -50 -15 0 0 50 15 100 30 150 45 200 60 250 75 300 90 350 105 400 120 450 135 500 150 550 165 600 180 650 195 700 210

peak pressure and impulse for all three of the front gauges, and the overestimation increases along the height of the column. Initially, the case of a shock wave striking a column may seem similar to that of a shock wave striking a wall; however, the fact that BEL and BlastX increasingly over-predict pressure and impulse as the distance from the column base increases may be evidence that clearing for a column is more complex than previously thought. Because columns are significantly more slender than walls, the empirically derived three-transits- to-the-edge rule, defined in Section 2.2.1 (Departments of the Army, Air Force, and Navy and the Defense Special Weapons Agency, 2002), may prove to be inaccurate for such slender members. Moreover, additional clearing at the free surfaces at the tops of these columns may contribute to the fact that BEL and BlastX increasingly over-predict pressures and impulses as the location along the height of the column increases. The data gathered during the Phase I non-responding column tests also show the importance of considering cross- sectional shape and standoff geometry when determining structural loads on columns. Figure 48 shows a plot of the net impulses on the circular and square columns for one of the tests, and the differences between the pressures and impulses along the height of the columns in these tests are clear. In gen- eral, the pressures and impulses acting on the circular column are less than those acting on the square column. This differ- ence can be 3%–34% for a series of cases with the same scaled 62 Figure 45. Schematic of geometry differences between two scenarios with the same scaled standoff but different actual standoff distances. R1 R'1 R2 R'2 α α' Figure 46. Front-gauge experimental and BEL-predicted pressure comparison.

standoff, and this difference depends on the actual standoff distance and the location of interest along the height of the column. In some cases, the pressures and impulses at the bot- tom gauge of the circular column are equal to or greater than their corresponding values for the square column. Caution must be exercised, however, when examining the Phase I data because small-scale tests can have a large relative error that makes direct comparisons between data sets difficult, and the small scale of the Phase I tests may have magnified very small inaccuracies in measurements to gauge locations and stand- 63 Figure 47. Front-gauge experimental and BEL-predicted impulse comparison. Figure 48. Net impulse comparison.

off distances, resulting in pressure–time histories that are dif- ficult to compare directly. Nevertheless, these observations are interesting and merit additional investigation, and future work should address these issues. Interestingly, the back-face pressures and impulses on the circular columns in the Phase I tests are typically equal to or larger than the corresponding pressures on the square columns. Figure 48 shows a comparison between the net impulses for the circular and square columns of a repre- sentative test, where the net impulse is defined as the differ- ence between the front- and back-face impulses at a given elevation. Typically, the net pressures and impulses on the circular columns of the Phase I tests are less than those on the cor- responding square columns; however, as was the case with front-face pressure–time histories, the results of some tests show that the net pressures and net impulses at the bottom gauges are greater for the circular columns than the square columns. Ultimately, based on a study of the collected test data and detailed analytical models, the researchers believe that several factors influence the pressure–time and impulse–time histories experienced by square and circular columns for a given charge weight and standoff distance. These factors include the difference between the standoff distances to the center of a col- umn and to the edge of the same column, the diameter of a column, the angle of the reflected pressure around the circum- ference of a circular column, and the difference in the clearing times for circular and square columns. In general, circular columns will experience less net load than square columns with the same projected area (i.e., when the edge width of the square column equals the diameter of the circular column). Several factors may influence the difference in pressures and impulses along the heights of square and circular columns that contribute to this difference in net loads. Reflected pressures, and more importantly reflected impulses, are functions of the angle of incidence and peak incident overpressure. The angle of incidence at a given loca- tion around the front face of a circular column is greater than that of a similar position along the front face of a square col- umn (i.e., locations with the same horizontal distance per- pendicular to the column centerline). Furthermore, the peak incident overpressure is less at a given location around the circumference of a circular column than at a similar position along the face of a square column. This reduction in peak incident overpressure is due to the increased standoff distance to the position on the circular column as compared to that on a square column. Both an increase in angle of incidence and a decrease in peak incident overpressure individually produce reduced reflected pressure and impulse. Therefore, a circular column will experience less load than a square column because the summation of the resulting reflected impulses around the front face of a circular column is less than that along the face of a square column. Additionally, the physics of clearing for circular columns is not well understood, and researchers do not know whether the shock wave clears from or simply flows around the front face of a circular column. Considering that circular columns do not have a flat edge to directly reflect a shock wave, one can safely assume that the reduction in load due to clearing will be greater for circular columns than for square columns, and the three-transits-to-the edge rule (Departments of the Army, Air Force, and Navy and the Defense Special Weapons Agency, 2002) may not be appli- cable to the case of a circular column. Finally, a shock wave will engulf a circular column more easily than a square col- umn, making the resultant impulse on the back face of a cir- cular column greater than that of a square column, further reducing the net resultant load a circular column must resist relative to a square column. While the exact relative contri- bution of all these factors is still unknown, experimental tests and fluid dynamics analyses confirm that the reflected pres- sures and impulses are less for circular cross-sections than for square cross-sections. 5.1.1.2 Methods to Predict Loads on Square and Circular Columns The data obtained during the Phase I blast tests provide a basis on which to evaluate a proposed method for predicting structural loads on square and circular columns. This method is outlined in a 2005 paper by Winget et al. The experimental data from Phase I of this research show that the method pro- posed by Winget et al. (2005) to adapt blast pressures on flat surfaces to predict blast pressures on circular columns is rea- sonably accurate in some cases; however, the method often yields erratic and erroneous results because it is very sensi- tive to several parameters that are difficult to define. For example, the current data show that pressures near the top of a column are significantly lower than those predicted by the proposed method. Furthermore, the current data also show that pressures at the bottom of a column are much higher and arrive earlier in time than pressures at the top of a column. This observation is expected due to the previously explained cross-sectional properties of a column and the geometry of the column location relative to the explosive charge. The proposed method, however, does not capture this behavior. Several factors may contribute to the inaccuracies intro- duced by the previously proposed method to predict struc- tural blast loads. The method proposed by Winget et al. (2005) uses the pressure predicted by BEL for several points along the height of a column to create a pressure–time history that varies with position along the height of a column but does not vary with time along the height of a column. The load- ing produced by this method would most likely result in a 64

flexural response of the column, whereas the loading due to the pressures recorded during the Phase I tests would most likely result in a shear-dominated behavior early in time. Moreover, because BEL increasingly over-predicts pressures as the height of the location of interest on a column increases, the error of the proposed method will also increase with the height of the location of interest on a column. Additionally, the method is very sensitive to the positive phase duration, which can be difficult to accurately define. The proposed method also relies on the three-transits-to-the-edge rule, which is likely not valid for circular and very slender mem- bers. Reflections along the front face of a circular column are not directly perpendicular to the blast source, and it is not yet clear how to calculate the clearing time for a circular column. Also, unlike square columns, the effective standoff distance and angle of incidence to positions along the circumference of a circular column varies. This difference is likely to be important for large columns subjected to explosive charges at small standoff distances. Although the previously outlined method can provide a good first approximation of structural loads on bridge columns, the current data do not support the accuracy of that method for the cases investigated in this study. 5.1.2 Phase II Tests The observations from each blast test on the half-scale, reinforced concrete bridge columns tested in Phase II of this experimental research program are summarized in this section. The entire Phase II test program was completed in October 2007 at the Southwest Research Institute test site in Yancey, Texas, and data processing was done by researchers at Protection Engineering Consultants and the University of Texas at Austin. The Phase II test program included ten half-scale, small standoff and six half-scale, local damage blast tests. Actual material properties for concrete and steel were determined from standard laboratory tests. Pressure and impulse data were used to determine equivalent TNT charge weights for consistent comparisons among the tests. The experimental observations were used to develop design and detailing guidelines for concrete highway bridge columns and analyt- ical tools to predict blast-load distribution and resulting col- umn response. 5.1.2.1 Material Properties The concrete and steel specified in this project were selected to represent that which is typically used in the construction of reinforced concrete highway bridge columns. Actual material properties were measured to improve calculations and analy- ses of the experimental test data. 5.1.2.1.1 Concrete Properties. Due to the variety of concrete strengths and types used in different states, the val- ues specified by the AASHTO LRFD were selected for the test program. Concrete with a strength of 4000 psi composed of Type-A cement and a maximum aggregate size of 3⁄8 in. was specified to accommodate the small rebar spacing used in sev- eral of the specimens. To verify that the concrete columns attained the specified strength, cylinder tests were conducted at various times to determine the compressive strength. The cylinder tests were performed using the Forney universal cylinder test machine according to ASTM C39 Standard Test Method for Com- pressive Strength of Cylindrical Concrete Specimens. Stan- dard 4-in. × 8-in. cylinders were capped with neoprene bearing pads and steel retaining rings before testing. For each age, a minimum of three compressive capacities were recorded and averaged, and the compressive strength was calculated using Equation 13. Table 10 lists the average compressive strength for the concrete columns, footings, and slab. where: f ′c = cylinder compressive strength (psi) P = applied load (lbs) D = diameter of cylinder (in.) 5.1.2.1.2. Mechanical Properties of Steel Reinforcement. This section presents the measured stress–strain properties of the reinforcing steel (rebar) used in this test program. All rein- forcement specified and used was uncoated steel bar. Standard deformed, Grade 60 reinforcing bars were specified for all rein- forcement except the continuous spirals that allowed the use of a smooth bar (AASHTO LRFD, 2007). To verify the actual yield strength, tension tests were per- formed in a 600-kip, hydraulically actuated, universal testing machine. Three different types of reinforcing bars were tested: #6 deformed longitudinal bars, #4 smooth spiral reinforce- ′=f P D c 4 13 2π ( ) 65 Column Footing Slab 0 000 14 3550 3600 4000 22 3900 4650 5250 23 3850 28 4150 35 4050 36 4100 38 4100 Specified 4000 4000 5000 Compressive Strength (psi)Days Table 10. Concrete compressive strength.

ment, and #4 deformed discrete ties and hoops. The stress– strain plot for each bar type is illustrated in Figure 49. The average yield strength for the deformed #6 reinforcement was 65 ksi, with an ultimate strength of 90 ksi. Testing of the smooth reinforcement was performed on straight pieces of the spiral reinforcement sent directly from the steel distributor. The smooth reinforcement had an average yield strength of 70 ksi and an ultimate strength of 98 ksi. The discrete hoops and ties consisting of #4 deformed rebar had an average yield strength of 50 ksi and an ultimate strength of 72 ksi. Thus, the #4 deformed bars had a lower strength than specified. For all subsequent calculations, the actual material properties were used. On average, all bars had a modulus of elasticity of 29,000 ksi. While Figure 49 shows a modulus for the smooth rebar that is slightly lower than the deformed bars, past tests have shown that it is not uncommon for the modulus to change slightly when a bar that was initially curved is straight- ened for the purposes of carrying out tension tests. Thus, it was assumed to have the same modulus as the other reinforc- ing bars. 5.1.2.1.3 Dynamic Strength Increase. The material properties described above were determined from standard laboratory tests using very slow loading rates, which are appro- priate for typical loads. Concrete and steel subjected to rap- idly applied blast loads, however, exhibit an increase in strength under the short duration loading. According to the ASCE (1997) document Design of Blast Resistant Buildings in Petrochemical Facilities, “These materials cannot respond at the same rate at which the load is applied. Thus, the yield strength increases and less plastic deformation will occur. At a fast strain rate, a greater load is required to produce the same deformation than at a lower rate.” Therefore, to account for this increase in strength, material properties were multiplied by the dynamic increase factors, as discussed in Section 2.2.3. The DIF amplifies the measured static mat- erial strength to account for high strain rates present in blast loading. 5.1.2.2 TNT Equivalency TNT equivalency is the ratio of the weight of an explosive to an equivalent weight of TNT. TNT equivalencies are used in the majority of research on blast effects to relate the energy output of common explosives to that of TNT. Table 11 sum- marizes the averaged free-air equivalent weights for different explosives based on peak pressure and impulse (Tedesco, 1999; Department of the Army, 1990). In this research program, a combination of ammonium nitrate and fuel oil (ANFO) was used as the explosive, and it 66 0 20 40 60 80 100 120 0.00 0.05 0.10 0.15 0.20 0.25 0.30 St re ss (k si) Strain (in./in.) #6 deformed rebar #4 smooth rebar #4 deformed rebar Figure 49. Tensile strength of reinforcing bars.

required a booster (a small quantity of C-4) to ensure reliable detonation. ANFO is commonly used in improvised explo- sive devices (IEDs) known as fertilizer bombs. Table 11 shows that ANFO is on average 82% as efficient as TNT. The actual TNT equivalency for each small standoff blast test was determined using the recorded free-field pressure and impulse data. While the equivalency factors listed in Table 11 are average values that are suitable for design, the actual equiv- alency varies as a function of explosive geometry, target ori- entation, and atmospheric conditions. Accordingly, to accu- rately capture the effective TNT equivalency from each blast test, a detailed investigation of the measured pressures and impulses was needed. Available literature and methods for predicting loads are based on TNT equivalency; therefore, TNT equivalency is used to ensure comparability of results with previous data and research projects. Each small standoff test employed three free-field pressure gauges at known distances away from the charge to determine the TNT equivalency for each test. First, the maximum side-on pressure, Pso, and impulse, is, were recorded from each pressure gauge. The impulse was calculated from the pressure data as the area under the pressure–time history curve. Next, know- ing the standoff, R (distance between the blast source and each gauge), the “spaghetti” charts and corresponding equa- tions from the U.S. Army’s TM 5-1300 Structures to Resist the Effects of Accidental Explosions were used to determine the charge weight, W, in pounds of TNT. Figure 5 illustrates the “spaghetti” chart for a hemispherical surface burst used in this test program. This chart, which is based on experimen- tal data, provides valuable information about blast loading parameters in a convenient, non-dimensional format. The ratio of the weight of ANFO to TNT was then calculated for each pressure and impulse to determine the efficiency, as shown in Table 12. The efficiencies at each free-field gauge location were then averaged together to determine the average efficiency of each blast test. The average efficiencies ranged from 72% to 94% for pressure and 40% to 66% for impulse. This wide range in efficiencies among different tests with the same explosive can be accounted for by considering the vari- ability in the composition of the ANFO, the exact location of the booster, the atmospheric conditions, and so on. Due to the lack of pressure gauges during the local damage tests, it was difficult to determine the efficiency for each of those tests. Atmospheric conditions should not have played a major role in the local damage tests because of the close 67 Table 11. Averaged free-air equivalent weights. Explosive Equivalent Weight, Pressure (lbm*) Equivalent Weight, Impulse (lbm*) Pressure Range (psi†) ANFO 0.82 -- 1 - 100 Composition A-3 1.09 1.067 5 - 50 1.11 0.98 5 - 50 1.20 1.30 100 - 1000 Composition C-4 1.37 1.19 10 - 100 Cyclotol (70/30) 1.14 1.09 5 - 50 HBX-1 1.17 1.16 5 - 20 HBX-3 1.14 0.97 5 - 25 H-6 1.38 5 - 100 Minol II 1.20 1.11 1.15 3 - 20 Octol (70/30, 75,25) 1.06 -- E 1.13 -- 5 - 30 1.70 1.20 100 - 1000 PBX - 9010 1.29 -- 5 - 30 PETN 1.27 -- 5 - 100 1.42 1.00 5 - 100 1.38 1.14 5 - 600 1.50 1.00 100 - 1000 Picratol 0.90 0.93 -- Tetryl 1.07 -- 3 - 20 Tetrytol (Tetryl/TNT) (75/25, 70/30, 65/35) 1.06 -- E TNETB 1.36 1.10 5 - 100 TNT 1.00 1.00 Standard TRITONAL 1.07 0.96 5 - 100 *To Convert pounds (mass) to kilograms, multiply by 0.454 †To Convert pounds (force) per square inch to kilopascals, multiply by 6.89 Source: Department of the Army, 1990 Composition B PBX - 9404 Pentolite Table 12. Efficiency of ANFO. Pressure Impulse Pressure Impulse Pressure Impulse Pressure Impulse 1A1 -- -- -- -- 0.84 0.43 84 43 1A2 0.93 0.38 0.76 0.54 0.78 0.48 82 47 1B 0.90 0.41 0.86 0.63 0.81 0.94 86 66 2A1 0.87 0.31 0.78 0.55 0.77 0.52 81 46 2A2 1.00 0.40 0.87 0.64 0.86 0.66 91 57 2B 1.00 0.39 0.84 0.66 0.98 0.67 94 57 2-Seismic 0.95 0.41 0.94 0.64 0.91 0.62 93 56 2-Blast 0.83 0.25 0.78 0.59 0.92 0.68 84 51 3A 0.93 0.37 0.76 0.65 -- -- 85 51 3-Blast 0.75 0.26 0.69 0.54 -- -- 72 40 Average % 91 35 81 60 86 62 85 51 Column Efficiency Gauge 1 Gauge 2 Gauge 3 Average %

proximity of the charge to the specimen. Also, the data in Table 12 suggests that as the charge gets closer to the target, efficiency increases in terms of pressure and decreases in terms of impulse. Therefore, due to the lack of pressure gauges and conflicting Gauge 1 efficiencies in terms of pressure and impulse, the ANFO efficiency for the local damage tests was assumed to be 82% as specified in Table 11. A constant effi- ciency is consistent with the decrease in variability due to atmospheric conditions. The average equivalent weight of TNT and standoff was then used to determine the scaled standoff, Z, with Equation 1. The scaled standoff is an indication of the intensity of the blast loading and enabled the comparison of different blast test results with previous testing on the response of structures subjected to blast effects. 5.1.2.3 Small Standoff Test Results The goal of the small standoff tests was to observe the mode of failure (i.e., flexure or shear) for ten columns with eight different column designs to aide in the development of blast-resistant design and detailing guidelines. The scaled standoff, Z, was varied depending on the column type to ensure that a failure was observed. Observations from each small standoff test are presented below. The high-speed video cameras were extremely helpful in capturing the blast front caused by each explosion, as shown in Figure 50. However, column response was hard to observe directly from the video due to the explosive “fireball” and large dust and smoke clouds caused by each test. Therefore, to record the column response after each test, pictures and post-test mea- surements were taken of each column. Overall, seven of the ten columns experienced a combination of shear and flexural cracking, while three columns sustained severe shear damage. Table 13 summarizes the column design, scaled standoff, and damage level for each small standoff test. Superficial damage entails surface cracking. Minor damage includes widespread surface cracking and some spall of concrete cover. Extensive damage is defined as significant deformation and spalling of concrete cover, though the majority of the con- crete core is still intact and able to carry some axial load. Failure, in this case, means that a shear failure occurred at the base. Specific charge weights and standoffs are not provided for security reasons. Therefore, the charge weight and standoff are shown in terms of the parameters w and z, respectively. 5.1.2.3.1 Column 1A1. The first small standoff test was completed on October 10, 2007. Column 1A1, an 18-in. diameter gravity design column with discrete hoops at 6 in. on center, was tested at a charge weight of 2.8w and the largest standoff, 5.8z, to determine the damage threshold for an 18-in. diameter column. Figure 51 illustrates the combination of shear and flexural cracking experienced by Column 1A1. Flexural cracking was observed at mid-height, while diago- nal shear cracking was seen near the column base. Cracking due to rebound of the specimen responding dynamically was expected for the blast loads; however, no cracks were noted on the front face of the column. Therefore, the column did not rebound far enough to produce tension and cracking on the front face. Overall, the column performed well at this scaled standoff and sustained only superficial damage (sur- face cracking). 5.1.2.3.2 Column 1A2. Column 1A2, with an identical design as Column 1A1, was the second column tested in the experimental research program. The previous test was used as a guide to determine the onset of inelastic deformation. Based on this analysis and due to a desire to understand col- umn response with large inelastic deformations, a 45% smaller standoff than on Column 1A1 was used for Column 1A2. During this test, a similar size charge at a location close to the column, 3.2z, was used to exceed the elastic response limit. The blast resulted in a direct shear failure at the base of 68 Figure 50. Blast-wave propagation during small standoff test. Table 13. Small standoff tests. 1A1 8.5 w 2.6 z superficial 1A2 gravity gravity gravity gravity failure 1B gravity w/ spiral minor 2A1 superficial 2A2 minor 2B gravity w/ spiral superficial 2-Seismic seismic superficial 2-Blast blast blast minor 3A gravity extensive 3-Blast 2.8 w 5.8 z 2.9 w 3.2 z 3.4 w 2.6 z 2.8 w 3.9 z 3.3 w 2.4 z 3.4 w 2.4 z 3.3 w 2.4 z 9.0 w 3.4 z 9.1 w 3.4 z extensive *w = charge weight parameter †z = standoff distance parameter Column Design Charge Weight*, W Standoff†, R Level of Damage

Column 1A2, as seen in Figure 52. A permanent deflection of 5 in. at a height of 21 in. above the ground was noted. Also, spall of the concrete cover was observed on the bottom 16 in. of the column. The brittle shear failure occurred due to the opening of three discrete hoops with standard hooks within the bottom 16 in. After the loss of concrete cover, the stan- dard hooks experienced an anchorage failure, resulting in the loss of core confinement and overall column integrity. The development of the failure zone prevented the column from fully rebounding and forming tensile cracks on the front face. 5.1.2.3.3 Column 1B. Column 1B, an 18-in. diameter gravity design column with a continuous spiral at a 6-in. pitch, was tested at a 17% larger charge weight, 3.4w, and 19% smaller standoff, 2.6z, than Column 1A2. A more intense loading than Column 1A2 was used to evaluate the benefit of spiral reinforcement over discrete ties. A combination of shear and flexural cracking is illustrated in Figure 53. Flexural cracking at mid-height was observed around the entire column, and shear cracking was observed at the base. The flexural cracking at mid-height on the front face of the column is most likely attributable to full rebound of the col- umn due to the intense blast load. Spall of the concrete cover at the base on the side closest to the blast was also noted. Overall, Column 1B performed well with minor damage noted, including widely distributed cracking and some spall. 69 Figure 51. Small standoff blast damage: Column 1A1. Figure 52. Small standoff blast damage: Column 1A2.

5.1.2.3.4 Column 2A1. The first 30-in. diameter col- umn was tested on October 15, 2007. Column 2A1, a grav- ity design column with discrete hoops at 6 in. on-center, was tested with a charge size of 2.8w at a large standoff, 3.9z, to determine the inelastic response threshold for a 30-in. diameter column. The standoff was 32% smaller than the standoff used on Column 1A1. Figure 54 illustrates the combination of shear and flexural cracking experienced by Column 2A1. Flexural cracking was observed at mid- height, while diagonal shear cracking was seen near the col- umn base. Rebound was illustrated by flexural cracking at mid-height on the front face of the column. Overall, the column performed well at this scaled standoff and only sus- tained superficial damage with cracking limited to the cover concrete. 5.1.2.3.5 Column 2A2. Column 2A2, with an identical design as Column 2A1, was tested with a 20% larger charge weight at a 38% smaller standoff than Column 2A1 to study the inelastic behavior of the column. A 3.3w size charge at a standoff of 2.4z was used. The column response was similar to that of Column 2A1, illustrating the robustness of the larger diameter columns exposed to a more intense blast loading. Flexural cracking was observed at mid-height, and diagonal shear cracking was observed at the column base, as seen in Figure 55. Rebound response was noted with flexural cracks on the front face of the column. Spall of the concrete cover closest to the blast at the column base was also observed. Overall, the column sustained minor damage, including wide- spread cracking and some spall. 5.1.2.3.6 Column 2B. Column 2B, a 30-in. diameter grav- ity design column with continuous spiral reinforcement at a 6-in. pitch, was tested at a similar size charge and stand- off as Column 2A2. A 3.4w size charge was placed close to the column at a standoff of 2.4z. A combination of shear and flexural cracking was noted, as shown in Figure 56. Flexural cracking at mid-height was observed on the front and back face of the column, indicating that significant rebound had occurred. Diagonal shear cracks were also observed at the base. Overall, Column 2B performed well with only superficial damage noted, including cracking of the concrete cover. 5.1.2.3.7 Column 2-Seismic. Column 2-Seismic, a 30-in. diameter seismic design column with a continuous spiral at a 3.5-in. pitch, was tested at the same charge weight and 70 Figure 53. Small standoff blast damage: Column 1B. Figure 54. Small standoff blast damage: Column 2A1.

standoff as Column 2A2. Again, a 3.3w size charge was placed near the column at a standoff of 2.4z. A combination of shear and flexural cracking was noted, as shown in Fig- ure 57. Flexural cracking at mid-height was observed on the front and back face of the column, indicating that rebound had occurred. Diagonal shear cracks were also noted at the base. Overall, Column 2-Seismic performed well, with only superficial damage noted, primarily cracking of the concrete cover. 5.1.2.3.8 Column 2-Blast. Column 2-Blast, a 30-in. di- ameter blast design column with a continuous spiral at a 2-in. pitch, was tested with a 2.7-times larger size charge and 1.4-times larger standoff than Column 2-Seismic. During this test, a 9.0w size charge at a physical standoff of 3.4z was used to create the large blast load. Thus, despite having a similar scaled standoff as Column 2-Seismic, the actual loading sce- nario involved a much higher quantity of explosives than the previous tests, thereby allowing for an evaluation of the use of scaled standoff as a design parameter for close-in blast tests on bridge columns. A combination of shear and flexural cracking was noted, as shown in Figure 58. Flexural cracking at mid-height was observed, as well as diagonal shear cracks at the base. Spall of the concrete cover occurred at the column base and at mid-height on the blast side of the column, indi- cating rebound of the column. Overall, Column 2-Blast per- formed well with minor damage noted, including widespread cracking and some spall. 71 Figure 55. Small standoff blast damage: Column 2A2. Figure 56. Small standoff blast damage: Column 2B.

5.1.2.3.9 Column 3A. Column 3A, a 30-in. square gravity design column with discrete ties at 6 in. on-center, was tested at a charge weight and standoff similar to Column 2-Blast. The column was tested with a 9.1w size charge at a standoff of 3.4z. The blast resulted in significant deformation due to shear at the base of Column 3A, as seen in Figure 59. A permanent deflection of 3 in. at a height of 9 in. above the ground was noted. Observations included complete spall of the bottom 17 in. of concrete cover and flexural cracking at mid- height on the front and back face of the column. Overall, the concrete core was still intact even though the column sustained extensive damage at the base. 5.1.2.3.10 Column 3-Blast. Column 3-Blast, a 30-in. square blast design column with discrete ties at 2 in. on-center, was tested at a 7% smaller charge weight and 24% smaller standoff than Column 3A. The small scaled standoff was com- prised of a charge size of 8.5w placed very close to the column at a standoff of 2.6z. Accounting for the actual charge weight and standoff distance, Column 3-Blast was subjected to the most severe loading of all the columns tested in the small standoff tests. The blast resulted in significant deformation due to shear at the base of Column 3-Blast, as seen in Figure 60. A permanent deflection of 6.5 in. at a height of 13 in. above the ground was noted. Also, complete spall of the bottom 16 in. of concrete cover and flexural cracking at mid-height was observed. Overall, the concrete core was still intact even though the column sustained extensive damage at the base, includ- ing significant deformation and spalling of cover concrete. 72 Figure 57. Small standoff blast damage: Column 2-Seismic. Figure 58. Small standoff blast damage: Column 2-Blast.

5.1.2.3.11 Summary of Column Failures. The small standoff tests enabled the observation of the mode of fail- ure for ten concrete columns with eight different column designs over a range of different scaled standoffs (i.e., blast scenarios). During the small standoff testing, three columns exhibited significant shear deformations at the base, includ- ing Columns 1A2, 3A, and 3-Blast, as the result of severe blast loadings. The other seven columns experienced a combina- tion of shear and flexural cracking and an overall less brittle response than the other three columns. The column design, standoff, charge weight, and damage level for each column are summarized in Table 13. The most common mode of fail- ure was shear; however, the majority of columns had ade- quate shear capacity to resist the applied loads, experienced limited spall of concrete cover and essentially no column breach, and were very robust. The blast columns were exposed to a more intense loading than their less reinforced counter- parts while experiencing similar responses. Thus, given the same blast loading scenario, the blast columns are expected to perform better than the respective gravity or seismic columns. 5.1.2.4 Local Damage Test Results In a local damage test, charges were placed very close to or in contact with the test column. The objective of the local damage tests was to observe the spall and breach patterns of blast-loaded concrete columns. Breach is defined as the com- plete loss of concrete through the depth of a given cross- section along the height of a column (i.e., a localized loss of cross-section). In most cases, columns with superficial or minor damage from the small standoff tests were re-used in the local damage tests. Table 14 summarizes the column design, blast load, and damage level of each local damage test. Pictures of each column illustrate the damage incurred by each blast. 5.1.2.4.1 Test 1. The first local damage test was conducted on October 17, 2007, on Column 2A2. Column 2A2 was a 30-in. diameter gravity design column with discrete hoops at 6 in. on-center that sustained minor damage from the small standoff tests. For the local damage experiment, the column was tested with a 3.7w size charge at a standoff of 0.4z. 73 Figure 59. Small standoff blast damage: Column 3A. Figure 60. Small standoff blast damage: Column 3-Blast.

below the blast location resulted in significant permanent deformation and imminent failure of the column. 5.1.2.4.3 Test 3. The third local damage test was on Column 1A1, an 18-in. diameter gravity design column with discrete hoops at 6 in. on-center that sustained superficial dam- age from the small standoff tests. Column 1A1 was tested at the same charge weight and standoff as Test 2. Figure 63 illustrates the complete breach of Column 1A1. A 100% loss of the con- crete core at least one column diameter above and below the blast location resulted in column failure. The concrete core within the splice overlap was completely rubblized, which led to the loss of column (concrete or steel) continuity in that region. 5.1.2.4.4 Test 4. The fourth local damage test was on Column 2B, a 30-in. diameter gravity design column with continuous spiral reinforcement at a 6-in. pitch that sustained superficial damage from the small standoff tests. For the local damage test, the column was tested with a 1.1w size charge at a standoff of 0.4z, identical to Tests 2 and 3. Figure 64 illus- trates the partial breach of Column 2B. The spiral was severed in two locations near the blast source, and the column lost approximately 90% of the concrete core at least one column diameter above the blast location. The continuous longitudi- nal reinforcement used in Column 2B remained in one piece; however, the longitudinal reinforcement sustained permanent deformation at the location of the blast. 5.1.2.4.5 Test 5. The last local damage test was conducted on Column 2-Blast and Column 3A simultaneously. Column 2-Blast was a 30-in. diameter blast design column with a con- tinuous spiral at a 2-in. pitch that sustained minor damage after the small standoff tests. Column 3A was a 30-in. square 74 Table 14. Local damage test summary. 2A2 gravity 0.0 w 0.0 x complete 2A1 gravity 2.8 w 5.8 x partial 1A1 gravity 2.9 w 3.2 x complete 2B gravity w/ spiral 3.4 w 2.6 x partial 2-Blast blast 2.8 w 3.9 x cover spall 3A gravity 3.3 w 2.4 x cover spall *w = charge weight parameter †x = standoff distance parameter ‡Assuming 82% efficiency of ANFO Column Design Charge Weight*, W Standoff†, R Level of Breach Figure 61. Local damage Test 1: Column 2A2. Figure 61 illustrates the complete breach of Column 2A2. The complete breach occurred in a very brittle manner. Loss of 100% of the concrete core at least one column diameter above and below the splice location resulted in the column failure. Therefore, Test 1 on Column 2A2 illustrates that a large charge placed sufficiently close will fail the column, reinforcing the idea that not all columns can be protected for all possible threat scenarios (Winget, 2003). 5.1.2.4.2 Test 2. The second local damage test was on Column 2A1, a 30-in. diameter gravity design column with discrete hoops at 6 in. on-center that sustained superficial damage from the small standoff tests. Noting the failure of Column 2A2 in the first local damage test, Column 2A1 was tested with a much smaller charge weight to determine the onset of localized damage or breaching. Figure 62 illustrates the partial breach of Column 2A1. A combination of three severed discrete hoops and the loss of approximately 70% of the concrete core at least one column diameter above and

gravity design column with discrete ties at 6 in. on-center that sustained extensive damage from the small standoff tests. Column 3A was chosen to allow the evaluation of column geometry during the local damage tests. The columns were tested with a 0.7w size charge at a standoff of 0.7z. 75 Figure 62. Local damage Test 2: Column 2A1. Figure 63. Local damage Test 3: Column 1A1. Figure 64. Local damage Test 4: Column 2B. Figure 65 illustrates the local blast damage on Column 2-Blast. The spall of concrete cover was noted on the column sides only, enabling the concrete core to stay intact. Widen- ing of previous cracks from the small standoff test was also observed. Figure 66 illustrates the local blast damage and final posi- tion of Column 3A. Additional spalling of concrete cover in the vicinity of the blast location was noted on the column sides only, while the back cover remained intact, enabling the concrete core to stay intact. Widening of previous cracks from the small standoff test was also observed. It is important to note that Column 3A would likely not have toppled over if the base was restrained. 5.1.2.4.6 Summary of Column Failures. The local dam- age tests allowed the observation of spall and breach pat- terns for blast-loaded concrete columns. Only two of the six columns (1A1 and 2A2) experienced a complete breach. Columns 2A1 and 2B experienced a significant loss of the concrete core, while the remaining two columns stayed intact. Column 2-Blast and 3A exhibited spalling of the side concrete cover. Table 14 summarizes the observations made during the local damage tests.

5.1.2.5 Discussion of Test Variables The test program included ten half-scale, small standoff and six half-scale, local damage blast tests on eight different col- umn designs. Column specimens were constructed with con- sideration given to five main test variables, including scaled standoff, column geometry, amount of transverse reinforce- ment, type of transverse reinforcement, and splice location. This section relates each test variable to the experimental observations and data to determine design and detailing rec- ommendations for concrete highway bridge columns. 5.1.2.5.1 Scaled Standoff. One of the best ways to improve the performance of blast-loaded reinforced con- crete highway bridge columns is to increase the standoff distance. Increasing the design scaled standoff will decrease the effects of blast loads on columns. Figure 67 compares two identical 18-in. columns tested at different standoffs. The gravity-designed columns used standard hooks, which are a poor detail for blast-loaded columns (see Section 5.1.2.5.4); however, when the standoff is sufficiently increased, detail- ing becomes less significant in controlling response. If only vehicle standoff is limited, small charges may still be placed in direct contact with a column, potentially causing localized damage or breaching of the concrete core. Results from local damage tests, shown in Figure 68, illustrate that increasing the standoff from the face of the structural col- umn by a small amount (on the order of inches) can increase a column’s chance of survival substantially in a situation involving close-in blast loads. Aside from physical barriers, standoff from a structural member can be increased by adding an architectural feature around the structural column. 5.1.2.5.2 Column Geometry. Column cross-sectional shape affects how a blast load interacts with a column. The use of a circular column is an effective way of decreasing the blast pressure and impulse on a column relative to a square col- umn of the same size. Therefore, the use of a circular column cross-section over a square cross-section is recommended. A square column provides a flat surface that reflects the over- pressure directly back toward the source, creating a large reflected pressure. The increase of reflected pressure on a circular column is not as significant as on a square column because the pressure around the perimeter of a circular column is reflected at an angle (with the exception of the centerline position) relative to the direction of shock wave propaga- tion from the blast source. Also, pressures do not clear around the edges of a square column as quickly as they wrap around and fill in behind a circular column. Therefore, a blast load interacting with a circular column will produce a less severe loading than one acting on a similarly sized rec- tangular or square column. Figure 69 illustrates the results of a blast wave propagating around a circular and square col- umn with the same cross-sectional dimension at the same scaled standoff. The square column was pushed over by the blast load, while the circular column remained standing, demonstrating the decrease in load on a circular column due to clearing. Clearing is the process by which a high reflected 76 Figure 65. Local damage Test 5: Column 2-Blast. Figure 66. Local damage Test 6: Column 3A.

pressure seeks relief toward lower pressure regions (free edges) through a rarefaction (or relief) wave that propagates from the low to the high-pressure region. Using a circular column cross-section over a square cross- section can lead to a decrease in the net resultant impulse the column must resist, as shown in Figure 70. Section 5.1.1.1 discusses the physical phenomena that contribute to the reduction in impulse experienced by a circular column as compared to that of a square column, and limited test data suggest that the reduction in impulse can be as much as 34% at certain locations for small standoffs. Although addi- tional testing is needed to validate these findings, a circular column with the same design and detailing requirements as a square column is expected to have less damage for a threat with the same scaled standoff. If the standoff distance cannot be increased to sufficiently decrease the effects of blast loads on columns, then the fol- lowing design and detailing provisions are recommended: 77 (a) (b) Figure 67. Importance of scaled standoff: a) large standoff (Column 1A1), b) small standoff (Column 1A2). (a) (b) Figure 68. Effectiveness of small-scaled standoff: a) Test 4, b) Test 5.

increasing the amount of transverse reinforcement, requiring continuous spiral reinforcement or discrete hoops with suffi- cient anchorage, and avoiding splices. These design provisions are covered in detail in the following sections. 5.1.2.5.3 Amount of Transverse Reinforcement. Experi- mental observations show that increasing the volumetric reinforcement ratio is beneficial to the response of blast- loaded columns because it increases the column ductility and shear capacity. Direct shear is a major concern for blast- loaded columns, as shown in Figure 71. To reduce the chances of a potential shear failure, the “blast” columns (specimens 2-Blast and 3-Blast) were designed to force the formation of plastic hinges (a flexural failure). Similarly to seismic design, all potential plastic hinge locations were considered to ensure the maximum possible shear demand. During the design process, a propped-cantilever column with a triangular load, 78 Figure 69. Clearing around a square and circular column (side view). IcircularIsquare (a) (b) Figure 70. Relative net impulse on column cross-section: a) square column, b) circular column. (a) (b) Figure 71. Shear deformation of blast-loaded columns: a) 18-in. round column (1A1), b) 30-in. square column (3A).

shown in Figure 72, was assumed to form two hinges before failure. The plastic hinge analysis determined a maximum shear of: where: MP = plastic moment capacity (kip-ft) L = height of column (ft) In comparison, a typical seismic column (shown in Fig- ure 73) that is fixed–fixed with a displacement at one end due to a laterally applied force will also form two plastic hinges. The maximum shear from a plastic hinge analysis is: where: MP = plastic moment capacity (kip-ft) L = height of column (ft) These plastic hinge analyses clearly demonstrate the large shear demand on blast-loaded columns, which can be greater than four times the shear demand from seismic loads. V M L base P = 2 15( ) V M L base P = 9 14( ) Using the shear equations given above, a required pitch of transverse reinforcement was determined by setting Equa- tions 14 and 15 equal to the shear design equations from the AASHTO LRFD (2007), modified to account for strain rate effects (ASCE, 1997), and solving for the spacing (Table 15). The plastic moment (MP), which is equal to the flexural capac- ity of the cross-section (Mn), should also account for the dynamic material strength, with dynamic increase factors for strain rate effects. Section 5.7.4.6 of the AASHTO LRFD (2007) requires a minimum transverse reinforcement ratio for a circular gravity- loaded column of: where: f ′c = specified compressive strength of concrete at 28 days (psi) fy = yield strength of reinforcing bars (psi) Ag = gross area of column (in.2) Ac = area of concrete core (in.2) Section 5.10.11.4.1(d) of the AASHTO LRFD (2007) requires a minimum transverse reinforcement ratio for a circular seismic column and minimum area of transverse ρs g c c y A A f f ≥ −⎛⎝⎜ ⎞ ⎠⎟ ′ 0 45 1 16. ( ) 79 INFLECTION POINT TRIANGULAR BLAST LOAD TRIANGULAR BLAST LOAD PLASTIC HINGE PLASTIC HINGE MP MP L ˜ 0. 55 L (a) (b) (c) Figure 72. Plastic hinge analysis for blast-loaded column: a) deflected shape, b) plastic hinge locations, c) plastic moment.

80 Table 15. Pitch of transverse reinforcement. End Condition Scale M N =M P (kip-ft) V u,base (kip) V c (kip) V c = 0 V c > 0 Pin-Fixed (blast) 1 : 2 584 592 90 0.9 1.0 Fixed-Fixed (seismic) 1 : 2 584 104 90 5.0 37.8 Seismic Column Capacity Pitch (in.) reinforcement for a rectangular seismic column, respec- tively, of: 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.) A sh f f sh c c y ≥ ′0 12 18. ( ) ρs c y f f ≥ ′0 12 17. ( ) Equation 16 controlled the transverse reinforcement design of all gravity-designed columns (Type-A and Type-B), while Equations 17 and 18 controlled the design of all seismic- and blast-designed columns. Though not initially anticipated, it is important to note that the Type-A columns did not meet the requirements of the AASHTO LRFD (2007) Article 5.7.4.6 as they were constructed because the actual steel strength (fy = 50ksi) was lower than the specified strength used for design, 60 ksi. Equation 17, with a coefficient of 0.18, is recommended as the minimum transverse reinforcement ratio for all circular blast-designed columns, while Equation 18, with a coefficient of 0.18, is recommended as the minimum area of transverse reinforcement for all rectangular blast-designed columns. Columns meeting these minimums tested at a small standoff sustained minor and extensive damage (see Figure 74); how- ever, the core still remained intact and the column could still carry load. Essentially, 50% more confinement is rec- ommended for blast-designed columns over current seismic provisions to improve the ductility, energy absorption, and dissipation capacity of the cross-section. Using the experimental test observations and data, a new minimum transverse reinforcement ratio and area of transverse reinforcement is recommended, with the following equations for a circular and rectangular blast-loaded column, respectively. (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 73. Plastic hinge analysis for seismic column: a) deflected shape, b) plastic hinge locations, c) plastic moment.

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.) A sh f f ssh c c y ≥ ′0 18 20. ( ) ρs c y f f ≥ ′0 18 19. ( ) 5.1.2.5.4 Type of Transverse Reinforcement and Dis- crete Tie Anchorage. Experimental observations show that continuous spiral reinforcement performs better than discrete hoops for small standoff tests. Figure 75 compares two 18-in. columns tested at a similar standoff, with the only variable being the type of transverse reinforcement. Figure 75 illustrates the benefit of spiral reinforcement in the response of blast-loaded columns. The continuous reinforcement better confines the core at the base where a shear failure is most likely to occur. Therefore, continuous spiral reinforcement is recommended for blast-loaded columns. Performance of discrete hoops can be improved to a level equivalent to that of the continuous spiral reinforcement 81 (a) (b) Figure 74. Columns meeting minimum transverse reinforcement require- ments: a) minor damage (2-Blast), b) extensive damage (3-Blast). (a) (b) Figure 75. Importance of continuous transverse reinforcement: a) discrete hoops (Column 1A2), b) continuous spiral (Column 1B).

with better anchorage into the concrete core. The column with discrete hoops in Figure 75 experienced a shear failure at the column base because the bottom three hoops were pulled open. The discrete hoop anchorage was designed in accor- dance with Section 5.10.2.1 of the AASHTO LRFD Bridge Design Specifications (2007) and used standard hooks with a “90° bend, plus a 6.0 db extension at the free end of the bar.” Section 5.10.2.1 defines standard hooks for transverse rein- forcement as one of the following: a) No. 5 bar and smaller: 90° bend, plus a 6.0 db extension at the free end of the bar b) No. 6, No. 7, and No. 8 bars: 90° bend, plus a 12.0 db exten- sion at the free end of the bar c) No. 8 bar and smaller: 135° bend, plus a 6.0 db extension at the free end of the bar where: db = nominal diameter of the reinforcing bar (in.) The standard hooks shown in Figure 76 do not suffi- ciently anchor the discrete hoops into the concrete core of a half-scale, blast-loaded column. Recent work by Bae and Bayrak (2008) on the seismic performance of full-scale, reinforced concrete columns demonstrated the opening of seismic discrete ties using hooks with a 135° bend, plus an extension of 8.0 db. The AASHTO LRFD (2007) Section 5.10.2 defines seismic hooks as a “135°-bend, plus an extension of not less than the larger of 6.0 db or 3 in.” Bae and Bayrak (2008) noted that, “unlike the full-scale concrete columns, the hooked anchorages often reach close to the center of the core concrete in scaled column specimens.” Similar obser- vations apply to the half-scale specimens used in this research program. Bae and Bayrak (2008) used a minimum hook length of 15.0 db for the remaining full-scale tests to prevent the opening of hoops. The researchers stated that the larger “hook length proved to be very effective, and opening of the 135° hooked anchorages of the ties was not observed in the other tests.” The blast-loaded column in Figure 77 also used discrete ties with a larger hook length than required by the current AASHTO provisions and did not experience any anchorage issues; however, construction of the specimen was challenging. The square column in the figure utilized sufficient transverse reinforcement and hooks with a 135° bend, plus a 20.0 db extension at one free end of the bar, reversing directions each spacing. To avoid anchorage pull- outs and to improve the performance of blast-loaded (and seismically loaded) columns with discrete hoops or ties, longer hook lengths than currently specified are recom- mended. Properly anchored hooks for blast loads should 82 6db (a) (b) Figure 76. Discrete hoop with standard hook (Column 1A2): a) before test, b) anchorage pullout. c D Dcore 6d b 20d b 90 ° 135° D (a) (b) Figure 77. Sufficiently anchored discrete ties (Column 3-Blast): a) discrete ties, b) discrete tie diagram.

consist of a 135° bend, plus an extension of not less than the larger of 20.0 db or 10 in. 5.1.2.5.5 Location of Longitudinal Splices. If the splice location is at or near the blast location, there is a possibility of breaching at the splice for cases of very small scaled standoffs. Breach is defined as the complete loss of concrete through the depth of a cross-section. Specific observations regarding the performance of longitudinal splices are not provided due to the limited data collected from these tests during the test pro- gram. Additional research is needed to fully characterize splice behavior at locations very close to applied blasts. 5.2 Observations from Analytical Programs Analytical models for load and response were used to sim- ulate the tests carried out during the Phase I and Phase II test programs and to carry out parametric studies. Results from these efforts are described in this section. Experimental observations show that direct shear early in time is a prominent failure mode for columns without ade- quate shear reinforcement, as seen in Figure 78. A simplified SDOF analysis was used to determine the shear reinforce- ment needed for each column to force a flexural mode of response. Strain data from the small standoff tests and data processing are presented below. The strain data were used to determine the actual boundary conditions and load distribu- tion of each small standoff test. 5.2.1 Strain Data Strain gauge data were collected from each small standoff test at six different locations. The strain gauge data were filtered manually by removing all outliers from the data set to deter- mine the maximum strain and time of occurrence. An outlier was defined in two ways: as a strain value exceeding the strain gauge capacity (30,000 µStrain) or as a maximum or minimum strain value with less than two points leading up to it. Other fil- tration methods, such as a low pass filter in Matlab (Mathworks, 2008) and DPLOT (HydeSoft, 2008), were evaluated to help smooth the data; however, these methods altered the data too significantly, cutting out multiple peaks. Therefore, the data were manually filtered as illustrated in Figure 79, which was more appropriate when focusing on the maximum strain val- ues of the strain–time history. To check the maximum values obtained in the manual fil- ter process, project researchers calculated likely times at which the maximum flexural response would occur, and it was apparent by the lack of data at those times that many of the gauges failed prior to a flexural response, indicating an early gauge failure. Tests on Columns 1A1, 2A1, 2A2 and 2B survived long enough and had readings that lasted through the calculated flexural response time to provide sufficient data for the evaluation of flexural response. Large early time responses were also noted near the time of arrival of the blast wave, before the pier had enough time to react to the impulse load. These early time responses were most likely gauge wire responses to the initial shock transmission through the columns. The early peak strain times were practically identical for all gauge locations because the strain gauge wires were bundled as they ran through the column base and were all excited at the same time as the shock crossed the bundle. Thus, the early time responses were essentially an indication of the time of arrival for the direct shock transmission and were not used in determining the peak strain. Flexural strain data from four of the small standoff tests (1A1, 2A1, 2A2, 2B) were used to evaluate the actual bound- ary conditions and load distribution. Filtered flexural strain data are shown in Figure 80; to better show the data trend, the rolling average over five points is also shown. The other six tests experienced early gauge failure and typically recorded a peak in less than one millisecond of starting data collection, as illustrated by Figure 81. The majority of the columns with early gauge failures experienced a shear failure at the base. Thus, a flexural response was not able to be developed within this millisecond time period of data collection before the shear failure occurred. 5.2.2 Actual Boundary Conditions Before completing the SDOF analyses to determine the shear capacity of each column, the experimental data were used to assess the actual boundary conditions experienced by each column during the blast tests. For design, a propped- cantilever was assumed as the boundary conditions for each column. To verify the actual boundary conditions developed 83 Figure 78. Direct shear failure.

84 -5000 -4000 -3000 -2000 -1000 0 1000 2000 3000 0 10 20 30 40 50 60 70 80 90 100 Time (msec) µS tra in Strain Data Filtered Strain Data Figure 79. Filtered longitudinal reinforcement (flexural) strain data. -1500 -1000 -500 0 500 1000 1500 2000 2500 0 10 20 30 40 50 60 70 80 90 100 Time (msec) µS tra in Filtered Strain Data Rolling Average Filtered Strain Data Figure 80. Longitudinal reinforcement (flexural) strain data.

85 -1500 -1000 -500 0 500 1000 1500 2000 2500 0 10 20 30 40 50 60 70 80 90 100 Time (msec) µS tra in Filtered Strain Data Rolling Average Filtered Strain Data Figure 81. Longitudinal reinforcement (flexural) strain data: early gauge failure. by the test frame, flexural strain data from four of the small standoff tests (1A1, 2A1, 2A2, 2B) were used. The peak flexural strain was recorded for each strain gauge on the longitudinal reinforcement. At the same time, a moment–curvature plot for each column, which depends upon the column shape, the cross-sectional dimension, and the amount of transverse reinforcement, was calculated using Response 1990 (Collins and Bentz, 1990). Actual material properties multiplied by dynamic increase factors for strain rate effects were used to develop each moment–curvature diagram. From the moment–curvature plots, the moment corresponding to the peak flexure strain data and the plastic moment capacity were determined. The columns were originally designed as propped- cantilevers; however, the measured strain data did not support this assumption. Figure 82 illustrates the moment diagram cor- responding to a linear load distribution acting on a propped- cantilever and a simply supported column. All of the measured flexural peak strain data were positive, resulting in only positive moments. A propped-cantilever column, however, creates a negative moment at the fixed end (i.e., compression on the back face of the column). From these data, it can be concluded that insufficient fixity was provided at the base to provide the propped-cantilever boundary conditions assumed for design. While some partial fixity was likely, it has been assumed for the purposes of analysis that the column behaved as though it were simply supported because rotation could occur at the base. This assumption greatly simplifies the response analyses and agrees well with the recorded data. For each column, the shape of the moment diagram was approximated using the location of the strain gauges and moments corresponding to the measured strains from the moment–curvature plots. 5.2.3 Actual Load Distribution The experimental data were also used to determine the actual load distribution experienced by each column during the blast tests. As indicated in the previous section, the boundary conditions for each column can be reasonably approximated with simple supports. Flexural strain data from four of the small standoff tests (1A1, 2A1, 2A2, 2B) were used to determine the approximate load distribution. The deformed shape from the static application of the assumed load profile is the primary deflected shape that the column is expected to experience during its dynamic response (Biggs, 1964). According to Biggs (1964), the mode shape cor- responding to the static application of the load is expected to be the dominant one for members that experience an inelas- tic response, which was true for all the columns tested during this project. To verify the actual load distribution developed by the blast-wave propagation, each column was analyzed using a uniform and triangular load distribution. Figure 83 illustrates the deflected shape and maximum moment for a simply sup- ported column with each load distribution. The type of load distribution affects the location of maximum moment.

86 W, TRIANGULAR BLAST LOAD MOMENT DIAGRAM ˜ 0 . 6 L W, TRIANGULAR BLAST LOAD MOMENT DIAGRAM L L ˜ 0 . 23 L ˜ 0 . 4 L wL 2 3 100 wL 2 8 125 wL 2 8 125 (a) (b) Figure 82. Moment diagram vs. boundary condition: a) propped-cantilever, b) simply supported. ˜ 0 . 58 L W, TRIANGULAR BLAST LOAD M MOMENT DIAGRAM ˜ 0 . 5L w, UNIFORM BLAST LOAD L L DEFLECTED SHAPE MOMENT DIAGRAM DEFLECTED SHAPE P M P (a) (b) Figure 83. Blast-load distribution: a) triangular, b) uniform.

The magnitude of the maximum moment was assumed to be equal to the plastic moment or column capacity from the moment–curvature plots. The moment at each strain gauge was then calculated using the gauge location and assumed moment diagram. The percent difference between the exper- imental moment and calculated moment for each support condition and load distribution were compared to determine the best match. Overall, a simply supported column best represents the boundary conditions in this test program. The load distribu- tion depends on the scaled standoff. For a large scaled stand- off (Z ≥ 2.0 ft/lb1/3, Column 1A1), a uniform load best approximates the blast load distribution; however, for a small scaled standoff (Column 2A1, 2A2, and 2B), a load that varies linearly from zero to wo, the maximum magnitude of an assumed load curve, along the height of the column is a bet- ter approximation than the uniform case. 5.3 Summary In this chapter, primary observations made from analytical studies and the experimental test programs were presented. The variation in blast loads on square and circular bridge columns were described and compared with predictions devel- oped from empirical relationships and available software. In general, it was found that the actual pressures and impulses experienced by blast-loaded columns were less than those predicted by available methods, particularly near the top of the columns studied. Furthermore, the test data suggest that the clearing of shock waves around slender components is a complicated process, and additional research is needed to better characterize these effects. Results from the Phase II experimental program were also presented in this chapter. The significance of each test vari- able, based on results from each small standoff and local damage blast test, was presented, and a description of dam- age observed from each test was given. The flexural strain data from the small standoff tests was presented and used to determine the actual boundary conditions and blast-load distributions that occurred during the test program. Based on the observations from the testing and analytical research, blast-resistant design and detailing guidelines for reinforced concrete highway bridge columns are presented in the next chapter. 87