Guidelines for Shielding Bridge Piers (2018)

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.

3 Literature Review 2.1 Guidelines and Specifications 2.1.1 AASHTO Bridge Design Specifications The guidance for protecting structures from vehicle impacts has evolved over the past several decades, as reflected in AASHTO bridge design specifications. The first edition of the AASHTO LRFD Specifications for the Design of Highway Bridges was published in 1994. The AASHTO Subcommittee on Bridges and Structures (SCOBS) voted to stop maintain- ing the Standard Specifications for Highway Bridges in 1999. Subsequently, AASHTO and the FHWA set a transition date of October 1, 2007, after which all new bridges were to be designed using the LRFD Bridge Design Specifications. Furthermore, by October 2006, the LRFD Specifications were to be used exclusively in the design of all replacement structures [FHWA 2013]. Twenty years of excerpts related to bridge piers from the Standard Specifications and LRFD Specifications are summa- rized in the following. These excerpts show the evolution of the design and protection of piers for vehicle impacts. (The 1992 and 2002 quotations are from the Standard Specifications, while the 1998 and 2007 are from the LRFD Specifications.) 1992 Standard Specifications: “When the possibility of collision exists from highway or river traffic, an appro- priate risk analysis should be made to determine the degree of impact resistance to be provided and/or the appropriate protection system” [AASHTO 1992]. 1994 LRFD: “Abutments and piers located within a distance of 30.0 ft to the edge of roadway, or within a distance of 50.0 ft to the centerline of a railway track, shall be designed for an equivalent static force of 400 kip” [AASHTO 1994b]. 1998 LRFD: “Pier columns or walls for grade separation structures should be located in conformance with the clear zone concept as contained in Chapter 3 of the AASHTO Roadside Design Guide, 1996. Where the practical limits of structure cost, type of structure, volume and design speed of through traffic, span arrangement, skew, and terrain make conformance for the Roadside Design Guide impractical, the pier or wall should be protected by the use of guardrail or other barrier devices. The guardrail or other device should, if practical, be independently supported, with its roadway face at least 600 mm from the face of pier or abutment, unless a rigid barrier is pro- vided. The face of the guardrail or other device should be at least 600 mm outside the normal shoulder line” [AASHTO 1998]. 2002 Standard Specifications: “Pier columns or walls for grade separation structures shall generally be located a minimum of 30 ft from the edges of the through-traffic lanes. Where the practical limits of structure cost, type of structure, volume and design speed of through traffic, span arrangement, skew, and terrain make the 30-ft offset impractical, the pier or wall may be placed closer than 30 ft and protected by the use of guardrail or other barrier devices. The guardrail or other device shall be independently supported with the roadway face at least 2 ft 0 in. from the face of pier or abutment.” The face of the guardrail or other device should be at least 2 ft 0 in. outside the normal shoulder line” [AASHTO 2002b]. 2007 and 2010 LRFD: “Unless protected . . . abutments and piers located within a distance of 30.0 ft. to the edge of roadway . . . shall be designed for an equivalent static force of 400 kips, which is assumed to act in any direc- tion in a horizontal plane, at a distance of 4.0 ft. above ground” [AASHTO 2007]. “In order to qualify for this exemption, such barrier shall be structurally and geo- metrically capable of surviving the crash test for Test Level 5. . . .” [AASHTO 2010]. Although a risk analysis was suggested in 1992, the nature of that analysis was not specified. The suggestion to perform C H A P T E R 2

4 a risk analysis, however, implies that quantifying the likeli- hood of a catastrophic impact is a key consideration, and that consideration would certainly depend on the particular site conditions. The 400-kip design load that first appeared in 2007 was based on information from crash tests with an 80,000-lb tractor-trailer truck and load estimates for train impacts developed by Hirsch [Hirsch 1985]. The specification also suggested that the pier should be at least 10 ft behind a shielding barrier to prevent vehicles that lean or roll over the barrier from striking the pier. The fifth edition of the AASHTO LRFD Bridge Design Specifications retained this same language [AASHTO 2010]. In March 2012, the sixth edition of the AASHTO LRFD Bridge Design Specifications made significant changes to the design for impact resistance of bridge piers, largely based on work by Buth et al. at the Texas Transportation Institute [AASHTO 2012, Buth 2010, Buth 2011]. For example, the design load for pier components was increased from 400 kips to an equivalent static force of 600 kips, and the height of load application was increased from 4 to 5 ft above the ground. The impact load was changed from any direction in the fourth and fifth editions to between 0 and 15 degrees from the traveled way in the sixth edition. The requirement to design for train collisions was removed in the sixth edition, although at least one state (Massachusetts) was considering keeping the train provision. The pier protection provisions of the eighth edition [AASHTO 2017] have not been changed since the sixth edition. Prior to the sixth edition, designing for structural resistance to collision forces was not necessary if a MASH Test Level 5 (TL-5) barrier was provided to protect the pier system. A new provision was added in the sixth edition that allows for the consideration of the annual probability of the bridge pier to be hit by a heavy vehicle [AASHTO 2012]. When the annual frequency of a heavy-vehicle collision is less than 0.0001, the pier does not need to be protected or designed for impact resistance. The sixth edition includes a method for assessing the annual frequency of a heavy-vehicle impact (i.e., AFHBP) [AASHTO 2012]: = i i iAF 2 ADTT 365HBP HBPP where AFHBP = Annual frequency for a bridge pier to be struck by a heavy vehicle; ADTT = One-way volume of trucks per day, where the percent trucks is assumed to be 10% of the total average daily traffic (ADT); and PHBP = Probability of a bridge pier being struck by a heavy vehicle. A table is provided in the commentary of Article 3.6.5 that contains values of the PHBP listed by ADTT and highway type. The probability PHBP is constant for each type of highway. For example, the PHBP for a divided tangent highway is 1.09E-09, and for an undivided tangent highway is 3.457E-09. The tabulated values do not account for the position of the pier with respect to the roadway or characteristics of the highway that are expected to affect the frequency of encroachments (e.g., grade, horizontal curvature, lane width), so the method does not adequately account for traffic conditions and site layout. In addition, the sixth edition method does not account for any potential redundancy in the pier system or continuity of the bridge superstructure that would affect the likelihood of a bridge collapsing in a heavy-vehicle impact. The changes in the design loading in the sixth edition (i.e., 400 kips at 4 ft above the ground from any direction to 600 kips at 5 ft above the ground at an impact angle of between 0 and 15 degrees) were based on two full-scale crash tests performed at the Texas Transportation Institute (TTI) that showed that an 80,000-lb tractor-trailer truck striking an instrumented bridge column at 0 degrees and 50 mph produced forces just slightly over 600 kips [Buth 2011, AASHTO 2012]. In the past 25 years, the LRFD Bridge Design Specifications have evolved toward a more probabilistic approach to dealing with heavy-vehicle impacts with bridge piers, but the current method does not consider many important characteristics that are likely to affect the frequency and severity of pier col- lisions. Traffic volume, traffic mix, and speed are important predictors, as are highway geometrics such as grade and hori- zontal curvature. The position of the pier components with respect to the travel lanes is also a notable feature that will affect the likelihood of a pier crash. While the probabilistic method in the sixth edition was a significant step forward, it still is relatively simple with respect to what is known about how and when vehicles encroach on the roadside and the effects of site, traffic, and roadway conditions on the likelihood of vehicles striking the piers. 2.1.2 The States The states have adopted a variety of approaches to pro- tecting bridge abutments and piers, as documented in state bridge design manuals. A survey of bridge design engineers included in Appendix C: Survey of Practice indicated that approximately 60% of the respondents used the AASHTO LRFD Bridge Design Specifications procedures of Article 3.6.5 for pier protection guidance, and 22% stated that they used something else (note: some of the respondents were inter- national). The current guidelines, therefore, are widely if not universally applied. A review of several state guidelines provides some understanding of the different approaches

5 currently employed and the material referenced in the devel- opment of these approaches. Chapter 13 of the Wisconsin Department of Transportation Bridge Design Manual states that piers shall resist loads applied directly to them, including vehicle impact loads [WIDOT 2013a]. For new structures, the Wisconsin Department of Transportation Bridge Design Manual states that it is preferred that the required clear zone be provided so that a barrier is not needed. W-beam barriers may only be used for passenger- vehicle occupant protection (i.e., not structure protection) and must be offset a minimum of 4.5 ft from the pier. Concrete safety shaped barriers may be used for passenger-vehicle occupant or structural pier protection and must be offset 4.5 ft from the pier. If used for structural pier protection, barriers must be a minimum of 54 in. tall; if used for vehi- cle occupant protection, barriers must be a minimum of 42 in. tall. A 51-in. vertical wall may be used for passenger- vehicle occupant protection with zero offset from the pier [WIDOT 2013b]. The Florida DOT design manual references the AASHTO LRFD specifications when a pier or bent is less than 30 ft from the edge of a traffic lane (i.e., within the clear zone). A TL-5 concrete safety shaped barrier is installed for a minimum of 50 ft parallel to the roadway upstream of the leading edge of the pier. The barrier must be 54 in. tall, and there must be a minimum offset of 2 ft from the face of the barrier to the face of pier column or pile [FDOT 2013a]. Allowances are made in FDOT index 410 when the minimum offset from the barrier to the pier cannot be achieved [FDOT 2013b]. The Ohio DOT design manual states that abutments and piers located within 30 ft of the edge of the roadway should be designed for impact from a vehicle through an equivalent static force of 400 kips applied at 4 ft above the ground, unless the pier system is redundant, protected by an embankment, or protected by a structurally independent TL-5 barrier. When the barrier is within 10 ft of the abutment or pier, the barrier must be 54 in. tall, while it may be 42 in. tall if the barrier is located more than 10 ft from the abutment or pier. Redundant pier systems should be shielded to protect passenger vehicle occupants when within the clear zone [ODOT 2013]. Minnesota distributed a memo in 2007 to bridge engi- neers explaining that MNDOT “considers Section 3.6.5 of the 2012 LRFD Bridge Design Specifications to be overly restrictive because . . .” consideration is not given to the probability of vehicle collision, the vehicle mix, or the travel speed [MNDOT 2007a, AASHTO 2012]. MNDOT exempted designers from protecting piers with a minimum of three columns, with design speeds less than or equal to 40 mph, or with design speeds over 40 mph but with ADTT of less than or equal to 250 per day. Designers were instructed to assume, when vehicle mix is not available, that ADTT is 10% of ADT. For non-exempt bridges, options were provided for piers with one, two, or three or more columns. The options included (1) protecting the piers with a 54-in.-tall TL-5 bar- rier if the distance between the pier and the barrier is 10 ft or less, or a 42-in.-tall TL-5 barrier if the face of the pier is more than 10 ft from the barrier; (2) design the columns to have an area greater than 30 ft2 and for a 400-kip collision load; or (3) provide a “crash strut” between pier columns, and design the strut for a 400-kip collision load. When the piers have three or more columns and are considered redundant, the preference of MNDOT was to provide a crash strut designed for a 400-kip collision load. The next preference was to design each column for impact or protect the piers with TL-5 barriers. Designers could have also verified that the structure would not collapse if any single column was removed [MNDOT 2007a; MNDOT 2013]. MNDOT updated this guidance in 2016 [MNDOT 2016]. The state of New Jersey designates the type of longitudinal barrier warranted based on median width. Provided that the median protection warrant is met, for median widths of up to 12 ft, a TL-4 concrete barrier may be used. A TL-4 concrete barrier is the preferred treatment for median widths ranging from 13 to 26 ft, but w-beam or thrie-beam barriers may also be used. For median widths above 26 ft, w-beam or thrie- beam barriers are preferred. When a guardrail is used and piers are present but vertical curbs are not, the minimum pier offset from the edge of roadway, regardless of shoulder width, should be 8.25 ft, with 4 ft from the back of the guardrail to the pier, or 4.75 ft if the rail is attached to the pier. When the TL-4 concrete barrier is warranted and piers are present, a minimum 3.25-ft offset from the face of the barrier to the face of the pier should be observed to prevent vehicles that roll over from striking the pier [NJDOT 2013]. Most DOTs treat bridge piers as they would any other fixed object (e.g., trees, utility poles, high-mast lighting) in the clear zone, stating that piers within the clear zone should be shielded by a barrier. Iowa DOT suggests that on high-speed, multilane facilities, piers located outside of the clear zone should also be considered for shielding [IDOT 2013]. South Dakota DOT provides specific barrier warrants for all fixed objects within the clear zone on road- ways of various traffic volumes and speeds [SDDOT 2013]. The New York State DOT also uses the clear zone as the determining factor for installing longitudinal barriers but notes that the LRFD guidance for protecting piers from truck impacts was anticipated at the time the document was pub- lished [NYDOT 2013]. 2.1.3 Summary There is some variation among the states in addressing the need for protection of bridge piers, although much of the variation appears to be related to which edition of the

6 AASHTO LRFD Bridge Design Specifications has been used as the basis for the state bridge design standards. A few states provide different test-level barriers for different offsets of the piers and travel lanes. No specific guidance was found for the treatment of bridge piers on existing structures that cannot be designed for impact with heavy vehicles since they are already constructed. All states consider bridge piers as hazardous fixed objects that require shielding when they are located within the clear zone and address such situations with the guidance contained in the AASHTO RDG. The RDG in this context is focused on protecting vehicle occupants from the piers. Less atten- tion has been given to protecting the piers from impacts with vehicles, particularly heavy vehicles traveling at high speeds. Examples of efforts used to protect piers from vehicles are designing piers for direct impact forces and using offsetting barriers in front of piers to allow for vehicle roll rotation. 2.2 Capacity, Design, and Impact Loading of Bridge Piers 2.2.1 Pier Component Capacity A bridge pier can be composed of a variety of structural components. The pier could be a simple wall that supports the bridge superstructure, or it could be a series of columns that support a pier cap in a bent arrangement that in turn supports the bridge superstructure. Failure or collapse of the pier system, then, is dependent on the strength of the component that is struck in an impact, the redundancy of the pier system, and the continuity of the superstructure. For example, a three-column bent system might be designed such that the pier is still stable even if one of the columns collapses. On the other hand, a two-column bent system will generally not be stable if one column collapses. Buth et al. calculated the shear capacity of a variety of circular pier columns using the fourth edition of the LRFD Bridge Design Specifications; these are summarized in Figure 3. Buth et al. also estimated the shear capacity of the pier columns for each of the 19 crashes summarized later in Table 2. Most of the real-world crashes investigated by Buth et al. involved circular, 30-in.-diameter columns with eight #9 longitudinal bars and #2 spiral stirrups with a 6-in. pitch using Grade 60 steel and 4 ksi concrete. These columns had an estimated unfactored lateral capacity of about 88 kips, far below the 400-kip recommendation of the fourth edition or the 600-kip recommendation of the sixth edition. 2.2.2 Pier Impact Loading El-Tawil examined the impact forces experienced by two types of bridge columns commonly used in Florida: (1) a 54-in. square column with 24 #11 longitudinal bars and #5 stirrups spaced at 5 in., and (2) a 43-in.-diameter circular bridge column with 14 #11 longitudinal bars and #5 stirrups spaced at 5 in. [El-Tawil 2004]. Unfortunately, El-Tawil did not calculate the capacity as per the LRFD Specifications but rather compared the dynamic forces to the equivalent static force (ESF), which he defined as the static force that results in the same deformation of the structure at the point of load application. El-Tawil found that the computed equivalent static force varied linearly with the approach speed of pickup trucks and single-unit vehicles in a 2004 study of vehicle collisions with bridge piers using the nonlinear dynamic finite element code LSDYNA (see Figure 4). El-Tawil concluded Figure 3. Pier column capacity as a function of column diameter [after Buth 2010].

7 that the AASHTO vehicle collision provisions were not adequate because the peak dynamic force (PDF) is always more than the AASTHO collision provisions, and the ESF exceeds the AASHTO provisions between impact speeds of 75 and 100 km/h. The PDF, however, is not a suitable force for comparison to the quasi-static force used in design since it is of extremely short duration. Most impact researchers prefer to use a force based on the 50-ms average acceleration (i.e., PFMSA in Figure 4 and El-Tawil) to estimate the quasi-static forces. The line labeled “AASHTO” in Figure 4 corresponds to the 400-kip recommendation of the fourth edition of the LRFD Specifications. El-Tawil concludes that the 400-kip design load is not adequate since the ESL exceeds this value at velocities over about 80 km/h (50 mph), but it is significant that the peak 50-ms average force remains below 1,800 kN (i.e., 400 kips) even at a velocity of 85 mph (135 km/h). The loads shown in Figure 4 are based on a standard pickup truck, so they suggest relatively little about the impact loads associated with heavier vehicles. Buth et al. conducted finite element analyses of heavy- vehicle impacts with bridge piers, concluding that the “impact force experienced by the pier is much larger than that stated in the AASHTO LRFD vehicle collision provisions. The values of the imparted force from the engine block impact ranges from 480 kip to 600 kip, while the values of the imparted force from the ballast impact (albeit through the squeezing of the cab) ranges from 480 kip to more than 2,000 kip” [Buth 2010]. Buth et al. also concluded that the forces vary with speed, as shown in Figure 5 and Figure 6. Buth, like El-Tawil, concluded that the 400-kip design force was prob- ably lower than some reasonably likely vehicle impact forces. Buth’s analysis also shows that the forces are affected by the character of the ballast in the truck. Rigid ballast resulted in 50-ms-average forces that were on the order of three times higher than the forces observed when the ballast was deform- able. This shows that the forces experienced by a bridge pier in an impact are not only determined by the mass, speed, and impact angle of the vehicle but are also dependent on the rigidity of the cargo and its ability to deform or shift in the trailer. A heavy-vehicle impact is not one impact but a sequence of impacts with the truck tractor, the trailer, and the trailer’s load. Buth et al. also were able to perform two full-scale crash tests using a 36-in.-diameter rigid instrumented bridge col- umn [Buth 2011]. The instrumented column was struck by an 80,000-lb tractor-trailer truck traveling at 50 mph. In the first test, the truck struck the column with its centerline misaligned by about 2 ft, so a second test was run where the centerline of the truck corresponded with the center of the instrumented column. As shown in the force–time history for the second test in Figure 7, the force exceeded 600 kips for a very short time at two times early in the impact event. Based on these results, Buth et al. recommended that the 400-kip design load used in the fourth edition of the LRFD Specifications be increased, and this was the basis for the change in the sixth edition to the current 600-kip design lateral capacity [Buth 2011, AASHTO 2012]. Buth et al. also found that the height of load applica- tion in the crash test was at 5 ft, which formed the basis for the force application recommendation in the sixth edition. Referring back to the unfactored design capacities shown in Figure 3, Buth’s results would indicate that typical cir- cular columns smaller than 50 in. may fail when struck by an 80,000-lb tractor-trailer truck traveling at 50 mph. El-Tawil’s Figure 4. Forces from a 14-kN pickup truck striking a 54-in. square pier [El-Tawil 2004].

8 Figure 5. Force velocity relationship for single-unit trucks (SUTs) [Buth 2010]. Figure 6. Force velocity relationship for tractor-trailer trucks [Buth 2010].

9 work suggests that even pickup trucks might pose a risk of bridge column collapse if the pickup truck is traveling very fast (i.e., 85 mph or more) and strikes a circular column less than about 40 in. in diameter. 2.3 Barrier Crash-Testing Guidelines In 1993, NCHRP Report 350 was published [Ross 1993], superseding the previous crash-testing guidelines contained in NCHRP Report 230 [Michie 1981]. One major change in NCHRP Report 350 was that six different test levels for roadside hardware were added for longitudinal barriers. The intent was to provide test guidelines for developing a range of barriers that could be used in different situations. Test Levels 1 through 3 related to containment of passenger vehicles (e.g., small passenger cars and pickup trucks) and varied by impact speed, with increasing impact speeds defined for increasing test levels. The “basic” test level for longitudinal barriers was TL-3. The structural adequacy test for this test level con- sisted of a 2,000-kg (4,409-lb) pickup truck striking a barrier at 100 km/h (62 mph) and 25 degrees. At a minimum, all barriers on high-speed roadways on the National Highway System are required to meet at least TL-3 requirements. Test Levels 4 through 6 also included consideration of pas- senger vehicles, but additionally incorporated consideration of assorted sizes of heavy vehicles. For example, many state transportation departments require that bridge railings meet at least TL-4, and TL-5 median barriers are becoming more common, especially on routes with higher percentages of trucks in the vehicle mix. The AASHTO LRFD Bridge Specifications require that TL-5 barriers be used when heavy-vehicle and railway static collision forces are not accounted for in the design of piers. A TL-5 test involves an 80,000-lb van-type tractor-trailer truck (TT) strik- ing the barrier at a speed of 50 mph and an angle of 15 degrees. TL-6 uses the same impact conditions but incorporates an 80,000-lb tractor-tanker trailer. Barriers meeting these higher containment levels are sometimes used when site and traf- fic conditions warrant. Site-specific factors that might justify use of a high-containment barrier include a high percentage of heavy truck traffic or truck-related crashes or an unusually high risk associated with barrier penetration. Such barriers are necessarily taller, stronger, and more expensive to construct. The higher test levels were intended for locations with a higher percentage of trucks and where the consequences of trucks penetrating or rolling over a barrier would be more severe. While NCHRP Report 350 provided the testing sug- gestions, no specific guidance was provided about what field conditions would indicate the need for a higher test-level railing. Since the publication of NCHRP Report 350 in 1993, changes have occurred in vehicle fleet characteristics, Figure 7. Impact force–time history for an 80,000-lb tractor-trailer truck striking an instrumented bridge column [Buth 2011].

10 operating conditions, technology, and so forth. NCHRP Project 22-14(2), “Improvement of Procedures for the Safety- Performance Evaluation of Roadside Features,” was initiated to take the next step in the continued advancement and evo- lution of roadside safety testing and evaluation. The results of this research effort culminated in 2009 with the first edi- tion of the AASHTO Manual for Assessing Safety Hardware (MASH) [AASHTO 2009]. MASH was updated in 2016, superseding NCHRP Report 350 as the industry-standard crash test and evaluation pro- cedure [AASHTO 2009]. MASH includes essentially the same test-level approach, with some modifications to the impact conditions for the higher-capacity longitudinal barrier test levels [AASHTO 2009; Ross 1993]. These modifications were primarily related to the size of the test vehicles. Impact conditions associated with the six test levels tend to be cali- brated from impact conditions associated with TL-3 impact conditions. TL-3 is intended to represent barrier applications on typical high-speed, high-volume roadways. Impact speeds and angles for TL-3 have traditionally been selected to be equal to the 85th percentile impact speed and 85th percentile impact angle of run-off-road (ROR) crashes. Further, vehicle masses are normally selected to be equal to the 95th and 5th percentile values from the passenger car fleet. However, in recognition of the recent increase in the size of passenger vehicles and the expectation that then-high gasoline prices might push vehicle masses down, the light truck vehicle mass was reduced to the 90th percentile, and the small car mass was reduced to the 2nd percentile of the 2002 new vehicle fleet [AASHTO 2009]. Even with these adjustments, the severity of the TL-3 test condition was increased significantly. The weight and body style of the pickup truck test vehicle changed from a 4,409-lb, ¾-ton, standard-cab pickup to a 5,000-lb, ½-ton, four-door pickup. This change in vehicle mass of approximately 15% was deemed to produce an impact severity that was more severe than the NCHRP Report 350 TL-4 single-unit truck test. The primary concern was that if TL-3 and TL-4 con- verged, highway agencies would lose one of the longitudinal barrier options. It was decided to increase the impact energy associated with TL-4 test 4–12 conditions by increasing the single-unit truck mass from 17,637 lb to 22,000 lb and the speed of the test vehicle from 50 mph to 56 mph. The 57% increase in impact severity for MASH TL-4 has resulted in a larger design impact load for TL-4, which will require stronger barriers and an increased overturning moment, leading to increased barrier height to prevent the single- unit truck (SUT) from rolling over the top of the barrier. The impact conditions for TL-5 and TL-6 involving the tractor-trailer and tractor-tanker trucks remained essentially unchanged from NCHRP Report 350 to MASH. 2.4 Crash Data Studies 2.4.1 Crash Data McDonald et al. analyzed the crash history at bridges over state-maintained, high-speed, multilane divided roadways in Iowa [McDonald 2009]. The crashes were categorized into bridge piers that were protected, partially protected, and unprotected. The severity of these crashes is summarized in Table 1, where crash severity is represented using the KABCO scale. In that scale, K is taken to equal a fatal crash, A is an incapacitating injury crash, B is a non-incapacitating injury crash, C is a possible injury crash, and O is a property- damage–only crash. As shown in Table 1, the highest crash severity percentage was found for unprotected piers, where more than 26% of the crashes involved severe or fatal inju- ries. The lowest crash severity was for fully protected piers, where 8.5% of the crashes involved severe or fatal injuries, indicating that shielding the piers resulted in a substantial reduction in crash severity. These results are also interesting because only unprotected piers in Iowa would be located outside the clear zone according to Iowa DOT policy. Perhaps the most comprehensive review of bridge pier collapses was performed by Buth et al. [Buth 2010], who col- lected pier crash data while conducting a study for TXDOT and FHWA. Table 2 shows the distribution of heavy vehicles which struck the piers in the state of Texas during 1998 through 2001. Buth et al. did not separate piers by their protection characteristics, but the data from both Iowa and Texas clearly show that impacts with piers are very hazardous collisions for vehicle occupants. Buth’s data also includes several complete bridge-pier component failures as well as some bridge col- lapses due to heavy-vehicle impacts. The cases collected by Buth et al. will be discussed in more detail in Section 2.5.3. 2.5 Exemplar Bridge Pier Crashes Although not useful from a statistical standpoint, bridge pier crashes reported in the media, investigated by the NTSB, or found in the literature help illustrate the outcome of these Severity Full Protection Unprotected Partial Protection Crashes Percent Crashes Percent Crashes Percent K 1 0.5 2 10.5 2 1.3 A 17 8.0 3 15.7 15 9.7 B 23 10.8 4 21.1 21 13.6 C 40 18.9 2 10.5 36 23.4 O 131 61.8 8 42.1 80 51.9 Total 212 100 19 100 154 100 Table 1. Crash history at Iowa bridges over high-speed, multilane divided roadways [after McDonald 2009].

11 exceptional events and provide anecdotal information that is useful in developing an improved understanding of the factors that influence catastrophic failure and the develop- ment of the guidelines. Cases from a variety of sources will be examined in the following sections. 2.5.1 Crashes in the Media 2.5.1.1 Worthington, Minnesota During the early morning of June 2, 2003, a truck crashed into a pier column of the bridge that carries Nobles County Road 9 over I-90 just west of Worthington, Minnesota, after a tire blowout caused the driver to lose control of the truck. The driver and his passenger suffered minor injuries. MNDOT officials already had the bridge closed for scheduled rehabili- tation; however, the I-90 detour was in place for more than 2 weeks. Figure 8 shows that the truck was able to get behind the guardrail, indicating that the length of need of the guard- rail protecting the pier may not have been adequate, allowing the truck to strike one of the pier columns. It appears that the guardrail was not struck during the crash and that the truck traveled behind the guardrail, striking the pier system [MNDOT 2003]. 2.5.1.2 Litchfield, Illinois A bus carrying 72 passengers and two drivers ran off the left side of the road after a tire blowout and entered the median on Thursday, August 2, 2012, at 1:20 p.m. The bus struck a rectangular bridge-pier column in the median and came to rest in contact with the leading pier column. One passenger was fatally injured, 47 passengers were taken from the scene to hospitals, and several others were treated for minor inju- ries [Stone 2012; Blakley 2012]. The pier had limited guardrail protection that was not sufficient to redirect the bus or prevent it from striking the pier. The pier and bridge do not appear to have been seriously damaged in the collision, although the potential for a catastrophic failure is apparent. 2.5.1.3 East Dallas, Texas A tractor trailer ran off the road while traveling on I-30, vaulting a concrete safety shaped barrier and striking a rectangular bridge-pier column on Sunday, June 10, 2012, at 4:45 p.m. The driver was fatally injured. There were no other reported injuries, but one of the pier columns was seriously damaged in the impact [Hardwick 2012]. 2.5.1.4 Hamilton County, Ohio On the evening of May 20, 2008, a flatbed trailer hauling a locomotive broke free from its hitch, ultimately crashing into a pier supporting I-74 eastbound over I-275. Two of the three columns were destroyed during the impacts with the trailer and the locomotive. The bridge did not collapse, and ODOT reopened the bridge within 2 months at a cost of $600,000 [WLWT 2008]. One of the interesting aspects of this crash was that the locomotive being hauled on the trailer was a very heavy, essentially rigid object and, as shown earlier by Buth et al., the rigidity of the cargo can increase the load experi- enced by the pier significantly in a collision [Buth 2010]. Severity Undivided Tangent Undivided Horizontal Curve Divided Tangent Divided Horizontal Curve Total Events Percent K 0 0 5 1 6 6.2 A 0 0 8 9 17 17.5 B 1 0 6 17 24 24.7 C 0 0 16 9 25 25.7 O 1 0 11 13 25 25.7 Total 2 0 46 49 97 100 Table 2. Heavy-vehicle crashes with piers [after Buth 2010]. Figure 8. Truck crash near Worthington, Minnesota [MNDOT 2003].

12 2.5.2 Crashes Investigated by the National Transportation Safety Board The NTSB investigates and determines the probable cause of significant crashes on highways and in other modes of transportation, with the goal of promoting transportation safety and preventing future similar crashes. In total, the NTSB investigates approximately six highway crashes per year, with each investigation lasting approximately 20 months. The fol- lowing sections contain brief summaries of crashes involving bridge piers that have been investigated by the NTSB in the last 30 years. 2.5.2.1 Indianapolis, Indiana On October 22, 2009, at 10:38 a.m., a tanker truck hauling approximately 9,000 gal of petroleum rolled over on I-69 in Indianapolis, Indiana. The driver was exiting I-69 south- bound in the right lane of a two-lane ramp that curved to the left toward I-465. The truck engaged in a series of erratic movements after entering the left lane, which was occupied by a Volvo. Ultimately, the tank trailer became decoupled from the truck, penetrated a steel w-beam guardrail, and struck the concrete pier that supported the southbound I-465 overpass. A fire started. The truck driver and vehicle driver sustained serious injuries as a result of the crash, and three occupants of a passenger vehicle traveling on I-465 received minor injuries from the fire. The bridge pier was completely displaced as a result of the crash. The NTSB “determined that the probable cause of this accident was the excessive, rapid, evasive steering maneuver that the truck driver executed after the combination unit began to encroach upon the occupied left lane.” The bridge design, “. . . including the elements of continuity and redundancy,” mitigated the outcome of the crash and prevented the struc- ture from collapsing [NTSB 2011]. NTSB had the following recommendations for the FHWA and AASTHO: To the Federal Highway Administration: Work with the American Association of State Highway and Transportation Officials to develop guidance for a bridge pier protection pro- gram that will allow state transportation agencies to conduct risk-based assessments of bridges located within highway inter- changes. At a minimum, the program should consider each structure’s redundancy, continuity, and the distance of bridge pier columns from the edge of traveled ways. Additionally, consider traffic volumes, traffic type, and the percentage of commercial vehicles transporting bulk liquid hazardous materials in identify- ing and prioritizing initiatives for preventing vulnerable bridges at high-risk interchanges from collapsing if struck or otherwise damaged by a heavy vehicle [NTSB 2011]. To the Federal Highway Administration: Once the guidance for a bridge pier protection program as described in Safety Recommendation H-11-16 has been developed, require that it be applied to bridges that are vulnerable to collapse if struck by a heavy vehicle [NTSB 2011]. 2.5.2.2 Evergreen, Alabama A tractor hauling a trailer with a bulk cement tank was traveling southbound on I-65 on May 19, 1993, at 1:35 a.m. when it left the paved road, vaulted a guardrail, and struck a bridge pier supporting County Road 22. The driver suffered serious injuries. Two spans of the overpass collapsed, leading to a car and tractor trailer striking the collapsed bridge. Both drivers of the subsequent crashes were killed. The NTSB determined that the cement truck driver may have fallen asleep and had been operating under the influence of marijuana, both of which contributed to the original crash. The “. . . nonredundant bridge design, the close proximity of the column bent to the road, and the lack of protection for the column bent from high-speed heavy-vehicle collision” contributed to the second collision [NTSB 1993]. NTSB had the following recommendations for the FHWA and AASHTO: To the Federal Highway Administration: Request states to identify and assess bridges that are vulnerable to collapse from a high-speed heavy-vehicle collision with their bridge columns and develop and implement countermeasures to protect the structures (Class II, Priority Action) (H-94-5). In cooperation with the American Association of State Highway Transportation Officials, ensure that the bridge management program guidelines include information on evaluating which bridges are vulnerable to high-speed heavy-vehicle collision and subsequent collapse (Class II, Priority Action) (H-94-6) [NTSB 1993]. To the American Association of State Highway Transportation Officials: In cooperation with the Federal Highway Adminis- tration, ensure that the bridge management program guidelines include information on evaluating which bridges are vulnerable to high-speed heavy-vehicle collision and subsequent collapse (Class II, Priority Action) (H-94-7) [NTSB 1993]. 2.5.2.3 Sacramento, California A Greyhound bus collided with a concrete overpass sup- port column on I-880 on November 3, 1973. The highway was a relatively flat, straight, six-lane divided highway. The lanes were 12 ft wide, and the paved shoulders were 10 ft wide. The piers were protected by a w-beam guardrail with wooden posts installed on a curbed median. The top of the guardrail was 21 to 23 in. above the height of the curbing. The bus penetrated the guardrail and struck the piers. NTSB made the following recommendation to the FHWA. To the Federal Highway Administration: Promulgate manda- tory national performance standards for traffic barrier systems. Those standards should contain criteria for dynamic testing or

13 analytical procedures substantiated by such test for each design to increase the compatibility of barriers with both light and heavy vehicles. The standard should also contain requirements regarding the placement of the barriers in the field to assure that compatibility of the vehicle/barrier is not compromised by adja- cent environment [NTSB 1973]. 2.5.3 Crashes in the Literature Buth et al. reviewed 19 heavy-vehicle crashes with bridge piers or at locations near piers and the outcome of those crashes [Buth 2010]. Crashes 13 and 14 identified by Buth et al. were actually not pier crashes. Crash 13 occurred near a pier, and crash 14 occurred on an overpass. Both involved tanker trucks that caught fire and damaged the bridges, but neither involved an impact with a bridge pier. The remain- ing 17 crashes investigated by Buth et al. are summarized in Table 3. Table 3 also includes a summary of crashes reported by El-Tawil and summarizes the crashes discussed earlier and investigated by the NTSB. The last column in the table includes a literature reference for the crash. When the crash was reviewed by multiple sources, multiple references are reported. Year/ Location Events Injuries Image Ref/ Source 1965/bridge over I-45, Dallas County, Texas A tractor trailer with an unknown load entered the median and struck the first column of a two-column pier. The column failed, and the bridge collapsed as a result of the impact. Unknown No photograph available Buth #6 [Buth 2010] 1973/I-880 overpass, Sacramento, California A passenger bus penetrated a 23-in.-tall w-beam guardrail and struck a pier. The pier was damaged; the bridge did not collapse. The driver and 12 passengers were killed, and 33 others were injured. No photograph available [NTSB 1973] 1989/Murphy Hollow Road over I-24, Marion County, Tennessee A box truck entered the median and struck a two- column pier. The column failed; however, the bridge did not collapse as a result of the impact. Unknown No photograph available Buth #16 [Buth 2010] 1993/ County Road 22 over I-65, Evergreen Alabama A cement truck vaulted a guardrail and struck a pier. Two spans of the bridge collapsed. A car and tractor trailer struck the collapsed bridge. Two were fatally injured. No photograph available El-Tawil #1 [El-Tawil 2004] [NTSB 1993] 1994/FM 2110 over I-30, Texarkana, Texas An 80,000-lb tractor trailer traveling about 60 mph carrying coils of steel crashed into the easternmost column of a two-column pier. The collision caused two spans of the bridge to collapse. The truck driver and passenger were fatally injured. Buth #1 [Buth 2010] 2002/SH 14 over I-45, Corsicana, Texas A tractor trailer carrying paper struck the southernmost column of the median two - column pier. The column failed, and the bridge collapsed. The collision killed the driver. Buth #8 [Buth 2010] El-Tawil #2 [El-Tawil 2004] Table 3. Summary of pier crashes found in the literature. (continued on next page)

14 Year/ Location Events Injuries Image Ref/ Source 2003/I-275 north ramp bridge at I-40 east, Knoxville, Tennessee A tractor trailer overturned on the I-275 north ramp toward I-40 east. The vehicle fell to the roadway below, striking a ramp support. The support was slightly damaged. Unknown No photograph available Buth #12 [Buth 2010] 2003/I-80 Bridge, Big Spring, Nebraska A tractor trailer struck the columns protected by a guardrail. The pier and bridge failed. Unknown El-Tawil #3 [El-Tawil 2004] 2003/I-90 bridge, #53812, Worthington, Minnesota A single-unit truck struck a column, causing the column to fail. The bridge did not collapse as a result of impact. Unknown Buth #17 [Buth 2010] [MNDOT 2003] 2004/ Tancahua Street over I-37, Corpus Christi, Texas A tanker loaded with compressed gas crashed into the easternmost 30-in.- diameter column of the center three-column pier located in the median of I-37. The easternmost column failed; however, the bridge did not collapse as a result of impact. The driver was fatally injured. Buth #3 [Buth 2010] 2004/Pyke Road over I-10, Sealy, Texas A tractor trailer carrying steel sheet piling struck the westernmost column of the median two-column pier. The bridge did not collapse. The driver was fatally injured. Buth #7 [Buth 2010] 2005/bridge at I-35 and U.S. 77, Red Oak, Texas A tractor trailer with an unknown cargo entered the I-35 median and struck the northernmost column of a three-column pier. The collision caused the column to fail; however, the bridge did not collapse. The driver was fatally injured. Buth #4 [Buth 2010] Table 3. (Continued).

15 Year/ Location Events Injuries Image Ref/ Source 2007/ Chatfield Road over I-35, Navarro County, Texas A tractor trailer carrying home construction materials struck the northernmost 30-in.-diameter column of the two-column pier after the driver fell asleep and drifted into the median. The bridge did not collapse; however, the collision caused severe cracking of the column. The driver was fatally injured. Buth #2 [Buth 2010] 2007/bridge over I-70, Grand Junction, Colorado A tractor trailer carrying barrels of a flammable liquid struck a median bridge-pier column. The cause of the crash is unknown. Unknown Buth #9 [Buth 2010] 2007/I-20 over Rabbit Creek, Longview, Texas A tractor trailer with an unknown load struck an exterior column of an interior three-column pier. The column failed; however, the bridge did not collapse as a result of the impact. Unknown Buth #10 [Buth 2010] 2007/bridge I-240 over I-40, Memphis, Tennessee A truck tractor trailer loaded with produce struck an exterior pier. Unknown No photograph available Buth #11 [Buth 2010] 2008/FM 1401 bridge over I-30, Mount Pleasant, Texas A truck tractor trailer loaded with car parts struck a pier of the bridge carrying FM 1401 over I-30. The westernmost 30-in.-diameter pier of the three-pier bent located on the shoulder of the eastbound lanes of I-30 was struck. The pier failed; the bridge did not. The collision killed the driver. Buth #18 [Buth 2010] Table 3. (Continued). (continued on next page)

16 Year/ Location Events Injuries Image Ref/ Source 2008/ Milepost 519 bridge over I-20, Canton, Texas An unloaded tractor trailer struck the westernmost column of the two-column pier located in the shoulder of I-20 eastbound. The column failed; however, the bridge did not. Unknown Buth #19 [Buth 2010] 2008/Exit 111 bridge over I-24, Manchester, Tennessee A tractor trailer carrying pies struck the pier. Damage to the pier was minor. Unknown No photograph available Buth #15 [Buth 2010] 2008/I-74 over I-275, Hamilton County, Ohio A flatbed trailer hauling a locomotive broke free from its hitch and struck a pier. Two of the three columns were destroyed. The bridge did not collapse. Unknown No photograph available [WLWT 2008] 2009/I-465 over I-69, Indianapolis, Indiana After a series of erratic movements, a tanker truck hauling petroleum penetrated a w-beam guardrail and struck a concrete pier. A fire was started. The bridge pier was completely displaced. There were two serious injuries and three minor injuries. No photograph available [NTSB 2011] 2012/ Litchfield, Illinois A passenger bus entered the median, penetrated a w-beam guardrail, and struck a pier. The pier and bridge do not appear to have been seriously damaged in the collision. One passenger was fatally injured, 47 were seriously injured, and several others had minor injuries. No photograph available [Stone 2012] 2012/ Dolphin Road over I-30, East Dallas, Texas A tractor trailer vaulted a concrete safety shaped barrier and struck a rectangular column. The column was seriously damaged. The bridge did not collapse. The driver was fatally injured. No photograph available [Hardwick 2012] Year unknown/ FM 2207 over I-20, Tyler, Texas A tractor trailer carrying structural steel struck the easternmost 30-in.-diameter column of a two-column pier located on the shoulder of the westbound lanes of I-20. The collision caused failure in the 30-in.-diameter column. The bridge did not collapse as a result of impact. Unknown Buth #5 [Buth 2010] Notes: Numbers in the source column refer to crashes in Buth 2010. For example, Buth #6 is the sixth crash listed in Buth 2010. FM = farm-to-market road, SH = state highway. Table 3. (Continued).

17 2.5.4 Summary These crashes illustrate that heavy vehicles have more than sufficient energy to cause considerable damage to bridge piers that can result in the structural failure of pier components and, in some cases, the catastrophic collapse of a bridge. Pro- tecting the bridge structures from heavy-vehicle, high-energy crashes is essential for maintaining the structural integrity of a bridge, especially for bridges with nonredundant pier systems or noncontinuous superstructures. The crashes in Table 3 are further summarized in Table 4 with respect to the approximate impact conditions and pier column structural details. Five of the cases described by Buth et al. summarized in Table 3 and Table 4 involved collapse of the bridge. Interest- ingly, all five of these cases involved two-column pier systems on bridges that were apparently not continuous [Buth 2010]. Three of the five cases reported by Buth where the bridge col- lapsed involved 30-in. piers with #9 longitudinal reinforcing steel and #2 spiral stirrups; the structural details of the other two cases are unknown [Buth 2010]. Buth estimated the unfactored design capacity of the columns where the bridge collapsed to be between 80 and 88 kips and, as shown earlier in Figure 6, the impact load measured in a crash test of an 80,000-lb tractor-trailer truck striking an instrumented column at 50 mph was just over 600 kips. While the impact conditions for most of the cases in Table 4 are unknown, it is interesting that many of the cases involved 80,000-lb tractor- trailer trucks traveling at speeds of between 50 and 60 mph. Roughly speaking, the cases in Table 4 where the bridge collapsed and the structural details are known appear to have experienced impact loadings almost seven times higher than their unfactored quasi-static design load. These cases also show that vehicle impacts with bridge piers can result in serious crashes that cause severe injuries and death to vehicle occupants. Of the 23 cases reported in Table 4 by Buth and El-Tawil, nine (i.e., 40%) involved at least one fatality. Many of the cases in Table 4 also illustrate lb. mi/h ft-kips lb.-sec. Buth #1 TT 80,000 60 9,620 218,634 Circ 2K 30 2 Buth #2 TT 80,000 60 9,620 218,634 Circ 1C 30 2 Buth #3 TT 72,000 55 7,275 180,373 Circ 1K 30 3 Buth #4 TT 40,000 60 4,810 180,373 Circ 1K 30 3 Buth #5 TT - - - - Circ - 30 2 Buth #6 TT - - - - Circ - 30 2 Buth #7 TT 80,000 50 6,680 109,317 Circ 1K 30 2 Buth #8 TT 80,000 - - - Circ 1K 30 2 Buth #9 TT - - - - Circ - Unk 2 Buth #10 TT 80,000 75 15,031 273,292 Circ - 24 3 Buth #11 TT - - - - Circ O 30 Unk Buth #12 TT - - - - Unk - Unk Unk Buth #13 TT - 25 - - Circ - 36 Unk Buth #14 TT - - - - Unk - Unk Unk Buth #15 TT - - - - Unk - Unk Unk Buth #16 TT - - - - Sqr - 24 2 Buth #17 SUT - - - - Circ - 32 Unk Buth #18 TT 80,000 - - - Circ 1K 30 2 Buth #19 TT - - - - Circ - 30 2 El-Tawil #1 TT - - Unk 2K Unk Unk El-Tawil #2 TT - - Unk 1K Unk Unk El-Tawil #3 TT - - Unk 1K Unk Unk C ra sh R ef er en ce V eh ic le A pp ro xi m at e V eh ic le W ei gh t A pp ro xi m at e Im pa ct S pe ed A pp ro xi m at e Im pa ct E ne rg y A pp ro xi m at e M om en tu m O cc up an t C ra sh S ev er it y C ol um n Sh ap e C ol um n Si ze N um be r of C ol um ns in P ie r R ei nf or ce m en t D et ai ls P ie r C ol um n F ai lu re ? B ri dg e F ai lu re ? #9 long. Bars and #2 spiral Yes Yes #9 long. Bars and #2 spiral Cracking No #9 long. Bars and #2 spiral Yes No #9 long. Bars and #2 spiral Yes No #9 long. Bars and #2 spiral Yes No Unk Yes Yes #9 long. Bars and #2 spiral Yes No #9 long. Bars and #2 spiral Yes Yes Unk Yes Yes #7 long. Bars and #2 spiral Yes No Unk No No Unk No No Unk No No Unk Unk Unk Unk No No #10 long. Bars and #4 spiral Yes No #9 long. Bars and #4 spiral Yes No #9 long. Bars and #3 spiral Yes Yes #9 long. Bars and #3 spiral Yes No Unk Yes Partial Unk Yes Yes Unk Yes Yes Note: TT = tractor-trailer truck; PDO = property damage only; Unk = unknown; Circ = circular; Sqr = square. Table 4. Pier details and impact conditions in crashes found in the literature [Buth 2010, El-Tawil 2004].

18 insufficient barrier shielding, where the length of need was too short or a TL-3 barrier was not adequate for a truck impact. Details of barrier selection, length of need, and bar- rier placement are important considerations when designing a bridge pier protection strategy that will not only protect the bridge but protect the motoring public. 2.6 Bridge Pier Risk Analysis In Section C3.14.5 of the LRFD Bridge Design Specifica- tions, in the discussion on collisions with ships and barges, risk is defined as “the potential realization of unwanted con- sequences of an event. Both a probability of occurrence of an event and the magnitude of its consequences are involved. Defining an acceptable level or risk is a value-oriented pro- cess and is by nature subjective” [AASHTO 2012]. Risk, then, involves estimating the probability of an undesirable event like a bridge collapse occurring. While the choice of select- ing an acceptable level of risk is a subjective, “value-oriented process,” the calculation of the risk itself is based on a con- ditional probability. The acceptable level is easily modified based on the quantifiable assessment of the probability of a catastrophic failure. 2.6.1 LRFD Bridge Specifications Probabilistic Method Buth et al. estimated the risk of heavy-vehicle collisions with bridge piers in the states of Texas and Minnesota. The study focused on principal arterials and collectors because the researchers believed there are higher risks of catastrophic failure at greater speeds [Buth 2010]. The crash risk for indi- vidual piers was modeled as well as the crash risk as a func- tion of the roadway characteristics. There was no distinction made as to whether the piers were protected or unprotected from errant vehicles on the undivided roadways or where the pier was located (i.e., median or roadside or offset from the roadway). Specifically, Buth et al. found that the probability of a bridge pier to be hit by a heavy vehicle (PHBP) is equal to: • 1.09 × 10–9 for divided highways in Texas, • 2.18 × 10–9 for curves on divided highways in Texas, • 1.35 × 10–8 for undivided roads in Minnesota, and • 2.19 × 10–8 for divided roads in Minnesota. It is notable that there is an order of magnitude difference between the coefficients for Texas and Minnesota, so it is unclear what value a state might use most appropriately or what the reason for the difference might be. Given the appro- priate value chosen for PHBP, the annual frequency of bridge pier crashes can be calculated as follows: = i iAF TAADT 365HBPP where AF = the annual frequency, TAADT = the heavy-vehicle volume per day, and PHBP = the probability of the bridge pier to be hit by a heavy vehicle. Buth et al. further found that striking a pier is a condi- tional probably, only possible when a vehicle has already run off the road [Buth 2010]. This has been noted in countless prior studies and documented in the RDG for some time [AASHTO 2002a; AASHTO 2006; AASHTO 2011]. Using the Texas data, the estimated frequency of ROR crashes by site can be found as follows: = lnu e L Fi Bo i iBi where ui = the estimated number of ROR crashes per year for site i, ln Bo = constant (–6.354 for undivided and –4.676 for divided), Li = the length of segment i in miles, Fi = vehicles per day (ADT) for segment i, and Bi = the flow (0.645 for undivided and 0.501 for divided). The truck volume can be used in place of general traffic volume to determine the estimated frequency of truck ROR crashes per site. This method is quite simple; however, it does not consider the offset of the pier from the travel lanes, the speed of the encroaching vehicles, the traffic mix, or the possible pro- tection of the pier. It also does not consider the width of the median, the terrain of the median, the number of lanes, the horizontal curvature or vertical grade of the road, and other factors that have been noted to affect the ROR crash rate. Obviously, a pier located 30 ft from the road will have a much different probability of being struck than one 10 or 15 ft from the road. Likewise, an unshielded pier and a pier shielded by a TL-5 concrete barrier will have much differ- ent probabilities of being struck. This regression equation, therefore, implicitly includes the typical offsets and shielding policies in place in the states where the data were collected— in this case, Texas and Minnesota. The prediction equation would have little value in a state where piers have different offsets or shielding policies. A method that includes the offset of the pier from the travel lanes, the speed of the encroaching vehicles, the traffic mix, the possible protection of the pier, the width of the median, the terrain of the median, the num- ber of lanes, the horizontal curvature or vertical grade of the road, and other factors that have been noted to affect runoffs should be used to develop national pier protection guide- lines. In the survey of bridge design professionals shown in

19 Appendix C: Survey of Practice, respondents indicated that “engineering judgement” was the method most often used for accounting for site and traffic conditions. The respondents indicated that the most important characteristics were (in priority order) the traffic volume, the percentage of trucks, speed limit, highway type, and the number of lanes. Another difficulty with this regression equation is that it was based only on heavy-vehicle volumes without respect to the traffic mix at the site. A particular truck volume might represent a vehicle mix with 10%, 30%, or 50% trucks, but the prediction would be the same. While truck volume is a good measure of exposure, the overall ADT is also a measure of the potential for conflicts on the road. A road with 1,000 heavy vehicles/day and an overall ADT of 2,000 vehicles/day (i.e., 50% trucks) would experience traffic conflicts at a much different rate than a road with 1,000 heavy vehicles/day and overall traffic of 10,000 vehicles/day (i.e., 10% trucks). Since it is these conflicts that often precipitate crashes, the total annual average daily traffic (AADT) and percent trucks should be included. 2.6.2 Roadside Safety Analysis Program Method Encroachment-based conditional probability models have been used in roadside safety analysis methods since the 1977 AASHTO Barrier Guide [AASHTO 1977, Appendix E]. The computer program Roadside was an encroachment- probability–based software tool used for roadside design in the first RDG [AASHTO 1989a]. The program BCAP was a further development of the method that was used in the 1989 AASHTO Guide Specifications for Bridge Railings [AASHTO 1989b]. BCAP was further improved and modified into the computer program RSAP (Roadside Safety Analysis Pro- gram), which was included in the 2002 and subsequent RDGs [AASHTO 2002a]. RSAP was extensively updated and revised in 2012, resulting in a new version of the software (i.e., RSAPv3) [Ray 2012]. Additional updates have been made to RSAPv3 under NCHRP Project 22-12(03) to allow RSAPv3 to include consideration of heavy vehicles’ properties and variations of heavy-vehicle encroachments as well as risk assessment [Ray 2014b]. Updates have been made under NCHRP Project 17-54 to allow for consideration of heavy- vehicle trajectories during encroachments [Carrigan 2017]. These updates have combined to make RSAPv3 the state-of- the-art software tool for roadside encroachment modeling. RSAPv3 and its predecessors are used to assess the prob- ability of a roadside feature being struck, the severity of the crash if one has occurred, and the resulting crash costs. These programs were designed to assess the risk of a road- side feature being struck and the subsequent benefit–cost of varying the roadside design. The following conditional probability model is used for each alternative on each seg- ment [Ray 2012]: ( ) ( ) ( ) ( ) ( ) = i i i i i CC ADT Encr Cr Encr Sev Cr CC Sev ,E L P P P E N M N s s where E(CC)N,M = expected annual crash cost on segment N for alternative M for a particular vehicle encroachment, ADT = ADT in vehicles/day, LN = length of segment N in miles, P(Encr) = the probability a vehicle will encroach on the segment, P(Cr|Encr) = the probability a crash will occur on the segment given that an encroachment has occurred, P(Sevs|Cr) = the probability that a crash of severity s occurs given that a crash has occurred, and E(CCs|Sevs) = the expected crash cost of a crash of severity s in dollars. This equation represents the expected annual crash cost for a particular encroachment on segment N for alternative M. An RSAPv3 analysis is composed of four major steps for assessing each alternative: • Encroachment probability, • Crash prediction, • Severity prediction, and • Benefit–cost analysis. Using a series of conditional probabilities, RSAPv3 first pre- dicts the number of encroachments expected on a segment. Given that an encroachment has occurred, the likelihood of a crash is assessed by examining the location of roadside features and comparing those locations to a wide variety of possible vehicle paths across the roadside. If a crash is pre- dicted (i.e., one of the possible trajectories intersects with the location of a roadside hazard), the severity is estimated and converted to units of dollars. RSAPv3 proceeds by simulating tens of thousands of encroachment trajectories and examining which trajectories strike objects, the probability of penetration or rolling over the object, and the likely severity of those collisions. The passenger vehicle trajectories used in RSAPv3 were gath- ered from reconstructed ROR crashes performed in NCHRP Project 17-22 [Mak 2010]. NCHRP Project 22-12(03) incor- porated a method to account for the differences in pas- senger vehicle and heavy-vehicle encroachment rates and developed capacity values for concrete median barriers and

20 bridge railings to be used to estimate the probability of barrier penetration. NCHRP Project 22-12(03) also added output features to support benefit–cost and risk-based analysis methods. These features and data have been incor- porated into RSAPv3 and can be used in evaluating the effectiveness of barriers protecting bridge piers [Ray 2014b]. The probabilistic benefit–cost approach long used in road- side design is capable of modeling variations in roadside and median terrain, traffic mix, highway geometry, pier location, pier capacity, and pier protection. Furthermore, RSAPv3 is capable of modeling minor variations in pier size, number, and location. This robust state-of-the-art program has expe- rienced multiple recent updates, making it the best approach for examining the problem of bridge pier protection and shielding. For example, for a tangent section of divided highway with a two-way AADT of 20,000 vehicles/day, the method from the sixth edition of the AASHTO LRFD Bridge Design Specifica- tions estimates an annual bridge-pier heavy-vehicle collision frequency of 0.0009 collisions/year. RSAPv3’s encroachment model, which is based on the so-called Cooper data [Cooper 1980], estimates that there will be 1.3091 encroachments/mi/ year on a divided highway with a two-way AADT of 20,000. Most encroachments in the Cooper data are less than 300 ft long, or 300/5,280 = 0.0568 miles, and using an assumption of 10% truck traffic would indicate that RSAPv3 would expect 1.3091 • 0.0568 • 0.1 = 0.0074 encroachments near a bridge pier/year, and of these only 10% (i.e., 0.0007 heavy-vehicle pier collisions/year) would have a lateral extent greater than 30 ft, so the rough estimates are consistent. The sixth edi- tion procedure, however, has several important assumptions about the location of the pier and the characteristics of the highway built into the procedure that mask the importance of critical design variables like pier offset, grade, curvature, and other site characteristics. On the other hand, RSAPv3 can be used to explicitly examine the effects of length of need, offset from the roadway, and barrier type on the frequency and severity of bridge pier impacts. Each of these variations is critical to protecting existing and proposed structures. 2.7 Benefit/Cost Versus Risk In roadside safety, the incremental benefit–cost ratio is the present worth over the project life of the reduction in the societal costs of implementing a safety improvement divided by the present worth over the life of the project of increase in construction, maintenance and repair costs of implementing the safety improvement. = − − BCR CC CC AC AC 1 2 2 1 where BCR = (incremental) benefit–cost ratio, CCi = present worth of the annual societal crash costs of alternative i over the project life, and ACi = present worth of the owner agency construction cost and the present worth of the annual expected maintenance and repair costs of alternative i over the project life. Benefit–cost methods were used to develop the guidelines in the RDG, but recent research has shown that using costs presents some difficulties, as shown in the following sections. It should be recognized that the numerator of the BCR equa- tion is really an estimate of the average annual crashes over the life of the project multiplied by the average crash cost of that type of crash. A benefit–cost analysis, therefore, has imbedded within it a risk assessment that has been transformed into units of dollars. In a study to develop selection guidelines for bridge rails, Ray et al. outlined some of the difficulties in using benefit/ cost to develop national guidelines [Ray 2014b]. Ray et al. found that, while construction costs had gone up and down considerably in the previous decade due to economic condi- tions, the societal cost of highway crashes had only increased. In addition to temporal variations in construction costs, costs also vary geographically. For example, Ray et al. showed that construction costs in the state of New York were more than 3.5 times higher than the national average, while construction costs in some states, such as Arizona, Mississippi, and Montana, were about half the national average [Ray 2014b]. While benefit–cost methods have been widely used in road- side safety for decades, the disparities in how costs change by region and in time and how construction costs vary differ- ently from crash costs are a cause for concern. What is cost beneficial today may not be tomorrow, and what is cost beneficial in one region may not be in another. For these reasons, benefit–cost methods are handicapped in terms of use for national guideline development. Benefit–cost methods are actually a risk assessment method that is used to estimate reductions in anticipated crash costs (i.e., the benefits), which are then used to perform an incremental benefit–cost analysis that includes agency costs like construction, maintenance, and repair over the life of the project. Another approach common in many other types of engi- neering fields and becoming more common in roadside safety is risk analysis. In risk analysis, the risk of experiencing a par- ticular type of event is quantified using probabilistic models. An acceptable level of risk is established, and then the system is engineered to ensure that the predicted in-service risk is below the targeted acceptable risk. There are advantages and disadvantages to each approach. The benefit–cost method has the advantage that it includes

21 both societal benefits and agency costs such that the benefits are maximized while making the best possible use of agency funds. The disadvantage is that, since costs are explicitly included, regional and temporal variations in the cost elements can make the same solution cost beneficial in one region and not cost beneficial in another. Another disadvantage is that the risk is not necessarily uniform, so one cost-beneficial solution can have a different inherent risk than another with the same benefit–cost ratio. On the other hand, risk analysis sets a specific risk objec- tive that is uniform across regions and through time such that the risk of an unacceptable event is always kept the same. The disadvantage is that the best risk-based solution may not always be cost beneficial, especially with respect to the agency costs. The important point is that construction costs vary by region and in time, so any guidelines developed based on a cost–benefit model will likewise vary by region and in time. This is not desirable from the point of view of developing national guidelines that are meant to be used in all regions of the country and are expected to have a reasonably long life. Risk-based approaches avoid this problem by setting the safety-performance goal in terms of the risk of a collision of a particular severity occurring sometime during the design life of the structure. Ray et al. have used this risk-based approach in developing guidelines for the selection of bridge railings [Ray 2014b]. Both the TCRS and SCOBS T7 committees endorsed the risk-based approach over the more traditional cost–benefit approach in reviewing the results of NCHRP Project 22-12(03). A risk-based approach was used in this research for the reasons outlined. 2.8 Summary The preceding review of the literature has shown that bridge piers are prone to creating two distinct safety prob- lems: (1) the difficulty of protecting vehicle occupants from impacts with piers because bridge piers are one of the most hazardous features of the roadside, and (2) the problem of protecting bridge piers from impacts with vehicles, in particular heavy vehicles, because high-energy impacts have the potential to significantly damage piers and thereby com- promise the structural integrity of bridges. Research into the crash capacity of bridge piers has indi- cated that the AASHTO LRFD’s suggested equivalent static force of 600 kips may not be adequate to ensure that bridge piers do not fail in some high-energy, heavy-vehicle impacts [AASHTO 2012]. Prior research has demonstrated that there is a strong prob- abilistic nature to the problem of bridge pier protection. The risk associated with any particular bridge pier is a function of: • Its location (e.g., offset from the road), • The importance of the bridge and/or under-crossing route, • The traffic characteristics (e.g., overall AADT, percent trucks, and speed), • Highway characteristics (e.g., horizontal curvature, grade, lane width), • Structural characteristics of the bridge pier and bridge superstructure (e.g., pier capacity, pier arrangement, bridge redundancy), and • Any shielding by barriers (e.g., type of barrier, barrier height, barrier offset). Probabilistic design methods have a long history in road- side safety, going back to the 1977 AASHTO Barrier Guide. Additions to the AASHTO LRFD Bridge Design Specifications have likewise adopted probabilistic methods to quantify the risk of a catastrophic bridge-pier crash occurring. RSAPv3 is the current encroachment-probability–based roadside design tool recommended by the RDG. This research used the risk tools available within RSAPv3. This approach allows for the incorporation of traffic and highway characteristics along with the precise location of bridge piers on the roadside. The following chapters document the research, which developed (1) guidelines for designing bridge pier com- ponents for heavy-vehicle impacts or determining if they require shielding for structural protection, and (2) guidelines for determining if bridge piers should be shielded to provide passenger-vehicle occupant protection.