Fracture-Critical System Analysis for Steel Bridges (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.

6 2.1 Summary of the State of the Practice Because of the 1967 collapse of the Silver Bridge, AASHTO published the first fracture control plan (FCP) for steel bridges in 1978 (AASHTO 1978, Lichtenstein 1993). The original FCP specified material toughness requirements and restrictions on acceptable detail types and welding practices, with the objective of preventing fracture and fatigue-related failures. It also established the fracture-critical terminology and phi- losophy that characterizes the provisions currently used. The FCP evolved over the years, resulting in the fabrication pro- visions currently contained in AASHTO/AWS D1.5M/D1.5 Bridge Welding Code (AWS D1.5) (AASHTO and AWS 2015), fatigue design and fracture toughness requirements described in the AASHTO LRFD Bridge Design Specifications (AASHTO LRFD BDS) (AASHTO 2014), and inspection mandates contained in the NBIS (FHWA and U.S. DOT 2017). The NBIS defines an FCM as “a steel member in tension, or with a tension element, whose failure would probably cause a portion of or the entire bridge to collapse” (FHWA and U.S. DOT 2017). The AASHTO MBE gives a similar definition: “fracture critical members or member components are steel tension members or steel components of members whose failure would be expected to result in collapse of the bridge” (AASHTO 2011A). These definitions are vague and subjec- tive, as a particular member might be in tension or compres- sion depending on the load case, and there is no universal definition of collapse. The concept of redundancy is effectively embedded in the definition of an FCM. Redundancy could be defined broadly in engineering as duplication of critical components or functions of a system with the intention of increasing reliability of the system. Hence, proving redun- dancy nullifies fracture-criticality. In other words, if it can be demonstrated that a structure can perform adequately after the failure of any tension member, by definition, there are no FCMs in the structure. In 2012, FHWA published a memorandum underlining pol- icies regarding FCMs (Lwin 2012). At the time—with regard to consideration of redundancy—only load-path redun- dancy might be considered in design and fabrication, while internal or member-level redundancy was not recognized in either design or fabrication. The use of refined analysis proce- dures to demonstrate bridge redundancy was perceived as the adequate analysis method. The memorandum introduced the definition of SRM as “a member that requires fabrication according to the AWS FCP, but need not be considered a FCM for in-service inspection.” It also states that “system- redundant members should be designated on the design plans with note to fabricate them in accordance with AWS Chapter 12” and comments on the need to monitor the bridge condition should future changes in bridge condition result in an SRM becoming an FCM. The authors of NCHRP Synthesis 354 surveyed several transportation agencies and found that there was no con- sensus with regard to the designation of FCMs (Connor et al. 2005). While two-plate girder bridges were unanimously considered to contain FCMs, three-plate girder bridges or twin-tub girder bridges were controversial. Addition- ally, there was some confusion with regard to fabrication of FCMs, as some state transportation agents believed that fabrication of steel tension members in high-performance steel would result in the member not being designated as an FCM (which is not true). Steel bridge owners could clearly benefit from provisions to evaluate the redundancy of steel bridges and to identify FCMs and SRMs. While fabrication in accordance with the FCP does not result in very high cost premiums, bridges with FCMs are up to five times more expensive to inspect. Additionally, there is a lack of consensus with regard to the designation of FCMs. Refined analysis techniques can be used to identify FCMs (and SRMs) so that owners can make properly informed decisions related to their inven- tory of steel bridges. C H A P T E R 2 Research Approach

7 2.1.1 Current Mandates and Specifications for FCMs The NBIS, which are part of the CFR, establish a base inspec- tion interval of 24 months, extendable to 48 months through approval of FHWA for routine inspections. Some bridges may require shorter inspection intervals because of corrosion, traf- fic, and so on. Bridges with FCMs must undergo additional FCM hands-on inspections every 24 months, and there is no possibility to extend the interval beyond 24 months. An FCM is defined as “a steel member in tension, or with an element in tension, whose failure would probably cause a portion or the entire bridge to collapse.” Currently, an effort is being made to include the definition of SRM in the NBIS, which may be subjected to inspection requirements that differ from the ones for FCMs (FHWA and U.S. DOT 2017). AASHTO LRFD BDS specifies tabulated toughness require- ments for steel members that are in tension under the Strength I load combination. With regard to FCMs, it states that the engineer is responsible for determining whether a steel mem- ber is an FCM; in such a case, that member must be fabricated in accordance with Section 12 of the AWS D1.5. It also states that FCMs shall be delineated in the plan, unless strength and stability of a bridge with FCMs are confirmed through “rigor- ous analysis with hypothetical cracked components.” However, there is no clear advice on how to perform such analysis except for the limited guidance provided for acceptable methods of structural analysis discussed in Chapter 4 of the AASHTO LRFD BDS (AASHTO 2014). Currently, the redundancy of a structure is considered in design and evaluation of structures in a general manner. The AASHTO LRFD BDS use a redundancy factor that is applied to the load side of the design equation. This factor can be down to 0.95 for redundant structures and up to 1.05 for nonredun- dant structures (AASHTO 2014). The AASHTO MBE uses a system factor that modifies the resistance side of the design equation. This factor can be down to 0.85 for nonredundant structures and up to 1.20 for redundant structures (AASHTO 2011A). It should be taken into account that the approaches are not consistent with each other and may result in a struc- ture designed in accordance with the AASHTO LRFD BDS failing a first load rating if the procedures in the AASHTO MBE are followed. The AASHTO LRFD BDS do not provide any guidance with regard to the identification of nonredundant structures (AASHTO 2014), but the AASHTO MBE provides tabulated system factors for a variety of structures in Table 6A.4.2.4-1 and a list of steel bridges that may have FCMs in the Commen- tary C4.11 (AASHTO 2011A). The AASHTO MBE underlines the importance of developing an inspection plan for FCMs and identifies the FCM hands-on visual inspection as the method for detecting flaws. 2.1.2 Redundancy in Bridges with Failed FCMs Failures of tension components are not commonplace, and the instances in which they result in structural collapse are exceptional. NCHRP Synthesis 354 only lists two such cases in the past 50 years, with the most recent being more than 30 years ago (Connor et al. 2005). In contrast, there have been a num- ber of field cases in which bridges traditionally considered to be nonredundant did not collapse after significant damage of one or several primary steel tension members. A list of some of these instances is provided for informational purposes. • The Lafayette Bridge carrying US 52 through Saint Paul, Minnesota, consists of two parallel two-girder bridges continuous over three spans (Fisher et al. 1977). In 1975, a main girder of the bridge was discovered to be completely fractured. The crack originated from lack-of-fusion defects at welds between the web and lateral bracing gusset plate and propagated in a brittle manner throughout the web and bottom flange. The structure was able to maintain service- ability in the damaged state. Holes were drilled at the cor- ners of problematic welds to prevent propagation of cracks and to relieve constraint. • The Neville Island Bridge carrying I-79 over the Ohio River near Pittsburgh, Pennsylvania, underwent fracture of the fascia girder in a continuous three-span, two-plate girder bridge in 1976 (Fisher et al. 1980B). The fracture was caused by defective electroslag weld repairs, occurred in the middle of the center span, and was discovered by a tugboat captain. The deflection of the bridge was not perceivable by the passing traffic, and the structure did not show any signs of instability. • The Dan Ryan Transit Elevated Structure in Chicago, Illi- nois, is composed of continuous and suspended-plate girders with a cast-in-place concrete ballast (Fisher et al. 1980A). Several spans of the structure were supported on steel box-girder caps, also referred to as bents. The bottom flange and almost the entire web were welded to the bent, resulting in high constraint at the welds that—in combina- tion with cold temperatures—led to fracture at three bents. After the discovery of cracks in 1978, train service was dis- continued. However, the structure maintained service while the bents were fractured. • The Green River Bridge carrying I-26 through Henderson County, North Carolina, consists of two parallel two- girder bridges continuous over five spans (McGormley et al. 2000). In 1992, cracks were revealed in the girders: two transverse cracks on the bottom flanges and several shorter cracks in web-to-flange fillet welds. The bridge maintained serviceability in the described conditions. As the cracks were not removable by coring, cover plates were bolted to provide alternative load paths.

8 • The Hoan Bridge carrying I-794 over the Milwaukee River in Milwaukee, Wisconsin, underwent failure of two girders in a continuous three-span, three-plate girder bridge in 2000 (Connor and Fisher 2002, Fisher et al. 2001). The girder failures were caused by constraint-induced fracture (CIF). In spite of the significant damage and the 4-ft sag, the bridge carried live load for a short period following the multiple-girder fracture as motorists continued to cross the bridge. • The bridge carrying US 422 over the Schuylkill River in Pottstown, Pennsylvania, is composed of two parallel two- girder bridges continuous over six spans (Kaufmann et al. 2004). In 2003, a fracture initiated at one of the lateral gusset plate connections and propagated downward, thereby severing the entire bottom flange of a girder. The fracture also extended about 9 in. above the gusset plate in the web. The cracks were concluded to be caused by details prone to CIF. The structure sustained traffic loads while the fracture existed. • The Diefenbaker Bridge carrying Highway 2 and High- way 3 over the North Saskatchewan River in Prince Albert, Saskatchewan, Canada, is composed of parallel two-girder bridges continuous over seven spans (Ellis and Connor 2013). In 2011, a major crack was observed in the south- bound structure, extending through the bottom flange and the web to near the top of the web. The cracks were concluded to initiate at connections details susceptible to CIF. Engineers determined that the structure had adequate redundancy levels in the damaged condition. • The Mathews Bridge carrying US 90 Alt over the Saint John’s River in Jacksonville, Florida, is a six-span truss structure. In 2013, a Military Sealift Command ship col- lided with the bridge’s center span. The impact severed several members, including one of the bottom chords. The structure did not collapse and did not experience an observable increase in deflection. 2.1.3 Studies on Post-Fracture Redundant Capacity NCHRP Report 319: Recommended Guidelines for Redun- dancy Design and Rating of Two-Girder Steel Bridges is one the earliest significant studies stating the need for redun- dancy (Daniels et al. 1989). The overall conclusion is that simple linear elastic structural analyses, such as girder-line analysis models, are not adequate to evaluate the capacity of a structure after the failure of a main load-carrying member, as alternative load paths must be considered. These alterna- tive load paths are provided by the slab through catenary action, lateral bracing, and floor beams (or cross-frames and diaphragms) conforming a pseudo-truss. Redundancy rating equations were developed for specific bridge types, and a redundancy rating for evaluating bridges in which fracture has occurred was proposed, in addition to existing inventory and operating rating procedures. Discussion on the expected post-fracture performance presented the issue of an extended service period in which the fracture had not yet been discovered. An important full-scale test to study redundancy was reported in Idriss et al. (1995). The structure was a two- girder bridge carrying I-40 over the Rio Grande River in Albuquerque, New Mexico. It was scheduled for demolition, which facilitated destructive testing. It had three continuous spans. The end spans were 131-ft long and the interior span was 136-ft long, with the two girders spaced at 30 ft with stringers supported on floor beams. The fracture was simu- lated by torch-cutting one of the girders at mid span of the interior span. The structure was shored during the cut and slowly released, resulting in quasi-static loading. Idriss et al. (1995) reported a vertical deflection of 11⁄16 in. under dead load, and an additional ½ in. when a single 82-kip tractor trailer truck was located over the fracture. Yielding was not observed based on strain gage measurements. Although NCHRP Report 319 provides insight with regard to redundancy in girder systems, as demonstrated by the testing described by Idriss et al. (1995), inertial effects that arise during the fracture event were not explicitly consid- ered. Before the occurrence of a fracture, all members are subjected to an initial stress sinitial. After sudden failure takes place, members experience a maximum stress smax caused by the free vibration of the structure. Once the dynamic behav- ior dissipates, the members are subjected to a final stress sfinal, as shown in Figure 1. It is typical to describe inertial effects through a dynamic amplification factor DAR , calculated as ( )= σ − σ σDAR max final final The redundant capacity and dynamic response of steel bridges with members traditionally considered as FCMs have been analyzed in several computational studies. The follow- ing is the most significant research found during the litera- ture review: • Lai (1994) investigated the redundant capacity and dynamic response of a tied-arch truss structure after failure of one tie. The analysis was linear elastic and did not include any damping. Results from static analysis concluded that the structure was able to carry its dead load and 1.3 times the HS-20 truck load model. Results of the dynamic analysis concluded that the dynamic amplification factor was larger than 1.0, based on displacements rather than stresses. During the dynamic analysis, the structure was subjected to its dead load only.

9 • Goto et al. (2011) investigated the dynamic response of a truss structure after sudden failure. The authors’ analysis included material nonlinearity, large deformation effects, and assumed 5% damping ratio. Dynamic amplifica- tion factors were calculated between 0.2 and 0.4. In their approach, they considered two sources of inertial effects caused by sudden fracture. • Cha et al. (2014) investigated the dynamic response of a truss structure after sudden failure of a tension member. The authors’ analysis included material nonlinearity, large deformations, and assumed 4% damping ratio. The finite element models were benchmarked against experimental data from the Milton–Madison Bridge connecting Milton, Kentucky, and Madison, Indiana, reported by Digglemann et al. (2012) (Digglemann 2012). The maximum dynamic amplification factor for a main truss member was under 0.36 for both experiment and FEA. Results showed that the truss system exhibited large redundant capacity after failure of an FCM. • A large research effort with regard to redundancy was carried out at the University of Texas at Austin and focused on twin steel box-girder bridges. Williamson et al. (2010) developed detailed finite element models used to evaluate behavior of box-girder bridges. These models were cali- brated with data from full-scale tests on a single span twin box-girder bridge. The finite element methodology was detailed in Hovell and Williamson (2008), and the detailed experimental testing was described in Neuman (2009). The dynamic response was monitored so that a value of 0.3 was computed for the dynamic amplification factor. More weight was applied to the structure, which exhibited large post-fracture redundant capacity. 2.1.4 Reliability and Redundancy Current AASHTO design and evaluation procedures are based on LRFD procedures (AASHTO 2014, AASHTO 2011A). These load and resistance factors are implemented to account for the bias and variability of the estimated loads and resis- tances. As it is recognized that load and resistance are not deter- ministic quantities, they should be described through statistical parameters. Further, a probability of failure, characterized by a target reliability index, must also be tolerated. The calibration of these factors is described in NCHRP Report 368: Calibration of LRFD Bridge Design Code and includes six steps: (1) selecting representative bridges, (2) establishing a statistical database for load and resistance parameters, (3) developing load and resis- tance models, (4) developing the reliability analysis procedure, (5) selecting a target reliability, and (6) calculating load and resistance factors (Nowak 1999). In NCHRP Report 368, the Rackwitz and Fiessler iterative procedure (Nowak 1999, Rackwitz and Fiessler 1977) was used to calculate reliability indices of existing AASHTO spec- ifications at the time. With the existing values of reliability, a choice for a target reliability index of 3.5 (probability of fail- ure less than 2.33 • 10−4) was made. The then-proposed load and resistance factors were computed, assuming lognormal distribution of resistance and normal distribution of load. The integration procedure described in the report resulted in the following equation for a resistance factor: ( )φ = λ +1 kVR R and for a load factor: ( )γ = λ +1 kVQ Q Figure 1. Stresses considered in computation of dynamic amplification factors.

10 where lR and lQ are the biases of resistance and load, VR and VQ are the coefficients of variation of resistance and load, and k is a scalar that increases with the target reliability level and is uniformly applied to both the load and resistance factors. The reliability principles used in the calibration of the AASHTO LRFD BDS have been employed to develop pro- cedures for the assessment of redundant capacity (AASHTO 2014). A similar approach was described in NCHRP Report 406: Redundancy in Highway Bridge Superstructures and NCHRP Report 776: Bridge System Safety and Redundancy (Ghosn and Moses 1998, Ghosn et al. 2014). NCHRP Report 406 consid- ered redundancy through the application of system factors, in addition to the factored nominal resistance. In this research, simple finite element models were used to calculate loads that resulted in member failure (LF1), undamaged system failure (LFu), functionality loss (LFf), and damaged system failure (LFd) for several types of systems. The computed loads were used to calculate ratios, which were applied to log- normal distributions of load and resistance to calculate reliability indices. A calibration was then performed to develop system fac- tors that satisfy the following: (1) the difference between reli- ability indices for undamaged system failure and member failure is greater than 0.85, (2) the difference between reliabil- ity indices for functionality loss and member failure is greater than 0.25, and (3) the difference between reliability indices of a damaged system failure and member failure is greater than −2.70. Additionally, for bridges not covered in their research, they recommended that (1) LFu/LF1 ≥ 1.30, (2) LFf/LF1 ≥ 1.10, and (3) LFd/LF1 ≥ 0.50 (Ghosn and Moses 1998). Reliability principles and the calibration methodology applied in NCHRP Report 406 were used in NCHRP Report 776 to estimate bridge system factor for lateral and vertical loads. System factors were calculated to characterize redundancy in intact and damaged structures subjected to vehicular live loads. However, in NCHRP Report 776, the redundancy of deficient and overdesigned bridges was considered, while in NCHRP Report 406, it was assumed that the structures closely met mem- ber strength design requirements (Ghosn et al. 2014). 2.1.5 International Fracture-Related Practices After a review of bridge engineering practices outside of the United States, researchers found that the fracture-critical phi- losophy is nonexistent in international codes. Outside of the United States, there does not appear to be any stigma attached to nonredundant structures, which are designed and inspected using the same engineering criteria as any other structure. This approach makes it difficult to directly find code provisions that relate to fracture-critical practice in the United States. The FHWA scanning tour report states that “perhaps the most significant design-related observation of the scan team was the rest of the industrialized world’s view of the importance of redundancy. Two-girder bridges, as well as other structure types considered nonredundant and fracture-critical in the U.S., are not discouraged and, in fact, are used extensively as safe and cost-effective bridge designs.” The report goes on to state that “the U.S. design philosophy for nonredundant bridges should be reconsidered, based on these observations and improvements in steel toughness.” (Verma et al. 2001) The Eurocode presents a reliability-based design philoso- phy similar to the LRFD procedures used in the AASHTO LRFD BDS. The details of calculating load and resistance factors are somewhat different between both specifications, but the methodology is similar. No reference with regard to redundancy or fracture-criticality was found in the Eurocode. Although there is not a philosophical mandate to provide redundancy, it appears that redundancy is partially considered in the calculation of force actions. In the Eurocode, the frac- ture limit state is handled through toughness requirements and connection design, as in AASHTO LRFD BDS. However, the toughness requirements demanded more complex evalua- tions in which lowest member temperature, stress level, resid- ual stress, assumptions for fatigue crack growth during a given inspection interval, strain rate, and presence of cold-forming are considered (European Committee for Standardisation 2005, European Committee for Standardisation 2006, Comité Européen de Normalisation 2007). The Standard Specification for Steel and Composite Structures issued by the Japan Society of Civil Engineers requires a verification method based on ratios between design response and design resistance factored by a structural (importance) factor. Load factors and structural analysis fac- tors are applied to force effects to obtain the design response. Material factors and structural member factors are applied to calculated capacities to obtain the design resistance. As in the AASHTO LRFD BDS, the fracture limit state is handled through toughness requirements and adequate connection design, referencing the Japan Road Association and Hokkaido Development Bureau with regard to material requirements, and recommending the use of high-performance steel. No specific procedures were found with regard to redundancy; however, the Japan Society of Civil Engineers acknowledges that strength can be verified through nonlinear structural analysis and that such analysis methods can be applied in the design for redundancy (Japan Society of Civil Engineers 2009). 2.2 Research Methodology Four major efforts constitute the approach taken to attain the objectives of the current research. First, an FEA method- ology for redundancy evaluation of steel bridges was devel- oped. Second (once a benchmarked analysis methodology

11 was developed), finite element models of existing bridges were developed to calculate the dynamic amplification factors after the failure of a primary steel tension member. Third, steel bridge fabrication and design procedures were reviewed to develop a set of recommendations to avoid occurrence of frac- ture. Finally, a comprehensive set of specifications was written with the intent of providing bridge engineers with a meth- odology for the identification of FCMs and SRMs through redundancy evaluations and a set of requirements to prevent fracture in steel bridge members. 2.2.1 Finite Element Analysis Methodology for Redundancy Evaluation of Steel Bridges The first effort is the development and validation of a steel bridge evaluation framework, based on the application of FEA. For this purpose, analysis procedures, techniques, and inputs to evaluate steel bridges with failed primary tension members were studied. The nonlinear modeling features of the commercially available general purpose FEA program Abaqus were used to construct detailed finite element models of three steel bridges that underwent failure of a steel tension member (Simulia 2017). Field experimental data was avail- able and used for benchmarking purposes. These structures are as follows: • The Neville Island Bridge: noncomposite three-span con- tinuous two-plate girder highway bridge. • The Hoan Bridge: noncomposite three-span continuous three-plate girder highway bridge. • The University of Texas Bridge: composite single-span twin-tub girder highway bridge. Additionally, the following two bridges whose finite ele- ment models were developed and benchmarked by Cha et al. (2014) and Cha (2014) were studied: • The Milton–Madison Bridge: simple span symmetric Pratt truss highway bridge. • The White River Bridge over US-41 near Hazelton, Indi- ana: noncomposite two-span continuous two-girder highway bridge. These five structures were selected because they are typi- cally considered to have FCMs or to have low levels of redun- dancy. Further, each bridge represented a different type of superstructure with its own unique behavior after the failure of a steel tension member. Once models were built and vali- dated by comparing to available data, they were used to study member-out scenarios representing steel tension member failure in different locations. Once the necessary FEA procedures, techniques, and inputs for the redundancy evaluation of steel bridges were established, the FEA methodology was completed with (1) the development of load cases characterizing the loading conditions during and after the failure of a steel tension member and (2) the establish- ment of performance benchmarks that guarantee redundant capacity in the faulted state. The reliability principles used in the development of loading combinations described in the AASHTO LRFD BDS and the AASHTO MBE were used to cal- culate loading combinations applicable to nonlinear FEA for the sudden failure of a primary steel tension member and for an extended period of service between the member failure and its discovery. Performance benchmarks considering strength, stability, and serviceability of the faulted structure were com- piled to set minimum performance requirements for typical steel bridges after the failure of a tension member. 2.2.2 Calculation of Dynamic Amplification Factor The second major effort is related to the characterization of the loading conditions in a steel bridge after the failure of a tension member. Sudden failure (i.e., brittle fracture) of a pri- mary steel tension member is a high-rate dynamic event that subsequently results in oscillation of the bridge. The inertia of the oscillating bridge will amplify the static forces present in the faulted bridge under static load conditions. It is prob- ably not practical to perform a dynamic system analysis to identify FCMs in all cases. Therefore, determining an accurate dynamic amplification factor that accounts for these inertial effects that could conservatively be applied to most cases was desired. Some modelers have used a dynamic amplifica- tion factor of 1.0, assuming the bridge acts like a weight on a spring, while the majority of the reviewed computational and experimental studies suggest a dynamic amplification factor lower than 0.5. In this research effort, the five structures previously benchmarked—and one additional structure—were modeled to characterize dynamic amplification. Specifically, the failure of a steel cross girder (bent) found in January 1978 on the Dan Ryan Expressway Bridge in Chicago, Illinois, was also modeled. The steel cross girder supported four plate girders that carried light rail traffic. This structure was also used to perform a study where failure of a steel member was mod- eled dynamically and subjected to simulated fracture events to evaluate the amplification that could be expected during the sudden failure of a cross girder. The only data that could be used to fully benchmark the analytical approach were from the testing of the twin-tub girder bridge at the University of Texas at Austin (Williamson et al. 2010, Hovell and Williamson 2008, Neuman 2009) and the Milton–Madison Truss Bridge tested by Purdue University

12 (Cha et al. 2014, Digglemann 2012, Digglemann et al. 2012). The objective of all of these analyses was to establish an appropriate dynamic amplification factor for steel bridges in which sudden failure of a primary steel tension member occurred. As will be discussed, this amplification factor was then applied to the load combination characterizing the dynamic failure event. 2.2.3 Strength, Fatigue Resistance, and Fracture Control in New Steel Bridges The third research effort was a study of steel bridge design and fabrication practice(s) that must be required to ensure strength, fatigue resistance, and fracture control in new bridges. Recommended detailing guidance was developed, based on information available in the literature, input from industry experts, and experience of the research team. In addition, details that have experienced distortional problems were identified as examples to avoid. Details without distor- tional issues can readily be evaluated, based on the load-based fatigue provisions currently described in AASHTO specifica- tions. This library of details is also used to provide a screen- ing basis to evaluate fracture vulnerability of a structure as it may not be prudent to exempt a structure from FCM hands- on inspection, based on system analysis if there are multiple poor details. Depending on bridge age, poor details heighten the potential for fatigue cracks and increase the possibility of multimember fractures during a fracture event. 2.2.4 Research Implementation in Bridge Engineering Practice The fourth and final research effort was the creation of the necessary documents to ensure that the research could be applied by practicing bridge engineers. A set of specifica- tions has been prepared in a format such that it can be used by bridge engineers to evaluate the redundancy of typical steel bridges with members that may be considered as FCMs. The specifications were written with the intent of becoming AASHTO guide specifications—focused on the evaluation of steel tension members in bridges traditionally considered as nonredundant—so that they can be designated as either FCMs or SRMs. While the proposed guide specifications con- tained herein are implementable, the authors recognized that the guide specifications will only be implemented by individ- uals with specialized skills in the area of nonlinear FEA. This is deemed appropriate at this time since such progressive collapse-type analysis is new to the bridge industry, and there is little experience with analyzing bridge structures for vari- ous member-out scenarios. While simplifications will likely be acceptable and appropriate in the future, at this point such simplifications have not been fully benchmarked. Further, the parametric studies that would be required to codify such simplifications could not be performed within the existing project constraints. To aid the users, evaluations of existing bridge designs have been provided as examples to illustrate application of the new methodology. The following four examples were developed: • Composite continuous two-span twin-tub girder bridge, • Composite single-span through-girder bridge, • Composite tied-arch bridge, and • Noncomposite continuous three-span three-girder bridge. These cases will be sufficient to illustrate proper applica- tion of the new specification provisions, as they cover a vast range of the existing steel bridge inventory.