CHAPTER 4

Knowledge Gaps

The literature review and its attendant synthesis indicated several areas of bridge rail practices for which further investigation would be beneficial to guide and substantiate proposed revisions to AASHTO LRFD BDS Section 13.

Proposed Design Loads

The proposed design loads outlined in Chapter 3 and included in the proposed Section 13 of AASHTO LRFD BDS may require further investigation, comparison, and validation. Proposed design load estimates were derived in three ways: (1) computer simulation in LS-DYNA (Silvestri-Dobrovolny et al. 2017; Bligh et al. 2017), (2) computational extrapolation from instrumented wall tests (Beason et al. 1989), and (3) a theoretical impulse-momentum method (Ritter et al. 1993). A comparison of these estimates is presented in Table 41. As shown, proposed loads are widely distributed for TL-3, TL-4, TL-5, and TL-6. It should be noted that the instrumented wall extrapolation method was used to formulate current AASHTO LRFD BDS requirements, but test results were scaled to NCHRP Report 350 criteria, rather than MASH 2016. Alternative design loads should be compared to impact loads measured in full-scale crash tests to form a relative hierarchy of accuracy. A preliminary comparison of these loads to impact loads measured in full-scale TL-3, TL-4, and TL-5 crash tests indicates relatively good agreement. The shaded row is the primary reference used to establish recommendations for this report.

Static Equivalent Design Loads

It is common for tested hardware to perform adequately even when subjected to peak dynamic loads far in excess of the system capacity. Although there are several possible explanations for this phenomenon—including DIFs, expected material strength, and alternative failure mechanisms—one possible explanation is that the duration of the impulse load is simply too short to result in significant deformations in the system. Williams, Abu-Odeh, and Bligh (2019) proposed this phenomenon as a possible explanation for a bridge rail system performing adequately under an estimated overburden of 167%. For example, for a concentrated mass dynamically excited by a triangular impulse load, the duration of the load must reach roughly 33% of the structure’s natural period for the dynamic load to have the same effect as a static load of the same magnitude. A preliminary investigation performed herein (Chapter 2) estimates the duration of impact load pulses between 50 ms and 100 ms and the natural period of concrete barriers around 30 ms. These results suggest that dynamic response factors for bridge rails subjected to traffic impacts will not be less than one. However, this investigation was rudimentary, and further research

Table 41. Summary of lateral impact force estimates.

^aIf the height of the rail is 36 in. for TL-4 or 42 in. for TL-5.

^bLS-DYNA estimates from NCHRP Project 20-07(395) (Silvestri-Dobrovolny et al. 2017) and NCHRP Web-Only Document 326 (Bligh et al. 2017) are the most recent estimates.

^cIf height of rail is >36 in. for TL-4 or >42 in. for TL-5.

NOTE: NA = not available.

is recommended to explore this possible explanation for the discrepancy between design loads, impact forces, and observed damage.

Effect of Mass Activation

Mass activation is another possible explanation for the discrepancy between design strength and peak dynamic loads. It is possible that barriers that are unable to support the impact load are only able to contain impacting vehicles with little damage by virtue of the barriers’ mass. This effect has been investigated by Schrum et al. (2016) and Badiee (2014), though further research is recommended to investigate this behavior in greater detail. The results of impact testing of the RESTORE barrier at MwRSF (Schmidt et al. 2015) suggest that mass activation plays a significant role in vehicle redirection. This system, which was designed to deform significantly under an impact load, imparted peak forces to the impacting vehicle that were similar to those of tests involving rigid barriers. While the system was designed with low lateral stiffness, it is possible that, by virtue of its mass, the system acts as an effectively rigid barrier for short pulse loads.

Concrete Bridge Rail Design Methods

Several alternative design methods have been proposed for solid concrete bridge rails. These methods include (1) a trapezoidal yield-line mechanism, (2) inertial activation, (3) scaling for impact height, and (4) punching shear capacity. Capacities determined in each of these methods should be compared to known failure loads measured in full-scale, dynamic, or static testing to determine the methods that are in best agreement with physical testing.

Although not explicitly investigated in any research effort, damage observations strongly indicate that torsional behavior plays a role in the limit state of concrete barriers. As such, the potential for torsional failure may be a key aspect in future evaluations of concrete barriers.

Additionally, a significant knowledge gap was identified in the evaluation of open concrete rails. Currently, researchers at MwRSF are investigating alternative yield-line mechanisms for

open concrete rails, and the results of this study (Delone and Schmidt 2019) will be monitored and included in future project reports, time permitting.

Transmission of Impact Loads to Deck Structure

When a vehicle strikes a bridge rail system, impact loads are transmitted through the rail and into the deck structure. Even with the focus of a research effort performed by Frosch and Morel (2016), the manner in which impact loads are accepted by the deck remains elusive and requires further investigation. Physical testing with instrumented bridge deck specimens, such as that performed by TTI on the William P. Lane, Jr. Bridge Rail (Williams, Menges, and Kuhn 2018), may provide valuable insight to better describe the transmission of impact loads into the superstructure.

Effect of Deck Flexibility on Barrier Capacity

Research efforts performed by Badiee (2014) and Matta and Nanni (2009) suggest that accounting for deck flexibility effectively increases the capacity of the attached bridge rail. This increase in strength is due to the current yield-line analysis method’s assumption that all external work must be absorbed by the barrier. With these modified methods, strain energy absorption occurs within both the bridge rail and the deck in cantilever bending. Further research is recommended to investigate this topic.

Required Deck Overhang Strength

Current AASHTO LRFD BDS guidance suggests that the flexural capacity of the deck should be greater than or equal to the cantilever bending capacity of the bridge rail, M_c. Several research efforts have demonstrated that decks with flexural capacities much lower than M_c can perform adequately, with the bridge rail failing before significant damage is sustained by the deck. The lowest ratio achieved was 0.45M_c in a static test performed by Alberson et al. (2005) on a TL-4 F-shape barrier. However, prescribing a factored M_c that could be deemed universally adequate is not currently possible, as this behavior is a function of the ratio of the barrier’s cantilever bending strength, M_c, to its wall bending strength, M_w, as well as stiffness effects afforded by the barrier at the edge of the deck. Further research is required to determine the required deck capacity to support bridge rail impacts.

Barrier Anchorage

Current AASHTO LRFD BDS guidance requires that anchor bars for bridge rail systems are able to develop their yield strength. This requirement can create proportioning problems for existing or unconventional bridge decks, and systems not satisfying this requirement have proven adequate. Additionally, AASHTO LRFD BDS commentary suggests that anchorages to concrete should be designed in accordance with ACI 318. However, ACI 318-19 Chapter 17, which addresses the design of anchorages to concrete, does not apply to anchorages subjected to impact loading (ACI 2019). As such, further investigation into barrier anchorages, particularly for cast-in-place and precast concrete barriers, is warranted.

Ground-Mounted Railing

NCHRP Report 663 (Bligh et al. 2010) and NCHRP Web-Only Document 326 (Bligh et al. 2017) investigated bridge rails secured to MSE walls. However, no guidance was identified in this review regarding the installation of bridge rail systems on grade beams, footings, or other unconventional foundations, nor is this discussed in AASHTO LRFD BDS.

GFRP-Reinforced Concrete Bridge Decks

No literature was identified documenting MASH crash testing of bridge rails secured to GFRP-reinforced concrete bridge decks. One report was identified in which a GFRP-reinforced deck supporting an open concrete rail was subjected to two tests according to NCHRP Report 350 test designation 3-11. Two more recent studies were identified in which GFRP-reinforced decks supporting post-and-beam railings (one concrete, one steel) were subjected to quasi-static testing. The most recent available guidance from AASHTO is found in the 2nd edition of the LRFD Bridge Design Guide Specification for GFRP-Reinforced Concrete (AASHTO 2018). For deck overhang design, Section A5.4 of the Guide Specification directs the reader to the LRFD BDS Section 13 Appendix, Article A13.4.1; and Section 3, Article 3.6.1.3.4. Further research is recommended to validate the design guidance provided in the 2nd edition Guide Specification in full-scale MASH crash testing.

Alternate Methodologies

The force-based equivalent static method is well established and has a long history of providing crashworthy, although perhaps nonoptimized, barriers. Alternative methodologies may provide avenues to realize improved material and construction efficiencies while also meeting barrier crashworthiness criteria.

Rigorous Nonlinear Dynamic Method

A rigorous nonlinear dynamic analysis may be used for novel systems and unusual situations that are not adequately addressed in existing crash-testing records. Some discussion of this approach could be included in the Commentary or an Appendix for a rewritten Section 13 to summarize general guidelines and best practices when using advanced modeling software packages (e.g., LS-DYNA or Abaqus). Supplementary requirements may be appropriate if relying on this method, similar to particularly significant or unusual building designs for which a peer review is required for a design to be deemed acceptable.

Simplified Nonlinear Dynamic Method

It may be possible to realize some degree of the benefits from a rigorous nonlinear dynamic analysis from a simplified nonlinear dynamic approach. A simplified nonlinear dynamic method could provide a more realistic representation of vehicle impacts and demands on bridge railings than a conventional equivalent static force-based method and may enable refinement of design demands without compromising an acceptable margin of safety. A simplified nonlinear dynamic method would also potentially be suitable for evaluating or designing systems excluded from equivalent static inelastic methods, such as timber, FRP, or barriers incorporating elastomeric components. This method would be intended to provide a dynamic analysis methodology with a more simplified representation of the impacting vehicle and structural system than would be required for a rigorous nonlinear dynamic approach. The simplified method could potentially be implemented in practice using routine, accessible analytical tools for practice, such as Excel, SAP2000, and RISA-3D. Additionally, a simplified dynamic method could provide an approximate validation tool that could be coupled with a rigorous dynamic method to provide reassurance of modeling validity for more complicated systems and computational tools. Matta and Nanni (2009) used a simplified nonlinear analysis method in their study on open concrete rails.

A simplified nonlinear dynamic procedure is envisioned to focus primarily on TL-3, with extensions to SUT and tractor-trailer vehicles to be evaluated pending a viability and utility assessment at TL-3. The vehicle would be modeled as a concentrated mass with rigid elements

to establish a spatial boundary. Modeling the vehicle with rigid elements will neglect crushing in the vehicle body and associated energy dissipation and will therefore provide a degree of conservatism. The railing would be modeled independently to characterize elastic and potentially inelastic (pushover) force-displacement response. The interaction of the simplified impacting vehicle and impacted railing would then be evaluated through numerical integration in time. As the vehicle impacts the barrier, the translational momentum of the vehicle will be converted into a lateral impulse force and vehicle angular momentum manifested as yaw, roll, and pitch. The analysis could conservatively neglect mass activation of the barrier, an assumption that is currently implicit in existing Section 13 methods, or it may more realistically account for railing mass inertia. Deck design would be similar to the conventional and updated methods for an equivalent static force-based approach incorporated in the current project. Decks would need to be capacity-protected to ensure that nonlinearity develops in the railing to conform to analysis assumptions, or the formulation would need to explicitly incorporate deck nonlinearity.

Displacement-Based Equivalent Static Method

A further simplification to a displacement-based equivalent static method could potentially be developed by approximating the simplified nonlinear dynamic approach. The primary aspect of simplification would be in the static versus dynamic nature of the procedure. Characteristics of the barrier (e.g., height, stiffness) would be mapped into an expression to adjust loading demands from a rigid barrier (which is used for the equivalent static force-based method) to a barrier with some degree of flexibility.

Protection for Superstructure Elements

AASHTO LRFD BDS provides specifications addressing protection of structures, but these are limited to substructure elements. AASHTO LRFD BDS also has provisions addressing impact loading of sound barriers that are located on or behind traffic railings. Guidance is lacking regarding protection of superstructure elements immediately behind and/or extending above traffic railings. Example superstructure elements include arch and suspension bridge hanger cables, through-truss members, and stay cables.