specifications was warranted. The proposed solution by the subcommittee was to transition to a more coherent method of design where applicable failure and serviceability conditions could be evaluated considering the uncertainties associated with loads and resistance. Work began on the new document in 1989 with the objectives of creating a state-of-the-art, comprehensive specification that was consistent with itself and multidisciplinary in approach (Withiam, et. al., 1998). In 1994, the first edition of the AASHTO LRFD Bridge Design Specifications was approved and released for use by designers. This significant effort utilized the Ontario bridge LSD code developed in 1979 as a starting point, and the migration from ASD to LRFD for all bridge design began (Allen 2013). The early versions of the LRFD specification contained comprehensive design and construction guidance for both structural and geotechnical features. However, it was quickly realized that the approach used for structures was not fully compatible with geotechnical design needs. To address some of these early issues, FHWA led a 1998 scan tour of Canada, Germany, France, Denmark, Norway and Sweden to review and document developments in LRFD for geotechnical engineering features. More specifically, the team wanted to obtain information on the history of LRFD use, development, implementation and performance (DiMaggio, et al., 1999). This led to a complete rewrite of the foundations portion of the AASHTO LRFD Bridge Design Specifications, which was published in 2005. In 2000, at the advice and recommendation of the AASHTO SCOBS, the FHWA issued a Policy Memorandum announcing its decision regarding a timeframe for transition to the use of LRFD for the design of new bridges on projects funded through the Federal- aid Highway Program. The policy was that all new bridges would be designed by the LRFD specifications after October 2007, and that all new culverts, retaining walls and other standard structures would be designed by the LRFD specifications after October 2010. Today, the AASHTO Standard Specifications (last published in 2002) are no longer being maintained in favor of the AASHTO LRFD Bridge Design Specifications, which most would agree are the most complete standard for geotechnical design. The standard is very widely used and referenced inside and beyond the transportation community. Transitioning to LRFD LRFD represents an approach in which applicable failure and serviceability conditions can be evaluated considering the uncer- tainties associated with loads and material resistances assuming a target reliability index. Previous practice using ASD required engineers to place all uncertainty in terms of a single factor of safety to account for the variation of applied loads to the foundation and geotechnical capacity of the soil and rock. The factor of safety was empirically developed, but is generally arbitrary and subjective. ASD is primarily used to reduce the potential for poor performance, but it is not inherently related to the reliability of the design in terms of probability of failure. Depending on the project type, design model and experience of the designer, the factor of safety could range from approximately 1.2 to 6 (Withiam et al., 1998). Figure 1 illustrates the design approach for ASD. The simplified figure highlights one of the principal limitations of ASD in that the values of load and resistance, Q and R, are assumed deterministic and have a probability of occurrence of one. Also, the safety factor or margin is more qualitative than quantitative in that designers will broadly consider sources of variability resulting in an inconsistent level of safety in design. Figure 1. Allowable stress design approach It is acknowledged that geotechnical engineers have been somewhat slow to embrace LRFD given the difficulties with quantifying the variability of geotechnical properties and design models. One could argue that LRFD is not superior to ASD for geotechnical applications and that there are benefits to ASD, such as its simplicity and the large experience base. However, LRFD does have the potential to provide a more rational approach to design since uncertainties can be incorporated quantitatively into the design process. LRFD incorporates analysis and design methodologies with calibrated load and resistance factors based on the known variability of applied loads and material properties. These load and resistance factors are calibrated from statistics to ensure a uniform level of safety, and the risks and uncertainties associated with the safety of the system can be defined in mathematical terms. Figure 2 illustrates the LRFD design approach where various load effects are multiplied by appropriate load Figure 2. Load and resistance factor design approach DEEP FOUNDATIONS • MAY/JUNE 2015 • 13
