As the state of practice moved from Allowable Stress Design (ASD) to Load and Resistance Factor Design (LRFD), a new philosophy was introduced by addressing uncertainty in load and resistance separately and quantifying the variability associated with the parameters used in design. While this approach allowed engineers to have a higher degree of control over how much weight to assign to individual variables, the entire framework still rests on what is a subjective decision, namely, the probability of failure associated with the system. The way LRFD has evolved from a geotechnical perspective since the 1970s has taken the direction of calibrating design methods to a target reliability that is correlated to the probability of failure that was considered acceptable in ASD, depending on the mode of failure being analyzed (e.g., bearing resistance, side friction). This approach is more comprehensive than previous practice given that the variability of the predictive method is addressed in the resistance factor; however, there is an important aspect that has not been investigated in depth and has to do with the compatibility of the probability of failure used to develop LRFD for foundations versus the superstructure. While the basic formulation of LRFD equates factored resistance to factored loads, compatibility of reliability has not been clearly addressed by previous work. Typical probabilities of failure of one in a hundred or one in a thousand are used to develop resistance factors for redundant and non-redundant foundations, respectively, which do not match the values used in superstructure design. For non-redundant foundations, this discrepancy highlights the need to research the effect of discontinuities in target reliability to ensure the overall philosophy of LRFD is addressed comprehensively. Another critical aspect of design that requires a more in-depth analysis is the way engineers determine and analyze redundancy. A framework for the analysis of the gray area between single elements of support and fully redundant foundations, that evaluates probabilities in terms of the conditional behavior of the members of the group, has not been fully developed and would provide a continuation of the work initiated in NCHRP Report 458: Redundancy in Highway Bridge Substructures.

There is a need for research focused on both the evaluation of the reliability between system components and creating a framework of analysis for the development of resistance factors for single elements and non-redundant groups. The treatment of foundation redundancy in the AASHTO LRFD Bridge Design Specifications should be amended to provide flexibility across various foundation types, and additional clarification may aid consistency and compatibility with superstructure design. The increasing prevalence of large-diameter foundation elements requires more explicit consideration of foundation redundancy issues that has raised questions regarding current provisions and highlighted inconsistencies among different foundation types. Engineers need a practical and consistent means to address the complex issue of foundation redundancy for all foundation types and to provide resistance factors that are probabilistically compatible with the design philosophy used for the superstructure.

The objective of this research is to develop a guide for practitioners that (1) establishes criteria for the assessment of redundancy of superstructures and foundation components of bridge structures; (2) provides a consistent means to establish resistance factors for single foundation elements in line with the load demands and the probabilistic models used in LRFD; (3) includes a framework for the analysis of non-redundant foundation groups to fill the gap between single elements of support and fully redundant foundations; and (4) includes suggested language for code revisions suitable for consideration by the AASHTO Committee on Bridges and Structures.