Assessment of reliability-based design for a spectrum of geotechnical design problems

Reliability analyses were performed for a set of benchmark problems consisting of footings, piles and retaining walls, each designed according to the limit states design method outlined in EN 1990:2002 and EN 1997:2004. Close correspondence was found between reliability index values from Monte Carlo calculations and those from the first-order reliability method, which affirms the latter's utility in reliability-based design. The results further show relatively uniform reliability indices among the variety of problems considered, and are somewhat lower than the target reliability (β) of 3·8 as specified by the Eurocodes. Finally, the analyses illustrate the robustness of the limit-state design approach implemented in EN 1997:2004, but highlight the importance of choosing reasonably conservative characteristic parameter values, as well as the need for caution when designing for ground where the friction angle in particular shows a high degree of variance.