Performance-based evaluation of seismic-displacement resistance of geosynthetic-reinforced soil walls using FFDM
Abstract
The force-equilibrium-based Finite Displacement Method (FFDM) and the conventional Newmark’s method are reviewed and applied to three revisited sites of geosynthetic reinforced soil retaining walls (GRSW) subjected to strong ground shaking. The strengths and limitations of both methods are examined through their predictions of seismic displacements. For the GRS FHR wall (Tanada wall) in the 1995 Hyogo-ken Nambu earthquake and the GRS MB walls at Sites 1 and 3 in the 1999 Chi-Chi earthquake, the FFDM rendered seismic displacement estimates consistent with those observed or inferred from field performance. Results using Newmark’s method show that displacement estimates inherently span an order of magnitude, even when properly calibrated soil strengths, expressed in terms of the friction angle (φ) and critical seismic coefficient (khc), are used. For GRSWs with relatively high khc, displacement evaluations using Newmark’s chart may become infeasible. In contrast, the FFDM produces seismic displacement resistance curves that span a wide displacement range (10 -3– 10 -1 m) under small to high levels of shaking, making it suitable for pre-earthquake hazard mitigation planning for displacement sensitive soil structures, such as GRSWs for transportation applications. The differing displacement prediction capabilities of the two methods stem from their fundamental assumptions: Newmark’s method employs simplified soil strength parameters (c, φ) with complex seismographs, whereas the FFDM uses deformation related soil parameters with simplified pseudo static seismic forces. Given the potential of the FFDM for displacement-sensitive soil structures, selecting appropriate deformation related soil parameters is key to reliable seismic displacement prediction. This need can be addressed through parameter sensitivity studies, back calculations and by establishing a database for stress-displacement related soil parameters.