Guidance on conducting 2D linear viscoelastic site response analysis using a finite element code

Various software packages are available to conduct one-dimensional (1D) and two-dimensional (2D) site response analyses (SRAs). In this article, a finite element program is tested with the purpose of assessing the importance of several aspects on the obtained results and verifying the software. Abaqus (Dassault Systèmes) is compared with the 1D SRA software STRATA for simple 1D models to understand the influence of the boundary conditions; as being a 1D SRA program, STRATA does not require vertical boundary conditions. For Abaqus, the subroutine by Nielsen is used to implement free-field boundary conditions. In addition, we test the influence of mesh dimension and Rayleigh damping as well as the importance of buffer-zone width. 2D SRAs with Abaqus and FLAC3D, which is commonly used for geotechnical analyses, are compared as part of the assessment. Similar results are obtained from the two programs, but Abaqus is preferred as it is more efficient for linear elastic analyses than FLAC3D, which, on the contrary, performs well for soil presenting strongly non-linear behaviour and effective stress. We demonstrate that reliable results can be achieved, not only for simple uniform sites but also for complex sites with multiple layers and dipping stratigraphy.