SeisSol is a well-established simulation tool for seismic wave propagation and earthquake source dynamics at extreme scale. To be able to simulate more realistic scenarios additional material models are needed. In my PhD project I extend SeisSol to anisotropic and poroelastic materials. Anisotropy is the generalization of isotropic material behavior. In contrast to isotropy the wave velocities depend on the direction of propagation. Where the Riemann problem can be solved analytically for isotropic materials, in the anisotropic case we need a numerical solver. In poroelastic media an elastic frame interacts with a fluid, which fills the pores. The coupling between fluid and solid is modeled through a source-term, that renders the system of equations stiff. To overcome the stability issues, we introduce an element-local space-time predictor. The update scheme now requires the solution of a linear system with a few thousand unknowns. This system has to be solved for every element in the mesh. To solve this element-local system efficiently, we employ the block triangular structure of the system. The solution procedure consists of a sequence of matrix-matrix multiplications. Code generation is used to achieve high performance. Wave propagation in anisotropic media is validated and fully supported by SeisSol. Currently, we are validating the simulation results for the poroelastic case against various benchmark tests. After the wave propagation part is validated, source mechanisms in poroelastic and anisotropic material will be investigated.     «

SeisSol is a well-established simulation tool for seismic wave propagation and earthquake source dynamics at extreme scale. To be able to simulate more realistic scenarios additional material models are needed. In my PhD project I extend SeisSol to anisotropic and poroelastic materials. Anisotropy is the generalization of isotropic material behavior. In contrast to isotropy the wave velocities depend on the direction of propagation. Where the Riemann problem can be solved analytically for iso...     »