Seismo-VLAB (SVL): An Open-Source Finite Element Software for Meso-Scale Simulations
We present, Seismo-VLAB (SVL), a new open-source finite element software designed to optimize meso-scale (i.e., order of km) simulations in the context of structural and geotechnical engineering. High-fidelity soil-structure interaction (SSI) simulations require advanced models that can capture the nonlinear behavior of soils and structures, and parallel computing capabilities to optimize the cost associated with large scale problems. These features are not generally available in open-source softwares, and more importantly, their performance degrades as the size of the problem increases, a drawback very frequently associated with SSI problems.
Since SVL is not intended to be a commercial software, it lacks of some sophisticated element and material models present in commercial applications. However, the purpose of SVL is to provide a simple interface in which enginners as well as researcher can test, modify, share, and implement not only new element and material, but also new solvers, integration schemes, and nonlinear solution algorithms. Such features, not provided by most commercial applications, are necessary as researchers to improve the modeling capabilities of the software and to use state-of-the-art nonlinear structural analyses. Currently, SVL is developed to meet two important functionalities that we believe will become important in the near future (a) can perform spatial variability of soil properties for uncertainty quantification in linear and nonlinear models of civil structures, and (b) can be coupled to high-level languages such as MatLab and Python to perform system identification for parameter estimation in nonlinear structural finite element models of civil structures.
Seismo-VLAB Modelling Capabilities
- Simulation with multi-stage analysis phases
- Dynamic nonlinear solvers for time-domain analyses
- Cutting-edge parallel linear system solvers
- Message Passing Interface (MPI) parallelization
- Domain decomposition for optimal parallel efficiency
- Perfectly matched layer (PML) absorbing boundary conditions
- Domain reduction that incorporates wave-field incoherency in truncated computational domains