Overview:

The Virginia Tech Large-Eddy Simulator, or ViTLES, has been developed to simultaneously study the mathematics of various LES models, evaluate these models in different flow regimes, and use it in conjunction with ongoing research in control and optimization algorithms for complex flows. Although primarily a research code, ViTLES is designed for robustness, flexibility as well as performance.

ViTLES is developed at the Interdisciplinary Center for Applied Mathematics at Virginia Tech and runs on the Virginia Tech Terascale Computing Facility (System X) and minime and has been ported to Linux clusters such as The Phoenix cluster at FSU's SCS.

ViTLES is written on top of PETSc (the portable, extensible toolkit for scientific computing) developed at Argonne National Laboratory. It makes use of MPI, linpack and the blas. We have also used the automatic differentiation tool, ADIC, to compute Jacobians. This allows us to rapidly implement different boundary condition and closure models as well as perform sensitivity analysis. Routines have been developed to convert ViTLES format to several visualization routines including Tecplot, VTK and VU. Links to these packages can be found under the Resources link.

Beta Release 2.3 2006-01-05

ViTLES is currently available for local developers. We intend public release of a version of ViTLES once documentation is complete and development stabilizes.

Features of ViTLES 2.3

Compatible with PETSc version 2.3.0
Parallel code with good scalability.
Implicit, second-order time stepping.
Several Preconditioning strategies (mostly from PETSc).
Finite element-based for general two and three dimensional flows.
Benchmark meshes generated by amiral (D. Pelletier at Ecole Polytechnique de Montreal)
Several sub-grid-scale models. (Smagorinsky and Deconvolution)
Time dependent boundary conditions. (ADBC - Approximate Deconvolution Boundary Conditions)
Built-in sensitivity analysis. (Compatibility with ADIC for automatic differentiation)

Contributors

ViTLES is an integral part of many ongoing and proposed research projects and is used by many current and former graduate and undergraduate students. A list of people involved in the development of ViTLES is given below.

Math Faculty

Graduate Students

Peter Hou: Studied unknown numbering for ILU preconditioning
Alexey Miroshnikov: Added sensitivity analysis via ADIC and passive and active scalars
Miroslav Stoyanov: Parallel Riccati solver for flow control (current)
Kay Vugrin: Built many benchmark meshes in amiral

Undergraduate Students

Grant Boquet: Studied domain decomposition based on weighted adjacency information
Jason Marley: Built benchmark meshes in Matlab