This training is an introduction to continuous modeling with FLAC2D and FLAC3D. At the end of the course, participants will master the graphical interface, documentation and the main modeling steps. Concepts are illustrated using a tunnel excavation example, from building the model geometry to results analysis. This introductory course provides the foundation for more advanced use of the software, which can be covered in more specific training modules.
FLAC3D 7 saw some impressive solution performance improvements over previous versions, but with FLAC3D 9 we've pushed this even further by optimizing the zone stiffness stability calculations, so that static and dynamic models run approximately 2-3 times faster for zones using the SOLVE command.
Maxwell Damping, an improved form of damping, is now available for zones, structures, and deformable links. With Maxwell damping, FLAC3D is a dynamics dynamo - making the analyses of large 3D site response and soil-structure interaction practical, while maintaining the high accuracy FLAC3D is known for.
Implicit solvers for saturated-fluid flow and thermal calculations have been implemented for much faster analyses. For example, the new solvers provided the same response for a nuclear waste repository thermal model run over 10,000 years of simulated time, but required less than 30 minutes to run in FLAC3D 9. The same model requires 43 hours of modeling time using FLAC3D 7. Fast analytical temperature calculation for specified sources has also been implemented.
For math calculations, assigning stresses, and accessing properties FLAC3D 9 has a modest 10% boost in performance compared to Version 7. However, if you are still running Version 6 or older, update to Version 9 and see performance improvements of between 2 and 100 times using FISH splitting (more efficient than looping) and FISH operators (multithreaded FISH functions). Actual performance gains will depend on the FLAC3D version and your computer hardware.
Zone plotting is now multi-threaded for 10x faster plot generation. And the saving/restoring algorithm is faster, especially for models with many groups and slots.