H_SIM-2D

Main features

• Implicit approach → enables larger time steps (fast calculation).
• Very good parallelization and vectorization (developed and tested on supercomputers with up to 256 processors), test results on request.
• Extremely short computing times: on newer CPUs / workstations approx. 10 to 20 times faster than explicit models.
  • works very well for operational use.
  • Very large high resolution models (up to 50 million points / elements and more) can be calculated relatively quickly .
  • Long-term simulations can be carried out successfully in a short time .
• Unstructured meshes , consisting of triangular and square elements, enable precise adaptation to structural courses in the area.
• Conservative formulation of the shallow water equations → conservation of mass and momentum.
• ENO method in combination with streamline diffusion → high accuracy and very robust in use.
• Simultaneously occurring subcritical and supercritical flow states - with and without hydrautic jumps - can be calculated.

A runoff simulation with 5 million points (point spacing around 1m, simulation time: 90,000 seconds) is completed e.g. on an older workstation * in a few hours. Please do not hesitate to contact us for more detailed information about the available computing times and the process itself.

* Workstation with 2 x Intel (R) Xeon (R) CPU E5-2690 v3 @ 2.60GHz; (12 physical cores per CPU thread (s) per core: 2.

Additional information on the H_SIM model family

• Due to the short computing times, a majority of the existing 2d models could already be used for real-time simulation or for operational use directly .
• The model creation runs fully automatically , similar to our earlier model HYDRO_AS-2D, but without such strong thinning of laser data (if at all).
• Updating of such a model, e.g. as a result of current laser data, is much easier and would mostly take place automatically. The user should only check whether something has changed at the points where boundary conditions and structures are defined and make adjustments there if necessary.
• The model operation takes place via the SMS interface , similar to the previous one. Otherwise, it is currently being examined together with a sales partner of ARCView to what extent a direct connection to these systems could be possible, especially with regard to flash floods and operational use.
• The current model uses shared memory (so-called shared memory model). A MPI model version (so-called distributed memory model), which runs on several computers with their own memory, is expected to be completed in the middle of next year. As a result, a significant increase in computing speed can also be expected.
• A GPU version should also be ready by the middle of next year.
• In the current program version only flow modeling is implemented. In the following years, the development of the sediment transport models (long-term simulation for longer flow stretches) follows.
• We are also working together with the Technical University of Rijeka on a 3d or a coupled 2d / 3d model (completion expected in 2 to 4 years).