FAQ – flox-GPU

flox-GPU is highly optimized for speed and fast simulations using modern NVIDIA GPUs. The software was developed from practicioners with need for speed. If you need large simulations, with millions of cells, running in real-time: flox-GPU is your friend.
flox-GPU allows for easy and comfortable pre- and postprocessing using grids. With standard GIS-software you can easily modify and visualize grid-data.
flox-GPU ships with fast pre- and postprocessors for mesh generation and 3D visualization, optimized for very large data amounts.
flox-GPU has proven its feasibility for various challenging and large-scale practical engineering applications. The software is in practical use for more than a year.

The use of structured meshes becomes attractive again. Structured meshes were largely replaced by unstructured meshes in the last years, due to inaccurate representations of the topography by too large structured cells. This problem is solved however nowadays, thanks to the ‘number-crunching’ GPU. The possibility to discretize the domain with millions of cells and still maintain high performance, allows for very accurate topography representation.
The GPUs are especially suited and effective for models based on structured grids.
Structured grids heavily simplify pre- and postprocessing of large model domains. The creation of unstructred high-quality meshes usually requires more time, than modelling the flow. Using flox-GPU, you can save much of this time and concentrate on the modelling part. This aspect was one of our main reasons for the development of flox-GPU.
The free preprocessing tool meshmaker, enforces breaklines (as dams, embankments) on the structured grid.
High-resolution DEM data is provided more and more by many institutions, often even as free data. Such data is specially suited for models based on structured grids, like flox-GPU.

Yes, to run flox-GPU you need to have a modern NVIDIA GPU. See the hardware section for more details. Test your system using the demo-version.

No, flox-GPU ships with comfortable free pre- and postprocessors. Further, it supports common GIS raster formats. Hence, advanced data processing can be done completely within GIS (like ArcGIS oder open source QGIS).

Open a console on windows. Type ‘flox -dongle license_file.txt‘.
Replace ‘license_file.txt‘ with the license file you obtained.

Yes, we provide support in case of problems or bugs with the software during the license time span.

The demo-version is in principal fully functional. All tutorials and model tests can be run using the demo-version. It is limited, however, to a single source, a single culvert, the global friction value and a restricted set of output options.

Yes, if the 2x2m-resolution is sufficient for your purpose, you can directly model on the SwissAlti3D-raster. Just specify the raster in your input-file, add some sources and you have created your 2D-model within a few minutes!

You can easily prepare a mesh with given water depths as initial condition for flox-GPU using standard GIS operations for manipulating meshs. However, be aware that the water depth is given as “cell-centered” data and, hence, has a slightly different shape as the dtm-mesh. To obtain a “cell-centered” mesh with the correct shape, run flox-GPU with the input parameter “output B”. You will obtain a mesh named “B.tif” with the correct shape.

If the bridge does not get pressurized, a simple and convenient representation of a bridge may be an artificial increase in the roughness value. The increased roughness accounts for the additional resistance due to the bridge piers.
If the bridge may get pressurized, you can use a culvert to simulate the bridge behaviour.
Depending upon your mesh resoultion, you may add bridge structures to the DTM.

No, values marked as NODATA values in the mesh are omitted from the computational domain. Removing parts of the mesh not taking part in the simulation by marking them as NODATA values reduces memory usage and may speed-up flox-GPU simulations.

Yes, the boundary conditions always apply to “straight” edges in your coordinate system (north, south, west, east); hence, you should clip your mesh with “straight” edges. This is due to performance reasons. At the interface with NODATA values in your mesh, the water will flow out of the domain (outflow boundary).

You can set a global strickler-value in the input file.
You can specify a grid with strickler-values (the grid must have the same cell size as the DTM and the extents of the ‘B.tif’). The simplest way to create the strickler grid, is to use a Polygon-shape-file with the strickler-values specified as attribute. Then use the software meshmaker, to automatically create the strickler-grid during DTM generation.