Embedded building 3D with MPI level-set based refinement

Author: Marc Núñez Corbacho

Kratos version: 8.1

MMG version: 5.5.2

ParMMG version: 1.3.0

Source files: MPI Embedded building 3D

Application dependencies: FluidDynamicsApplication, LinearSolversApplications, MeshingApplication, MetisApplication, TrilinosApplication

To run this example, please first unzip the file:

skin_building.zip

And then execute:

export OMP_NUM_THREADS=1
mpirun -n nprocs python3 MainKratos.py

Where nprocs is the number of processors to use in the MPI run.

This example adaptively refines a background mesh according to the distance to a building. The refinement is then driven by the distance field only, and we can customize the size of the elements according to this, depending on how close/far are they from the geometry. Check the instructions below to modify these parameters.

This considerably improves the representation of the building on the mesh. It then start a a fluid dynamic analysis with a constant logarithmic inlet, with the geometry embedded and represented by a distance field.

Initial discretization

Final discretization

Velocity field

How to use

In ProjectParameters.json:

• “end_time” to change the total length of the simulation.
• “time_step” to change the time step used.

In RemeshingParameters.json:

• “number_of_iterations” to change the total number of consecutive remeshing steps.
• “minimal_size” to set the minimal size of the mesh. This will be the size set at `distance=0.0’.
• “maximal_size” to set the maximal size of the mesh. This will be the size set at `distance=boundary_layer_max_distance’.
• “boundary_layer_max_distance” to set the distance up to where the refinement will be performed. Elements outside this distance will keep its initial size.
• “interpolation” to set the interpolation set between the minimal_size and the maximal_size. Possible interpolation settings are: constant, linear, exponential.