Windtunnel simulation

A toy model that solves the Navier Stokes equations for the flow of a wing around a wind.

Compiling

These instructions have been tested on with:

• cmake/3.25.2
• nvhpc/22.2 The nvhpc suite provides the fortran openmpi cuda aware implementation.

You will need a Fortran compiler and CMake installed. You will also need an MPI library. If building with CUDA support (for an NVIDIA GPU), the MPI library needs to be CUDA aware. In most cases , it should be enough to run:

mkdir build
cd build
FC=mpif90 cmake ../

If building with GPU support, then you should define the USE_CUDA variable:

FC=mpif90 cmake ../ -DUSE_CUDA=ON

Running

An example input file is present in the file config.txt. The main variables are given in the &SHAPEPARAMS section where:

• ALPHA is the angle of attack of the wing against the wind direction,
• M is the maximum camber of the wing and
• T is the thickness of the wing section (as a faction of the wing length).

Additional variables in this section are nx_global and ny_global which are the number of cells in the x and y direction respectively.

The calculation can be launched on a GPU by setting device = .TRUE. and device= .FALSE. if you do not want to compile for a GPU.

If you run the code from the build directory, you will need to:

cp src/windtunnel .  # Copy the executable to the local directory
cp ../config.txt .   # Copy the configuration file to the local directory.
export OMP_NUM_THREADS=${OMP_THREADS}
mpirun -np ${NUM_RANKS} ./windtunnel

You can then run the program by assigning the environment variable OMP_NUM_THREADS with an integer less or equal to the number of cores available on your machine and NUM_RANKS can be set as the number of MPI processes to run the code on.

export OMP_NUM_THREADS=1 # Number of OpenMP threads to use
export NUM_RANKS=4       # Number of MPI processes to run on
mpirun -np ${NUM_RANKS} windtunnel

The program will write the output binary files:

• output.dat
• potential.dat: the potential flow

The output files contain a binary dump of arrays containing grid and field information. Also several information is printed out in the standard output, including drag and lift.

Wind shape

The shape of the wing is a cambered 4-digit NACA airfoil https://en.wikipedia.org/wiki/NACA_airfoil . An implementation is found in the areofoil routine at vars.f90:159.