NuSiF-Solver/README.md

70 lines
2.0 KiB
Markdown
Raw Normal View History

2023-02-05 07:02:36 +01:00
# NuSiF-Solver
2023-02-07 14:06:08 +01:00
This is a broad collection of C implementations of Navier-Stokes Fluid simulator as
described in [this
book](https://epubs.siam.org/doi/book/10.1137/1.9780898719703) by Michael
Griebel et al.
## Build
Every variant comes with a makefile. First configure the toolchain by editing
the variabel `TAG` in `config.mk`. For the Intel tool chain set it to `ICC`, for
LLVM clang using Open MPI set it to `CLANG`. If required adapt the respective
flags and include and library paths in `include_[CLANG|ICC].mk`.
You can uncomment the `-DVERBOSE` line for additional dignostic output.
Build the code with:
```
$ make
```
Clean intermediate build products with:
```
$ make clean
```
Clean all build products with:
```
$ make distclean
```
The makefile provides the additional build targets:
* `$ make info`: Output CFLAGS and the compiler version for documentation
purposes.
* `$ make tags`: Run `ctags -R`.
* `$ make format`: Run `clang-format` on all source files.
## Run
All examples come with two test cases:
* dcavity: Lid driven cavity.
* canal: Channel flow with inflow on one side.
To start a simulation for sequential implementations execute (here for CLANG toolchain):
```
$ ./exe-CLANG dcavity.par|canal.par
```
To start a simulation for parallel implementations execute (here for CLANG toolchain):
```
$ mpirun -np <NUMPROC> ./exe-CLANG dcavity.par|canal.par
```
## Configuration options
Important option in the test case parameter files:
* `re`: Desired reynolds number. Higher for a less viscous fluid and lower for a
more viscous fluid.
* `imax`, `jmax`, `kmax`: The resolution of the grid. Increase resolution for
higher Reynolds numbers.
* `te`: The final simulation time.
* `itermax`: The number of maximum iterations performed in every Poisson
pressure equation step.
* `eps`: Stopping tolerance for the pressure equation solver.
* `omg`: Relaxation parameter for SOR solver. Default is 1.7 to 1.9. For MPI
parallel runs omega needs to be reduced to 1.3 to guarantee convergence.
2023-02-07 16:08:48 +01:00
The other parameters should not be changed.
2023-02-07 14:06:08 +01:00