BEMPPSolver

Basic utilities for running Boundary Element Method (BEM) simulations on loudspeaker surface meshes and generating directivity and impedance plots.

The project is organized as a simple four-step workflow:

1. cleanmesh.py
2. bemppsolver.py
3. prepare_visualization_data.py
4. visualizer.py

What It Does

Given a loudspeaker mesh in Gmsh .msh format with two surface groups comprising a rigid enclosure and a driven diaphragm, this project can:

• clean and stitch a triangle surface mesh for BEM use
• run a frequency sweep with bempp-cl
• compute normalized horizontal and vertical polar response
• compute acoustic impedance
• generate directivity and impedance plot images

Requirements

This project is written in Python and depends on:

• numpy
• meshio
• scipy
• matplotlib
• bempp-cl
• pyopencl

Example install:

pip install numpy meshio scipy matplotlib bempp-cl pyopencl

To utilize OpenCL solving, you will need to download the Intel OpenCL runtime from this page: https://www.intel.com/content/www/us/en/developer/articles/technical/intel-cpu-runtime-for-opencl-applications-with-sycl-support.html, which is compatible with both AMD/Intel CPUs.

Mesh Expectations

The solver expects a triangular surface mesh with physical groups in the .msh file.

• One surface group should represent the rigid enclosure.
• One surface group should represent the driven diaphragm or throat surface.
• In bemppsolver.py, the driven surface is selected with tag_throat, which defaults to 2.

If your mesh uses different physical tag values, update the configuration in bemppsolver.py before running the solver.

Workflow

1. Clean the mesh

Use cleanmesh.py to merge coincident vertices, remove degenerate or duplicate triangles, and write a cleaned mesh.

Default input/output:

• input: samplemesh.msh
• output: samplemesh_clean.msh

Example:

python cleanmesh.py samplemesh.msh samplemesh_clean.msh --merge-tol 1e-9

If you run it without arguments, it uses the editable constants near the top of the file.

2. Run the BEM simulation

Use bemppsolver.py to run the acoustic simulation over a frequency sweep.

Default mesh:

• samplemesh_clean.msh

Default output:

• pressure_data.npz

Important settings are still defined in SimulationConfig in bemppsolver.py, including:

• frequency range and number of frequency points
• worker count for process-based parallel execution
• sound speed and density
• observation distance
• axial offset for polar origin
• tag_throat
• mesh scale factor

Run with defaults:

python bemppsolver.py

Run with CLI overrides:

python bemppsolver.py samplemesh_clean.msh --output-npz-base-path pressure_data --freq-min 200 --freq-max 20000 --freq-count 72 --polar-angle-step-deg 2.5 --polar-angle-min-deg -180 --polar-angle-max-deg 180 --observation-axial-offset-m 0.116 --workers 4

The solver currently supports these CLI inputs:

• positional mesh_file
• --output-npz-base-path
• --freq-min
• --freq-max
• --freq-count
• --polar-angle-step-deg
• --polar-angle-min-deg
• --polar-angle-max-deg
• --observation-axial-offset-m
• --workers

--workers uses a spawn-based process pool and splits the frequency sweep into equally sized chunks. This is process-based rather than thread-based so it works consistently on Windows, macOS, and Linux.

3. Prepare visualization data

Use prepare_visualization_data.py to convert raw solver output into plot-ready arrays.

Default input:

• pressure_data.npz

Default output:

• pressure_data_formatted.npz

Run with defaults:

python prepare_visualization_data.py

This step also applies clipping, interpolation, and fractional-octave smoothing for the isobar plots.

Run with CLI overrides:

python prepare_visualization_data.py pressure_data.npz pressure_data_formatted.npz --min-db -30 --max-db 0 --isobar-angle-samples-smooth 250 --isobar-freq-samples-smooth 500 --isobar-octave-smooth-fraction 24 --horizontal-reference-angle-deg 10 --vertical-reference-angle-deg 10

The prep stage currently supports these CLI inputs:

• input_polar_npz
• output_npz
• --min-db
• --max-db
• --isobar-angle-samples-smooth
• --isobar-freq-samples-smooth
• --isobar-octave-smooth-fraction
• --horizontal-reference-angle-deg
• --vertical-reference-angle-deg

4. Generate plots

Use visualizer.py to generate PNG outputs.

Default input:

• pressure_data_formatted.npz

Default outputs:

• horizontal_isobar.png
• vertical_isobar.png
• acoustic_impedance.png

Sample output previews:

Run with defaults:

python visualizer.py

The visualizer supports these CLI inputs:

• positional input_npz
• --output-horizontal-png
• --output-vertical-png
• --output-impedance-png
• --isobar-interp-freq-factor

Typical Run Sequence

python cleanmesh.py
python bemppsolver.py
python prepare_visualization_data.py
python visualizer.py

Notes

• cleanmesh.py writes Gmsh 2.2 format for compatibility.
• bemppsolver.py is configured to use CPU devices by default via OpenCL