Air-Gap Force FFT

Python postprocessor for motor air-gap flux-density results. It reads Br and Btheta, calculates Maxwell stress, evaluates integrated force and torque, and extracts time-space harmonic modes with 2-D FFT.

The included sample is the healthy-position result for the GL80 PMSM.

The project is open source under the MIT License.

Repository Layout

.
|-- src/                         # Processing scripts and reusable modules
|   |-- airgap_force_fft.py       # CLI entry point
|   |-- airgap_io.py              # JSON/CSV readers and sector expansion
|   |-- airgap_fft.py             # Maxwell stress, FFT, torque comparison
|   `-- airgap_plotting.py        # Plotting utilities
|-- tests/                       # Unit tests
|-- data/
|   `-- GL80/healthy_position/   # Public sample input and generated outputs
|-- docs/                        # Discussion notes and analysis reports
|-- config.yaml                  # Default GL80 example configuration
|-- pyproject.toml               # Packaging metadata
`-- requirements.txt             # Runtime dependencies

Quick Start

Install the published package from PyPI:

python -m pip install airgap-force-fft

For development, create an environment and install the repository dependencies:

python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt

Run the default GL80 example:

python .\src\airgap_force_fft.py --config .\config.yaml

The main outputs are written under data/GL80/healthy_position/fft_output by default.

Input Formats

Supported air-gap flux-density inputs:

• EMSolution/pyemsol gapB JSON
• Long-form CSV with columns such as time, theta_deg, Br, and Btheta

For EMSolution JSON, geometry and symmetry metadata can be read from the file when available. For CSV input, set the missing metadata explicitly in config.yaml, especially:

• input.sector.periods
• input.sector.symmetricity
• geometry.radius_m
• geometry.stack_length_m

For full 360-degree CSV data, use input.sector.periods: 1 and input.sector.symmetricity: 0.

Torque Reference Files

Torque comparison supports two reference formats:

• The earlier transient_results.json style with Time.data and Torque.*.data
• pyemsol output_transient.json, where torque is read from postData.forceNodal.forceNodalData entries whose propertyNum is stator or rotor, using forceMZ

The comparison plot uses rotor-positive convention: -Ftheta integrated, -stator, and rotor.

Reports

Key analysis reports are in docs/:

• GL80 baseline analysis
• Original discussion notes
• Implementation plan

Tests

Run the unit tests with:

python -m unittest discover -s tests -v

Packaging

For editable development installation:

pip install -e .

After installation, the CLI can also be invoked as:

airgap-force-fft --config config.yaml