Skip to main content

SKILLS / MCP + Coding AI

eMachineSim is used through a Python API and JSON input control files. In real motor development, however, engineers also need to reason about mesh regions, property IDs, loss sources, cooling paths, thermal equivalent circuits, gap coupling, and output diagnostics.

The eMachineSim SKILLS workflow is intended to make that process easier to start and easier to audit with coding AI. SKILLS provide reference knowledge for how an AI assistant should reason about eMachineSim workflows. The MCP server gives AI agents structured access to Examples, input JSON files, and diagnostic CSV outputs.

What Coding AI Can Help With​

TaskTypical AI-assisted work
Choose an exampleSelect the closest thermal, rotor structural, or modal/NVH example.
Prepare input.jsonModify losses, boundary conditions, output settings, and coupling options.
Run an analysisGenerate a small Python script using eMachineSim.run_file() or Session.solve().
Inspect resultsRead heat balance, FEMH outflow, gap heat flow, and surface assignment CSV files.
Summarize findingsProduce a short Markdown summary of conditions, key values, and remaining assumptions.

When the MCP server is available, an AI agent can also use tools to inspect:

MCP-assisted checkTypical use
input JSON overviewCheck metaData, BDF import, volume heat sources, gap coupling, and interface resistance settings.
Example discoveryFind thermal, realistic motor, rotor structural, modal/NVH, and electrostatic examples.
Thermal heat balanceSummarize thermal_global_heat_balance.csv and thermal_fem_volume_heat.csv.
Surface assignmentCheck FEMH/gap overlap, unassigned external area, and gap-only area.
Thermal pathsSummarize gap coupling, shared-node interface resistance, and cooling path CSV files.
JSON variantsCreate focused variants for gap h, gap Nu, or PM total_heat settings.

Detailed MCP server setup, connection, and tool names are documented in the public repository under mcp_servers/emachinesim/README.md. This page focuses on what the workflow enables.

Typical Motor Analysis Flow​

  1. State the engineering target, for example: "run steady thermal analysis with coil copper loss and PM loss."
  2. Ask the coding AI to select the relevant eMachineSim SKILL and Examples.
  3. Review the proposed input.json changes.
  4. Run the analysis through the Python API.
  5. If the MCP server is available, let the AI inspect CSV diagnostics through structured tools.
  6. Adjust FEMH, gap coupling, or thermal-network parameters as needed.

Engineering Responsibility​

SKILLS and the MCP server help prepare and inspect analysis workflows. They do not replace engineering judgment. Users should still verify material properties, mesh quality, boundary conditions, loss scaling, and any calibration against test data.

The MCP server assumes the public Python API / pyd workflow. User-facing analysis execution should use eMachineSim.run_file() or Session.solve(), not an internal debug executable.

For thermal motor examples, always inspect at least:

  • thermal_global_heat_balance.csv
  • thermal_component_heat_balance.csv
  • thermal_volume_heat_sources.csv
  • thermal_surface_assignment_summary.csv
  • thermal_cooling_path_summary.csv
  • thermal_gap_coupling_summary.csv and thermal_gap_coupling_flow_summary.csv when gap coupling is used