Enterprise Engineering Platform
Trox — Electric Motor Design Platform
- Problem
- Decades of motor-design knowledge lived in legacy FORTRAN — accurate, but opaque, hard to extend, and unreachable from modern tools.
- Solution
- A modern .NET platform that re-implements the physics as a tested, typed engine and exposes it as an API for engineers, automation, and AI agents.
Architecture
- Electromagnetic engine with SI units enforced by typed structs — every formula traced to a published reference
- Lumped-parameter thermal network solver with a parametric sweep endpoint
- MCP server so AI assistants can call the engine as a tool
- OAuth 2.0 / JWT SSO with Kerberos constrained delegation for per-user SAP access
- C#
- ASP.NET Core
- EF Core
- SQL Server
- MCP
- SolidWorks API
Architecture Details
Trox — Electric Motor Design Platform
The goal was not a rewrite for its own sake: it was to keep proven engineering results while making the math inspectable, testable, and callable from anywhere.
Legacy FORTRAN routines were reverse-engineered, then rebuilt in a dedicated core library with formulas sourced to standard motor-design texts (Pyrhönen, Boldea & Nasar, and others).
- Core
- MotorDesign.Core class library — winding factors, magnetic circuit analysis, equivalent-circuit parameters, and iron-loss modelling. Units are typed structs, so a dimension error fails at compile time instead of on the test stand.
- API
- ASP.NET Core Web API over EF Core and SQL Server with a repository/service layer, seeded reference data, Swagger UI, and a parametric sweep endpoint for design-space exploration.
- Thermal
- LPTN (lumped-parameter thermal network) solver exposed alongside the electromagnetic engine.
- CAD
- SolidWorks COM API integration for automated STEP export.
- Identity
- OAuth 2.0 JWT bearer tokens via enterprise SSO, with Kerberos Constrained Delegation propagating the user principal to SAP.
- AI
- Model Context Protocol server (ModelContextProtocol.AspNetCore) so agents can query designs and run calculations. In progress: an in-app AI chat built on a backend-orchestrated agent loop over the same tools.
- Telemetry
- Event-sourcing-style action log built on Channel<T> and a BackgroundService, so instrumentation never blocks a request.
- C#
- ASP.NET Core
- EF Core
- SQL Server
- MCP
- SolidWorks API