MTKPowerConverters.jl: Converter Modeling and Simulaiton in the Julia SciML Ecosystem

Sergio A. Dorado-Rojas, Nicolas Dimate, Patrick Panciatici, Sorin Olaru, Daniel K. Molzahn

• Author's copy

• URL

bibtex Entry

@inproceedings{dorado-rojas2026a,
  title = {{{MTKPowerConverters}}.jl: {{Converter Modeling}} and {{Simulation}} in the {{Julia SciML Ecosystem}}},
  booktitle = {2026 {{IEEE}}/{{CCAE Symposium}} on {{Advanced Electrification Systems}} ({{SAES}})},
  author = {Dorado-Rojas, Sergio A. and Dimate, Nicolas and Panciatici, Patrick and Olaru, Sorin and Molzahn, Daniel K.},
  date = {2026-09},
  doi = {10.1109/SAES71201.2026.11766937},
  url = {https://ieeexplore.ieee.org/document/11766937},
  eventtitle = {2026 {{IEEE}}/{{CCAE Symposium}} on {{Advanced Electrification Systems}} ({{SAES}})},
}

Abstract

MTKPowerConverters.jl is a proof-of-concept open-source Julia library for the simulation of power electronic converters. The library adopts the design philosophy of the Modelica language: acausal, equation-based components compose without rewriting, and physical model specification stays cleanly separated from the numerical solver. Built on ModelingToolkit.jl and relying on the solver catalog of DifferentialEquations.jl, the models are natively embedded in the Julia SciML ecosystem, potentially enabling automatic differentiation and physics-informed analysis workflows unavailable in closed tools. Three experiments validate the concept. Smooth surrogates of the PWM carrier waveforms are compared against their ideal discontinuous counterpart to quantify the accuracy versus solve-time trade-off. An open-loop buck converter switching at 100 kHz is cross-validated against MATLAB/Simscape, with close steady-state agreement and a smaller end-to-end solve time. Finally, an open-loop singlephase inverter is benchmarked against the same reference under numerically demanding conditions, confirming that the approach is robust and comparable in accuracy to commercial tools.

Back