Laboratory Framework for the Initialization of Modelica Models in Power System Time-Domain Simulation
Vicenc Gaitan, Milenko Halat, Pere Gimenez Febrer, Quentin Cossart, M. Chiaramello, Mathilde Bongrain, Adrien Guironnet · 2022 IEEE Power & Energy Society General Meeting (PESGM) · 2022
Initialization of large differential-algebraic systems as those found in power system simulations requires specific techniques to make the starting state computable. Modelica simulation tools such as OpenModelica or Dymola offer different methods, for example homotopy or tearing, but they lack flexibility for experimentation since the algorithms are hard-coded. In this work, a Python-based full pipeline that serves as a laboratory for testing methods on solving the initialization problem of Modelica models is presented. It is explained how a) equations are extracted to a symbolic environment for manipulation and b) the framework solves these through a state-of-the-art implementation of tearing decomposition. Finally, the pipeline is applied to power systems examples extracted from the Dyna$\omega$o Modelica library.