Towards a coupled (MI)NLP solver for coupled energy networks

  Tanja Clees
  Fraunhofer SCAI

In addition to power grids, gas grids play a decisive role in current and future energy scenarios. For both, circuits and pipe networks, simulation can be based on Kirchhoff's laws and corresponding nonlinear systems of constraints. We developed a system for simulation, analysis and optimization of energy networks, called MYNTS (multiphysical network simulator). While its predecessors were developed for microelectronic circuits, the current system already works for large gas networks, water cooling circuits as well as power grids with a selection of classical elements. The talk gives an overview on modeling the networks, formulating in NL and setting up strategies for their efficient numerical solution via ipopt. In particular, we discuss physically oriented "ramping" strategies for the NLPs, graph mining and simulation vs. optimization. Results for realistic network examples are presented. An outlook is given on combined gas&power grid modeling and MINLP solving.