Modeling and optimizing dynamic networks: Applications in process engineering and energy supply

We deal with the modeling, simulation, and optimization of dynamic networks from application domains as diverse as power grids and district heating on the one hand, and batch distillation processes on the other hand. In order to do so, we use the tailor-made dynamic network software AD-Net. AD-Net consistently uses automatic differentiation in the solution of the differential-algebraic network equations and, therefore, sensitivities of all process quantities w.r.t. user-specified parameters come for free. In particular, all arising dynamic optimization problems can be solved efficiently by means of external gradient-based optimizers. In this contribution, the focus is on the version of AD-Net for batch distillation processes. Apart from the simulation of batch distillation columns with liquid-liquid equilibria, it allows for the efficient estimation of thermodynamic mixture parameters like NRTL parameters and for the optimal design of reflux ratio profiles and fractionation schemes so as to achieve desired product purities.