Order reduction for nonlinear dynamic models of district heating networks.

This thesis focuses on the formulation of reduced order models for a numerically efficient simulation of district heating networks. Their dynamics base upon incompressible Euler equations, forming a system of quasi-linear hyperbolic partial differential equations. The algebraic constraints introduced by the network structure cause dynamical changes of flow direction as a central difficulty. A control system is derived allowing to analyze essential properties of the reduced order model such as Lyapunov stability. By splitting the problem into a differential part describing the transport of thermal energy and an algebraic part defining the flow field, tools from parametric model order reduction can be applied. A strategy is suggested which produces a global Galerkin projection based on moment-matching of local transfer functions. The benefits of the resulting surrogate model are demonstrated at different, existing large-scale networks. In addition, the performance of the suggested model is studied in the numerical computation of an optimal control of the feed-in power employing a discretize-first strategy.