Adaptive linearized models for optimization of gas networks

We are interested in simulation and optimization of gas networks. Usually, a gas network consists of various components like compressors and valves connected by pipes. The aim is to run the network cost efficiently whereas demands of consumers have to be satisfied. This results in a complex nonlinear mixed integer problem. We address this task with methods provided by discrete optimization. Therefore, the gas dynamics in all pipes and at compressors must be described by piecewise linear constraints. We introduce an adaptive approach for the linearization process to handle the complexity on the one hand and the aimed accuracy on the other and present numerical simulation and optimization results based on our model.