Gas Network Optimization by MINLP

Jesco Humpola
Seminar

We present a procedure for topology optimization of large-scale real-world natural gas distribution networks. It decides which combination of network extensions such as additional pipelines, compressors or valves should be added to increase the network's capacity or enhance its operational flexibility. We formulate this as a mixed-integer nonlinear problem. A sub-problem has different convex reformulations. Hence we use a combination of linear outer approximation and NLP solution techniques to solve the MINLP. We show that every dual solution of the convex reformulations allows to generate capacity inequalities (or cutting planes) which reduce the overall solution time when added to the formulation. The dual solution also enables the measurement of infeasibility level of the scenario. Furthermore we give a primal heuristic for our model. We present computational results that are obtained by a special tailored version of the solvers SCIP and IPOPT.
 
Short Bio:
2004-2006 Undergraduate Studies TU Bremen (GER)
2006-2009 Diploma University of Bonn (GER)
2009-2014 PhD at TU Berlin / Zuse Institute Berlin (GER)