Decomposition algorithms for global solution of deterministic and stochastic pooling problems in natural gas value chains

Abstract

In this thesis, a Benders decomposition algorithm is designed and implemented to solve both deterministic and stochastic pooling problems to global optimality. Convergence of the algorithm to a global optimum is proved and then it is implemented both in GAMS and C++ to get the best performance. A series of example problems are solved, both with the proposed Benders decomposition algorithm and commercially available global optimization software to determine the validity and the performance of the proposed algorithm. Moreover, a two stage stochastic pooling problem is formulated to model the optimal capacity expansion problem in pooling networks and the proposed algorithm is applied to this problem to obtain global optimum. A number of example stochastic pooling problems are solved, both with the proposed Benders decomposition algorithm and commercially available global optimization software to determine the validity and the performance of the proposed algorithm applied to stochastic problems.