9622106 Boyd This research focuses on models and solution methodologies for minimizing fuel consumption in compressor-driven flow networks. Recent advances in the fields of discrete optimization and large-scale integer programming will be employed, and solution algorithms will be developed on parallel computing platforms. The domain of feasible solutions prescribed by the models is mathematically complicated by virtue of the existence of both nonlinearities and discontinuities. Classical techniques are wholly inadequate for many of the large and growing pipeline networks in operation today. Necessary mathematical results will be developed for obtaining optimization algorithms, first, for the steady-state case, and then, for the transient case. Intelligent search procedures will be developed and mixed-integer linear programming models will be developed for the steady-state case. Based on the experience gained in the steady-state analysis, an algorithm for solving the transient case will be developed. Results of this research can be used to operate existing networks efficiently, to evaluate tradeoffs between safety and throughput, and to design safe and energy-efficient new networks. The direct impact of the research will be applicable methods for reducing operating costs and pollutants generated by the thousands of gas pipeline networks in operation throughout the world. A better understanding of a little-studied class of network flow problems with broad applicability in a variety of industries will also result.
