Optimal operation of large gas networks: MILP model and decomposition algorithm

As of today, natural gas is one of the most widely deployed energy sources and its transport relies on large-scale infrastructures managed by the expertise of transmission system operators. In this framework, this paper proposes a Mixed-Integer-Linear-Programming model and a decomposition algorithm to optimize the operation of gas networks. The model aims to minimize the CO2 emissions from compressor stations by optimizing the unit commitment of the gas-turbine-driven compressors, their loads and the dynamic operation of the network. The formulation includes the detailed linearization of the performance maps of the machines, their technical limitations, the dynamic conservation equations of the pipes, and the operating constraints of control valves. On top of this, the algorithm is capable of handling flow reversals in pipes, being suitable also for networks with cyclic topology. The resulting large scale MILP is extremely challenging to solve, and therefore we developed a decomposition algorithm to speed up the computational time. The decomposition algorithm is composed by two MILP levels, where the master level contains a simplified model of the compression stations and of pipe friction, while the lower level is the actual detailed model. The aim of the master level is to remove unused compression stations and to fix the flow direction in pipes in the lower level, thus reducing the size of this second problem. At the end of each bilevel iteration, an integer cut is added to the master problem to explore different combinations of committed stations. The algorithm is effectively tested on the Italian gas network case study, featuring 112 pipes and 9 compressor stations, each composed by 3-5 units.