Abstract

Optimal operation of a compressor station is important since it accounts for 25% to 50% of a company's total operating budget. In short-term management of a compressor station, handling demand uncertainty is important yet challenging. Previous studies either require precise information about the distribution of uncertain parameters or greatly simplify the compressor model. We build a two-stage robust optimization framework of power cost minimization in a natural gas compressor station with non-identical compressors. In the first stage, decision variables are the ON/OFF state of each compressor and discharge pressure. The worst-case cost of the second stage is incorporated in the first stage. First-stage decision variables feasibility is discussed and proper feasibility cuts are also proposed for the first stage. We employ a piece-wise approximation and propose accelerate methods. Our numerical results highlight two advantages of robust approach when managing uncertainty in practical settings: (1) the feasibility of first-stage decision can be increased by up to 45%, and (2) the worst-case cost can be reduced by up to 25% compared with stochastic programming models. Furthermore, our numerical experiments show that the designed accelerate algorithm has time improvements of 1518.9% on average (3785.9% at maximum).