A modified priority list-based MILP method for solving large-scale unit commitment problems

This paper presents a two-step modified priority list (MPL) based mixed integer linear programming (MILP) method for improving the computational speed of unit commitment (UC) programs while preserving optimality. In the first step, the heuristics of UC results for a given generation fleet are investigated to develop the MPL. A subset of the generators are determined to be online (committed) or offline (uncommitted) within a planning period (e.g., a week), based on the demand curve and generator priority list. Then, for generators whose on/off status is predetermined, the corresponding binary variables are removed from the MILP solving process. After this simplification, the remaining problem can be solved much faster using an off-the-shelf MILP solver, based on the branch-and-bound algorithm. Scale factors are used to adjust the tradeoff between solution speed and level of optimality. Simulation results show that the proposed method can significantly speed up the large-scale UC problem with negligible compromise in optimality by selecting appropriate scale factors.