Solving the security constrained unit commitment problem: Three novel approaches

Abstract

This work proposes three novel approaches to speed up the solution of the Security Constrained Unit Commitment problem: an improvement of an active-set iterative approach taken from literature, an approach using solver callback functions for the evaluation of system and security constraints in the branch-and-bound tree, and one based on a shrinking horizon decomposition integrated with the use of callback functions. The three approaches were tested over five different case studies and compared against an approach taken from literature to assess scalability and performance. Results show that the modified iterative approach is always faster than the original one reported in the literature (between −58% and −93% run time), while the callback-based method does not reduce the computational time of large-scale instances. Finally, the shrinking-horizon-based approach was proved to be the fastest (up to −98% less time) despite not guaranteeing optimality (about 1% suboptimal).