Skip to Main Content
Pursuing optimal solutions for large scale transmission network planning problems is a formidable task due to their combinatorial nature and also due to the nonconvexities involved. Successful approaches using hierarchical Benders decomposition incur in a high computational cost mainly due to the need to solve a large integer program (the investment sub-problem) for every Benders iteration. In this work the authors propose to use heuristics within the decomposition framework, therefore avoiding to solve to optimality each integer sub-problem. The global computational effort is substantially reduced, and allows coping with large problems that would be intractable using classical combinatorial techniques. Case studies with the 6 bus Garver test system and a reduced Southeastern Brazilian power network are presented and discussed.