Reinforcement scheduling convergence in power systems transmission planning

The paper presents a new approach to the reinforcement scheduling problem in transmission power systems planning. This problem is typically formulated as a large scale multistage decision under uncertainty problem. We represent uncertainty by load/generation scenarios and propose the use of Gaussian search techniques to solve the scenario deterministic subproblems. The solution of each subproblem is a schedule of reinforcements. As information is different from scenario to scenario, scheduling conflicts may appear at optimal subproblem decisions. We solve these conflicts with a progressive hedging algorithm. Decision convergence on reinforcement scheduling is achieved for the first decision stage by progressively modifying the individual scenario subproblems. We present a rule for modifying scenario subproblems, discuss the algorithms presented, and illustrate their application with an example built over the IEEE 14-bus network.