Optimization-based inter-utility power purchases

Power transactions are important activities among electric utility companies. Effective transaction decisions can result in significant savings since marginal generation costs of neighboring utilities could be quite different. Transactions, however, are coupled with the scheduling of units through system demand and reserve requirements, and are confined by many accustomed rules. How to make effective transaction decisions is therefore a very difficult problem. In this paper, an optimization-based method for the integrated consideration of power purchase transactions and the scheduling of thermal units is presented based on the augmented Lagrangian decomposition and coordination method. After the system-wide demand and reserve requirements are relaxed by using Lagrange multipliers and penalty coefficients, the overall problem is decomposed into purchase and thermal subproblems. For a purchase subproblem, the optimal purchase level for each purchase interval is first determined. The subproblem is then efficiently solved by using the dynamic programming approach without discretizing purchase levels. Thermal subproblems are solved by extending our previous method recently reported in the literature. The multipliers and penalty coefficients are then updated at the high level so that system demand and reserve requirements are gradually satisfied over iterations. The augmented Lagrangian decomposition and coordination method avoids the solution oscillation difficulties associated with linear cost functions of purchase subproblems and speeds up algorithm convergence. Numerical testing results based on modified Northeast Utility's data show that the algorithm is efficient and significant savings can be obtained.