Towards a more rigorous and practical unit commitment by Lagrangian relaxation

A Lagrangian relaxation algorithm for power system generator unit commitment is proposed. The algorithm proceeds in three phases. In the first phase, the Lagrangian dual of the unit commitment is maximized by standard subgradient techniques. The second phase finds a reserve-feasible dual solution, followed by a third phase of economic dispatch. A mathematically based, systematic and generally applicable procedure to search for a reserve-feasible dual solution is presented. The algorithm has been tested on systems of up to 100 units to be scheduled over 168 hours, giving reliable performance and low execution times. Both spinning and time-limited reserve constraints are treated