Dynamic optimal dispatch of real power for thermal generating units

A primary objective in electric power system operation is the scheduling of the real-power outputs of generators to minimize total fuel consumption while supplying all load requirements. This scheduling process is known as economic dispatch. As practiced today, economic dispatch is a static optimization procedure since costs associated with the act of changing the outputs of generators are not considered in the optimization problem. A practical method for considering change-related costs in economic dispatch is presented here and is called dynamic optimal dispatch. The dynamic optimal dispatch method uses forecasts of system load to develop optimal generator output trajectories. Generators are then driven along the optimal trajectories by the action of a feedback controller. Optimal trajectories are first calculated for nominal load forecasts using quadratic programming or gradient projection methods; then are updated using neighboring optimum methods as load forecasts are revised through time. The dynamic optimal dispatch method which has been developed has a number of important advantages over a previously reported scheme: (1) the method is applicable to a practical system having a large number of generators, (2) the method does not require a basic change in present system control philosophy, and (3) all parameters used in the method are measurable by field test. Studies of a small sample system indicate that dynamic optimal dispatch may be economically attractive. Savings of several tenths of one per cent of total fuel consumption seem entirely possible if dynamic rather than static optimal dispatch is employed. However, final assessment of the economic attractiveness of dynamic optimal dispatch must await field tests to determine the magnitude and functional form of change-related costs for generators. Suggested methods for making such field tests are outlined