DYNAMIC ECONOMIC-DISPATCH FOR LARGE-SCALE POWER-SYSTEMS - A LAGRANGIAN-RELAXATION APPROACH

The dynamic multi-period economic dispatch problem for large-scale power systems is modelled as a linear programming problem. The model considers loading and deloading rates, limits on generators outputs, spinning reserve requirements and group power import-export limits. The solution algorithm is based on Lagrangian relaxation and on exploiting the intimate relationship between optimizing the dual Lagrangian function and Dantzig-Wolfe decomposition. The relaxation is carried out so that the relaxed problem is decomposable to a number of subproblems corresponding to the periods in the dispatch horizon. These are solved simply by using priority lists. The dual Lagrangian function is optimized using subgradient optimization. If an overall solution feasible in all constraints and sufficiently close to a computed best lower bound is discovered during subgradient optimization, it is deemed optimal. Otherwise, Dantzig-Wolfe decomposition is invoked, using almost all the information generated during subgradient optimization to ensure a speedy conclusion. The computational efficiency of the algorithm renders it suitable for on-line dispatch.