Constrained dispatch of active power by linear decomposition

The decomposition of the constrained allocation problem for active power generation is considered, and a problem formulation developed which enables the grouping of the constraints to form a structure suitable for linear decomposition. Dantzig and Wolfe showed that a finite convergence algorithm is a suitable solution of the decomposed primal problem in which a series of linear, constrained subproblems are co-ordinated by a global problem of transformed variables. The solution of a large dispatch problem is presented and the computational requirements contrasted with conventional primal and dual linear-programming techniques. The technique is concluded to permit significant computational savings and also to be suitable for implementation on a multilevel or multiprocessor computer system.