A Sparsity-Based Technique for Identifying Slow-Coherent Areas in Large Power Systems

A sparsity-based technique is developed for the identification of coherent areas in large power systems. The technique, based on the slow-coherency approach, is novel in that it introduces small machines at the load buses to retain the system sparseness. Then the computation of the slow eigenbasis for the identification of slow-coherent groups of machines is performed by the Lanczos algorithm which is an efficient eigenfunction computation method for large, sparse, symmetric but unstructured matrices. The technique also groups the load buses into coherent areas, information that is useful for network reduction. Two large scale models of portions of the U.S. power system are used as illustrations. The computation time required is of the order of magnitude of that required for a few load flow solutions.