A novel approach to engineering computing in the area of the electric power transmission systems and other large textured systems is introduced and proposed for study and evaluation. The approach is novel in various ways. It exploits the "textured" character of this class of engineering systems by constructing their model and the associated algorithmic arrangement as one unit. Conventionally engineers supply the models and problem statement and then experts in numerical work solve this mathematical programming problem. Strengthening this interface creates a truly engineering computing approach, where the system is broken up into optimally sized multiple groups of active and passive elements. Disjointness and intersection of these groups is then alternately exploited in iterative parallel processing to solve problems of dynamic or static optimization or other computations. The computational speedup ranges (on the basis of preliminary work) into orders of magnitude and it is rapidly increasing with the size of the system given the appropriate number of parallel processors. Even serial processing produces very large speedup, however. The proposed project will 1) refine the concept of textured system, based on the power system, 2) investigate convergence, computation time, applicability and accuracy, 3) examine parallel processing considerations including communications overhead, and explore links with existing computer architectures, e.g. interconnection networks, 4) evaluate effectiveness and limitations, and 5) conduct preliminary software development and testing.
