IGF::OT::IGF OTHER FUNCTIONS:A MAJOR CHALLENGE IN THE MULTIDISCIPLINARY DESIGN ANALYSIS AND OPTIMIZATION (M…

Public award description

IGF::OT::IGF OTHER FUNCTIONS:A MAJOR CHALLENGE IN THE MULTIDISCIPLINARY DESIGN ANALYSIS AND OPTIMIZATION (MDAO) FIELD IS THE HIGH LABOR EFFORT REQUIRED TO SETUP A COMPLEX ANALYSIS MODEL. THE ADVENT OF A NEW GENERATION OF MDAO FRAMEWORKING TOOLS HAS ENCOURAGED ENGINEERS TO BUILD SYSTEMS ANALYSES WITH INCREASING NUMBERS OF DIFFERENT SOFTWARE TOOLS. WHEN MODELS GET LARGE, THE WORK INVOLVED IN LINKING UP ALL OF THE INPUTS AND OUTPUTS BETWEEN CODES CAN GET VERY LARGE. THIS LINKING PROCESS CAN BE THOUGHT OF AS BUILDING UP THE DATA DEPENDENCY GRAPH FOR THE SYSTEM. IN SOME CASES, THERE MAY NOT EVEN BEEN A SINGLE UNIQUE DATA DEPENDENCY GRAPH SINCE MULTIPLE GRAPHS COULD ALL BE FEASIBLE. THIS ALREADY HARD PROBLEM IS MADE MORE DIFFICULT WHEN YOU CONSIDER THE USE OF MULTI-LEVEL MDAO OPTIMIZATION ARCHITECTURES, SUCH AS BLISS2000 OR ANALYTICAL TARGET CASCADING, ON SUCH LARGE SYSTEMS ANALYSES. THUS IT IS NECESSARY TO ESTABLISH A SET OF ALGORITHMS AND TOOLS, BACKED BY ESTABLISHED GRAPH THEORETIC APPROACHES, TO MANAGE THE COMPLEXITY IN LARGE MODELS AND REDUCE THE TIME AND LABOR COSTS OF THEIR INITIAL SETUP. RESEARCH OBJECTIVES THE OBJECTIVES OF THIS RESEARCH ARE TO: INVESTIGATE THE REPRESENTATION OF MDAO PROBLEMS, AND THEIR DATA DEPENDENCIES, USING DIRECTED CYCLIC GRAPHS (DCG) AND DIRECTED ACYCLIC GRAPHS (DAG). DEVELOP A SET OF GRAPH MANIPULATION TECHNIQUES TO DETERMINE FEASIBLE DAGS FROM A GIVEN DCG INVESTIGATE POSSIBLE METHODS OF SELECTING A PARTICULAR DAG FROM A SET OF POSSIBLE DAGS, GIVEN A CERTAIN USER/MODELER DEFINED GOAL FOR THE SIMULATION (E.G. LOWEST COMPUTATIONAL TIME, HIGHEST PARALLELIZATION, MOST HIERARCHICAL DESIGN INTEGRATE THE DEVELOPED GRAPH MANIPULATION TECHNIQUES INTO THE OPENMDAO FRAMEWORK AND DEMONSTRATE THEIR EFFECTIVENESS AT REDUCING SETUP COST ON A TEST PROBLEM.