Multidisciplinary Design Optimization (MDO)

Multidisciplinary Design Optimization (MDO) is about optimizing the performance and reducing the lifecycle costs of complex systems involving multiple interacting disciplines, such as those found in aircraft, spacecraft, automobiles, industrial manufacturing equipment, various consumer products, while developing the necessary mathematical and computational design methodologies and tools.

MDO is a broad area that encompasses design synthesis, sensitivity analysis, approximation concepts, single and multi-objective optimization, rule-based design, and mathematical programming, all in the context of integrated design dealing with multiple disciplines and/or subsystem interactions. The goal is to "fine tune" the configuration of a system to be as good as it can be, while satisfying all constraints and boundary conditions. Some more recent research in design "synthesis" is trying to auto-generate system architectures using a set of simple rules.

Contributions to multidisciplinary design optimization include:

(*) Isoperformance -- a method for obtaining sets of designs that meet a vector of performance targets within some numerical tolerance, while satisfying secondary objectives
(*) Adaptive Weighted Sum (AWS) multiobjective optimization - an improvement of classical weighted sum opimization to find Pareto frontiers efficiently and uniformly
(*) Integrated System Level Optimization for Concurrent Engineering (ISLOCE) - an approach to system optimization that does not replace but integrates human designers
(*) Coupled vehicle design and network flow optimization for complex transportation systems - co-designing vehicles and the transportation networks in which they operate
(*) Variable chromosome length genetic algorithm (VCLGA) - the length of the chromosome changes over time as more detail is added to the definition of the configuration
(*) A modular state-vector based modeling and optimization approach. Application to diesel exhaust after-treatment architectures, easily swapping in and out and evaluating different combinations of component technologies
(*) An approach to maximizing expected performance and availability of extreme long-endurance systems so that they can operate in partially degraded state.