Multidisciplinary Design Optimization (MDO) can be defined as a formal methodology for the design of complex coupled systems in which the synergistic effects of coupling between various interacting disciplines/phenomena are explored and exploited at every stage of the design process.

Why MDO ? To enable the design of high performance products, Balance product performance considerations with manufacturing, economics, and life cycle issues, Achieve design process timetable compression - Economic competitiveness, Respect the problem physics of coupled systems - towards a physically meaningful design practice.

Must Read : Current State of the Art on Multidisciplinary Design Optimization - A White Paper by the AIAA MDO TC. See also IIan Kroo's White Paper on MDO and Computation-Based Design, and Optimization: An Essential Tool for Decision Support, a presentation by John Dennis at the MPS Symposium.

Obstacles to MDO Practice

Computational Cost : (1) The requirement of large-scale computationally expensive models for high fidelity analysis in disciplines such as aerodynamics and nonlinear aeroelasticity. (2) The requirement of iterations for analysis of coupled systems.

Organizational Challenges and Misconceptions : Hurdles in Analysis Code, Data and Human Team Organization; Incompatibility Between Disciplinary analysis codes; Complexities in Software Integration; Difficulties in using existing stand-alone analysis and design packages with rich user interfaces in MDO; Defining the roles to be played by the various departmental design teams within a MDO setting; Acceptability of new methodologies by Industry - work culture shock - how do we get around it ? The notion of automated integrated multidisciplinary design tools is still prevalent. Are the mammoth cost and man hours required for the development of integrated analysis and design tools worth the effort ? Can MDO retain the advantages of division of labor and autonomy of disciplinary specialists ? MDO is "NOT" a push-button approach : Eradicating the confusion between MDO, and the misguided futile attempts made earlier for replacing the human element and do automatic design. MDO is not an alternative to traditional Knowledge Based Engineering practice - MDO complements KBE tools!

Some Recent Developments

Consensus on the best short term way ahead: Methodologies which fit into existing industry organizational setups, exploiting the well known advantages of division of labor. See the Technical Report Archive for a collection of online papers on various research topics related to MDO.

Approximation Concepts: Response Surface Methodology, Design and Analysis of Computer Experiment (DACE) Modeling, Approximation Concepts for systems governed by linear PDEs, Constant-free bounds on numerical solution of PDEs.
MDO Architectures: Collaborative and Concurrent Subspace Optimization, Coevolutionary Optimization Architectures, Improvements in Heuristic Procedures for Decomposition of MDO Problems, Formal Classification and theoretical analysis of MDO architectures, Comparison of MDO formulations on various test problems.
Sensitivity Analysis: Ever growing focus on development of disciplinary sensitivity analysis capabilities in analysis software, Automatic Differentiation Tools, Adjoint Methods.
Optimization Algorithms and Theory: Convergence Theories for Variable-Complexity Trust Region-Based Approximate Optimization Approaches and a Class of Multilevel Optimization Algorithms, The renaissance of Pattern Search Algorithms - Convergence theory for problems with general nonlinear constraints ! Balanced Local-Global Search Algorithms - Towards Dichotomous Optimization Algorithms, Multiobjective design, DSP, Advanced Design Parameterization Strategies.
Software Infrastructure: Experiences in use of Heterogeneous and Parallel Computing Architectures, Distributed Object Computing Technology - CORBA, DCE, Seamless Integration of MDO in IPPD, commercial software supporting MDO, Smart Product Modeling, Software Development and Computational Infrastructures for Data Management and Visualization,
Analysis Methodology: Theoretical Developments in Coupled Field Formulations and Solution Strategies - The creation of new interdisciplinary fields.

Some Areas for Further Research

More Flexible MDO Architectures for tackling problems with broad coupling bandwidth; Efficient techniques for solving mixed variable optimization problems; Convergence theories for general multilevel optimization algorithms; Development of Nongradient Search Algorithms for Multilevel Optimization Problems; Overcoming the curse of dimensionality in Response Surface Methodology by employing principles from statistical learning theory; Physics-based surrogate modeling; Dichotomous Search Algorithms for Optimum Seeking, Approximation, and Archiving of large nonconvex design spaces; Research on Design Grammars/Parameterization Strategies for novel concept exploration - Exploiting advances in Feature Based Solid Modeling in MDO; Taking Robustness characteristics into account at every stage of the design process; Further work on Theoretical Formulation and Solution Methodology for Coupled Field Phenomena; Truly multiobjective design; Interpretable and Transparent Approximation Models; Reduced-Order Models for systems governed by nonlinear PDEs; Formulations for Concurrent Response and Sensitivity Analysis; Algorithmic Frameworks for Management of a Hierarchy of Variable-Fidelity Analysis Models in DSS; Balancing design considerations - Incorporating Models of entire product life-cycle from requirements phase to decommissioning in MDO; Problem-Solving Environments for MDO; MDO Technology Standards - Making the transition from sequential to concurrent design practice easier in industrial environments; Benchmarking MDO methods on Realistic Industrial Problems.

This page contains a collection of links to resources and research groups working on topics related to this field. Ongoing efforts include classification of the various publications made available online by several research groups.