Multidisciplinary System Design Optimization
Multidisciplinary system design optimization: decomposition, gradient and heuristic algorithms, sensitivity, multiobjective and robust design.
Designing a complex engineering system means balancing many disciplines at once, and a rigorous quantitative method has to work alongside the creative side of design. This graduate course teaches multidisciplinary system design optimization: how to break a system into disciplinary models and reintegrate them, explore the design space, apply gradient-based and heuristic optimization, and interpret results through sensitivity and tradeoff analysis. Teams apply the method to a design problem of their own across the term.
What You Will Achieve
- Decompose a complex system into disciplinary models and manage their interfaces
- Choose between gradient-based and heuristic optimization algorithms
- Perform sensitivity and post-optimality analysis on design results
- Compute Pareto fronts for multiobjective design problems
Program Overview
Problem Formulation and Modeling
Setting up a design optimization problem.
- Problem formulation
- Modeling, simulation and problem decomposition
- Design space exploration and visualization
Optimization Methods
Algorithms for finding good designs.
- Numerical optimization and sensitivity analysis
- Simulated annealing and genetic algorithms
- Multi-objective optimization and isoperformance
- Approximation methods
Design for Value
Robustness, value and flexibility in design.
- Post-optimality analysis and robust design
- Design for value and flexibility
- MDO in industry practice
Course Completion
Job Outlook
Design optimization is used in aerospace, automotive and energy engineering, and in product development for complex systems.
Good for: Graduate engineering students with a computational background who design complex products or systems.
Keep in mind: A spring 2010 course with lecture notes, five assignments with solutions and five example team projects with papers and presentations; the course assumes a prior course in computational science and engineering.
Bottom line: A complete, project-driven course on optimizing multidisciplinary engineering designs.
Frequently Asked Questions
What do the example projects cover?
Five team projects with papers and presentations: a wind turbine blade, a supersonic business jet, a barge, the Space Shuttle external tank and an airbag system for Orion.
Which algorithms are taught?
Gradient-based methods such as sequential quadratic programming, and heuristic methods such as simulated annealing and genetic algorithms.
How is the team project graded?
A rubric scores problem formulation, modeling and validation, optimization and tradeoff analysis, and presentation clarity, alongside a short final paper.



There are no reviews yet.