MIT16_842F09_sw06

MIT16_842F09_sw06 - Benchmarking Multidisciplinary Design Optimization Algorithms 16.842 Fundamentals of Systems Engineering Alessandro

Info iconThis preview shows pages 1–8. Sign up to view the full content.

View Full Document Right Arrow Icon
Benchmarking Multidisciplinary Design Optimization Algorithms 16.842 Fundamentals of Systems Engineering Alessandro Aliakbargolkar, Rhea Liem, Brian Yutko October 16, 2009 Tedford, Nathan, and Joaquim R. R. A. Martins. "Benchmarking Multidisciplinary Design Optimization Algorithms." (2007): 1-30. 1
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Intro/Objective Traditional sequential optimization Can’t find true optimum in MD systems Consider interdisciplinary interactions true optimum Various MDO architectures developed Monolithic (MDF, SAND) Good for small systems – scale poorly Poor for industrial settings where disciplines act largely indep System-level (CO, BLISS) Method selection is typically done ad hoc Performance dependent on implementation Comparison of results between studies is difficult Provide a framework to facilitate benchmarking architectures Describe the problem once and implement MDO methods automatically πMDO
Background image of page 2
Sequential Optimization Wing sweep is a global variable Wing thickness is local to structures Wing twist is local to aerodynamics Image by MIT OpenCourseWare.
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Sequential Optimization (cont) Always results in an elliptical lift distribution . Aerodynamic Optimization Max Range w.r.t twist s.t. lift = weight Structural Optimization Max Range w.r.t thickness s.t. stress constraints Forces Drag Displacements Weight
Background image of page 4
Non-hierarchical vs Hierarchical MDO Architectures Distributed Analysis – Non-hierarchical analysis performed in disciplinary models “centralized” optimization system level e.g., MDF, IDF, SAND Distributed Design – Hierarchical disciplinary model performs design tasks bilevel optimization e.g., CO, CSSO System Level Optimizer Analysis 1 Analysis 2 Analysis 3 System Level Optimizer Opt. 1 Opt. 2 Opt. 3 An. 1 An. 2 An. 3
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Multidisciplinary Feasible (MDF) MDA solves all governing eqns using design variables until coupling variables converge Objectives and constraints then computed Requiring MDA solution at each design point iteration is MD Feasible Optimizer Discipline 1 Discipline 2 Discipline 3 z1, x1 z2, x2 z3, x3 f, c y1 y1 y2 y2 y3 MDA x: Local Design Variable z: Global Design Variable y: Discipline states and coupling variables
Background image of page 6
Aero/Structural MDF MDF method differs up to 33% with non-MDO analysis Development of MDA is costly Sensitivity analysis is especially expensive No parallel computation outside MDA module 6 5 4 3 2 1 0 2 4 6 8 10 12 14 16 18 20 Spanwise distance (m) - [Root at left, Tip at right] x 10 4 Lift (N) MDF Elliptical Distribution Sequential Image by MIT OpenCourseWare.
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 8
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 11/08/2011 for the course AERO 16.810 taught by Professor Olivierdeweck during the Winter '07 term at MIT.

Page1 / 22

MIT16_842F09_sw06 - Benchmarking Multidisciplinary Design Optimization Algorithms 16.842 Fundamentals of Systems Engineering Alessandro

This preview shows document pages 1 - 8. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online