Date of Award

Summer 1996

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Mechanical & Aerospace Engineering

Program/Concentration

Mechanical Engineering

Committee Director

Gene Jean-Win Hou

Committee Member

Jen K. Huang

Committee Member

Yozo Mikata

Call Number for Print

Special Collections; LD4331.E56 T39

Abstract

Design optimization provides an organized, systematic, and efficient method for obtaining design improvements based upon a specified measure of design performance. In the multidisciplinary design optimization (MDO) environment efficient optimization techniques allow complex designs to be improved by managing conflicting disciplinary design goals so as to improve the global design. This thesis studies the implementation issues surrounding MDO techniques and strategies. To facilitate this investigation two testbed problems have been selected for study. These problems consider two approaches to the conceptual design of an aeroelastic wing. The first approach is based on an idealized, two-dimensional representation of an airfoil under the action of an aerodynamic load. The second approach considers a three-dimensional finite element plate model coupled with a Computational Fluid Dynamics (CFD) aerodynamic model based on linear full potential flow equations.

Three MDO strategies were evaluated in this thesis: a loosely-coupled (sequential) disciplinary approach; a tightly-coupled multidisciplinary approach and an uncoupled (parallel) approach that provide coupling via a separate coordination scheme based on prescribed coupling parameters. A code was created that included modules for aerodynamic and structural analysis, sensitivity analysis and optimization. Structural optimization results are provided by the research software, ADS and are based on an efficient sensitivity analysis methodology built upon the ADIFOR symbolic processor system. A separate chapter is provided to discuss sensitivity analysis.

Recent trends have been toward monolithic engineering analysis codes that require huge computational resources often on the order of a supercomputer. For MDO a coupling between large scale disciplinary codes means that available computer resources cannot be taken for granted. This thesis considers the opportunities for coarse-grained parallelization in the MDO environment using the new task-parallel Fortran preprocessor, Fortran-M.

Results show that the coarse-grained parallel implementation is effective for specific MDO strategies where the analysis and optimization operations of individual disciplines can proceed concurrently. Results from the application of the various MDO strategies show very similar values for loosely-coupled (disciplinary) and the tightly coupled (multidisciplinary) optimums for the thesis problem. The sensitivity analysis implementation for this thesis was found to be efficient and robust. Further, sensitivity results demonstrated the utility of the discrete adjoint formulation for MDO.

Rights

In Copyright. URI: http://rightsstatements.org/vocab/InC/1.0/ This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).

DOI

10.25777/07xp-kg79

Download

Share

COinS