Date of Award

8-2011

Document Type

Thesis

Degree Name

Master of Science (MS)

Legacy Department

Mechanical Engineering

Advisor

FADEL, GEORGES M

Committee Member

MILLER , RICHARD S

Committee Member

QIAO , RUI J

Abstract

The ever increasing demands towards improvement in vehicle performance and passenger comfort have led the automotive manufacturers to further enhance the design in the early stages of the vehicle development process. Though, these design changes enhance the overall vehicle performance to an extent, the placement of these components under the car hood also plays a vital role in increasing the vehicle performance. The study of the placement of these components in an automobile underhood forms a 3-Dimensional packaging problem. In the past, a study on the automobile underhood packaging problem was conducted and a multi objective optimization routine with three objectives namely, minimizing center of gravity height, maximizing vehicle maintainability and maximizing survivability has been setup to determine the optimal locations of the underhood components. Also in the past, another study was conducted which asserted the need for the inclusion of the thermal performance of the vehicle Underhood as an objective to the optimization routine proposed earlier.
This study makes an assessment of the several available thermal analyses that are performed on the automotive underhood to evaluate the thermal objective. The assessment conducted in this study indicates that these thermal analyses, when included into the rigorous optimization routine, increase the computationally expense by a large amount. Thus an approximate thermal model is presented to evaluate the thermal performance of the vehicle underhood, which when included as an objective into the optimization routine, does not make it computationally expensive. The approximate thermal model is a Neural Network approximation of the CFD analysis conducted over the automotive underhood. The results obtained from the neural network are compared with the CFD results, showing good agreement. The Neural Network model is now included in the multi objective optimization routine and the results are obtained. A non-deterministic evolutionary multi-objective algorithm (AMGA-2) is used to perform the optimization process.

Included in

Engineering Commons

Share

COinS