Date of Award

12-2011

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Mechanical Engineering

Advisor

Summers, Joshua D

Committee Member

Fadel , Georges M

Committee Member

Mears , Laine

Committee Member

Mocko , Gregory M

Abstract

ABSTRACT
This dissertation presents a design method to reduce engineering changes caused due to change propagation effect. The method helps designers to systematically plan a verification, validation, and test (VV&T) plan. The rationale behind such a method is founded on a well-accepted principle that a robust validation plan can reduce design changes. However, such method has not yet been developed in mechanical engineering domain, so a method from software engineering has been adopted and extended to address the limitations in the existing design evaluation tools.
Tools extensively used in industry, such as FMEA, and in academia have been reviewed to determine if they can identify different propagation pathways including variant, behavior, organization, and geometric pathways. As a result, it is found that variant and organizational pathways are not identified in any of these tools -- propagation in these pathways have caused major product failure in commercial vehicle and automatic fire sprinkler manufacturing industries.
A seven-step VV&T method is proposed to address the aforementioned gap in which each step is tailored to suit mechanical engineering needs. The major contribution is developing the construct to identify variant and organization pathways and a prescriptive method. It has been validated in a leading commercial vehicle manufacturer, one of the passenger car manufacturing giants, and an automatic fire sprinkler manufacturer. The results from these three companies indicate the proposed VV&T method enables designers to identify variant and organizational pathways and evaluate them, which in turn can reduce design changes due to propagation effects. Objective evidence obtained from the fire sprinkler manufacturing company supports this claim.
'If we know what assembly combination to test with, testing is not a problem...and if it can prevent a failure of this magnitude --I think this method can --it can be extremely beneficial...'
- Project engineer, commercial vehicle manufacturer

Share

COinS