Towards design representations for product space exploration in early design

Srikanth Devanathan, Purdue University

Abstract

Mechanical design is a complex activity involving creative reasoning, decision-making and problem solving. Although a lot of work has been done in trying to automate several aspects of the design process, design nevertheless involves enormous human involvement. Therefore, any design-support system should ease the effort needed by the designers in understanding design information and expressing design decisions. Designers reason about the design problem in multiple abstractions such as pure functional, geometric, and analytical representations. Moreover, each of these representations can themselves be represented in different levels of detail; for example, schematics, freehand sketches, 2D drawings, and 3D models can all be used to represent the geometry of a design at different stages. Designers are known to be visual thinkers and use several visual media such as diagrams, charts, and drawings for reasoning about the design problem. Yet existing design tools do not actively use these “visual tools” in a cohesive manner to support mechanical design. Apart from 3D geometric models, there are very few visual representations of parametric designs. A model of the knowledge involved in early design, what is termed working knowledge, is developed in this work to provide a basis for connecting these visual tools. The Working Knowledge Model (WKM) is developed by extending the state-of-the-art product representation models to include representations for multiple abstractions of product description and alternatives that are encountered during early design. The elements of the WKM are then used to describe commonly used visual tools in design in order to facilitate iterative exploration in early design. A prototype system has been developed that includes SysML (Systems Modeling Language) requirements & parametric diagrams, House of Quality, Hierarchical Function Structures, Morphological Matrices and 2D drawings. In order to support visual exploration of parametric design sets this thesis introduces a new polytope-based representation for design spaces. These polytopes approximate the locally connected design space around an initial feasible point. The algorithm for constructing the design space is developed by adapting consistency algorithm for polytope representations. By transforming the parametric design problem into a geometric problem, this approach paves the way for using geometric processors and algorithms for design support. The work on polytope representation of design spaces can be viewed independently of the WKM. The design of an automotive coolant flow control valve is used as a case-study to illustrate the use of WKM in supporting design exploration and creation of polytope approximations of the design space.

Degree

Ph.D.

Advisors

Ramani, Purdue University.

Subject Area

Mechanical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS