A DEVELOPMENT METHODOLOGY FOR PARAMETRIC SYNTHESIS TOOLS

ABSTRACT Software to support the solution generation phase of the engineering design process has been developed in academia for decades. Computational synthesis software enables generation of solutions on both conceptual and embodiment level. This paper focuses on the class of parametric design, such as documented in mechanical engineering handbooks. Examples include machine elements such as bearings, springs, fasteners, transmissions, etc. A parametric synthesis tool automates the engineering design process from functional requirements to quantified solutions, for a single machine element. Since the amount of machine elements is vast and software development time should be low, a generic methodology is helpful to speed up this process. This paper discusses such a methodology to develop synthesis tools for the class of parametric designs. It includes an analysis-oriented approach to formalize the design process' parameters in terms of embodiment, performance and scenario. Mathematical constraint solving techniques are used to generate candidate solutions. Graphical presentation and exploration of the solution space is done with interactive plots. A standardized layout for the graphical user interface is suggested to allow uniform and intuitive use. A demonstrator is developed using the described methodology and several challenges are discussed for improved constraint solving techniques, more advanced visualization and handling problems with higher complexity. Although small in size, parametric design processes are time consuming due to their reoccurring nature. Developing synthesis tools for these designs will allow engineers to save time and improve design quality. INTRODUCTION The industrial product design process is experiencing an increasing amount of pressure to cut cost and reduce time to market [1], [2], while product complexity increases. Both from an engineer's and manager's point of view, uncertainties and risks in the engineering design process are being reduced where possible Software support to increase the efficiency of the design process has been developed for decades. Ullman [4] describes an ideal support system for the mechanical engineering design process: it should allow the designer to work from functional requirements to a feasible product. The system should provide insight in the relationship between the set of requirements and the eventual product. This should be an automated process, guided by the preferences of the engineer, leading him/her towards the best possible design. Support systems that provide multiple possible solutions give an overview of alternatives and allow the engineer to select the best solution, rather than find one acceptable design. Academia have recognized the need for support systems to generate (alternative) solutions. Decades of research have proven successful in exploring different types of support in different phases of the design process using a number of approaches and techniques. For example, computational synthesis offers support for design processes ranging from engineering design problems to shape driven (architectural) design