Development of an Automatic Design and Optimization System for Industrial Silencers

Current computer-aided design (CAD) software tools focus on rapid production of computer models, which usually takes place after the product design is completed. The product design process, which has more significant influence on product life-cycle costs, is not fully supported. This work documents the development of an automatic design and optimization system for industrial silencers. The developed system greatly reduces the silencer design time from one day to a few minutes. Moreover, the system proves the feasibility of developing an open-architecture CAD system supported by design of experiment (DOE) based optimization methods to integrate product life-cycle considerations into the design. It is expected that the developed system can help the development of similar systems for other products. Through the development of this system, some further research issues are identified.
Introduction
Today's global market demands quicker, cheaper, and better product designs for manufacturing industries. The computer-aided technologies help to reduce production time and costs. However, most current CAD tools function as a productivity aid to help speed up the modeling and drawing generation process. The product design, which generally influences 70-80% of product life-cycle costs (Boothroyd, Dewhurst, and Knight 1994), has not been directly supported by current CAD tools.
In the framework of concurrent engineering, much research has been done recently in the area of design automation with integrated optimization tools in CAD systems. One contribution proposed by the research team from Brigham Young University (Rohm III et al. 2000) provides a methodology of integrating parametric design with a programmatic toolkit to optimize product design. The methodology was applied to the design of a jet component using an interactive CATIA® environment together with the programmatic program, the CATIAIUA language. The complicated free-form surfaces of the airfoil and impeller of a turbine blade are designed using the CATIA programmatic toolkit. Also, the Brigham Young team tried to develop a common graphical user interface (GUI) to ease the programming and communication between various CAD packages. Line and Steiner (2000) in their research proposed a concept of automatic calculation of product architecture metrics using a solid modeling program, I-deas®, as the modeling tool. A program that uses internal I-deas functions is created to find all of the joined parts. The strength of each joint is calculated for adjusting the parts connectivity and the average joint strength to obtain a satisfactory design. Line and Steiner define the architecture of the product as the scheme in which functions are mapped to physical components. The result is a modular architecture (a one-to-one mapping of function to component) and an integral architecture (a many-to-one mapping of function to component). Then, Line and Steiner's method of architecture calculation is coded and integrated with the CAD tool to perform the calculation of architecture metrics automatically. A similar concept on design automation was developed by Chan and Lewis (2000), which involved integration of manufacturing and cost information into the engineering design process. Their research was proposed to create a design for manufacture (DFM) system called the DFM-C system for small to medium-sized enterprises (SMEs) to seek manufacturing and economic benefits. The DFM-C system was designed to apply to the conceptual phase of the design process, and its structure was based on an expert system called CLIPS (C-Language Integrated Production System). There are three basic components in the DFM-C system: a knowledge base that contains designs and design-related knowledge; the material selection (MS), manufacturing process (MP), and cost estimation modules to select appropriate materials and processes; and an inference engine. The program is written in C and is integrated with the CLIPS expert systems programs for filtering incompatible processes and optimizing selected processes. Bras and Kalyan-Seshu (1997) and KalyanSeshu and Bras (1998) integrated design for "X" tools with CAD systems (I-deas and Pro/E) to achieve optimal design ("X" stands for manufacturing, assembly, environment, and so on). Due to the unavailability of programming tools, the integration was semi-automatic. Forster, Boufflet, and Lecouvreur (2000) proposed an automatic design method to construct tolerance chains of a mechanical assembly. Esche, Chassapis, and Manoochehri (2001) addressed the product and process design using knowledge-based module in hot forging processes. They used the combination of concurrent engineering and a knowledge-based system to motivate the development of automated concurrent engineering software (ACES) for part design and manufacturing. The system combines deterministic and empirical knowledge of a variety of product aspects to provide decision-making power to the designer. The ACES is capable for material and machine selection, process design, die design, and early cost estimation. The user first specifies the basic design requirements. The part geometry is generated in an associated CAD system. Then, the relationship between customer requirements and the part model is established with constraints and is imposed automatically on the system. The system will alert the user of any design constraint violation during the design and modification phase. Thus, the overall design process is iterative with integrated design, manufacturability analysis, and cost analysis procedures. In a brief summary, it is generally believed by many researchers today that the commercial CAD systems should be advanced to a "design" tool considering product life-cycle aspects rather than being a mere modeling tool.