Research conducted in the Caltech Engineering Design Research Laboratory is concerned with developing a fundamental understanding of Engineering Design. The two major projects are related to a) the development of formal methods for representing and manipulating imprecision in design and b) structured methods for micro-electro-mechanical system (MEMS) design.
Our structured methods for MEMS design work is based on simulation of fabrication methods, coupled with an iterative stochastic optimization. Using this approach, we can specify a desired 3-D shape, and automatically synthesize a mask-layout and fabrication process that will produce the desired shape. Integrating this approach with a simulation of function will mean that MEMS designers can concentrate on the desired behavior of a device, rather than on the details of its shape or fabrication.
We are applying our MEMS design research results in a project which is developing micro-propulsion devices for micro-spacecraft. These devices consist of arrays of small one-shot thrusters, fabricated with MEMS techniques.
The primary goal of our design imprecision research is to provide representational methodologies to aid engineering designers, particularly for choosing between alternative solutions during the preliminary design phase. Design choices of this nature are the most critical (and likely to be the most costly if in error). Conventional CAD systems are of little use in this highly imprecise environment. Our approach has been to aid design decisions with methods developed for efficient set-based computations on imprecise information. Among other attributes, this approach permits the designer to formally specify a design trade-off strategy to develop a multi-criteria objective function. Close technical interactions with the NASA Jet Propulsion Laboratory and Industrial research sponsors provides real-world motivation and examples.
Every stage in the design process includes a corresponding precision, or level of approximation. The level of imprecision in the description of design elements is typically high in the preliminary phase. As the design process proceeds, the imprecision with which each design parameter is known is reduced. Our approach has been to aid preliminary design decisions with analysis tools developed for computations on imprecise parameters. To provide these basic tools, we are applying the mathematics of fuzzy sets, and have developed efficient algorithms for performing design computations on imprecise design descriptions. Among other attributes, this approach permits the designer to formally specify a design trade-off strategy (or combination of strategies) to develop a multi-criteria objective function. The notion of strategies can be incorporated into other design methods, including matrix methods.
In addition to the research outlined above, a design contest is held annually as part of one of the undergraduate classes (ME 72). Each student is provided with an identical bag of parts and materials, and a detailed description of a task to perform. At the end of the term a highly publicized competition is held to determine the best design. The contest has attracted considerable industrial sponsorship, and is the focus of the undergraduate design activities.