Many different terms have been introduced in the past few years to describe this emerging field: Computational Prototyping, Multidisciplinary Design and Optimization, Simulation-Based Design. Many more interpretations have been given to what these terms mean, but the basic idea involves a combination of simulation, modeling, and design tools that are required for the design of complex systems.
An example of computational prototyping, at one end of the spectrum, is texture mapping a label around a new bottle shape to visualize a new product. In the middle of this spectrum, we might walk through a 777 opening doors to make sure there is no interference. At the other end of the spectrum we must integrate results from finite element and CFD with manufacturability considerations in the early phases of a new design. The latter situation is where computational prototyping has the potential for making the biggest difference in engineering design. It is also farthest from being realized.
An excellent example of the uses of computational prototyping is the Boeing 777. It has been noted that the 777 was designed, test flown, and repaired before a single component was manufactured. But this example also illustrates how far we are from fully exploiting computation-based design. By the time the 777 team put together the first virtual prototype of the airplane, the decisions had been made that would lock in more than 70% of the life-cycle cost of that product. The wing planform shape for this airplane was designed primarily by the high speed aerodynamics group with only heuristic considerations given to low-speed performance or structures, and essentially no input from the stability and control group. One of the goals of simulation-based design is to be able to incorporate multidisciplinary and cross-functional requirements and objectives in the early stages of the design process where tools such as computational prototypes and optimization can make the biggest difference.
The field is still in its infancy, although the emerging methods are being applied in the design of new aircraft, ships, electronic systems, automotive systems, and spacecraft. Industry has adopted some of the better-developed tools of this trade including some new developments in CAD and visualization, specific disciplinary analysis and optimization methods, and some related new ideas for system planning and management. However, many of the required strategies for dealing with such complex systems have yet to be developed. One cannot simply link a CFD simulation to a CAD program and create a system for wing design. Collaboration and communication tools are still evolving rapidly; strategies for mediating conflicting goals of many design teams are still a topic of research; and computational design methods capable of dealing with conceptual-level design decisions are still being formulated.
Advances in "CAD", and disciplinary simulations, such as computational fluid dynamics and computational mechanics, have outpaced the technology for exploiting their results - design remains very ad hoc. While the details of the wing loft of a modern airplane design may be refined using parallel supercomputers, optimization, and sophisticated simulations, the "important" design parameters (e.g. overall size, sweep, thickness) were selected using extremely simple analyses and heuristics. This is not because these parameters should be selected this way, but rather because we just do not know how to integrate multidisciplinary simulations and preliminary design decisions (except in the simplest cases) in much more than an ad-hoc fashion. It is easier to integrate the disciplines when all that is involved is a scheduling or fitting task, much harder when a disciplinary simulation involves sophisticated numerical simulation such as aerodynamics or 3D E&M.
Integration of computational simulations and multidisciplinary design is the key to more rapid development and more competitive products in aerospace. If one cannot efficiently incorporate more refined simulations in the early design process, one must rely exclusively on statistical methods and previous experience. This immediately increases the risk associated with new concepts (because we have no experience with them and cannot analyze them) and we find ourselves making only evolutionary changes to an existing design, rejecting potentially better ideas because they are just too risky.
Aero/Astro graduates need some background in this emerging field, both through course work and in research. Recent graduates with a focus in these areas (along with disciplinary strengths) are already playing important roles in NASA and industry programs such as the HSCT (a next generation SST), and interplanetary spacecraft.
This arena has the potential for unifying the work of many researchers in the SOE. In our department alone, important work in nonlinear control synthesis, smart structures, and aerodynamic design is accomplished with too little attention paid to the integration of these technologies. The level of fidelity in these modeling efforts enables exciting new possibilities for aircraft design.
Three examples:
Our department has received considerable recognition for research that involves the use of small-scale flight test systems (e.g. GPS aircraft and helicopters, oblique wing flight tests, satellite systems). We expect that this work will continue and would be greatly aided by advances in simulation-based design methods and computational prototyping.
Aircraft design has led the field of computation-based design because it relies so heavily on complex simulation, but the importance of SBD, MDO is being increasingly recognized in electronics, automotive, marine, and mechanical design. Advances in this field will have immediate impact in these domains. It is critical to exploitation of new research areas in manufacturing, MEMs, system engineering. We are particularly excited about the opportunity to combine work these areas with our simulations in a realistic design environment which includes considerations of development time, manufacturing and the discrete nature of many design decisions.
Listed below are some examples of research areas based on categories identified at a recent NSF workshop.
A. Collaborative Design tools and techniques Tools/ Infrastructure / methods for distributed and collaborative design Seamless representation of information from concept to detail Design/management of teams B. Prescriptive models, design methods & normative theories Design methodology /tools/environment development Rapid concept generation Method for choosing analysis tools with respect to cost and needs Methods for modeling and optimizing design process in different domains Dealing with complexity and complex phenomenon Automatic methodology for innovation C. System integration and infrastructure tools Design for complete life cycle Integrated process design through robust design simulation Integrated design and manufacturing Integration of modeling, analysis & simulation. Overcoming the curse of dimensionality: complexity & size management Virtual manufacturing Integrating design knowledge Methodological foundations of concurrent engineering D. Design automation systems/ tools CAD packages to handle levels of uncertainty Function based design Smart interfaces Multi-disciplinary sensitivity tools How to integrate new technologies Product layout design tools Perceived risk vs. real risk (tools) Tools for conflict resolution. Knowledge acquisition and integration tools Information capture and representation tools Innovative/ creative design support tools New generation of CAD design tools for configuration design Automation design detailer KBS for creative design Methods for capture of design rationale Requirements tracking and design change management E. Analysis, Simulation, Optimization Tools Engineering design based on multidisciplinary design optimization (MDO) Satisficing and optimal solution Mixed continuous and discrete variables Use of statistical analysis (better models) Qualitative evaluation methods Distributed optimization Evolutionary Programming, genetic algorithms, simulated annealing F. Formal Models of Design Process/ Design theories Normative theory of design behavior Unified theory of design Unified design process model Identification of common core design process Formalize models (products/process/facility/environment). Models of conceptual design G. Design Information Access and Support Systems Designers notebook; Designer specific portable knowledge bases Design information standard to share information. Corporate memory for design Methods and tools for cross functional design information exchange Accessibility/dissemination Intelligent Design document Methods for capturing company specific information Information flow modeling