Progress in Civil and Structural Engineering Computing
Edited by: B.H.V. Topping

Chapter 3

Structuring the Design Process from Phenomenological Models to FEA via Knowledge Engineering

S. Chandra and J. Dhayanidhi
Advanced Research Group, Structures Division, National Aerospace Laboratories, Bangalore, India

Keywords: knowledge based engineering, phenomenological models, structural design, FEA, tacit knowledge, PDM, browsers, OOPS.

There have been number of attempts to map the design process associated with the structural design of mechanical and aerospace components to enable design automation. It has been argued that this would reduce the time to market costs, provide repository knowledge to certification authorities and designers through the entire product life cycle and more importantly be useable in the synthesis of new evolutionary designs. During the eighties, the emphasis was on representation of `tacit' knowledge using knowledge representation methodologies developed as part of artificial intelligence research. In the nineties, the issues that occupied the industry were related to data management using COTS (commercial of the shelf) databases and on developing a product data model. However, the integration of structural design into the entire design process is still not seamless. Attempts have been made to develop STEP models that provides an interface between CAD and FEA as well [1]. It is clear that there are islands of research activity in early conceptual modelling, FEA modelling and product modelling. We find that the design process can be structured into preliminary models (PM) which are driven in large part by phenomenological models and detailed FEA with possibilities of interoperability between both.

Many design methodologies developed before the extensive use of FEA were driven by a number of simplifying assumptions. For example, design of a lifting surface component like the airplane wing would have been by the use of shear flow theories, loads carried through spars and buckling of skin panels [2]. However, the extensive use of FEA has enabled considerable insight into local behaviours as well as overall load paths. Interestingly, examination of earlier design methodologies shows that they were driven by descriptive phenomenological models. A mathematical description was then provided to these models. Based on these simple mathematical models, design sizing was established. On examination of the design process, it was found while expert knowledge was important, the process for most part depended on structuring of data from the initial design till detailed design. We have found that PMs are cognitive models that can be regarded as object oriented, decomposable and runnable [3,4]. Using IDEF0 diagrams, we map the design process from preliminary modelling to FEA and discuss the use of STEP formats like AP203 and AP209.

The process that appears to be established is by developing initial sizing via these simple approaches and consequently validating it using a detailed finite element analysis. The integration of these early models to a detailed FEA model is required to enable consistency in understanding physical behaviour. In this paper, while the early mental models of design are examined briefly, the emphasis is on structuring the preliminary design, CAD and FEA processes with the use of a product model based on a feature based approach that uses the simple approaches to establish sizing.

Over the past few years, after the introduction of federated browser technologies, development of a product data management structure and more importantly acceptance of the difficulties with the use of expert systems in their present form as the panacea for automating design, there is recognition that there is clearly a need to use convergence technologies as a strategy to design an architecture for knowledge based design automation. In this paper, we describe a framework by illustrating the design of a vertical tail of an aircraft that uses the present web and database technologies for the routine design of a typical aerospace component that takes into account preliminary design procedures, product modelling and FEA in the design process. We argue that the use of feature based modelling from the preliminary stage provides interoperability and describe an architecture by illustrating the design of a vertical tail of an aircraft that uses the present web and database technologies for the routine design of a typical aerospace component integrating preliminary models and FEA. Such architecture, if replicated for other major aerospace components, will provide the framework for creating repositories of design knowledge and data over the entire product life cycle.

References

1
Hunten, M., ``CAD/FEA Integration with STEP AP209 Technology and implementation", MSC Aerospace Users Conference, September 1997.

2
Bruhn, E.F., ``Analysis and design of flight vehicle structures", Tri-state offset company, Cincinnati, U.S.A., 1965.

3
Williams, M.D., Holans, J.D., and Stevens, A.L., ``Human reasoning about physical systems", in Gentner, D., and Stevens, A.L., (Eds), Mental models, Lawrence Erlbaum Associates, NJ, 1983.

4
Chandra, S., and Blockley, D.I, ``Cognitive and computer models of physical systems, Int J. of Human Computer studies, 43, 539-559, 1985.

return to the contents page