Why is Design Automation Software not Everywhere?
Editor: Norell Bergendahl, M.; Grimheden, M.; Leifer, L.; Skogstad, P.; Lindemann, U.
Author: Schotborgh, Wouter; Kokkeler, Frans; Tragter, Hans; van Houten, Fred
Section: Design Information and Knowledge
This paper discusses the question how to increase the amount of design automation software in industry. First, the contemporary industrial context is outlined to motivate the need for automation of parametric routine design problems. Second, the knowledge engineering (KE) activity is identified as a bottleneck that prevents large scale deployment of design automation software. Knowledge engineering is seen as the process to construct a mathematical model of a design problem. Literature and our own research results are used to illustrate that the KE activity for new problem automation is often unpredictable in terms of the activities to execute and time it will take. A correlation seems to be missing between the problem size and the required knowledge engineering effort. Possibly, this is a result from the diversity of design problems and the tacit nature of expert knowledge. A route toward more prescribed KE is suggested by tightly integrating theories from cognitive design, knowledge engineering and constraint satisfaction. The goal is to closely relate knowledge acquisition (interviewing of experts) with the knowledge modeling activity and automation algorithm.