Collaborative desktop engineering

Abstract

This paper discusses conclusions from development, testing and comparison of three AI based applications that support multidisciplinary analyses of complex engineering problems. Our approach to automating multidisciplinary analyses involves integrating independent software tools (services) into unified systems that reason about shared computer interpretable models. Early desktop engineering systems integrate a fixed set of services and are limited in their flexibility to modify the types of analysis included in the integrated system. Next generation systems face the challenge of providing users flexibility to choose services freely based upon their analysis needs while providing a structured framework for controlling interactions and data sharing among the chosen services.This paper starts with a case example illustrating pre-project planning (PPP) challenges that today's practitioners confront. We then describe our approach to integration and automation as exemplified by the Facility Alternative Creation Tool (FACT), a system that utilizes structure, behavior, and function representation to implement integration and automation of PPP.We discuss how service reasoning exploits the function, behavior, and structure representation to process project information from multiple perspectives in a general way. We describe how circle reasoning enables dynamic addition and deletion of services and automated sequencing of changing sets of services. These reasoning approaches together with a PPP ontology enable greater flexibility for integrating and automating PPP. We compare FACT to two other applications, focusing on five issues: purpose, representation, reasoning, interface, and testing. From experimenting with these systems, we conclude that, like desktop publishing, desktop engineering enables significant change in engineering practice.