Design Model Sharing in Concurrent Engineering J. S. M. Vergeest Subfaculty of Industrial Design Engineering, Delft University of Technology, Jaffalaan 9, NL-2628 BX Delft, The Netherlands j.s.m.vergeest@io.tudelft.nl I. Horváth Subfaculty of Industrial Design Engineering, Delft University of Technology, Jaffalaan 9, NL-2628 BX Delft, The Netherlands i.horvath@io.tudelft.nl Abstract The transfer of design model data between two participants of an engineering team is a fundamental subprocess in concurrent engineering (CE). The problems encountered in data transfer are abundant. It is understood that data transfer should be enhanced toward knowledge transfer. Although it is possible to define what knowledge transfer is in general terms, it is crucial to specify what knowledge transfer is at the subprocess level. This comes down to determination of the conditions for the data transfer subprocesses. In this paper we investigate which conditions are specific for CE. They are then formulated using the finite automaton formalism. From these conditions it can be derived which measures should be taken in order to avoid data exchange problems and hence to achieve knowledge sharing. Examples are provided and an outlook to further applications is discussed. 1 Introduction The most fundamental subprocess in CE is the communication of design models among members of the team. Successful CE depends on proper management of the design project, at very different levels. Yet, already on the level of data transfer between agents, there are significant, long-standing problems, disabling effective CE. Progress in conquering this problem is known to be slow. In the domain of geometry, the data exchange process is fairly well understood and a procedure of writing and reading model data based on an external (i.e., system non-specific) representations is available through the International Organisation for Standardisation [1]. The majority of CAD data exchange in industry, however, still proceeds via the IGES standard and the DXF format and, although perhaps technically less advanced, their use is expected to remain widespread in, at least the near-term, future [2]. The fundamental issue of CAD model sharing is known to be related with the mismatch between the required semantics of the object models and the semantics they actually posses [3,4]. The reason for this mismatch is the lacking description of the semantics of the CAD models.. This situation is not surprising, since the process of defining CAD model semantics is poorly understood and seems to be a huge task. On the other hand, if the semantics is not considered then the CAD communication problem cannot be properly solved. Ultimately, successful data transfer would require 1) a full specification of the semantics for each of the CAD systems involved and 2) a formal unification of the obtained collection of semantics [5]. For several reasons this would be a difficult goal to achieve. It is therefore relevant to question whether this unification is in all situations necessary, and what can be achieved without it. In this paper we explore which data transfer can be successful without a full unification of the semantics of the CAD systems involved. In section 2 we will provide, for each individual member of the collaborative design team, a description of the CAD system which is to his or her disposal together with the semantics belonging to that CAD system. In section 3 we define a criterion for the successful data transfer between any two team members. This criterion is evaluated in section 4. Conclusions will be drawn in section 5. 2 Engineering Information, its Representation and its Semantics When the design model needs to be communicated to (or to be shared with) a different party then the main interest