Knowledge Representation in Process Engineering

In process engineering, as in many other application domains, the domain specific knowledge is far too complex to be described entirely using description logics. Hence this knowledge is often stored using an object-oriented system, which, because of its high expressiveness, provides only weak inference services. In particular, the process engineers at RWTH Aachen have developed a frame-like language for describing process models. In this paper, we investigate how the powerful inference services provided by a DL system can support the users of this frame-based system. In addition, we consider extensions of description languages that are necessary to represent the relevant process engineering knowledge. The application domain Process engineering is concerned with the design and operation of chemical processes that take place in large chemical plants. This engineering task includes activities like deciding on an appropriate flowsheet structure (e.g. configuration of reaction and separation systems), mathematical modeling and simulation of the process behavior (e.g. stating mathematical equations and performing numerical simulations), sizing of components (like reactors, heat exchangers etc.) well as budgeting and engineering economics. These highly complex tasks can be supported by building computer models of the chemical plants and processes, using appropriate software tools such CAD, decision support systems and numerical tools. Rather than designing each new model from scratch, one wants a system that offers standard building blocks that can easily be put together. Standard building blocks [Marquardt, 1994; Bogusch&Marquardt, 1995] are objects representing ̄ material entities such as reactors, pipes, control and cooling units, ̄ models of these devices such as device-, environment-, and connection-models, ̄ interfaces between these models and socalled implementations describing their behaviour, ̄ symbolic equations specifying these implementations and variables occuring in these equations, which are related to each other as specified in the interfaces. Since there is a great variety of different building blocks, they must be stored in an appropriate database. Since process engineering is a quickly evolving field, the number of standard building blocks increases constantly. Hence it must be possible to define new building blocks in a comfortable way. The process engineers at the RWTH Aachen we are cooperating with have developed a framelike language for describing these standard building blocks, [Bogusch& Marquardt, 1995; Marquardt, 1994]. This language allows to group building blocks into classes, and to order the classes in an is-a/specialization-of hierarchy. It should be noted that this hierarchy is explicitly given by the person defining the classes (the knowledge engineer), and not automatically inferred from the definition of the class. As the complexity of the database increases, navigation in the class hierarchy becomes difficult, and modifying or extending the hierarchy becomes dangerous. More precisely, the knowledge engineer is faced with the following prob74 From: AAAI Technical Report WS-96-05. Compilation copyright © 1996, AAAI (www.aaai.org). All rights reserved.