Ontology engineering approach to support process model integration
Linsey Koo, Nikolaos Trokanas, Franjo Cecelja
PRISE Group, Chemical and Process System Engineering, Faculty of Engineering and Physical Sciences, University of Surrey, Guildford, Surrey, U. K.
Process modelling and simulation is a vital tool to plan, evaluate, assess, and develop different alternatives for the design of products and processes. The complexity of problems as well as heterogeneity of modelling methods make process modelling and simulation challenging, time consuming and often tedious process requiring a wide range of expertise. Inconsistencies in model development are the main cause for redundant work. Models remain implicit to the engineers who have built them, which further limits the potential of reusability.
The only model integration framework in use, the CAPE-OPEN, addresses the issue of standardisation of interfaces to enable interoperability between simulator software components from different sources. It is the framework built around a middleware, the Common Object Request Broker Architecture (CORBA) that hosts communication between unit operations defined for a specific function and the process modelling environments. The standard specification is defined as a property package which is needed for a thermodynamic or physical property calculation. The interoperability of models, such as model selection, parameter identification, and experimental work is enabled through the connection related to the unit operations and physical properties. It is not necessary to match all parameters in order to facilitate Input-Output (I-O) matching. However, the shortcoming of the CAPE-OPEN is in the need for identifying key variables for each unit operation.
A new approach for model integration however builds upon the CAPE-OPEN framework proposes the use of ontology and replaces the CORBA object bus with more flexible semantic repository (Koo and Cecelja, 2015). Models are described by Semantic Web Services (SWS) using Ontology Web Service Description (OWL-S) as an enabler of web services through service discovery, selection, composition, and execution stages, as shown in Figure 1. The discovery stage allows formation of an integrated model through matching requests from a public repository(ies). The best match that satisfy the requestor's functionality is selected in the model selection stage. The model composition stage then formulates the chain of integrated models and execution of integrated model takes place during the execution stage.
Figure 1. Semantically described model
This paper focuses on the matching parameters related to the domain of process system engineering, with emphasis placed on the role of physical properties and unit operation. We propose an extension of classification of variables defined by CAPE-OPEN with adaptation to ontology. To this end, each model representing a device (e.g. unit operations, flowsheets, subflowsheets etc.) is semantically described in domain ontology including domain assumptions and descriptions of the functionality of the model. The domain ontology guides the process of registering models and instantiation of ontology through ontology parsing. The domain ontology makes the model and data explicit and sharing terminology improves consistency. The devices in a process are connected via streams that transmit information through multiple inlets and outlets from one unit to the other. The connection between devices are described in ontology by introducing the concept of ‘ports' and ‘connections' (CO-LaN, 2014). The ‘ports' generally describe inlets and outlets of devices and three different types of streams are distinguished: material, energy, and information, which are further described by object properties. The ‘connections' are the object that is responsible for establishing a link between two ports, which contains information regarding methods, types, quantities, and units of streams.
The structural concept and classification of matching parameters defined by CAPE-OPEN have been employed in ontology. It intends to guide through the complexity of modelling methods and to facilitate I-O matching, while demonstrating the flexibility of model integration. To highlight the main feature of the ontology approach model integration, manageable scale models related to bio-refining process are used to verify the performance of the proposed approach.
CO-LaN, 2014, Conceptual Design Document for CAPE-Open Project, www.co-lan.org [Accessed on 20/04/2015]
Koo, L., Cecelja, F., 2015, Model Integration Using Ontology Input-Output Matching, 25th European Symposium on Computer Aided Process Engineering Proceeding Book, Copenhagen, Denmark, pp. 2567
See more of this Group/Topical: Computing and Systems Technology Division