(374a) A Semantic Based Decision Support Framework to Enable Model and Data Integration

Process modelling and simulation are employed as vital tools to design and evaluate biorefining technologies. With an increasing number of models and data built by various computer aided process engineering (CAPE) tools in the domain of biorefining, the researchers who stand to benefit from such an updated understanding are manifold. However, this large number of models and data distributed from CAPE tools results in heterogeneity and further requires expertise to establish interoperability.

The emergence of developing a common interface has led to the CAPE-OPEN initiative. CAPE-OPEN is a standard that defines rules and interfaces that allows CAPE tools and/or models to interoperate at a software engineering level. The CAPE-OPEN mainly focuses on the interoperability of flowsheet simulation. The interoperability is established between the process modelling components (PMCs) that are developed in various process modelling environments (PME) using a “middleware” The middleware is a connectivity mechanism that hosts communication across heterogeneous modelling environments by providing their functions, as well as input and output parameters (CO-LaN consortium, 2003). Despite the awareness and visibility of CAPE-OPEN within chemical engineering community, most end users without much software engineering expertise provided feedback that some found it difficult to use as it requires expert level support (Baur, 2018). To gain a fuller ability to increase awareness and to retain the most suitable model and/or data amongst existing models and data, the development of a systematic approach is required.

In this paper, a new approach for model integration however builds upon the CAPE-OPEN framework proposes the use of ontology and replaces the middleware with more flexible semantic repository ((Koo, Trokanas and Cecelja, 2017). Models are described by Semantic Web Services (SWS) using Ontology Web Language for Services (OWL-S) as an enabler of web services through service discovery, selection, composition, and execution stages (Figure 1). By The same token, datasets are also described by the OWL-S through their outputs and auxiliary precondition for their execution, i.e. extraction, modification and deletion (Figure 2). The process of model and data discovery using input/output matching has been supported by the structure of domain ontology. The matching process was employed as a method to perform the flexible discovery of models and data, which allows partial matching for inputs and outputs. Users are then given choices of ranked discovered models and data based on the requirements to assist them to make an inform decision. Once the relevant models and data sets are chosen, the demonstration of establishing interoperability is performed by composition and execution process.

This paper mainly focuses on the key parameters used in the formation of model integration through input/output matching based on semantic relevance between the models and data. The heterogeneous models and data in the domain of biorefining are semantically described in a domain ontology, OntoCAPEmodel ontology. The ontology acts as a framework that serves as a blueprint for knowledge dissemination and exchange by defining the basic terminology and interaction between models and data. The structure of unit is configured by a coupling through the different inlet and outlet ports where a unit can be connected to another unit, which is separated from the functional behaviour of the unit model. A set of data to be exchanged from one unit to another is distinguished by three different types: material, energy, and information streams. In addition, the stream properties are characterised by thermodynamics and physical properties, e.g. composition, temperature, pressure, flow, etc., with associated variables describing quantitative properties, e.g. physical dimension, value, unit, etc. The structure of the ontology further guides a process to record details about the models and data in an explicit format, and concomitantly supports registration and instantiation of the models and data.

The whole process is verified by a case of model integration for a supply chain modelling in biorefining.

References

1. Baur, R. (2018) CAPE-OPEN Survey CAPE-OPEN Survey - Objectives, CAPE-OPEN 2018 Annual Meeting, CO-LaN. Available at: http://www.colan.org/presentation/cape-open-survey-2/.
2. CO-LaN Consortium (2003) CAPE-OPEN Methods & Tools Integrated Guidelines, Cape-Open.
3. Koo, L., N. Trokanas, et al. (2017). "A semantic framework for enabling model integration for biorefining." Computers & Chemical Engineering 100: 219-231