Framework for Integrated Model-Centric Process Support

• Published in: Industrial & engineering chemistry research Vol. 50; no. 18; pp. 10533 - 10548
• Main Authors: Aragon, D, Rolandi, P. A, Romagnoli, J. A
• Format: Journal Article
• Language: English
• Published: Washington, DC American Chemical Society 21.09.2011
• Subjects: Algorithms; Applied sciences; Chemical engineering; Computer programs; Computer simulation; Continuously stirred tank reactors; Dynamical systems; Dynamics; Exact sciences and technology; Mathematical models; Process Design and Control; Visual programming languages; Modeling
• ISSN: 0888-5885; 1520-5045
• DOI: 10.1021/ie1021442

In spite of the abundance of development efforts in the area of modeling and simulation of chemical processes, current process modeling environments, PMEs, continue to have two important drawbacks. First, they have been designed for modeling experts; and second, they have not been designed to formulate real process engineering problems involving raw plant measurements and their inherent high volume of data. In response to these shortcomings, the theoretical framework for support of process systems for realistic formulation of engineering model-based problems (Rolandi, P. A.; Romagnoli, J. A. Integrated model-centric framework for support of manufacturing operations Part I: The framework. Comput. Chem. Eng. 2010, 34, 17–35) was reformulated. This innovative formulation allows for the incorporation of new mechanisms and novel algorithmic approaches for outlier detection and dynamic data reconciliation, allowing for the treatment of real world plant data. A practical system was devised in Visual Basic and gPROMS and tested by means of two case studies, namely, two continuous stirred-tank reactor (CSTR) in series and a distillation column, demonstrating the great flexibility of the developed system. The performance of the framework and the practical software application were evaluated with simulated data as well as real data, demonstrating the effectiveness of the proposed methodologies for outlier detection and dynamic data reconciliation, which displayed high efficiencies and superiority to their traditional counterparts, offline as well as online.