An environment for topology analysis and data reconciliation of the pre-heat train in an industrial refinery. (25th January 2019)

• Record Type: Journal Article
• Authors: Chebeir, J.; Webb, Z.T.; Romagnoli, J.A.
• Is Part Of: Applied thermal engineering. Volume 147(2019)
• Page Start: 623
• Page End: 635
• Publication Date: 2019-01-25
• Subjects: Topology analysis -- Data reconciliation -- Gross error detection -- Industrial heat exchanger network; Heat engineering -- Periodicals; Heating -- Equipment and supplies -- Periodicals; Periodicals; 621.40205
• DOI: 10.1016/j.applthermaleng.2018.10.081
• Languages: English
• ISSNs: 1359-4311

Abstract

Highlights: A framework is proposed for data reconciliation of an industrial system. The systematic error present in different sensors is isolated and eliminated. A performance test for each heat exchanger is developed based on the analysis. The performances of heat exchangers are monitored for different time periods. Abstract: Monitoring the operating performance of the pre-heat train in an industrial refinery is of primary concern due to its effect on the production costs as well as on emissions and safety issues. The estimation of the true state conditions of a preheat train from the raw measurements is essential to achieve optimal process monitoring, control, and optimization. An integrated approach for data reconciliation involves a set of procedures applied on the process measurements. The case of a pre-heat train typically leads to the resolution of a large-scale problem and the manipulation of countless process data. In this paper, an environment is proposed to expedite the creation of the matrix systems and automatically execute the data reconciliation (DR) procedures. The implementation of the environment for the plant topology analysis, variable classification, bias detection/estimation, data reconciliation, and prediction of unmeasured process variables is discussed in terms of an application to a pre-heat train in an industrial refinery. The data used consists of a snapshot of the actual operating data and, in this specific application, an average during three hours of operation.