Research On Industrial Application Of Data Rectification For Process Industries

In the past four decades, data rectification has become a significant accomplishment of the process systems engineering (PSE), both in academic research and engineering applications. With broadening the scope of PSE, the PSE area is now closely tied to mathematics, operations research and computer science. Data rectification has been widely used in petrochemical, steel and other process industries to deal with multi-scale data, as one important part of computer integrated manufacturing system. But the performances of data rectification systems consequently is now facing various challenges arising from incomplete measurement conditions, inaccurate modeling, invalid information systems and so on.With petrochemical industry as the research background, this thesis focuses on the issues of multi-scale data rectification in the process industry and carries out the following work.By analyzing the demand of simulation research of multi-scale data rectification applications, we propose the design of a multi-scale modeling and simulation system. The functions of the system are defined and implementation details are given. The system is designed as a flexible platform for testing and studying of plant-wide multi-scale data rectification applications.Based on the multi-scale material balance demands of petrochemical companies, a rectification method based on a multi-scale material model is introduced for multi-scale material data. This method combines data from other scales in the rectification model, contributing to redundancies of measurements and the consistence of data used for different tasks. Material balance based on rectified data is carried out from high levels to lower levels based on the assumption that there are some measurements which are more accurate on higher levels in plants.Parameters of refinery production planning optimization models are closely related to plant measurements. A method for integrating refinery production planning and plant data rectification is introduced. Process measurements are often corrupted by gross errors and distort planning optimization solutions. To overcome the problem and improve plant efficiency, data rectification is adopted to rectify and reconcile history process data. Then planning model is updated and the planning problem is solved integrated with reconciled values. Sensitivity analysis is carried out for assessing the effectiveness of data rectification's contribution.