Application of data reconciliation for steel pickling process

Abstract

Process measurements are taken in chemical plants for the purpose of evaluating process control or process performance. However, not all variables needed are generally measured due to technical infeasibility or cost. Furthermore, the measurements are often contaminated in the sense that random noise may be present due to result of miscalibration or failure of the measuring instruments. Thus, data reconciliation is frequently required before the data can be used for evaluating process control or process performance. In this research, the benefits of combining linear dynamic data reconciliation (DDR) with Generic Model Control (GMC) are demonstrated on two example problems, a steel pickling process and an exothermic batch reactor. The robustness of the proposed control strategy is investigated with respect to changes in process condition, modeling error and disturbance variable for both set point regulation and set point tracking. The results show thata rudimentary treatment of measurement errors and an estimation of unknown quantities have proved extremely effective. Therefore, DDR approach is an important adjunct to advanced control.