Dynamic Real-Time Optimization with Product Quality Considerations and Closed-loop Prediction: Application to the Tennessee Eastman Benchmark

Articles in Press

Document Type : Research Article

Authors

1 Process Systems Engineering Laboratory, Department of Chemical Engineering, Amirkabir University of Technology (Tehran Polytechnic), PO Box 1591634311, Tehran, Iran.

2 Department of Chemical and Petroleum Engineering, Sharif University of Technology, PO Box 11365-8639, Tehran, Iran.

3 Process Engineering Department, Faculty of Chemical Engineering, Tarbiat Modares University, PO Box 14115-111, Tehran, Iran.

Abstract

Conventional real-time optimization (RTO) algorithms provide steady-state set points at which the process would operate economically. However, the process may undergo severe economic losses (e.g., off-specification production) when transitioning from the nominal steady state to the optimal one. In this paper, a dynamic RTO (DRTO) strategy accounting for the total cost of set point transitions is developed for the Tennessee Eastman process. The objective function of the DRTO is defined as the integral of the sum of operating costs and the costs of producing off-specification products. A closed-loop prediction model is employed to estimate future process outputs required to evaluate the DRTO objective function and constraints. The results reaffirm that the off-specification cost should not be ignored in the optimization as it constitutes up to 80% of the total cost. Moreover, the proposed DRTO is shown to strike a reasonable balance between the operating and the off-specification costs, yielding significant economic saving (up to 28%) over conventional RTO that considers the steady-state operating cost only. The saving is more pronounced when the process faces sustained disturbances. Interestingly, the DRTO finally reaches the same steady-state operating cost as the conventional RTO, ensuring no extra costs imposed on the long-term plant operation.

Keywords

Main Subjects

Scientia Iranica

Articles in Press, Accepted Manuscript
Available Online from 23 June 2026

Files

History

  • Receive Date: 01 July 2025
  • Revise Date: 02 August 2026
  • Accept Date: 20 May 2026

Share

How to cite

Statistics