Retrofit self-optimizing control of Tennessee Eastman process

Abstract

This paper considers near-optimal operation of the Tennessee Eastman (TE) process by using a retrofit self-optimizing control (SOC) approach. Motivated by the factor that most chemical plants in operation have already been equipped with a workable control system for regulatory control, we propose to improve the economic performance by controlling some self-optimizing controlled variables (CVs). Different from traditional SOC methods, the proposed retrofit SOC approach improves economic optimality of operation through newly added cascaded SOC loops, where carefully selected SOC CVs are maintained at constant by adjusting set-points of the existing regulatory control loops. To demonstrate the effectiveness of the retrofit SOC proposed, we adopted measurement combinations as the CVs for the TE process, so that the economic cost is further reduced comparing to existing studies where single measurements are controlled. The optimality of the designed control architecture is validated through both steady state analysis and dynamic simulations.