Linear Combination of Gradients as Optimal Controlled Variables

Abstract

In this paper, we show that optimal economic operation can be achieved using feedback control, by controlling the right variables that translate economic objectives into control objectives. We formulate a generic framework for selecting the controlled variables based on the Karsh-Kuhn-Tucker (KKT) conditions, that can be used to select the optimal controlled variables for different operating conditions. The proposed generalized framework is given as a linear combination of cost gradients. Furthermore, we also show that, the proposed linear gradient combination framework can be used to select the economically optimal controlled variables for parallel operating units. The proposed linear gradient combination framework can be used with any gradient estimation scheme. A benchmark Williams-Otto reactor example is used to demonstrate the effectiveness of the proposed CV selection framework.