(197e) Market Responsive Control: A Second Order Approach to Economic Based Controller Design

Authors: 
Chmielewski, D. J. - Presenter, Illinois Institute of Technology
Mendoza-Serrano, D. - Presenter, Illinois Institute of Technology
Yang, M. W. - Presenter, Illinois Institute of Technology


The economics of plant operations are greatly influenced by the cost of utilities. While the price of fuel is typically a slowly varying parameter, the cost of electricity tends to change quickly. While classic control schemes have no ability to respond to utility costs, advance controllers typically are capable, but only through the Real-Time Optimization (RTO) algorithm, which takes a steady-state perspective on the cost of utilities. Thus, in the case of electricity price variations, which have a time scale on the order of most chemical processes, an RTO approach will be insufficient to capture the fast changes associated with electricity markets. The notion of Market Responsive Control aims to exploit measurements and the fairly regular cycle of electricity price to achieve a maximum in average profit (by defraying energy intensive operations until prices are low).

The effort focuses on the design of a high level (supervisory) control system that is capable of market responsiveness. That is, design a controller that uses the electricity spot price as a disturbance input to the decision making process. However, in contrast to traditional controller designs, where the objective is to attenuate disturbances, the Market Responsive Controller will judiciously amplify the spot price disturbance. We have found that a Linear Quadratic Gaussian controller can be forced to have such behavior if the cross-term weights of the objective function are utilized. However, it is a non-trivial task to select weights such that profit is maximized, while observing the equipment limitations. Thus, the core result of the paper is the development of a profit maximizing scheme for the selection of LQG weights subject to process equipment limitations. The resulting optimization problem will be shown to be convex and thus will yield a globally optimal solution. The methodology will be illustrated by application to an Integrated Gasification Combined Cycle (IGCC) as well as a building HVAC system. If time permits, a globally optimal scheme for the sizing and selection of energy storage units will also be presented.

Topics: