(21c) Development of an Advanced Control Infrastructure for Subcritical Power Plant Online Power Demand Tracking

Many subcritical coal-fired power plants that usually operate in baseload conditions are migrating to load-following operations as more renewables are penetrating into the power grid. As renewables, such as solar and wind, deliver intermittent power, dispatchable energy systems such as coal-fired power plants are required to cycle their load to accommodate power demand fluctuations. However, most of the subcritical plants were not designed to operate under load-following conditions, resulting in additional emissions and increased power plant damage. Given that cycling may become the new operating mode for coal-fired power plants, there is a critical need to properly control these plants. In particular, to operate in different power dispatching modes, a power plant will need to produce steam at certain flows, temperatures and pressures as required by the turbine. The highly variable resulting plant dynamics will require predictive control algorithms capable of maintaining the process within the operating regions of interest.

In this work, an advanced control structure is developed for online control of the total power demand of a subcritical coal-fired power plant cycling around the setpoint of 200 MW. This structure involves the combination of proportional-integral-derivative (PID) and advanced model predictive control (MPC) algorithms [1] for real-time control of power demand. Specifically, temperatures of the primary and secondary desuperheaters are controlled by a 3 input x 3 output MPC that manipulates 2 spray valves and coal flowrate as inputs. Also, the PIDs (2) control the pressure and water level of the drum boiler by manipulating the water valve and steam-to-turbine valve, respectively.

A communication infrastructure encompassing the power plant model [2], real power plant data obtained from online databases (e.g., ERCOT [3]), and the control algorithms is established using the OSIsoft PI system at West Virginia University (WVU). The PI system centralizes the information received from the power plant model in MATLAB Simulink and the online power demand from the web, and send the control actions calculated by the MPC in MATLAB back to the power plant for implementation. To evaluate the proposed infrastructure, several days of power plant online data with cycling profiles are considered. The developed infrastructure enables power plant control using real-time data to improve decisions and operations of existing and future fossil-fuel-based power plants.

[1] He, X.; Lima, F. V. A modified SQP-based model predictive control algorithm: application to supercritical coal-fired power plant cycling. 2020. Industrial & Engineering Chemistry Research, 59(35), 15671-15681.

[2] Agbleze, S.; Lima, F. V.; Indrawan, N.; Panday, R.; Pezzini, P.; Bonilla-Alvarado, H.; Bryden, K. M.; Tucker, D.; Shadle, L. J. Modeling and control of subcritical coal-fired power plant components for fault detection. In Proceedings of the 2020 ASME ICONE 28-POWER, August 2020.

[3] Electric reliability Council of Texas (ERCOT). (n.d.). Retrieved April 10, 2021, from http://www.ercot.com/