(756d) Design and Implementation of MPC Strategies for Supercritical Pulverized Coal-Fired Power Plant Cycling with Carbon Capture

In recent years, with increasing energy demand and focus on greenhouse gas emissions, concerns on whether coal should be used as a primary energy source are heightened. According to a reference case study for power generation from the years 2000 to 2040, coal-fired power plants would still account for more than 30% in future power generation [1]. However, as renewable energy sources increasingly penetrate into the grid, coal-fired power plants which were designed to operate at base-load conditions would need to cycle their load. This cycling includes ramping the power generation rate up/down for load-following, as well as shut down/start up according to load demand. Previous literature studies have shown the impact of wind generation when integrated into gas and coal generation [2]. In particular, a sudden decrease in the wind power generation would lead to increased emissions and efficiency reduction from the coal-fired power generation due to cycling. To systematically achieve the optimal profiles for cycling the power plant, the development of advanced control algorithms is needed.

In this work, the design and implementation of linear and nonlinear model predictive control (MPC) strategies are performed for base-load and cycling scenarios. The application system is a supercritical pulverized coal-fired (SCPC) power plant with post-combustion carbon capture. The linear MPC control strategy is based on the dynamic matrix control (DMC) method. For this method, a multiple-input-multiple-output dynamic matrix is obtained by performing step response tests in the SCPC plant. The proposed nonlinear MPC strategy is an extension of NLMPC, which is based on the direct transcription method [3, 4]. For the NLMPC implementation, the nonlinear programming (NLP) solver IPOPT is employed, which is an efficient interior point-based large-scale nonlinear optimization algorithm [5]. Also, an open-source automatic differentiation package ADOL-C is used for improving the accuracy and calculation speed of system derivatives [6]. For the extended nonlinear MPC strategy, sequential quadratic programming (SQP) algorithms are analyzed for potential computational improvements when solving the posed NLP problem related to the large-scale power plant application.

In this presentation, a number of scenarios for the SCPC power plant will be discussed including: (i) trajectory tracking associated with power generation demands according to cycling starting from the original power generation setpoint; (ii) disturbance rejection for maintaining the carbon capture rate. In this case, the power generation should be kept at the original setpoint while potential disturbances affect the system, such as the carbon content in the coal is changed; and (iii) combination of both setpoint tracking and disturbance rejection cases. Preliminary results on the implementation of the DMC strategy for the carbon capture sub-system show that the controller can successfully keep the carbon capture rate at the specified 90% setpoint when the fluegas flowrate entering the system acts as step and ramp disturbances. Results on the closed-loop responses for different scenarios related to the SCPC plant will be compared and analyzed considering the advanced linear and nonlinear MPC strategies.

 

References:

1. Annual energy outlook 2015 with projection to 2040. Contract No.: DOE/EIA-0383(2015).
2. Bentek Energy LLC. Wind power, and unintended consequences in the Colorado energy market. Evergreen, CO: Bentek Energy, LLC, 2010.
3. F.V. Lima, R. Amrit, M. Tsapatsis, and P. Daoutidis. Nonlinear model predictive control of IGCC plants with membrane reactors for carbon capture. In Proceedings of the American Control Conference. Washington, DC, June 17-19 2013.
4. F.V. Lima, Xin He, R. Amrit and P. Daoutidis. Advanced control strategies for IGCC plants with membrane reactors for CO₂ capture. Process systems and materials for CO₂ capture: modeling, design, control and integration, A.I. Papadopoulos and P. Seferlis (eds.), Wiley, 2017.
5. A. Wächter, L.T. Biegler. On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming. Mathematical programming. 2006;106(1):25-57.
6. A. Wälther, A. Griewank. Getting started with ADOL-C. Combinatorial Scientific Computing. 2012:181-202.