(635f) A Multistage Stochastic Optimization Approach to Power Plant Scheduling with Flexible Carbon Capture

CO₂ emissions from fossil-fueled power plants can be effectively reduced by carbon capture using solvent-based absorption. However, the high energy intensity of solvent regeneration and CO₂ compression stages reduces the net power output of the plant by 25-40% [1]. The dynamic variation of electricity prices offers an opportunity to reduce energy consumption in CO₂ capture, thereby making the technology attractive for widespread deployment. The capture unit can be scheduled to operate flexibly, wherein more CO₂ capture occurs during low electricity price periods and less CO₂ is captured when the price is high. Majority of the previous literature in flexible carbon capture scheduling of power plants assumes perfect foreknowledge of electricity prices [2-4]. The competitive electricity markets exhibit high uncertainty owing to several market factors, which necessitates the incorporation of price uncertainty in optimal decision-making.

To this end, we develop a multi-stage stochastic programming algorithm based on reinforcement learning principles to determine an optimal hourly schedule of power production and carbon capture operations in uncertain electricity markets [5]. We consider a pulverized coal-fired power plant retrofitted with a carbon capture unit, which varies its load with dynamic price variation in the day-ahead market. A deterministic optimization formulation for maximizing profit with perfect foreknowledge of electricity prices is extended to a stochastic model to incorporate price uncertainty. Moreover, hourly electricity prices can assume a range of values, resulting in a large number of price scenarios. To reduce the computational complexity in the optimization framework, we develop low-complexity surrogate models for optimal action policy at each stage through data-driven modeling. The results represent the optimal hourly action policy as continuous functions of electricity price enabling power plants to take cost-effective decisions under uncertainty. These models are then used to determine total optimal profit for different real-time scenarios of electricity price. The mean profit obtained under uncertainty is within 25% of the benchmark, maximum profit with CO₂ emissions being sufficiently below the threshold limit.

References:

[1] E. S. Rubin, C. Chen, and A. B. Rao, “Cost and performance of fossil fuel power plants with CO₂ capture and storage,” Energy Policy, vol. 35, no. 9, pp. 4444–4454, 2007.

[2] Q. Chen, C. Kang, Q. Xia, and D. S. Kirschen, “Optimal flexible operation of a CO₂ capture power plant in a combined energy and carbon emission market,” IEEE Trans. Power Syst., vol. 27, no. 3, pp. 1602–1609, 2012.

[3] R. Khalilpour, “Multi-level investment planning and scheduling under electricity and carbon market dynamics: retrofit of a power plant with PCC (post-combustion carbon capture) processes,” Energy, vol. 64, pp. 172-186, 2014.

[4] S. M. Cohen, G. T. Rochelle, and M. E. Webber, “Optimizing post-combustion CO₂ capture in response to volatile electricity prices,” Int. J. Greenh. Gas Control, vol. 8, pp. 180–195, 2012.

[5] M. S. Zantye, A. Arora, and M. M. F. Hasan, “Operational Scheduling of Power Plants with Flexible Carbon Capture under Uncertain Electricity Price: A Multistage Stochastic Optimization Approach,” Submitted, 2019.