There is a need for greater operational flexibility at thermal power stations in response to increasing renewable energy penetration on the electrical grid [1]. Many coal-fired power stations have responded by changing to load following operation instead of traditional baseload operation to aid in grid stability [2]. This has resulted in new operating regimes for all auxiliary equipment at the power station, including the cooling tower. Cooling towers represent a significant auxiliary energy expense for thermal power stations; however, with more attention given to optimizing the boiler [3], less work has been done to maximize cooling tower performance, particularly under these new operating conditions. Current operation is typically achieved through operating curves and PID controllers [4]. It has been shown that even simple logical rules can significantly increase the efficiency of multiple cooling towers that are in parallel [5]. Previous works have treated such cooling towers or sets of cooling tower cells as identical, but in practice cooling tower cells tend to have different efficiencies. This work demonstrates through simulation that a multi-cell cooling tower can be optimized in real-time through a novel closed-loop machine learning based particle swarm optimization. A neural network is trained using synthetic operating data, and the optimization is tested using a year of historic data. The results are compared to the same year of operations using the current control scheme. This work demonstrates that a data-driven machine learning based optimization algorithm can be built on top of existing cooling tower controls and yields as much as 6.7% energy savings compared to current practice, while still meeting the required outlet temperature. This has direct implications for increasing the overall efficiency of the power plant by decreasing auxiliary power usage. The results are shown to be more relevant as the plant more frequently experiences lower loads, which is more common with load following operation.
[1] Mararakanye, N., & Bekker, B. (2019). Renewable energy integration impacts within the context of generator type, penetration level and grid characteristics. Renewable and Sustainable Energy Reviews, 108, 441-451.
[2] Tuttle, J. F., & Powell, K. M. (2019). Analysis of a thermal generator’s participation in the Western Energy Imbalance Market and the resulting effects on overall performance and emissions. The Electricity Journal, 32(5), 38-46.
[3] Tuttle, J. F., Vesel, R., Alagarsamy, S., Blackburn, L. D., & Powell, K. (2019). Sustainable NOx emission reduction at a coal-fired power station through the use of online neural network modeling and particle swarm optimization. Control Engineering Practice, 93, 104167.
[4] Liao, J., Xie, X., Nemer, H., Claridge, D. E., & Culp, C. H. (2019). A simplified methodology to optimize the cooling tower approach temperature control schedule in a cooling system. Energy Conversion and Management, 199, 111950.
[5] Schleiâ€Peters, I., Wichmann, M. G., Matthes, I. G., Gundlach, F. W., & Spengler, T. S. (2018). Integrated material flow analysis and process modeling to increase energy and water efficiency of industrial cooling water systems. Journal of Industrial Ecology, 22(1), 41-54.
