Enhancing peak performance forecasting in steam power plants through innovative AI-driven exergy-energy analysis

This study aims to investigate and predict the performance of a 400 MW steam power plant operating on the Rankine cycle through a combined exergy-energy analysis and an artificial intelligence-based random forest regression model. The primary objective is to assess component-wise inefficiencies, identify key parameters influencing plant performance, and develop an optimized predictive model for performance evaluation. A mathematical formulation of energy and exergy balance equations is developed for each plant component and analyzed using the Engineering Equation Solver (EES). The study investigates temperature and pressure gradients, as well as mass flow rates, across all integral components. A parametric analysis is conducted to evaluate the impact of operational parameters on cycle efficiency, exergy destruction, and exergy losses. The results indicate that the boiler experiences significant temperature and pressure gradients, leading to higher irreversibility, whereas the gland steam condenser exhibits lower gradients, resulting in reduced exergy destruction. Among the plant components, the intermediate pressure turbine demonstrates the highest exergetic efficiency (90–93 %), while the condensate extraction pump has the lowest (20–26 %). Similarly, energy efficiency is highest in the intermediate pressure turbine (90–92 %) and lowest in the condensate extraction pump (18–22 %). The study further reveals that steam quality and reheat pressure at the low-pressure turbine outlet significantly influence overall power output and plant efficiency. The mass flow rates of steam through the high, intermediate, and low-pressure turbines follow a ratio of 110:124.3:143.6, with corresponding pressure ratios of 20:2.1:0.071. To enhance predictive accuracy, a random forest regression model is employed to forecast various performance indicators of the steam power plant. The model utilizes 100 decision trees with a maximum depth of 10, enabled bootstrapping, a fixed random seed of 42, and a minimum sample split of 2. The model's predictions for energy and exergy efficiencies are validated against experimental data, with root mean square error (RMSE) and coefficient of determination (R2) computed for accuracy evaluation. The study highlights that the random forest regression model can be utilized to predict and optimize the performance of steam power plants, thereby enhancing their efficiency and minimizing exergy losses.

Text 121252.pdf - Published Version Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (5MB) Official URL or Download Paper: https://linkinghub.elsevier.com/retrieve/pii/S2590...

Actions (login required)

View Item