Integrating Analytical Simulations With Regression Learning: Advancing Efficiency in Energy and Water Use in Sugar Production

Abstract

Energy and water consumption are critically important in the sugar industry. In this context, the heat exchanger network of a target sugar factory has been modeled and optimized, as this sector is the primary consumer of energy and water. A key innovation of this work lies in the coupling of interacting components within the model, leading to a more comprehensive framework compared to previous models in the literature. Some sections of the system are modeled using analytical interpretations, while others are developed through a regression learning process utilizing statistical data. This integration of analytical formulation and data-driven modeling represents another significant advancement in this research. The resulting model demonstrates acceptable accuracy for most measurable parameters, with an average deviation of approximately 4%. The optimization results indicate that certain parameters, such as the cooling pool evaporation rate, exhibit considerable flexibility, allowing optimization algorithms to converge more easily. Conversely, other parameters, such as the vapor fed to the exchangers, are more rigid, which restricts the freedom of the optimization process. Moreover, the effectiveness of the elements within the optimization target function is crucial for identifying the optimal point. Overall, minimizing energy consumption and water usage simultaneously presents a significant challenge, necessitating careful consideration in determining which optimal point is most practical.