Abstract Air-conditioning systems condense air water vapor to achieve dehumidification. This process entails excessive cooling followed by reheating, resulting in high energy usage. As a more energy-efficient alternative, liquid desiccant-based dehumidification eliminates the need for excess cooling by dehumidifying through an absorption process. Advanced internally cooled liquid-desiccant dehumidifiers enhance overall air conditioning efficiency by incorporating internal cooling mechanisms. These devices are three-stream heat and mass exchangers, involving air, liquid desiccant, and cooling water. Simulating them involves discretization and solving comprehensive partial differential mass and energy balance equations to capture the interconnected heat and mass transfer phenomena. The models require detailed dehumidifier information, are computationally expensive, and pose challenges in convergence. These requirements make them unsuitable for integration into Building Energy Simulation software or inclusion within a comprehensive air-conditioning system model. This study aims to generate models with less computational demands during simulation using artificial neural networks. A comprehensive finite difference model was used to generate training data for developing the neural network. Various network configurations were explored to assess their impact on prediction precision. The trained network, integrated into an independent code, underwent evaluation for prediction accuracy using a distinct dataset from the training stage, proving accurate to simulate the internally cooled dehumidifier, reaching R-values up to 0.96 for the 5 predicted outlet variables. This is the initial step towards incorporating neural network models into specialized Building Energy Simulation software, as the foundation for conducting system-level, transient, and long-term simulations.
Read full abstract