Dynamic modeling and economic model predictive control of a liquid desiccant air conditioning

As a good alternative to cooling-based air dehumidification, Liquid Desiccant Air Conditioning (LDAC) has drawn increasing attention for energy saving in large-scale public buildings. However, available investigations mostly pay attention to the dehumidification performance analysis and optimization based on steady-state models. Dynamic control issues for LDAC are rather important to the system performance and economic cost under the changing working conditions, such as the changing setting points of indoor air conditions and sensible/latent loads in occupied space. In this study, a dynamic model based on the Adaptive Neuro-Fuzzy Inference System (ANFIS) method is established to describe the system output air temperature and humidity ratio of LDAC dynamically with varied input conditions. The verification results show that a good agreement can be found between the prediction results from the proposed model and the experimental data with the relative errors less than 2% for outlet air temperature and less than 4% for outlet air humidity ratio, respectively. An EMPC strategy is designed based on the proposed dynamic model for LDAC by solving a receding optimization problem considering tracking error, control efforts rate, and energy consumption as the objective functions with Genetic Algorithm (GA) to get optimal dynamic response and improve the energy efficiency as well. The simulation analysis shows the proposed EMPC strategy outperforms Proportional–Integral (PI) control with smaller tracking error, faster response, and higher energy efficiency. The average energy saving by the proposed EMPC strategy can reach up to 9.5%.