Performance evaluation of a high-influx, bubble dehumidifier

Abstract

In hot humid climates, a major portion of air conditioning energy is used for dehumidification. In traditional centralized air conditioning systems, air is cooled below the dew point to remove the moisture. This process is energy intensive. Liquid desiccant dehumidification systems (LDDS) provide a promising alternative as they use low grade or free energy like solar energy. Majority of the existing direct-contact liquid desiccant dehumidification modules adopt the concept of low influx rates which ensures high residence time resulting in improved dehumidifier performance. There is limited knowledge with respect to high influx rates which is the topic of this paper. The overall objective of this study is to evaluate the performance of a high-influx, bubble dehumidifier. A prototype dehumidifier employing aqueous Lithium Chloride (LiCl) as desiccant that could accommodate high-influx rates was developed and experiments were conducted to evaluate its performance. Results of the experiments are analyzed in terms of performance parameters namely moisture removal rate (MRR) and dehumidifier effectiveness. The apparatus could achieve a maximum MRR of 0.28 g/s resulting in 66% dehumidifier effectiveness. Empirical equations to predict the MRR were developed through linear regression using experimental data. The input variables used in regression are the inlet depth, the inlet air vapor pressure, the desiccant solution vapor pressure and the diffusion coefficient of water in LiCl solution, along with the design parameters. The diffusion coefficient parameter which was ignored in other dehumidifier models was found to be imperative in this research. The developed empirical equations were validated by comparing the predicted and measured values on a new set of experiments, the data for which were not used for regression. It is found that the time-dependent nonlinear response of the system is predicted accurately by the empirical model. The root mean square errors of the observed and predicted values are less than 8%.