Investigation of thermal stratification in cisterns using analytical and Artificial Neural Networks methods

Abstract

The thermal characteristics of an underground cold-water reservoir are investigated analytically and using Artificial Neural Networks (ANN). An analytical solution is developed for the temperature distribution in the reservoir by assuming a linearized boundary condition at the water surface. For the general non-linear boundary condition, the temperature distribution is modeled using ANN. Very good agreements between the analytical and ANN results at various times during the withdrawal cycle are observed, ensuring the accuracy of the analytical and ANN procedures. The results show that a stable thermal stratification is preserved in the reservoir throughout the entire course of withdrawal cycle. As one important outcome of this research, two different regions are observed inside the thermally stratified tank during discharge cycle. The bottom region with a linear temperature distribution and the upper one in which a nearly exponential thermal stratification are developed. During withdrawal cycle, the outside temperature reaches as high as 42 °C, while cool water with the temperature varying from 12 to 13 °C is easily available from the underground water reservoir under investigation.