scope:

INTRODUCTION

The use of ice storage for air conditioning applications is increasing due to the need to reduce peak power demand. Many utility companies offer various rate structures to encourage energy use during off-peak periods and ice storage systems have been widely employed in many projects.

The performance of ice storage systems is more complex than that of traditional HVAC systems mainly because of the interaction between the chiller and ice tank. Therefore, a designer must know the transient variation of the fluid temperatures in the design stage to prevent mistakes and minimize risks in future operation. Simulation software facilitates the design process. However, few general-purpose software packages are available, so designers have had to use software provided by the specific manufacturers. These software packages tend to promote commercial products and contain limited technical information that does not allow optimal design and operative strategies to be determined.

Simulation software for the dynamic simulation of thermal storage systems, named TSTORS, is currently under development in the HVAC laboratory, Tsinghua University. The package is to be used as a design tool in feasibility studies, economic analysis, and control strategy determination for ice storage systems.

The ice tank is the most important component in the thermal storage system. For this reason, modeling of the ice tank is one of the key projects in the software development. The various models should not only accurately represent the real process, but should also minimize the computing time for the system simulation. Furthermore, the designers should only need to input variables with clear physical meaning that are understood easily by the HVAC designers such as the structural design and the tank size rather than heat transfer coefficients, so that they can be easily described by the designers without any further expert knowledge of the thermal processes inside the tank.

The encapsulated ice tank is either an erect or horizontal tank that is used in many situations. It is packed with ice balls at a packing rate of about 60%. Yamaha (1993) developed a detailed model for encapsulated ice tanks using phase change heat transfer theory and by separating the fluid flow pattern into a mixing region and a piston flow region. The validations showed that the simulation still did not accurately model the experiments. Arnold (1990, 1994, 1991) related the global heat transfer coefficient and IPF (Ice Packing Factor) using linear regression of experimental data. However, the model did not include the effects of either the flow-rate outside the ice ball or the ice ball diameter. Kamiya et al. (1997) studied the freezing process in a single ice ball using visualization to develop a model considering the volume expansion in ice formation.

The thermal processes inside a single ice ball were studied experimentally using visualization to develop a formalized model for a single ball (Zhao and Jiang 1994a,b). Li (1997) studied the charging and discharging performance of an encapsulated ice tank. This paper describes the development of a simple algorithm that requires less computing time and is adaptable to most of the products made by different manufacturers. The model adequately represents all thermodynamic and heat transfer processes, and is based on a detailed thermal process analysis. The analysis introduced in this paper is based primarily on the experimental results obtained by co-workers Li (1997) and Zhao and Jiang (1994a,b).

Two typical approaches are used to develop models: detailed models by purely theoretical analysis, and "black box" models by fitting an input-output curve from experimental data provided by the manufacturer. In a theoretical model, the thermal processes in the capsule are very difficult to describe because of the phase change inside the ball with both heat conduction and natural convection.

The "black box" models are always steady-state models, which only apply to a specific product. The models developed for TSTORS are "grey box" models, with the basic thermal processes described theoretically and some complex processes modeled using experimental data. TSTORS differs from other models by separating as much as possible the effects of different factors on the heat transfer coefficient, so that the model can be more accurately applied to different tank configurations and operating scenarios.

In addition, a regression tool was developed to obtain supplemental information from a manufacture's source book to account for special designs made by the manufacturer. This information will be used to modi@ some model parameters. The model will accurately model the real ice tank and can be used to simulate different encapsulated ice tanks from different manufacturers.