Simulation of Fault Characteristics for Commercial Refrigeration Systems via Semi-Empirical Models

Refrigeration systems can undergo many faults that could negatively affect their operation and performance. This research addresses the use of semi-empirical models to simulate the fault impacts on the operation of the commercial walk-in refrigeration systems and evaluates existing thermostatic expansion valve mass flow rate models. Semi-empirical models often require less computational effort than full forward models, and could be used in scenarios where detailed information is missing, such as in field-measured systems. An important characteristic of a typical walk-in refrigeration system is the existence of a liquid-line receiver after the condenser, which significantly changes the cycle behavior, in comparison to a receiver-less system. Components joined to form the cycle model are: a compressor, two heat exchangers, pipelines, and a thermostatic expansion valve. These models were identified and trained with a limited set of faulty and fault-free experimental data: heat exchangers fouling, liquid-line restriction, and compressor valve leakage. System model iterated through implicit solutions of the components until it is converged. Numerical results showed that the existing semi-empirical component models in the literature can successfully be applied to the major components of refrigeration systems, such as the compressor and heat exchangers, to predict important performance indices of a refrigeration system with good accuracy. However, the modeling of the TXV showed relatively large errors, due to the unreliable refrigerant property calculations at the valve inlet as well as the calculation of the opening force of the TXV from evaporator superheat value. The receiver causes additional modeling errors in comparison to a receiver-less system due to the difficulty in the calculation of two-phase refrigerant properties in the liquid line for some faulted conditions. It was found that the excess charge is stored in the receiver and the receiver may not be totally empty even when that runs out of the excess charge. The model was used to generate a larger dataset from the limited set of experimental data. Model challenges and future research for systems with a liquid-line receiver are outlined