Date of Award
Master of Science in Mechanical Engineering (MSME)
James E. Braun
Committee Member 1
W. Travis Horton
Committee Member 2
In a vapor compression system, the amount of refrigerant charge significantly affects the performance of the unit. Improperly charged systems run sub-optimally and thus result in higher operating costs due to increased energy usage. The unit can be undercharged due to a small leakage over the years. Similarly, the unit can be overcharged due to improper maintenance practices. Currently, there is no direct way to measure the amount of refrigerant in the system. In this thesis, virtual refrigerant charge sensors were developed and implemented for a rooftop unit using a previously developed approach. An open lab methodology with low cost electrical hardware was implemented for training the virtual refrigerant charge sensor which can be used as a substitute to expensive psychrometric chamber testing. The entire test methodology was automated which significantly reduced the need for human intervention. The open lab training results were validated by testing the RTU inside psychrometric chambers at different ambient conditions and charge levels. The accuracy of the virtual refrigerant charge sensor model trained using open lab methodology was within ±10% of the actual charge measurement. The concept of an automated charging kit was extended further to facilitate adding as well as removing refrigerant charge in the system. This apparatus was used to test the RTU at different charge levels in an attempt to locate an optimal operating charge for coefficient of performance as well as cooling capacity. The results indicate a relatively flat variation of COP and cooling capacity with charge around the optimal. The automated open lab training methodology can significantly lower testing costs for VRC sensor tuning as it eliminates the need for psychrometric chambers as well as human interference.
Patil, Akash, "Development and Evaluation of Automated Virtual Refrigerant Charge Sensor Training Kit" (2018). Open Access Theses. 1434.