Oak Ridge Researchers Publish Field Measurements on Refrigerant Charge Faults in Cooling Season
In the team's 2024 field report, a 3-ton R-410A heat pump held a 75-degree setpoint until it was 40 percent undercharged. At 50 percent low, supply air ran about 5 degrees F warmer. Overcharges up to 25 percent left supply air and runtime unchanged.
A team at Oak Ridge National Laboratory has published field measurements of how refrigerant undercharge and overcharge change indoor conditions and system performance during cooling season, in Applied Thermal Engineering volume 309. The authors are Yeobeom Yoon, Sungkyun Jung, Young Jae Choi, Piljae Im, Junjie Luo and Tim Dyer.
Elsevier's site blocked access to the article and no abstract could be retrieved, so its own results are not reported here. The same researchers documented the cooling-season test program in an open ORNL technical report, ORNL/TM-2024/3661, issued in October 2024, which covers "how refrigerant undercharge and overcharge faults affect building indoor conditions and HVAC system operation." The findings below come from that report. Because the journal title refers to cooling seasons in the plural, the paper may draw on more data than the 2024 report did.
A 3-Ton Heat Pump, Charged by Scale
Testing ran at ORNL's Yarnell Station research house in Knoxville, Tenn., in ASHRAE climate zone 4A. The two-story, 2,400-square-foot house, built in 2009, has a single-stage 3-ton R-410A heat pump rated SEER 14 and HSPF 8.2. The team set nominal charge by subcooling against the manufacturer's charging table, then removed or added weighed amounts. It ran undercharges of 10 to 50 percent and overcharges of 5 to 25 percent, each for about a week, between May 29 and Sept. 16, 2024, with the thermostat at 75 degrees F for cooling and no nighttime setback.
Low Charge Stayed Hidden Until 40 Percent
Both floors held the 75-degree setpoint through a 30 percent undercharge. Unmet hours began at 40 percent. Supply air ran 43.9 to 53.6 degrees F at baseline and 47.8 to 58.2 degrees at 50 percent low, about 5 degrees warmer. At an 80-degree outdoor temperature the other scenarios ran 30 to 60 percent of the time, while the 50 percent undercharge ran continuously and still fell behind.
The report credits the expansion valve for that plateau, saying the TXV held superheat steady from baseline through 30 percent low before superheat climbed sharply at 50 percent once refrigerant flow dropped. Compressor energy per minute fell as charge fell, so longer runtimes did not raise hourly compressor energy by a significant amount.
Overcharge Left Few Traces
No overcharge level produced unmet hours. Supply air stayed between 43 and 48 degrees F for most hours regardless of the overcharge, and runtime fraction did not change with it. In the body of the report, compressor energy per minute between 89.0 and 89.9 degrees outdoor rose as overcharge increased, which the authors tie to higher capacity and greater cycling losses. The report's executive summary says compressor energy decreased with overcharge, so its own text conflicts on that point.
A companion paper by the group in Energy and Buildings, using the 2024 undercharge data, found that rule-based and machine-learning algorithms using a minimal sensor set could detect undercharges of 30 percent or more and, after a mitigation step, produced no false alarms. In that dataset superheat tracked charge level most closely, with a correlation of minus 0.95, and the supply-to-return temperature difference followed at 0.90.
Why This Matters to Contractors
The Energy and Buildings paper cites earlier field work that found undercharge in 45 percent of new systems and 71 percent of older ones, and ORNL's report cites an estimate that charge and airflow faults in newly installed residential air conditioners and heat pumps waste about 20.7 terawatt-hours of site energy and add 9.6 million metric tons of carbon dioxide each year. Neither the thermostat nor the supply-air sensor registered overcharge in the Yarnell house, so subcooling measured against the manufacturer's table remains the check that finds it.
Sources
- Y. Yoon, S. Jung, Y.J. Choi, P. Im, J. Luo and T. Dyer, "Impact of refrigerant undercharge and overcharge faults on building indoor conditions and HVAC system performance during cooling seasons: field measurements," Applied Thermal Engineering, vol. 309, 133734, November 2026. doi.org
- Y. Yoon et al., "Residential HVAC Fault Detection: Field Data Analysis and Interviews with Smart Thermostat Manufacturers," ORNL/TM-2024/3661, October 2024. ornl.gov
- Y.J. Choi, Y. Yoon, P. Im et al., "Field-Based AFDD for Refrigerant Undercharge in Residential HVAC Systems: Enhancing Reliability Through False Alarm Mitigation," Energy and Buildings, 2026, accepted manuscript via OSTI. osti.gov