Industries / Industrial equipment and services / HVAC equipment
Testing and efficiency ratings
Testing and ratings is the lab work that measures how much heating or cooling a unit delivers and how much electricity it uses, and the rules and labels built on those numbers.
7
Milestones achieved, from 1911 to 2023.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1911
The math of moist air
Psychrometrics is the study of air and the water vapor in it. It sits under every capacity test, because a cooling coil both cools air and takes water out of it. Willis Carrier, head of engineering at the Buffalo Forge Company, published what Carrier's heritage site calls "the first psychrometric chart ever published" in a catalog in 1906, the forerunner of the charts used today. On 8 December 1911 he presented "Rational Psychrometric Formulae" at the annual meeting of the American Society of Mechanical Engineers (ASME); the site calls it the "Magna Carta of Psychrometrics".
Credit
Willis H. Carrier, Buffalo Forge Company
1987
A national efficiency floor
The National Appliance Energy Conservation Act, signed on 17 March 1987, set the first US federal minimums for central air conditioners and heat pumps. Split systems made from 1 January 1992 had to reach a seasonal energy efficiency ratio (SEER, cooling delivered over a typical season per unit of electricity) of 10.0, and split heat pumps a heating seasonal performance factor (HSPF) of 6.8. The law told the Department of Energy (DOE) when to review the standards, and a court later read it as allowing standards to be raised but not lowered.
Credit
US Congress
1999
Japan's Top Runner standard
From April 1999 Japan set appliance standards by the "Top-Runner Approach": the efficiency of the best products on the domestic market becomes the next standard, which every maker above a set volume must reach by a target year. For heat-pump room air conditioners the 2004 targets were met in every size class; units under 2.5 kW averaged a coefficient of performance (COP) of 5.33 against a standard of 5.27. Meeting them meant ending inefficient models, about 70 models and 2 million units a year in the 2.8 kW size alone. Japan then switched from COP, measured at full output, to an annual performance factor (APF) worked out over the whole heating and cooling season, because almost all units now used inverters that run below full speed most of the time.
LessonA rating taken only at full output stops describing a machine once its controls let it run slower most of the time.
Credit
Japan's Ministry of Economy, Trade and Industry; review by Yutaka Nagata of the Central Research Institute of Electric Power Industry
2004
A court blocks a rollback
In January 2001 DOE raised the minimum for central air conditioners to SEER 13, for units made from 23 January 2006. The Air-Conditioning and Refrigeration Institute (ARI), the makers' trade group, asked DOE to reconsider, and in May 2002 DOE withdrew the rule and adopted SEER 12. On 13 January 2004 the US Court of Appeals for the Second Circuit ruled for the challengers and restored SEER 13. It held that the 1987 law built in an "anti-backsliding" mechanism: efficiency standards "could be amended in one direction only, to make them more stringent".
LessonOnce a US efficiency standard is published it can be raised but not lowered, which gives makers a fixed target to design to.
Credit
the Natural Resources Defense Council, several states and consumer groups, who brought the case; the Second Circuit
2013
Ratings across the season
EU Regulation 206/2012 replaced single-point ratings for air conditioners up to 12 kW, including those that also heat, with seasonal ones from 1 January 2013. SEER for cooling and SCOP (seasonal coefficient of performance) for heating are built from the unit's performance at part load across a table of outdoor temperatures, weighted by the hours each temperature occurs in a season, and they count power used in standby, off and thermostat-off modes. The minimums were SEER 3.60 and SCOP 3.40, and about 10% lower for units using a refrigerant with a global warming potential (GWP, its heating effect relative to carbon dioxide) of 150 or less.
Credit
European Commission
2019
Testing under the unit's controls
Standard ratings run variable-speed units at fixed settings the maker supplies for the test. CSA Group's EXP07:19 is a "load-based" test: the lab imposes a simulated house's heating or cooling load, and the heat pump responds through its own built-in controls to hold the room temperature. NEEA had tested 13 heat pumps with the 2019 version by July 2020. A paper by Bruce Harley and colleagues reported that the method uncovered "substantial performance variability and substantial differences in relative ranking of products".
Credit
CSA Group; NEEA and its testing partners
2023
SEER2 and the ductwork
From 1 January 2023, US central air conditioners and heat pumps must be rated under DOE's test procedure Appendix M1, with new metrics called SEER2, EER2 and HSPF2. One change is external static pressure, the resistance of the ducts that the indoor fan must push air through, which raises the fan's power use. DOE had tested ducted systems at 0.10 to 0.20 inches of water column; after reviewing 20 field studies with 1,010 assessments of home systems, it set 0.50 inches for conventional systems, so the same unit now gets a different number under the new metrics. DOE adopted the procedure in January 2017.
Credit
US Department of Energy, with makers and advocates in its negotiating working group
Watch for
Checking variable-speed claims
From 7 July 2026, when DOE tests a central air conditioner or heat pump sold as variable-speed, it may run a controls verification procedure (CVP). The unit runs on its own controls instead of the maker's fixed test settings, and if it does not modulate as certified, it is rated as a single-speed or two-speed system. In comments to DOE in 2025, NEEA said its lab testing showed "a substantial fraction of systems tested would likely fail the procedure", and that Bonneville Power Administration field data showed one-third of 17 variable-speed units behaving as fixed-speed systems. Lennox backed the procedure. Other makers asked for more time, and DOE delayed enforcement by one year.
Achieved whenDOE or an outside lab publishes CVP results for a sample of certified variable-speed models, showing how many pass.
Claimed by
NEEA and Lennox, in comments summarized by DOE in September 2025
Open
01
Check window AC ratings
A low-cost kit and method that measure how much cooling a window air conditioner delivers as installed, and how much power it draws, plus a public table comparing many installed units with their rated capacity and efficiency. Temperature and humidity sensors at the intake and outlet grilles give the change in the air's heat content (its enthalpy), an airflow reading gives the volume of air, and a plug-in meter gives the power; capacity is airflow times the change in heat content. Indoor and outdoor conditions are logged so that results can be compared at the rating condition.
The briefCredit · unclaimed02
One ledger of load-based results
One public table of every published load-based or controls-verification test of home heat pumps and air conditioners, each unit listed with its certified ratings (SEER2 and HSPF2, or the EU's SEER and SCOP), and a test of how well the rating predicts the load-based result.
The briefCredit · unclaimed
Open 01 · the brief
Check window AC ratings
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A low-cost kit and method that measure how much cooling a window air conditioner delivers as installed, and how much power it draws, plus a public table comparing many installed units with their rated capacity and efficiency. Temperature and humidity sensors at the intake and outlet grilles give the change in the air's heat content (its enthalpy), an airflow reading gives the volume of air, and a plug-in meter gives the power; capacity is airflow times the change in heat content. Indoor and outdoor conditions are logged so that results can be compared at the rating condition.
Why it is still open
NREL lab-tested four window units in 2012. Three came within 5% of their rated capacity, but an old unit with a fouled coil delivered 3,497 Btu an hour against a 6,000 rating. NREL also noted that the rating test leaves out the maker's installation method, air leaking through the open window, and on-off cycling. New units match their labels in the lab: Oak Ridge National Laboratory measured a new 10,000 Btu/h GE window unit in 2015 within 3% of its published efficiency and 2.6% of its capacity. Searches in November 2026 found no public data set comparing installed room air conditioners with their ratings, and no low-cost field method checked against a lab.
Built with what exists
calibrated digital temperature and humidity sensors; a vane anemometer or a simple flow hood; a plug-in power meter; the CoolProp library for the properties of moist air; Python.
First step
Measure one window unit for an hour on a hot afternoon with two sensors, an anemometer and a power meter, and repeat the next day to see how much the result moves.
Done looks like
The kit's capacity reading falls within 10% of a laboratory measurement of the same unit for at least three models; then at least 50 installed units in at least 20 homes are measured on hot days, with a published table of measured against rated capacity and EER (energy efficiency ratio, Btu an hour of cooling per watt), and the method's uncertainty stated.
Open 02 · the brief
One ledger of load-based results
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
One public table of every published load-based or controls-verification test of home heat pumps and air conditioners, each unit listed with its certified ratings (SEER2 and HSPF2, or the EU's SEER and SCOP), and a test of how well the rating predicts the load-based result.
Why it is still open
The results sit in separate reports. NEEA tested 13 heat pumps under EXP07:19 by 2020 and six variable-speed heat pumps at UL's lab in Plano, Texas, in summer 2024. Purdue University's group has tested "multiple units from different manufacturers", and its research page links papers but no data. Germany's Federal Institute for Materials Research and Testing (BAM) publishes a load-based building model for heat pumps, not results. Harley and colleagues reported large changes in how products rank. Searches in November 2026 found no combined data set.
Built with what exists
the published reports and papers; an AI model that reads tables in PDF reports (see the Note "The jobs agents repeat"); the AHRI Directory of certified products and the NEEP cold-climate heat pump list for ratings; Python with pandas and SciPy for rank correlation.
First step
Copy the 13 heat pumps in NEEA's 2020 EXP07 interim report into a spreadsheet with the ratings the report gives for each.
Done looks like
At least 40 tested units from at least four sources in one open file, each row linked to its source page; the rank correlation between rated and measured efficiency published for each test type, with confidence intervals; and a list of the units whose measured rank differs most from their rated rank.