Every time refrigerant is added to a covered HFC appliance, the rule requires a leak rate: the percentage of the appliance's full charge it would lose over 12 months at the rate shown. You calculate it one of two ways. The annualizing method takes the pounds just added, divides by the full charge, and scales the result to a year using the days since the last addition (capped at 365). The rolling average method adds up every pound added in the previous 365 days and divides by the full charge. Either method gives a percentage you compare against the trigger for the appliance's category, which is 20% for commercial refrigeration.
Covered appliances are those with a full charge of 15 pounds or more of an HFC or substitute with a GWP above 53. The requirement took effect January 1, 2026. The arithmetic is simple. What goes wrong in the field is the inputs: the wrong full charge, a missing prior addition, a method switched halfway through the year, or a seasonal adjustment that was never written down.
This guide covers the formulas as the regulation states them, the edge cases that change the result, worked examples with checked numbers, and the records that have to back up every calculation.
When the Calculation Is Required
Under 40 CFR 84.106, a leak rate is calculated "every time refrigerant is added to an appliance unless the addition is made immediately following a retrofit, installation of a new refrigerant-containing appliance, or qualifies as a seasonal variance."
That leaves only three exceptions:
- Right after a retrofit. The first charge of the new refrigerant is not a leak event.
- Right after installing a new appliance. Commissioning charge doesn't count.
- A qualifying seasonal variance, covered in its own section below.
Every other addition requires a calculation: a top-off on a night call, a charge after a repair, a pound added during a preventive maintenance visit. The obligation applies to appliances in scope under EPA's leak repair fact sheet, which sets the triggers at 30% for industrial process refrigeration, 20% for commercial refrigeration, and 10% for comfort cooling and other appliances, and excludes the residential and light commercial AC subsector.
The Four Inputs and Where Each One Comes From
A leak rate is only as good as its inputs. Before doing any math, confirm you have all four.
| Input | Where it comes from | What goes wrong |
|---|---|---|
| Pounds added now | Cylinder weight before and after, on a scale | Estimated by pressure or sight glass instead of weighed |
| Full charge | One of four methods the regulation allows | Nameplate of the condensing unit used for a field-piped system |
| Date of last addition | The appliance's service record | Another contractor's addition is missing from the file |
| Additions in the window | All additions in the prior 365 days, or since the last clean follow-up verification | Window not reset after a successful repair |
Inputs as defined in 40 CFR 84.102; common errors are field observations, not regulatory text.
The annualizing method needs the first three. The rolling average method needs pounds added, full charge, and the full window of additions.
Getting Full Charge Right
Full charge is the denominator, and a bad denominator skews every calculation that follows. The regulation defines it in 40 CFR 84.102 as the amount of refrigerant "required for normal operating characteristics and conditions of the appliance," determined using one or a combination of four methods:
- The equipment manufacturer's determination. Straightforward for a self-contained or factory-charged system.
- Calculations "based on component sizes, density of refrigerant, volume of piping, and other relevant considerations." This is the realistic method for a field-piped rack or a remote condensing unit with long line sets.
- Actual measurements of the refrigerant added to or evacuated from the appliance, such as a documented full recovery weighed into a cylinder followed by a weighed recharge.
- An established range based on the best available data, "where the midpoint of the range will serve as the full charge."
Two practical points. First, a condensing unit's nameplate charge usually covers only the unit, not the evaporators, the line set, or the receiver at operating level. Using it as the full charge on a field-assembled system understates the denominator and inflates every leak rate you calculate. Second, if you use a range, the regulation fixes the full charge at the midpoint. You don't get to pick whichever end gives a better result.
Determine the full charge once, document the method, and put it in the appliance record. Every later calculation depends on it.
The Annualizing Method, Step by Step
The regulation sets out the annualizing method in four steps:
- Divide the pounds of refrigerant added by the pounds of refrigerant in a full charge.
- Take the shorter of the number of days since the last day refrigerant was added or 365 days, and divide that number by 365.
- Divide the result of step 1 by the result of step 2.
- Multiply by 100 to get a percentage.
As one expression:
Leak rate (%) = (pounds added ÷ full charge) ÷ (min(days since last addition, 365) ÷ 365) × 100
The 365-day cap
Step 2 is the part most online calculators leave out. The regulation says to use "the shorter of the number of days that have passed since the last day refrigerant was added or 365 days." If a system went 500 days between additions, you use 365, and the leak rate is simply the pounds added as a percentage of full charge. Without the cap, a long quiet period would divide the addition across more than a year and understate the rate.
What the method is sensitive to
The annualizing method measures one addition against one interval, so short intervals drive the result up quickly. A modest top-off only three weeks after the previous one can annualize well past 20% even when total refrigerant use over the year is low. That is intentional: it catches a leak that has just started. It also means a follow-up top-off soon after a charge that was left short will count against the appliance as if it were a new leak.
The Rolling Average Method, Step by Step
The rolling average method looks backward instead of projecting forward:
- Add up all refrigerant added over the previous 365-day period, or over the period since the last successful follow-up verification test showing all identified leaks were repaired, if that period is less than one year.
- Divide by the pounds of refrigerant in a full charge.
- Multiply by 100.
Leak rate (%) = (pounds added in the window ÷ full charge) × 100
The window resets after a verified repair
The parenthetical in step 1 is important. Once a repair has been completed and the follow-up verification test confirms all identified leaks were repaired, the rolling window starts over from that test if it happened less than a year ago. Refrigerant added before the repair no longer counts against the appliance. Without the reset, a rack that was properly fixed in March would carry its pre-repair additions until the following March and could trigger again on its first small top-off.
Recording the follow-up verification date isn't only proof the repair held; it's also an input to the next rolling average calculation.
Worked Examples
The examples below use illustrative figures chosen to show how each method behaves. They aren't data from real systems. All results were computed directly from the regulatory formulas.
Example 1: a walk-in cooler just over the 15-pound line
A walk-in on a remote condensing unit has a documented full charge of 18 pounds. A tech adds 3 pounds, 60 days after 2 pounds were added.
- Annualizing: (3 ÷ 18) ÷ (60 ÷ 365) × 100 = 101.4%
- Rolling average: (3 + 2) ÷ 18 × 100 = 27.8%
Both methods put the appliance over 20%. Small systems reach the trigger quickly because each pound is a large share of the charge. That's why walk-ins that were never regulated before now account for so many exceedances.
Example 2: a supermarket rack with a steady slow leak
A rack has a full charge of 1,500 pounds. A tech adds 40 pounds, 90 days after the previous addition. Over the past year, the rack also received 60 and 55 pounds.
- Annualizing: (40 ÷ 1,500) ÷ (90 ÷ 365) × 100 = 10.8%
- Rolling average: (40 + 60 + 55) ÷ 1,500 × 100 = 10.3%
Both results are under 20%, and the two methods agree closely. That's typical when additions are evenly spaced.
Example 3: where the methods disagree
A rack with a 1,000-pound full charge takes 30 pounds, 20 days after it took 25 pounds.
- Annualizing: (30 ÷ 1,000) ÷ (20 ÷ 365) × 100 = 54.8%
- Rolling average: (30 + 25) ÷ 1,000 × 100 = 5.5%
This is the divergence the diagram above shows. A short gap between additions makes the annualizing result spike, while the rolling average reflects only the 55 pounds actually used. Which one applies depends on the method the owner has adopted for this appliance, not on which number looks better that day.
Example 4: a long quiet stretch and the cap
The same 1,000-pound rack takes 50 pounds after 500 days without an addition.
- Annualizing, with the 365-day cap: (50 ÷ 1,000) ÷ (365 ÷ 365) × 100 = 5.0%
Without the cap, dividing by 500 days would give a lower number than the rule allows.
| Scenario | Annualizing | Rolling average | Against 20% |
|---|---|---|---|
| 18 lb walk-in, 3 lb after 60 days | 101.4% | 27.8% | Over under either method |
| 1,500 lb rack, steady additions | 10.8% | 10.3% | Under either method |
| 1,000 lb rack, 30 lb after 20 days | 54.8% | 5.5% | Depends on method in use |
| 1,000 lb rack, 50 lb after 500 days | 5.0% (capped) | Depends on window | Under |
Illustrative figures; results calculated with the formulas in 40 CFR 84.102.
Method 1 vs Method 2: Choosing and Sticking With One
Some vendor guides call these "Method 1" and "Method 2." The regulation calls them the annualizing method and the rolling average method. What matters is picking one deliberately and keeping it.
On switching, 40 CFR 84.106 allows a change of method essentially only after a facility is acquired from an owner who used the other method, and only under specific conditions. There is no option to run both and report the lower number.
How the methods compare in practice:
- Annualizing catches new, fast leaks early and is harsh on closely spaced additions. It fits owners who want early warning and are disciplined about charging systems fully on each visit.
- Rolling average reflects actual refrigerant use and is less affected by a single close pair of additions. A leak that starts slowly can run for months before the total crosses 20%, and the owner has to keep a complete addition history.
For multi-site owners, the practical rule is one method per portfolio, written into the service agreement, so every tech on every site calculates the same way. Get Freon's guide to pricing leak repairs under the new rules covers how to set that up contractually.
First Calculations After January 1, 2026
The regulation provides defaults for the first calculation on each appliance under the new program:
- Annualizing, first use: "substitute 365 days as the number of days since last refrigerant addition." The first addition is therefore simply pounds added as a percentage of full charge.
- Rolling average, first use: "substitute pounds of refrigerant added since January 1, 2026." The window starts on the rule's effective date, not at some earlier point.
For example, a 1,000-pound rack taking 30 pounds as its first addition of 2026 comes out at 3.0% under either default. Neither default is punitive. Both mean that any 2026 addition missing from the record understates the next result, which makes it a compliance problem rather than a benefit.
Seasonal Variance: The Legitimate Exception
Many racks and condensing units with head-pressure control need more refrigerant in winter than in summer. The regulation accounts for this. Seasonal variance is defined as removing refrigerant "due to a change in ambient conditions caused by a change in season, followed by the subsequent addition of an amount that is less than or equal to the amount of refrigerant removed in the prior change in season, where both the removal and addition of refrigerant occurs within one consecutive 12-month period."
Three conditions must all be met:
- The removal happened because of the season, not for a repair or as recovery for another reason.
- The addition is no more than what was removed. Anything beyond that is a normal addition and needs a calculation.
- Both happen within one consecutive 12-month period.
The documentation matters as much as the physics. Under 84.106(l)(12), owners using this flexibility keep records "stating that they are using the seasonal variance flexibility and documenting the amount added and removed." A fall addition with no recorded spring removal looks exactly like a leak, and in an audit it will be treated as one.
What to Record With Every Calculation
A calculation that isn't recorded doesn't help in an audit. Under 84.106(l)(2), each refrigerant addition's record includes the appliance's identity and location, the service date, the refrigerant type and amount added, the full charge, and "the leak rate and the method used to determine the leak rate." The leak rate and method aren't required for disposal, retrofits, new installations, or seasonal variance additions.
A practical service-ticket block for every addition on a covered appliance:
- Appliance ID and location
- Date of service and technician
- Refrigerant type and pounds added, weighed
- Full charge value and how it was determined
- Calculation method in use
- Date of the previous addition (annualizing), or the window start and total (rolling average)
- Resulting leak rate and the applicable trigger
- Whether a seasonal variance or other exception applies, with the supporting removal record
EPA's fact sheet requires these records to be kept for three years, on paper or electronically.
What the Result Tells You to Do Next
The calculation exists to trigger action. If a commercial refrigeration appliance comes out above 20%, the owner has 30 days to identify and repair the leaks, followed by initial and follow-up verification tests, or 30 days to write a retrofit or retirement plan. Get Freon's article on the HFC leak repair rules and the 20% threshold covers the full sequence, including extensions and follow-up inspections.
A result under the trigger is not a clean bill of health. A rack at 15% is still losing refrigerant that someone has to replace. With the HFC allowance cap dropping in 2029, as covered in Get Freon's analysis of the 2029 allowance cut, an appliance that stays below 20% is still consuming a supply that is about to get tighter.
A Separate Check: The 125% Calendar-Year Report
One more number is easy to confuse with the leak rate. EPA's fact sheet requires a report for appliances that leak "125 percent or more of the full charge in a calendar year," due by March 1 of the following year.
This isn't the annualizing or rolling average result. It's a calendar-year total: add up every pound added between January 1 and December 31 and divide by the full charge. A system can show a moderate rolling average in one part of the year and still cross 125% over the full calendar year if the owner kept topping off instead of repairing. Run this check at year-end for every covered appliance, using the same addition records.
Federal, Section 608, and State Programs Side by Side
The Part 84 calculation isn't the only one a commercial refrigeration contractor may need to run.
| Program | Covers | Commercial refrigeration trigger |
|---|---|---|
| AIM Act, 40 CFR Part 84 | HFCs and substitutes above 53 GWP, 15 lb and up | 20%, either method |
| Clean Air Act Section 608 | Ozone-depleting refrigerants such as R22, 50 lb and up | 20% |
| Washington State program | State-regulated systems | 16%, rolling average required for reporting |
Federal thresholds per EPA's Part 84 fact sheet and EPA's Section 608 leak repair page; Washington figures per Washington Ecology's leak rate guidance.
An older R22 rack is still judged under EPA's Section 608 leak repair requirements, which apply at 50 pounds rather than 15. State programs can be stricter. Washington Ecology's leak rate guidance requires the 12-month rolling average for reports to the state, lists a 16% threshold for commercial refrigeration, and treats the annualizing method as an internal monitoring tool only. In a state with its own program, run the state calculation alongside the federal one.
Common Mistakes That Change the Answer
Most wrong leak rates come from a short list of errors:
- Using the condensing unit nameplate as full charge on a field-piped system. This understates the denominator and inflates every result.
- Ignoring the 365-day cap in the annualizing method, which understates the rate after long gaps.
- Not resetting the rolling window after a successful follow-up verification, which leaves pre-repair additions counting against a fixed system.
- Missing another contractor's addition, which throws off both methods. The owner's record, not your company's ticket history, is the authoritative source.
- Estimating pounds added from pressures instead of weighing the cylinder.
- Claiming seasonal variance without a removal record, or adding more than was removed.
- Switching methods to get a lower number, which the regulation doesn't allow outside the narrow acquisition case.
Refrigerant additions, recovery, and leak repair on covered appliances must be done by technicians with appropriate EPA Section 608 certification.
Frequently Asked Questions
How do I calculate an EPA Subpart C leak rate in 2026?
Divide the pounds added by the full charge and annualize using days since the last addition, capped at 365 (annualizing method). Or divide the total added in the previous 365 days, reset after a successful follow-up verification, by the full charge (rolling average method). Multiply by 100 and compare with the trigger for the appliance's category.
What is the difference between Method 1 and Method 2?
They're vendor shorthand for the annualizing and rolling average methods. Annualizing projects one addition across a year and reacts to short intervals. Rolling average adds up actual additions over a year and reflects real use.
Can I use whichever method gives the lower number?
No. The owner uses one method per appliance, and 40 CFR 84.106 allows switching essentially only after acquiring a facility that used the other method.
Do I calculate a leak rate after recharging a repaired system?
Yes, unless the addition immediately follows a retrofit or the installation of a new appliance, or qualifies as seasonal variance. Recharging after a repair is an addition. The rolling average window resets only from the successful follow-up verification test.
What if the system went more than a year without an addition?
Under the annualizing method, use 365 days, not the actual interval. The result is simply the pounds added as a percentage of full charge.
How do I find the full charge on an old rack with no documentation?
Use the regulation's calculation method, based on component sizes, refrigerant density, and piping volume, or a documented weighed recovery and recharge. If you use a range, the midpoint is the full charge.
Is the 125% report the same as the leak rate?
No. It's the total refrigerant added over a calendar year divided by full charge, checked at year-end and reported by March 1 if it reaches 125%.
Why do calculating refrigerant leak rates matter if the system is under 20%?
Because the calculation record is the proof of compliance, and because every pound added is refrigerant bought from a supply that is shrinking. A tight system is the cheapest system to keep running.
The Number Only Works If the Inputs Are Right
The formulas in 40 CFR 84.102 fit on an index card. The work is in the inputs: a full charge determined by an accepted method, weighed additions, a complete addition history, a window that resets when it should, and seasonal adjustments recorded when they happen. Get those right on every ticket and the leak rate becomes a routine part of the service call, with a record that holds up if EPA asks for it.
