Calculating TRI releases when you have no monitoring data

This page explains what the rules and common practice call for. Every site is different, and what a survey or a filing needs at your facility is something we work out with you after we see it.

Most facilities do not have stack testing on every emission point, and the Toxics Release Inventory (TRI) does not require it. TRI is the annual toxic chemical release report filed with EPA, and Form R is its full reporting form.

The Emergency Planning and Community Right-to-Know Act (EPCRA) is clear on this. A facility may use readily available data that was collected under other laws. Where it has no such data readily available, it may use reasonable estimates of the amounts involved. Nothing in that part of the law requires monitoring or measurement beyond what other laws already demand. EPA's own current guidance states the standard this way: use your "best available information, or reasonable estimates," and take the "best reasonable approach" to the calculation.

So the question is never whether you have data. The question is which method you can best defend for each release point, meaning each place where the chemical leaves your process, and whether you can show your work.

EPA's "basis of estimate" codes

Every release quantity on Form R carries a code, which EPA calls the basis of estimate, that shows how you arrived at the number. The codes go in Section 5 or 6 of the form, and there are six codes in four families:

EPA's reporting rule sets no ranking among the codes. M1 is not automatically "better" than C, and a well-built mass balance on a closed system is often more accurate than periodic monitoring extrapolated across a year.

Even so, the statute itself sets a soft order of preference: readily available data where you have it, and reasonable estimates where you do not. So you cannot simply pick whichever method you like. Use the data you have and, where you have none, choose the estimate that fits the release point and document why.

Mass balance (code C)

What goes in must come out, so inputs equal outputs plus whatever accumulates on site.

Chemical in equals the sum of these:

• Chemical out in product
• Chemical out in waste
• Chemical consumed by reaction
• Chemical released
• Change in inventory

Then solve for the amount released.

Best for: solvent use, degreasing, coating operations, closed systems, and any situation where purchase records and product composition are reliable.

Where it breaks down: when releases are a very small fraction of throughput. If you purchase 100,000 pounds and 200 pounds is released, your release estimate is the difference between two large numbers, and a 1% error in either one swamps the answer. Mass balance works very well when releases are a meaningful fraction of throughput and is unreliable when they are not.

Worked example: solvent degreaser

A plant uses a listed solvent in a vapor degreaser.

Step 1. How much solvent was consumed? Take the quantity purchased, add the January 1 inventory, and subtract the December 31 inventory.

Solvent consumed = 42,000 + 3,100 − 2,400 = 42,700 lb

Step 2. How much was released to the air? Subtract the solvent shipped off-site in drums and the solvent that left on parts, and the remainder is the air release.

Air release = 42,700 − 34,800 − 900 = 7,000 lb

Step 3. Does the plant have to report? The plant "otherwise used" 42,700 pounds, which is the rule's term for any use that is not manufacturing or processing the chemical. The reporting threshold for that activity is 10,000 pounds. The plant is over it, so it reports. The air release is 7,000 pounds, with basis of estimate C.

Look at what supported the calculation: purchase records, two physical inventory counts, shipping manifests and one test of the solvent left on the parts, all of which already existed for other reasons. The inventory figures are the ones plants most often fail to capture, and without them the calculation will not stand up to review.

Emission factors (codes E1 and E2)

An emission factor is a published or site-developed number that relates emissions to a measure of activity, such as pounds per ton processed, per gallon applied, or per hour of operation.

E1 is for published factors, including EPA's AP-42 compilation, trade association factors and vendor data. E2 is for site-specific factors developed from your own testing or engineering analysis.

Best for: combustion sources, well-understood processes, and industries where EPA has published factors.

Where it breaks down: when your process differs materially from the one the factor was developed for. AP-42 factors carry quality ratings for a reason: a factor rated E, applied to a process unlike yours, is a guess with a reference number attached.

Worked example: coating line

The setup:

• The coating contains 18% by weight of a listed glycol ether solvent.
• 24,000 gallons are applied.
• The coating density is 9.1 pounds per gallon.
• The booth exhaust goes to a thermal oxidizer, which burns off the vapors. Capture efficiency is 90% and destruction efficiency is 98%.

Step 1. How much of the chemical was used? Multiply the gallons applied by the coating density, then by the chemical's weight fraction in the coating.

Chemical used = 24,000 gal × 9.1 lb/gal × 0.18 = 39,312 lb

(High-boiling glycol ethers used as coalescents can retain a measurable fraction in the film through a short bake. If your cure oven has its own exhaust, that split matters, so measure it and do not assume it.)

Step 2. How much do the controls remove overall? Multiply the capture efficiency by the destruction efficiency.

Overall control efficiency = 0.90 capture × 0.98 destruction = 88.2%

Step 3. How much was destroyed, and how much reached the air? Multiply the chemical used by the overall control efficiency, then subtract the destroyed amount from the chemical used.

Destroyed in the oxidizer = 39,312 × 0.882 = 34,673 lb

Air release = 39,312 − 34,673 = 4,639 lb

Two points need to be right here.

Capture efficiency and destruction efficiency are not the same number. Treating 98% destruction as though it were 98% overall control assumes perfect capture, which no booth achieves, so it understates your release.

The destroyed amount is not gone for reporting purposes. It is on-site waste treatment, and it belongs in Sections 7 and 8 of Form R.

The basis of estimate is E1 if the efficiencies come from vendor literature, E2 if from your own performance testing, and M2 if from stack testing.

Engineering calculations (code O)

These apply basic physics and chemistry to your specific equipment, for example:

• Evaporation losses, calculated from vapor pressure and Raoult's law
• Tank breathing and working losses
• Equipment leak estimates (fugitive emission factors) based on a count of valves, flanges, pumps and connectors

Best for: storage tanks, equipment leaks, evaporation from open surfaces, and any source with no published factor and no workable mass balance.

Where it breaks down: when the assumptions are not documented. An engineering calculation is only as defensible as its stated assumptions, and this is where recordkeeping most often fails.

Worked example: storage tank losses

A 12,000-gallon fixed-roof tank holds a listed chemical and has 18 turnovers per year, meaning its contents are replaced 18 times.

The tank loses vapor in two ways, and each is calculated separately with the equations in AP-42 Chapter 7 (Liquid Storage Tanks). Those equations use vapor pressure, tank dimensions, liquid temperature and turnover rate.

• Working losses: vapor displaced when the tank is filled
• Standing storage (breathing) losses: vapor expelled by daily swings in temperature and pressure

(Note that "breathing loss" refers to the standing component only. Using it as a label for the total is a small slip that a reviewer will notice.)

Suppose the calculation gives working losses of 340 pounds and standing losses of 95 pounds. The air release is the sum of the two.

Air release = 435 lb

The basis of estimate here is a judgment call worth documenting. The AP-42 tank equations are a published method, which points to E1. Here, though, they are applied as a site-specific engineering calculation, with your tank's dimensions and your liquid's properties, which points to O. You can support either one, but you cannot support choosing one without saying why, so record your reasoning next to the worksheet.

What makes this defensible is not the number but the saved worksheet. It shows the tank dimensions, the vapor pressure used and its source, the liquid temperature you assumed and why, and the version of the equation you used.

Choosing a method: the practical order

1. Do you have monitoring data for this release point? Use it, with code M1 or M2.
2. Is the release a meaningful fraction of throughput, and are your purchase, inventory and shipment records reliable? Use mass balance, code C.
3. Is there a published factor developed for a process genuinely like yours? Use an emission factor, code E1 or E2.
4. Otherwise, use an engineering calculation, code O.

Use different methods for different release points on the same chemical. Nothing requires one method per chemical, and applying one method across a whole plant because it is simpler is how estimates become impossible to defend. Stack releases might be M2, while equipment leaks are O and wastewater is C.

What gets checked

What gets checked is the basis, not the number.

EPA's rule requires you to keep records that support the basis of estimate behind every release figure and every off-site transfer figure. You must keep them at the facility for 3 years from the date you file. In practice, for every release quantity, that means documenting:

• Which method, and why that one
• The inputs, and where each came from
• Every assumption, written out
• The math
• Who did it and when

Picture two plants that each report 7,000 pounds, one with a saved worksheet and one that cannot reconstruct where its number came from. Their positions are very different, because the second plant has a reporting violation whether or not the number is right.

This is also the limit of what a software tool can do for you. Software will do the arithmetic, but the judgment calls are yours. Is your degreaser better described by a mass balance or by an engineering estimate? Does your assumption about how much of the coating evaporates hold? Can you defend an AP-42 factor that was developed for a different process? Those are the calls that get questioned.

The regulatory basis for this page:

EPA's industry guidance documents

EPA publishes estimation guidance for specific industries, and it helps to know both that these documents exist and how old they are:

• Semiconductor Manufacturing, an EPA guidance document (title: "EPCRA Section 313 Industry Guidance"). It replaces a January 1988 booklet.
• EPA's 2019 compilation of questions and answers (title: "EPCRA Section 313 Questions and Answers, 2019 Consolidation Document"). This is the single most useful general reference.
• Chemical-specific guidance for dioxin and dioxin-like compounds (2019), hydrochloric acid (2019), aqueous ammonia (2019) and sulfuric acid (2020)

The late-1990s industry guidance series has no volume for metal finishing or electroplating. That series covers semiconductor manufacturing, electricity generation, petroleum refining, chemical distribution and a few other sectors. An older issue paper does cover electroplating: "Section 313 Reporting: Issue Paper, Clarification and Guidance for the Metal Fabrication Industry," January 1990. It predates most of the rule changes since, however, so plating shops today should work from the 2019 questions and answers document.

→ Do you have to file? · What happens if you miss July 1