Apply the EPA’s average emission factor for a light liquid pump seal, and you get 0.114 kg/hr. Screen that same seal with an OVA at 500 ppmv and run the correlation equation, and you get roughly 0.002 kg/hr. Same pump, same seal, same day — estimate differs by 57x. The EPA itself acknowledges that its emission factors are “simply averages” where “approximately half of sources will have emission rates greater than the factor.” Yet most facilities still plug average factors into their inventories without questioning whether a more accurate method applies.
The estimation method you select is the most consequential decision in fugitive emission inventory development for sealed rotating equipment.
Why Method Choice Drives Your Emission Numbers
A non-leaking pump mechanical seal emits about 0.012 kg/hr; a leaking one emits 0.437 kg/hr — a 36x gap. The same leak/no-leak split runs across every component class the method covers: a gas valve in refinery service, for comparison, sits at 0.0006 kg/hr non-leaking versus 0.2626 kg/hr leaking. If your facility uses Tier 1 average factors, you’re blending both populations into a single number that accurately represents neither.
I’ve reviewed inventories where a facility with 100 valves and 250 flanges calculated 447 kg methane annually using average emission factors. Actual monitoring — one leak detected in October, repaired in 10 days — put real emissions at 36.6 kg. A 12.2x overestimate from a method that was supposed to be “conservative.”
Average factors were developed before modern leak detection existed, from component populations never inspected for leaks. But the reverse is also true: for facilities with large undetected releases — failed seals venting to atmosphere, piping vibration cracks — average factors can grossly underestimate actual losses. The same methodology errs in both directions depending on maintenance quality.
Your facility’s actual condition, not a default spreadsheet, should determine which calculation approach produces defensible numbers. Fugitive emission control mechanical seals exist specifically because this emission factor gap creates both compliance risk and cost exposure.
Average Emission Factors vs Correlation Equations
Three EPA estimation tiers exist, each producing different numbers from the same equipment.
Average Emission Factors (Tier 1)
Tier 1 is the simplest method: multiply a published emission factor by component count and operating hours.
Annual emissions = Emission Factor x Component Count x Operating Hours
Refinery emission factors for sealed equipment in light liquid service:
- Pump seals: 0.114 kg/hr
- Compressor seals: 0.636 kg/hr
- Valves: 0.0268 kg/hr
- Flanges: 2.50E-4 kg/hr
For 50 pumps running 8,000 hours annually: 0.114 x 50 x 8,000 = 45,600 kg/yr. That number goes into your Title V emission inventory.
These are industry-wide averages including poorly maintained equipment. A facility running dual seals with barrier fluid and quarterly LDAR monitoring will overstate emissions by an order of magnitude using Tier 1. The mechanical seal leakage rate feeding into these factors varies enormously by seal configuration — but Tier 1 ignores seal type entirely.
Screening Ranges (Tier 2 — Leak/No-Leak)
Tier 2 separates components into “leaking” and “non-leaking” populations at a 10,000 ppmv screening threshold, then applies different factors to each group.
For pump seals in light liquid refinery service:
- Leaking (≥10,000 ppmv): 0.437 kg/hr
- Non-leaking (<10,000 ppmv): 0.012 kg/hr
If 48 of your 50 pumps screen below threshold and two screen above: (48 x 0.012 x 8,000) + (2 x 0.437 x 8,000) = 4,608 + 6,992 = 11,600 kg/yr. That’s 75% lower than Tier 1 for the same equipment — because Tier 2 credits your facility for the 96% of seals that aren’t leaking.
Correlation Equations (Tier 3)
Tier 3 converts individual screening values directly to mass emission rates. For a refinery valve in light liquid service:
Emission Rate (kg/hr) = 2.29E-6 x (Screening Value)^0.746
Three tiers operate within the correlation method itself:
- Zero reading (0 ppmv): apply a default-zero rate. Refinery valve, light liquid: 7.8E-6 kg/hr.
- Measurable reading (above zero, below analyzer peg): apply the correlation equation with the actual screening value.
- Pegged reading (exceeds analyzer maximum): apply a fixed pegged emission rate. At 10,000 ppmv peg: 0.064 kg/hr. At 100,000 ppmv peg: 0.14 kg/hr.
A valve screening at 500 ppmv: 2.29E-6 x 500^0.746 = 2.29E-6 x 107.6 = 2.46E-4 kg/hr. That same valve using the Tier 1 average factor would be assigned 0.0268 kg/hr — 109x higher. Tier 3 rewards good screening data with estimates that reflect actual field conditions.
The Pegged Source Problem Most Facilities Overlook
When your analyzer reads beyond its maximum range, the screening value “pegs” — and the assigned emission rate jumps to a fixed value that may bear no relationship to actual emissions.
A standard OVA pegs at 10,000 ppmv, assigning 0.064 kg/hr for a refinery valve. Upgrade to an analyzer with a 100,000 ppmv range and that same component — now reading 30,000 ppmv instead of pegging at 10,000 — gets calculated through the correlation equation at approximately 0.005 kg/hr. The pegged value was 13x higher than what the correlation equation produces with the actual concentration.
It gets worse. An analyzer pegging at 30,000 ppmv assigns a reading of 30,001 to the 100,000 ppmv pegged factor. Converting that back through the equation yields an effective concentration of 2.6 million ppm — for a component actually reading 30,001. Your analyzer selection is not just an instrument decision; it’s an emission inventory decision.
Individual correlation equation measurements carry uncertainty “as much as a factor of 3 higher or a factor of 10 lower.” Across 100 components, uncertainty narrows to about ±1.4x — but for any single dry gas seal on a centrifugal compressor, the estimate can be wildly off. Acceptable for population-level inventory work, but dangerous for repair-or-replace decisions on individual components.
Which EPA Method Fits Your Facility
Every LDAR practitioner I’ve worked with has a default method — usually whatever their state agency accepted last time. That works until a consent decree, a corporate emissions target, or a Title V renewal forces a recalculation. The right method depends on three questions.
Question 1: What industry emission factors apply?
SOCMI and petroleum refinery factors cover different equipment populations. Using refinery factors for a chemical plant introduces systematic error from the start. The gap between SOCMI and refinery factors for the same equipment type can exceed 5x.
Question 2: Do you have component-level screening data?
If your LDAR program generates component-specific screening values, you can use Tier 2 or Tier 3 methods. These will almost always produce lower and more accurate emission estimates for well-maintained facilities. Without screening data, Tier 1 average factors are your only option.
Question 3: What does your regulatory context require?
State agencies vary widely. Texas (TCEQ) publishes specific fugitive guidance with its own factor tables. California (BAAQMD) has refinery-specific inventory guidelines. Consent decrees may specify which estimation method is acceptable. A facility I worked with switched from Tier 1 to Tier 3 during a permit renewal and reduced reported pump seal emissions by 80% — same seals, same leaks, different math.
A large refinery with 200,000 bolted-flange connections reduced fugitive emissions from 689 tons/year to 87 tons/year by upgrading to certified low-leak sealing technology — saving $682,000 annually. Using API 682 seal selection to specify dual seals with barrier fluid can shift your equipment from light liquid to effectively heavy liquid emission factor categories — but only if your estimation method recognizes the distinction.
Your Next Calculation Step
Match your method to your data. If you have screening values, use correlation equations — they’re more work per component but produce defensible numbers that reflect actual facility conditions. If you lack screening data, use Tier 2 leak/no-leak factors over Tier 1 averages whenever possible.
Before your next submission, verify three things: that your emission factors match your industry classification (SOCMI vs refinery), that your analyzer’s peg value isn’t inflating pegged source rates by orders of magnitude, and that your state agency hasn’t updated its preferred methodology since your last filing.
The API 682 standard specifies dual seals with barrier fluid for a reason — and that reason extends beyond reliability into emission accounting. A seal upgrade that moves equipment from the 0.114 kg/hr light liquid factor to the 0.012 kg/hr non-leaking category doesn’t just reduce actual emissions. It restructures your entire emission inventory calculation, and every permit and fee schedule that depends on it.
