Fugitive emissions from equipment leaks are calculated per component: count the pump seals, valves, connectors and other components by type and service, then multiply each count by an emission factor, the stream’s total organic compound (TOC) weight fraction and the hours in service. For an annual VOC figure the calculation is:
Annual VOC emissions = number of components × emission factor × TOC weight fraction × (VOC wt % ÷ TOC wt %) × hours in service
Add the results across streams and equipment types for the unit total. Where the plant has Method 21 screening values, each reading replaces the average factor and goes through a correlation equation that turns ppmv into kg/hr.
The procedure and the factors come from EPA’s 1995 Protocol for Equipment Leak Emission Estimates.
Step 1: Count and Classify the Components
List every component in the process unit by equipment type and by service (gas/vapor, light liquid or heavy liquid), and group the components into streams whose TOC weight fraction is about the same. For each stream, record the TOC concentration, the VOC or HAP concentrations if you will speciate, and the hours each component spent in that service.
Pump seals, agitator seals and compressor seals are separate equipment types in EPA’s source list. Agitator seals have no factor of their own, so EPA estimates them with the light-liquid pump-seal factor.
Build the count in the field. Fix the process-unit boundary, walk each process stream from a simplified flow diagram, tag every source as you list it, then sweep the unit in a grid for anything the stream walk missed. Seals that are unsafe or difficult to screen still go in the count and are estimated with the average factor.
A liquid is in light-liquid service when the components with a vapor pressure above 0.3 kPa at 20 °C make up at least 20 % of it by weight. Gas/vapor service means the fluid is a gas at operating conditions, and everything else is heavy liquid. Texas guidance uses the same pressure, written as 0.044 psia at 68 °F, but applies it to the stream’s vapor pressure as a whole, with no 20 % test.
EPA’s chemical-plant (SOCMI) average factor for a pump seal is 0.0199 kg/hr in light-liquid service and 0.00862 kg/hr in heavy-liquid service.
Step 2: Choose an Estimation Method
Use the EPA correlation approach for every seal that has a Method 21 screening value, and average emission factors only where no screening data exist. EPA lists four approaches, in order of increasing refinement:
- Average emission factor: component count × factor, with no screening data. It covers new units until readings exist, and any seal or valve that is not monitored.
- Screening ranges: needs screening data. EPA keeps it mainly for analysing old data sets collected when 10,000 ppmv was the leak definition.
- EPA correlation: each screening value goes into an equation that predicts that component’s leak rate. EPA prefers it whenever actual readings exist.
- Unit-specific correlation: needs bagging data on top of screening values. EPA advises developing one only when its own correlations do not give reasonable estimates for the unit.
Regulators can override this choice. In Texas the correlation equations are mandatory when a permit condition, rule or commission order requires you to keep screening data, and TCEQ accepts them for the emission inventory but not for permitting.
South Coast AQMD allows average factors only where no inspection and maintenance program under Rule 1173 or Rule 1176 is in place and no reliable site screening data exist. It also does not allow the screening-ranges method to be combined with the correlation method.
Choose the factor set by process unit type. EPA publishes one set for chemical plants (SOCMI) and, for the petroleum industry, one correlation set but separate average factors for refineries, marketing terminals, and oil and gas production. Under Texas guidance, a chemical unit inside a refinery, such as an MTBE unit, takes the SOCMI set.
If you use average factors at a unit with a leak detection and repair (LDAR) program, EPA’s control-effectiveness tables reduce the light-liquid pump-seal estimate at a chemical plant by 69 % for monthly monitoring and 45 % for quarterly monitoring, both at a 10,000 ppmv leak definition. EPA’s refinery credits for the same two programs differ by one point or less. The tables have no row for annual monitoring.
Step 3: Take Screening Values at the Seals
Take each seal’s screening value with a Method 21 traverse around the shaft where it leaves the seal housing, with the probe inlet within 1 cm of the shaft-seal interface and held perpendicular to it. Where the reading rises, search slowly for the maximum, hold the probe there for about twice the instrument response time (which Method 21 caps at 30 seconds), and record that maximum as the screening value.
The same traverse applies to agitator and compressor seals. Where the housing blocks a full circle, sample every accessible part of it. Also sample the other housing joints where leakage could occur, and screen seal-system degassing and accumulator vents with the probe at about the center of the opening.
At a wet or greasy seal, keep liquid and grease out of the probe. Over an obviously dirty surface, hold the tip just above it.
Many analyzers top out at 10,000 ppmv. For an emission estimate, fit a dilution probe and calibrate the instrument to the higher range so readings above 10,000 ppmv are measured; without one, the reading is pegged and takes a pegged rate in Step 4.
Check the response factor of the sealed fluid for your instrument and calibration gas at 500 ppmv and at 10,000 ppmv. If both response factors are below 3, use the readings as they are; if either exceeds 3, multiply the affected screening values by the response factor before Step 4. Calibrating the instrument with the fluid in the seal makes the response factor 1.
Correct each reading for background and keep one screening value per component. Readings are never averaged before they go into the correlation.
Neither a visible drip nor a measured liquid leakage rate goes into the emission estimate. The emission figure comes from the Method 21 reading through the correlation in Step 4.
Step 4: Convert Screening Values to Emission Rates
Put each non-zero, non-pegged screening value into the pump-seal correlation equation for your factor set; a zero reading takes the default-zero rate, and a pegged reading takes a pegged rate. The pump-seal values, all in kg/hr per seal with SV as the screening value in ppmv, are:
| Factor set | Applies to | Correlation (kg/hr) | Default zero (kg/hr) | Pegged at 10,000 ppmv (kg/hr) | Pegged at 100,000 ppmv (kg/hr) |
|---|---|---|---|---|---|
| EPA chemical plant (SOCMI), light-liquid pump | Chemical plants | 1.90E-05 × SV0.824 | 7.5E-06 | 0.14 | 0.62 |
| EPA petroleum industry, pump seals | Refineries, marketing terminals, oil and gas production | 5.03E-05 × SV0.610 | 2.4E-05 | 0.074 | 0.160 |
| CAPCOA revised, pump seals | California petroleum facilities | 5.07E-05 × SV0.622 | 1.9E-05 | 0.089 | 0.610 |
EPA applies the SOCMI light-liquid pump correlation and default-zero rate to agitator seals, compressor seals and heavy-liquid pumps as well, and the SOCMI pegged rates to agitators and compressors. EPA’s correlations predict total organic compounds, and the petroleum correlation includes methane and ethane.
A reading at or below background counts as zero. The default-zero rates hold only when the analyzer’s minimum detection limit is 1 ppmv or less above background; with a less sensitive instrument, enter half its minimum screening value into the correlation instead.
The rule for pegged and very high readings depends on who regulates the unit:
- EPA: use a dilution probe and the correlation for readings up to 100,000 ppmv. The 10,000 ppmv pegged rate is for old data or where a dilution probe cannot be used.
- Texas (TCEQ): the correlation covers readings from zero to 100,000 ppm, and above 100,000 ppm for the emission inventory. An instrument pegged at 10,000 ppm takes the 100,000 ppm pegged rate, never the 10,000 ppm rate.
- California (South Coast AQMD): with the CAPCOA factors, the correlation covers background-corrected readings up to 9,999 ppmv, or up to 99,999 ppmv where the local district allows it. Without that permission, a reading of 10,000 ppmv or more takes the 10,000 ppmv pegged factor.
Worked for one pump seal at a chemical plant, read at 12,000 ppmv with a dilution probe:
1.90E-05 × 12,0000.824 = 0.0437 kg/hr (0.096 lb/hr)
Over 8,760 hours, a SOCMI pump seal reading 12,000 ppmv gives about 382 kg of TOC. The same 12,000 ppmv reading at a refinery gives 0.0155 kg/hr (0.034 lb/hr) through the petroleum equation. Under South Coast AQMD rules, without district permission above 9,999 ppmv, the 12,000 ppmv reading takes the 10,000 ppmv pegged factor of 0.089 kg/hr.
EPA’s factors and correlations are meant for a population of equipment, not for one piece over a short period such as an hour, because individual leak rates span several orders of magnitude. Each factor set has one pump row for packed and mechanical seals alike, so a packed pump takes the same equation.
Step 5: Annualize and Speciate the Result
Multiply each period’s leak rate by the hours the seal held process material in that period, add the periods into an annual TOC figure, then multiply by each compound’s share of the TOC (its weight percent divided by the TOC weight percent) to get VOC, HAP or single-compound emissions.
Count the hours the equipment contains material, not the hours the process ran. Texas inventory guidelines require 8,760 hours a year for every stream unless the lines were purged during the downtime.
When a seal has several screening values in the year, calculate each monitoring period from its own reading and sum the periods. EPA applies each reading to all the hours since the previous one. TCEQ also accepts the reading applied to half the monitoring period plus the repair time; any other method needs TCEQ approval.
Speciate each compound from the TOC result:
E(x) = E(TOC) × (WP(x) ÷ WP(TOC))
WP(x) is the weight percent of compound x in the equipment and WP(TOC) the TOC weight percent. Using the VOC weight percent for WP(x) gives VOC emissions, which removes methane, ethane and the other organics not classed as VOC, such as methylene chloride and 1,1,1-trichloroethane.
Credit for Dual Seals and Sealless Pumps
A pump with a dual mechanical seal whose barrier fluid is held above the pumped-fluid pressure, as in an API Plan 53A system, or a sealless pump, has an approximate control efficiency of 100 % in EPA’s tables. The credited estimate is:
Controlled emissions = uncontrolled emissions × (1 − control efficiency)
The same 100 % applies to a compressor dual seal with the barrier above the compressed-gas pressure. A credited pump stays in the component count.
Texas gives its own credits, including one for dual seals that EPA’s table does not cover:
| Pump arrangement | EPA approximate control efficiency | Texas (TCEQ) credit |
|---|---|---|
| Sealless: canned-motor, magnetic-drive or diaphragm pump | 100 % | 100 % |
| Dual mechanical seal, barrier above process pressure | 100 % | 100 % |
| Dual mechanical seal, barrier seal pot vented to a control device | About 90 % (closed-vent system) | 100 % |
| Any other dual mechanical seal | No figure | 75 % |
Texas permitting guidance lets a monitored dual seal take the LDAR monitoring credit in place of the 75 %. The 75 % credit applies to any component with double mechanical seals, agitators included.