Introduction
A verifier does not test the total. It tests the path from each source to each reported tonne. Most oil and gas emissions inventories start life as a corporate reporting exercise: fuel used, a factor from a table, a total in a spreadsheet. That is adequate until someone asks how a particular tonne was calculated, and the question now comes from verifiers working to ISO 14064-3, from emissions trading schemes, from methane regulation and from buyers of the gas.
The engineering behind a defensible inventory is not exotic. Flaring, venting and fuel combustion are governed by the same gas compositions, meter reference conditions and depressuring behaviour that process engineers work with every day. The discipline is in keeping that engineering attached to the number.
This post walks from the source register to the reported tonnes for a single asset, and ends with what a verifier will ask to see.
Start with the boundary and the source register
Two decisions come before any arithmetic.
The first is the boundary. The GHG Protocol Corporate Standard and the IPIECA/API/IOGP sector guidance give three consolidation approaches: operational control, financial control and equity share. Operators commonly report operated assets on operational control and also need an equity view for partners. Write the choice down, per asset, with the date it applies from.
The second is the source register: the list of every emission source inside the boundary, built from the P&IDs, the equipment list, the metering philosophy and a walk of the plant. A verifier checks completeness against it, so it deserves the same document control as a line list. A useful register row carries at least the following fields:
| Field | Why it matters |
|---|---|
| Source ID and tag | Ties the source to the P&ID and equipment list |
| Category and type | Combustion, flaring, venting or fugitive; then turbine, HP flare, tank vent and so on |
| Gases emitted | CO₂, CH₄, N₂O, and HFCs or SF₆ where relevant |
| Activity data and meter | Which meter or count feeds it, with units and reference conditions |
| Gas analysis | Which sample point and how often it is analysed |
| Method and tier | The calculation approach and whether it is measured, calculated or factor-based |
| Factor and source | Edition, table and unit of any emission factor used |
| Owner | Who supplies the data each month |
A register built only from the meter list will miss everything that is not metered: tank flash, glycol and amine vents, compressor seal leakage, pneumatic devices, purge and pilot gas.
The four Scope 1 source families
For an upstream asset almost every Scope 1 tonne falls into one of four families, each with its own physics.
- Combustion. Gas turbines, reciprocating engines, fired heaters, boilers and diesel engines. CO₂ follows directly from the carbon in the fuel, so a fuel-gas meter and a gas analysis give a carbon mass balance. CH₄ and N₂O are small and are normally estimated with energy-based factors.
- Flaring. Routine and non-routine flaring, purge and pilot gas, well test flaring. CO₂ comes from the carbon in the flared gas, scaled by the combustion efficiency, plus any CO₂ already in the gas. Uncombusted methane is the fraction that slips through.
- Venting. Cold vents, tank and loading losses, blowdowns, compressor seal vents, gas-driven pneumatics, the glycol regenerator still vent and amine acid-gas CO₂. Venting is whole gas, so the methane content of the stream matters more than anything else.
- Fugitives. Unintended leaks from components such as valves, connectors, seals and flanges, quantified from component counts and factors or from leak rates measured in leak detection and repair (LDAR) surveys.
Refrigerant, SF₆ and fire-suppression losses are a separate line in the register, quantified by mass balance on top-ups rather than as component leaks.
Scope 2 is simpler but has its own trap. Purchased electricity, including power from shore, and purchased heat or steam are reported both location-based (grid average factor) and market-based (contractual instruments and residual mix) under the GHG Protocol Scope 2 Guidance. Keep both; different programmes ask for different ones.
Quantification methods under the API Compendium
The API Compendium of Greenhouse Gas Emissions Methodologies for the Natural Gas and Oil Industry (2021 edition) is the sector's reference for methods and factors. For most sources it offers a ladder of approaches from direct measurement down to generic factors. Use the best approach your data supports for the sources that matter, and record which approach was used. Typical approaches by source:
| Source | Typical higher-quality approach | Typical fallback |
|---|---|---|
| Fuel gas combustion | Metered fuel × carbon content from gas analysis | Metered fuel × default factor per energy |
| Diesel combustion | Measured volume × density × net calorific value × factor | Purchased volume × default factor |
| Flaring | Flare meter × composition × combustion efficiency | Estimated volume from mass balance × default factor |
| Blowdown events | Pressure, temperature, isolated volume and compressibility | Event count × factor per event |
| Pneumatic devices | Measured bleed rates | Device count × factor × hours in service |
| Equipment leaks | LDAR measured leak rate × leak duration | Component count × average factor |
| Tank flash | Process simulation or measured vent flow | Throughput × default factor |
The method has to match the data you actually hold every month, and it is a versioned thing: if the factor table changes, or a source moves from counts to measurement, last year's results keep the method they were calculated with.
OGMP 2.0 levels: measured versus estimated methane
OGMP 2.0, the UN Environment Programme's Oil and Gas Methane Partnership framework, gave the industry a shared vocabulary for how good a methane number is. It applies to methane only, but the same thinking serves the whole inventory. Its five reporting levels:
| Level | What is reported | Typical basis |
|---|---|---|
| 1 | A single consolidated figure for the venture | Whatever is available |
| 2 | Emissions by broad emission category | Generic emission factors |
| 3 | Emissions by source type | Generic, source-type emission factors |
| 4 | Emissions by individual source | Source-specific factors and direct measurement |
| 5 | Level 4 reconciled with site-level measurement | Aerial, drone, satellite or continuous monitoring |
Member companies commit to reporting at Level 4 or 5 within three years for operated assets and within five years for non-operated assets, and the EU Methane Regulation now writes a broadly similar measurement expectation into law for operators in the EU. For the inventory engineer the practical consequence is a tier recorded per source: is this number measured, calculated from site data, or a factor applied to a count? A portfolio view of that mix shows where the next measurement campaign will do most good.
Level 5 adds a step that is easy to get wrong: when a site-level survey disagrees with the bottom-up inventory, the difference has to be investigated and any adjustment recorded, not quietly absorbed.
Measured does not automatically mean better. An ultrasonic flare meter at the bottom of its range during low purge flow can be less accurate than a good engineering estimate, so record the meter's range, calibration status and how gaps were filled.
Gas analysis is an input, not an attachment
For combustion, flaring and venting, the gas composition sets the answer; for vented gas the methane mole fraction is the emission.
Treat each analysis as data with a scope:
- Which stream and which period. Record the sample point, the sample date and the months it is applied to.
- Normalisation. Compositions should sum to 100 mol% within a stated tolerance. A lab report that sums to 98.7 mol% because hydrogen sulphide or water was reported separately needs a decision, not silent rescaling.
- Reference conditions. Volumes must be converted to moles before the composition can be used. Standard cubic metres (15 °C, 101.325 kPa, the ISO 13443 basis) and normal cubic metres (0 °C, 101.325 kPa) differ by about 5.5%, and US standard conditions are 60 °F. A meter configured on one basis while the spreadsheet assumes another produces a silent error on every tonne from that source.
- Mole basis throughout. Carbon balances work on moles of each component.
As in fiscal and allocation metering, a number without its reference conditions and its sample is not a measurement.
Worked example: one month of flaring
Scenario (illustrative). An HP flare meter records 250,000 Sm³ for the month, at 15 °C and 101.325 kPa. The monthly analysis of the flare gas, in mol%, is:
| Component | C1 | C2 | C3 | iC4 | nC4 | C5+ (as nC5) | CO₂ | N₂ |
|---|---|---|---|---|---|---|---|---|
| mol% | 80.0 | 8.0 | 5.0 | 1.0 | 1.5 | 0.5 | 3.0 | 1.0 |
Combustion efficiency is taken as 98%, the default destruction efficiency used in the API Compendium and in US EPA Subpart W. It should be justified for the actual flare, because tip type, steam or air assist, crosswind and operation at low turndown can all reduce it.
Molar volume at 15 °C, 101.325 kPa (ideal gas)
V_m = R·T / P = 8.314 × 288.15 / 101.325 = 23.645 m³/kmol
Z taken as 1.000 (about 0.997 for this gas, a 0.3% effect,
well inside meter uncertainty)
Moles flared
n = 250,000 / 23.645 = 10,573 kmol
Hydrocarbon carbon per mole of gas
C_HC = 0.80×1 + 0.08×2 + 0.05×3
+ (0.01 + 0.015)×4 + 0.005×5 = 1.235
CO₂ emitted (combusted carbon plus CO₂ already in the gas)
CO₂ = n × (η × C_HC + y_CO₂) × 44.009 / 1000
= 10,573 × (0.98 × 1.235 + 0.03) × 0.044009 = 577.1 t
CH₄ emitted (uncombusted methane slip)
CH₄ = n × (1 − η) × y_C1 × 16.043 / 1000
= 10,573 × 0.02 × 0.80 × 0.016043 = 2.71 t
Summary for the month (250,000 Sm³ at 15 °C, 98% combustion efficiency, N₂O excluded):
| CO₂ (t) | CH₄ (t) | CO₂e, AR5 (t) | CO₂e, AR6 (t) |
|---|---|---|---|
| 577.1 | 2.71 | 653.1 | 658.0 |
Of the 577.1 t of CO₂, about 14.0 t is CO₂ that was already in the gas and passed through the flame; the rest is combustion. Uncombusted ethane and heavier components are not greenhouse gases in this accounting. N₂O from flaring is small and is normally added with an energy-based factor; it is left out here to keep the arithmetic visible.
Two sensitivities are worth showing whoever owns the flare:
- Reference conditions. If the meter were actually reporting normal cubic metres and the calculation treated them as standard, every result above would be about 5% low (22.414 / 23.645 = 0.948, so 5.2% low). The opposite mistake, standard cubic metres treated as normal, overstates by 5.5%.
- Combustion efficiency. At 95% instead of 98%, CO₂ falls slightly to 559.9 t but methane slip rises to 6.78 t. With an AR5 methane GWP of 28, total CO₂e rises from 653.1 t to 749.9 t, about 15% more. Poor flare performance increases reported emissions, which is why methane programmes now ask operators to justify the efficiency they use.
From gas mass to CO₂e: state your GWP
Store results as mass of each gas. Convert to CO₂e only when reporting, and state the GWP set every time, because different programmes still use different IPCC assessment reports. The 100-year values (dimensionless, t CO₂e per t of gas) are:
| Gas | AR4 GWP₁₀₀ | AR5 GWP₁₀₀ | AR6 GWP₁₀₀ |
|---|---|---|---|
| CO₂ | 1 | 1 | 1 |
| CH₄ | 25 | 28 | 29.8 (fossil) |
| N₂O | 298 | 265 | 273 |
For the worked example, the 2.71 t of methane is 76.0 t CO₂e at AR5 and 80.9 t CO₂e at AR6, giving the monthly totals of 653.1 t and 658.0 t CO₂e. Neither is wrong; an unlabelled figure is. AR6 also distinguishes fossil methane (29.8) from non-fossil methane (27.0). Keeping per-gas masses lets the same data be reported under AR5 to one programme and AR6 to another.
The evidence trail a verifier asks for
A verifier working to ISO 14064-3, or under an emissions trading scheme's verification rules, will typically review the boundary and the register, then select a sample of results and re-perform them. Have this ready before they ask:
- Source register under revision control, with a completeness check against P&IDs and the previous year.
- Activity data as exported from the meter or historian, with units, reference conditions, calibration records and a log of how gaps were filled.
- Gas analyses with sample point, date and the months each was applied to.
- Factors with their publication, edition, table and units, and the reason for choosing them.
- Method statements per source: the formula, its version, its tier and any assumptions such as combustion efficiency.
- Review, approval and change records: who approved each period, and restatements with their reason.
- Uncertainty assessment, at least by IPCC Approach 1 (error propagation), with its coverage stated.
If a result cannot be re-performed from this pack by someone who did not build the spreadsheet, the trail is broken somewhere. It is the same discipline that applies to annulus pressure management in well integrity: a status or a tonne is only as good as the record of how it was decided.
Common errors
- Reference conditions mixed up (see above). No total-level check will catch it.
- Double counting at the flare. A blowdown calculated as a venting event and also measured by the flare meter appears twice. Decide where each event is counted and record it.
- Missing purge and pilot gas because they are not on the flare meter, or counting them twice because they are.
- Stale compositions applied to a stream whose make-up changed with the operating mode.
- CO₂ in the feed forgotten, or counted again as combustion CO₂ (see above).
- Heating value basis mixed. Energy-based factors on a net calorific value basis applied to gross energy data, or the reverse, misstate CH₄ and N₂O by around 10%.
- Blank treated as zero. A missing month is not zero activity; it is a gap to be filled and disclosed.
- Edits to closed periods made in place, so last year's reported number can no longer be reproduced.
Where EmitLedger fits
We built EmitLedger, FEEC's Scope 1 and 2 emissions accounting software for oil and gas, around the approach described above. Each supported source is calculated with a versioned, source-specific method (flaring, for example, from volume, gas composition and combustion efficiency), gas volumes carry their reference conditions and are converted to moles explicitly, and each calculated result retains its activity data, any gas analysis used, the emission factor and its reference, the method version and the library release. Blank means "no data" and zero means zero activity, a distinction kept throughout. Results are stored per gas, and CO₂e is calculated at reporting time with the AR4, AR5 or AR6 GWP set selected for that year.
Reporting periods move through submit, review, approve and lock, and a change to a locked period goes through a restatement. For OGMP 2.0 Level 5, site-level measurement results are compared with the source-level inventory for the same asset and period, and adjusting the inventory always goes through a restatement. External verifiers can be given read-only, time-limited access. EmitLedger is designed to support verification to ISO 14064-3; it does not replace your verifier, and FEEC is not a verifier and does not provide assurance. There is more on our EmitLedger page.
Conclusion
A defensible inventory is not a bigger spreadsheet. It is a complete source register, a method per source that suits the data held, gas analyses and reference conditions treated as inputs, per-gas results with a stated GWP set, and periods that stay closed unless a recorded restatement opens them. Then, when a verifier picks a tonne at random, you can show where it came from.
If you are setting up a source register, preparing for a first verification or trying to work out which flaring and venting sources are worth reducing first, talk to us. Related reading: flare network sizing, emergency depressurisation and blowdown and fuel gas, flash and vapour recovery.
