Lean-burn natural-gas engines were built in part to suppress a pressing NOx problem. The tradeoff is methane that can survive combustion - and carry a climate penalty far larger than its concentration suggests.[5,12]

Why is this relevant today? On August 19, INNIO announced a planned expansion of more than $300 million at its Waukesha, Wisconsin site as demand for decentralized power grows, including from data centers. The machines are stationary natural-gas engines: familiar combustion hardware scaled into multi-megawatt power plants. Most fuel burns. Some methane survives and leaves through the exhaust.[1,2]

First, what is a stationary engine?

A stationary engine stays put. Its hardware looks familiar: pistons move in cylinders, a crankshaft turns, fuel and air enter, and combustion makes mechanical work.

The difference is the crankshaft's job. A car turns wheels. A stationary engine usually turns a generator or drives equipment such as a natural-gas compressor.

Take the Waukesha 16V-275GL+ in the material we reviewed. It has 16 cylinders and 285 liters of displacement, weighs about 52,400 kilograms, and at its 50-Hz rating produces 3.6 megawatts at 1,000 rpm. A passenger-car engine is measured in a few liters. This one is measured in hundreds.[2]

Stationary engines also live a different life. Many are built for continuous-duty service and can stay near a designed operating point for long stretches, making efficiency, reliability and service intervals especially important.[2,11]

We should disclose our interest. Endaris is developing catalyst technology intended to reduce methane in natural-gas-engine exhaust. We are early-stage and working toward field validation. We are pro-data-center and pro-responsible AI growth, and we expect natural gas to remain part of the power system for decades. Nothing below is a claim about our product's field performance.

Large stationary natural gas engines are often designed to run lean. Excess air lowers peak combustion temperature, suppresses NOx, improves knock margin and can support high efficiency. But cooler, more dilute combustion also gives a small fraction of methane more ways to escape oxidation. That trade is the story of methane slip.[5,11]

The fuel that survives the flame

In chemistry class, methane combustion is tidy: methane and oxygen become carbon dioxide, water and heat.

CH4 + 2O2 → CO2 + 2H2O

Real combustion chambers are not tidy. Piston-ring crevices, cool metal walls and mixture variations give the flame only milliseconds to finish the job.[3,4]

Methane can hide in crevices the flame cannot enter, quench near cool walls, or sit in pockets too lean to burn completely. Some mixture can also move toward the exhaust before combustion finishes.[3,4]

Some natural gas enters the engine, survives combustion and leaves through the exhaust still as methane. That is methane slip.

The phenomenon came before the name

Engineers understood the physics before they routinely called it methane slip. By 1970, researchers were describing wall quenching and combustion-chamber crevices as sources of unburned fuel. In 1978, Lawrence Berkeley Laboratory researchers filmed methane combustion at roughly 7,000 frames per second and measured the methane left unburned.[3,4]

The more revealing discovery came while engineers were solving the pollution problem of the time. NOx forms at high combustion temperatures, helps form smog and contributes to acid deposition. Running lean - with excess air - lowers combustion temperature and NOx. Engineers describe the air-fuel mixture with lambda, λ: 1 is stoichiometric; above 1 is lean and slightly below 1 is rich. [5,14,15]

In a 1980–1981 EPA-sponsored full-scale engine experiment, increasing air-fuel ratio cut corrected NOx from roughly 729 to 423 ppm. Methane moved the other way, from about 486 to 558 ppm—roughly a 15% increase.[5]

The researchers were solving NOx, not climate change. The fix worked: NOx fell while methane rose. Methane slip emerged as a side effect of solving a different pollution problem.

By the mid-1990s, researchers were identifying methane as the dominant hydrocarbon in natural-gas-engine exhaust. By 2010, a Norwegian government/MARINTEK report explicitly used "methane slip." We have not established who first coined the phrase. The physics came first; the terminology and climate accounting followed.[6,9]

Lean burn is not one methane number

A Colorado State University-led field campaign measured methane directly in operating natural-gas compressor engines. On a fuel-energy basis, the mean was 0.10 lb CH₄/MMBtu for four-stroke rich-burn engines and 1.15 lb CH₄/MMBtu for four-stroke lean-burn engines—an 11.5-fold difference.[7,8]

Even “lean burn” was not one number. The G3500-series mean was about 0.52 lb/MMBtu; the G3600-series mean was 1.41 lb/MMBtu, roughly 2.7 times higher.[7,8]

The CSU campaign did not test load as an experimental variable. Separate Norwegian marine measurements showed the effect of load: then-current lean-burn spark-ignited engines measured roughly 22.6-41.3 g CH4/kWh at 25% load versus 6.0-6.17 g/kWh at full load.[7,9]

Architecture, model, load, fuel, ignition, air-fuel control and maintenance all matter. The useful question is not "What is the methane-slip rate?" but "What is it for this engine, at this load, in this condition?"[7,9]

Why isn’t this already solved with a catalyst?

Cars trained us to expect a catalytic converter to clean up exhaust. Natural-gas engines complicate that.

The catalyst problem is exhaust chemistry. A rich-burn engine near stoichiometric can use a three-way catalyst to reduce NOx while oxidizing CO and hydrocarbons, including methane. Lean-burn exhaust contains excess oxygen and three-way catalysts cannot operate in these conditions. Therefore, lean-burn systems split the job: an oxidation catalyst handles CO and hydrocarbons; selective catalytic reduction (SCR) catalyst handles the NOx. Methane is harder. A 1994 study reported that conventional lean-burn oxidation catalysts needed above roughly 500°C for strong methane oxidation and the exhaust of lean-burn engines is significantly cooler than that, making them ineffective in reducing methane slip. [10,11]

Endaris is developing a full replacement for the existing oxidation catalyst in a lean-burn exhaust system. It is intended to do the oxidation catalyst's existing work on CO and hydrocarbons, including formaldehyde, while also oxidizing methane; SCR remains the separate NOx-control step described above. That is the gap we are targeting: methane that survives combustion but is difficult for conventional oxidation catalysts to convert at lean-burn exhaust temperatures. 

Small in the exhaust

A few percent of the fuel sounds large until excess air dilutes it. In the exhaust, the result is measured in parts per million.

Take a simple case: pure methane fuel, ordinary air and λ = 2. Let s be the fraction of feed methane that leaves unburned. The exhaust balance gives CH4 ppmv ≈ 1,000,000 x s / (1 + 9.52λ). Here 9.52 is the air balance: 2 moles O2 plus about 7.52 moles N2 per mole CH4 at λ = 1. Water vapor is included.

At λ = 2, 1% fuel slip is about 499 ppmv. Three percent is about 1,497 ppmv - just 0.15% of the exhaust by volume. Even 5% is only about 2,495 ppmv, or 0.25%. At 3%, roughly one methane molecule is mixed into 668 exhaust molecules.

This is not a capture-cost claim. The point is physical: slip is a dilute exhaust problem. Small concentration does not mean small climate impact.

Figure 1 - Illustrative methane concentration versus feed slip. Pure methane fuel, ordinary air, λ = 2, exhaust basis including water vapor. Calculated sensitivity curve, not a field measurement.

How much can a few percent change the carbon footprint?

Assume pure methane fuel. Every kilogram that burns produces 2.75 kilograms of CO2. The methane that escapes is weighted by its global warming potential:

Methane share of direct CO₂e = (GWP × s) ÷ [2.75 × (1 − s) + GWP × s].

Using IPCC AR6 fossil-methane values - 82.5 over 20 years and 29.8 over 100 - 1% fuel slip makes methane about 23% of direct CO2 + CH4 climate impact on the 20-year basis. At 3%, it is about 48%; at 5%, 61%. On the 100-year basis, 3% is about 25%.[12]

We lead with 20 years because this is a near-term methane argument; the 100-year result is shown beside it.

Three percent of the fuel did not become 48% of the exhaust. The methane that fails to burn simply carries far more near-term warming impact per kilogram than the CO2 produced when methane burns.

Figure 2 — Methane’s share of direct engine CO₂e versus methane fuel slip. Pure-methane sensitivity calculation using IPCC AR6 fossil-methane GWP20 = 82.5 and GWP100 = 29.8. Direct CO₂ + CH₄ only.[12]

Now look at a real engine specification

The Waukesha 16V-275GL+ lists typical laboratory emissions of 422 g CO2/bhp-hr and 4.73 g CH4/bhp-hr at its 50-Hz rating - almost 89 times more CO2 by mass. The 422 g CO2/bhp-hr is the CO2 coming directly out of the engine from fuel that burned. The 4.73 g CH4/bhp-hr is unburned methane that slipped through the engine. We use this engine because the public datasheet reports both gases on the same rated basis; it is an example, not a fleet average.[2]

At GWP20 = 82.5, 4.73 g methane becomes about 390.2 g CO2e, about 48.0% of combined direct CO2 + CH4 climate impact. At GWP100 = 29.8, it becomes about 141.0 g CO2e, about 25.0%.[2,12]

“But if you oxidize methane, you make CO₂”

Correct. One kilogram of methane has a 20-year warming weight of 82.5 kg CO₂e. Fully oxidizing it creates 2.75 kg CO₂, for a net reduction of 79.75 kg CO₂e.[12]

For every ton of CO2 created by fully oxidizing methane, about 29 tons of net CO2e are avoided on the 20-year basis. That is why methane-oxidation aftertreatment is worth pursuing: it targets the unburned fraction after the engine has made useful work. The reaction creates CO2; the net arithmetic still strongly favors oxidation.[12]

The number that belongs on the engine spec sheet

Operators already track NOx, CO, efficiency and load because those numbers affect permits, performance and maintenance. Methane should be measured with the same discipline.[11]

What would change our mind? Representative field data from current prime-power installations showing consistently low methane across actual load cycles. We do not have that evidence yet.

Until then, measure methane at the stack. Report kg CH4/MWh with engine model, fuel, load and operating condition. Distinguish measurements from generic factors and show both 20- and 100-year climate results.

A natural-gas meter tells us how much fuel the engine received. The exhaust tells us whether it finished the job.

References

1. INNIO N.V. “INNIO Expands Waukesha Site to Meet Growing Demand for Decentralized Energy Solutions in the U.S.” Press release, August 19, 2026. Source

2. INNIO Waukesha Gas Engines Inc. 16V-275GL+: Gas Engine for Power Generation. IWK-123020-EN, 2023. PDF

3. Daniel, Wayne A. “Why Engine Variables Affect Exhaust Hydrocarbon Emission.” SAE Technical Paper 700108, 1970. DOI

4. Ishikawa, Nobuhiko, and John W. Daily. “Observation of Flow Characteristics in a Model I.C. Engine Cylinder.” SAE Technical Paper 780230, 1978. DOI

5. Castaldini, C. Environmental Assessment of NOx Control on a Spark-Ignited, Large-Bore, Reciprocating Internal-Combustion Engine, Volume I: Technical Results. EPA-600/7-86-002a. U.S. Environmental Protection Agency, January 1986. Source

6. Stapper, Carole J. Methane Emissions from the Natural Gas Industry, Volume 11: Compressor Driver Exhaust. GRI-94/0257.28; EPA-600/R-96-080k. Gas Research Institute and U.S. Environmental Protection Agency, June 1996. PDF

7. Vaughn, Timothy, Benjamin Luck, Daniel Zimmerle, Anthony Marchese, Laurie Williams, Kindal Keen, Terri Lauderdale, Matthew Harrison, and David Allen. Methane Emissions from Gathering and Boosting Compressor Stations in the U.S., Supporting Volume 2: Compressor Engine Exhaust Measurements. Colorado State University Energy Institute, 2019. DOI

8. Vaughn, Timothy L., Benjamin Luck, Laurie Williams, Anthony J. Marchese, and Daniel Zimmerle. “Methane Exhaust Measurements at Gathering Compressor Stations in the United States.” Environmental Science & Technology 55, no. 2 (2021): 1190–1196. DOI

9. Nielsen, Jørgen B., and Dag Stenersen. Emission Factors for CH4, NOx, Particulates and Black Carbon for Domestic Shipping in Norway, Revision 1. MARINTEK Report MT22 A10-199 / TA-2746. Trondheim: Norwegian Marine Technology Research Institute, 2010. Source

10. Gluck, K. H., E. Löx, A. Schafer-Sindlinger, T. Kreuzer, and R. S. Muniz. “Catalyst Development for Stoichiometric and Lean Burn Natural Gas Engines.” SAE Technical Paper 942419, 1994. DOI

11. U.S. Environmental Protection Agency. Catalog of CHP Technologies: Section 2, Technology Characterization—Reciprocating Internal Combustion Engines. September 2017. PDF

12. Intergovernmental Panel on Climate Change (IPCC). Climate Change 2021: The Physical Science Basis. Working Group I Contribution to the Sixth Assessment Report. Cambridge: Cambridge University Press, 2021. Chapter 7, Table 7.15. Source

13. INNIO Waukesha Gas Engines. “Waukesha Powers Carbon Neutral Gas Storage Facility.” Enstor Gas case study, n.d. Source

14. U.S. Environmental Protection Agency. “Ground-level Ozone Basics.” Last updated February 18, 2026. Source

15. U.S. Environmental Protection Agency. “What Is Acid Rain?” Last updated February 13, 2026. Source

Calculation note. Methane-concentration and climate-impact calculations in this article are Endaris calculations based on stoichiometric methane combustion, the Waukesha emissions data cited above, and the IPCC AR6 fossil-methane global-warming-potential values cited above. Calculated values are illustrative and are not field measurements.