Methane and NOx emissions from natural gas stoves, cooktops, and ovens in residential homes

From the abstract:

We quantified methane released in 53 homes during all phases of stove use: steady-state-off (appliance not in use), steady-state-on (during combustion), and transitory periods of ignition and extinguishment. We estimated that natural gas stoves emit 0.8–1.3% of the gas they use as unburned methane and that total U.S. stove emissions are 28.1 [95% confidence interval: 18.5, 41.2] Gg CH4 year–1. More than three-quarters of methane emissions we measured originated during steady-state-off.

Note the wide confidence interval there. Still, it's quite a bit.

More about what they did:

We measured methane emissions from stoves in 53 homes in 7 California counties between January 2020 and May 2021 (Table S1). Our sample set included private homes, properties for sale or rent by real estate agents, and, because of the constraints of COVID-19, Airbnb rentals. We measured methane emissions from gas cooktops (18 unique brands) at all 53 homes and tested gas ovens and broilers in 40 and 31 homes, respectively. For stoves with information available, we found the ages of stoves ranged between 3 and 30 years. Cooktop burners ranged in heat output from 4500 to 25 000 British thermal units (BTU) h–1, and most oven burners ranged between 16 000 and 19 000 BTU h–1, as per product specifications.

It seems that some stoves are much more leaky than others when off:

Steady-state-on emission measurements were also long-tailed: the top 5 stoves (9%) emitted 51% of all steady-state-on emissions.

This is also true when they're turned on:

Similar to the steady-state emissions, these on/off pulses exhibited a long-tail distribution driven predominantly by a few cooktops─even individual burners within cooktops─that emitted most of the methane. The highest emitters were the cooktops that ignited using a pilot light (total of eight burners); these burners had an average on/off pulse of 258 mg CH4. One particularly large outlier─stove 19 (Figure 2C)─emitted on average 450 mg per pulse, nearly 4× higher than the second-highest emitter. This particular stove was one of the two stoves with a pilot light; on this specific unit, some burners took several seconds to ignite, while burners on other units with pilot lights took less time to ignite (thus releasing less methane), suggesting that delays in pilot light ignition may drive some emissions from on/off pulses.

They didn't find a simple rule for deciding which stoves leak the most. But it seems they don't have enough data to do that:

For the stoves in our dataset with a known manufacturing date, we found no evidence of a relationship between either the age of the stove or purchase price of the stove with methane emissions nor with NOx emissions normalized by burner energy (Figures S8 and S9; p values all >0.5).