BRENNEN

Industries / Industrial equipment and services / Industrial services / Leak detection

Fixed sensors

Fixed sensors are gas detectors, laser beams, sampling tubes and cameras installed in one place at a plant, a well site or a property line, which watch for leaks all the time instead of waiting for the next survey.

7

Milestones achieved, from 1958 to 2025.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1958

    A bead that burns gas

    A pellistor is a small porous ceramic bead coated with a catalyst and heated by a fine platinum coil. Flammable gas burns on its surface, the heat raises the coil's resistance, and the change shows how much gas is present. By the account of Ian Unwin, the UK Safety in Mines Research Establishment developed it in 1958 for coal mines, where it replaced fragile heated platinum wires. In 1974 a UK minister told Parliament that most instruments sold to give early warning of an escape of flammable gas, including one brought to the attention of chemical and gas plant operators, used a pellistor. Pellistors could lose their sensitivity for good after exposure to silicones, and their readings changed with the speed of the air flowing past them.

    LessonA fixed detector can fail without showing it, so it has to be checked with gas on a schedule rather than trusted because it is installed.

    pollutionsolutions-online.comapi.parliament.uk

    Credit

    UK Safety in Mines Research Establishment, Sheffield

  2. 2018

    Refineries watch their fence lines

    Under the Refinery Sector Rule that the US Environmental Protection Agency (EPA) issued in 2015, every US refinery rings its property line with passive diffusive tubes, small tubes that soak up benzene from the air for two weeks at a time, and reports each tube's result. For each two-week period the lowest reading is subtracted from the highest. If the average of that difference over the last 26 periods, a year, is above 9 micrograms per cubic meter, the "action level", the refinery must find the cause, such as leaking equipment, and fix it. Monitoring began in January 2018; EPA's dashboard says refineries began reporting to EPA in May 2019, and 118 refineries had reported by September 2021. EPA estimates that 85% of refinery benzene emissions are fugitive, meaning they escape from open sources rather than a stack.

    awsedap.epa.govoversight.gov

    Credit

    US EPA

  3. 2019

    Half the leaks went unseen

    Ryan Hilditch of the engineering firm WSP and John McNay of Micropack, a gas detection engineering company, analyzed the UK's records of offshore hydrocarbon releases from 1992 to 2015. Almost half, 48.5%, of the recorded gas releases were apparently not detected by the fixed gas detectors in place, while detectors spaced 5 m apart caught about 97% of simulated leaks in experiments. Of 27 real detector layouts from 18 facilities that they reviewed, fewer than half followed the "target gas cloud" method, which spaces detectors so that any cloud they miss is too small to cause a damaging explosion, and fewer than half of those met typical coverage targets.

    LessonThe authors asked that operators report their detector layouts with each release, because a detector network's record cannot be judged without knowing where the detectors were.

    elib.spbstu.ru

    Credit

    Hilditch and McNay, Ninth International Seminar on Fire and Explosion Hazards, 2019, using the UK's offshore release records

  4. 2023

    Eleven monitors, one test site

    A continuous monitor is a network of point sensors, or a scanning instrument, left at a site to flag leaks and estimate their size. At Colorado State University's Methane Emissions Technology Evaluation Center (METEC), a test site of real gas equipment with metered releases, Clay Bell, Chiemezie Ilonze, Aidan Duggan and Daniel Zimmerle tested 11 commercial solutions blind against releases of 0.4 to 6,400 grams of methane an hour. The release rate that each solution detected 90% of the time ranged from 3 to 30 kg an hour. False positive rates ranged from 0 to 79%, and at release rates of 0.1 to 1 kg an hour single estimates of leak size were off by between -97% and +2,077%.

    LessonThe authors wrote that each solution's performance should be well understood before anyone relies on its results.

    pmc.ncbi.nlm.nih.gov

    Credit

    Bell, Ilonze, Duggan and Zimmerle, Colorado State University; Environmental Science & Technology, April 2023

  5. 2023

    Exceedances that lingered

    EPA's Office of Inspector General reviewed the fence-line data from January 2018 to September 2021. It found that 13 of 118 refineries had benzene above the action level in 20 or more weeks after their first exceedance, that three of them had not brought their annual average back down, that five refineries had failed to report at least two weeks of data, and that EPA and state agencies had taken few formal enforcement actions. Some approved monitoring plans subtracted benzene from neighboring sources using modeling alone, which could hide an exceedance.

    LessonA fence-line number finds a problem only if someone acts on it; the Inspector General wrote that some refineries "may not have accurately identified the root cause".

    oversight.gov

    Credit

    US EPA Office of Inspector General, report 23-P-0030, 6 September 2023

  6. 2024

    No alarm, still leaking

    Z. Chen, Adam Brandt and colleagues at Stanford University tested eight commercial continuous monitoring systems blind against large releases meant to mimic venting from tanks, gas-driven controllers and unlit flares. All reported false positive rates below 10% second by second, and when a system reported an emission it was right more than 70% of the time. When a system reported no emission, it was right between 29.4% and 96.2% of the time, and the five systems that estimated daily release rates all underestimated them, by an average of 74.38%.

    LessonThe authors concluded that using these systems for emissions reporting or regulation "may be premature".

    eao.stanford.edu

    Credit

    Chen, El Abbadi, Sherwin, Burdeau, Rutherford, Chen, Zhang and Brandt, Stanford University; ACS ES&T Air, June 2024

  7. 2025

    Sensors stand in for rounds

    EPA's 2024 methane rule for new oil and gas sites (NSPS OOOOb) lets operators use approved detection technologies, under 40 CFR 60.5398b(d), in place of surveys on foot. In 2025 EPA approved fixed sensor networks for periodic screening: LongPath Technologies' network of open-path lasers, which send light across a site and measure the methane it passes through, on 21 February; Qube Technologies on 14 March; Sensirion Connected Solutions on 15 April; Project Canary on 16 April; and SLB's SOOFIE on 23 September. The approved thresholds run from under 1 to under 15 kg of methane an hour. LongPath's application cited a third-party validation study and its own published blind tests from 1 km away. EPA extended several of the rule's compliance deadlines on 31 July 2025, but kept approving methods: 22 in all, fixed, handheld, aircraft and satellite, by 17 November 2026.

    epa.govepa.gov

    Credit

    EPA Office of Air Quality Planning and Standards; the companies named

Watch for

  1. Thirteen thousand sensors

    TotalEnergies says that since 2025 it has run 13,000 sensors, including wireless gas detectors, fixed infrared cameras, flow meters and flare monitors, across all its operated upstream oil and gas sites, feeding a center in Pau, France, called MethaneLive, where about ten experts analyze the data and alert operators. It says methane emissions from its operated assets fell 65% from 2020 to 2025, to 22.5 kilotons.

    Achieved whenan outside party publishes the leaks the network found at a sample of these sites beside an independent survey of the same sites, or audits the 65% figure.

    totalenergies.com

    Claimed by

    TotalEnergies, May 2026

Open

  1. 01

    Find the units behind peaks

    A public analysis, refinery by refinery, that takes every two-week benzene result from every fence-line tube, pairs it with the hourly winds over the same two weeks, and estimates which part of the refinery the benzene most likely came from. For refineries that exceeded the action level, it sets that estimate beside the cause the refinery gave, where the cause is public.

    The brief
    Credit · unclaimed
  2. 02

    Ultrasonic listeners, measured

    A low-cost fixed ultrasonic leak listener, a microphone and small computer that listens for the hiss above human hearing that gas makes when it escapes a small hole under pressure, tested side by side with a commercial ultrasonic leak detector on a bench of calibrated compressed-air leaks. The result is a published curve, for each device, of how far away it detects each leak size against background noise.

    The brief
    Credit · unclaimed

Open 01 · the brief

Find the units behind peaks

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public analysis, refinery by refinery, that takes every two-week benzene result from every fence-line tube, pairs it with the hourly winds over the same two weeks, and estimates which part of the refinery the benzene most likely came from. For refineries that exceeded the action level, it sets that estimate beside the cause the refinery gave, where the cause is public.

Why it is still open

The tube-by-tube results are public, but each tube soaks up benzene for two weeks while the wind shifts, and the rule leaves finding the cause to the refinery. EPA's Inspector General found that 13 of 118 refineries stayed above the action level in 20 or more weeks after a first exceedance and that some "may not have accurately identified the root cause". In April 2026 Daniel Peters, Elizabeth Mohr, Lauren Padilla and Ramón Alvarez analyzed six years of results from 119 refineries and traced likely source areas at two of them, where hotspots kept recurring at the same tubes. EPA's own 2009 study at a Corpus Christi refinery found that downwind tubes read higher than upwind ones during six months of steady winds but not when the winds were mixed, so the method has known limits. No one has run the tracing at every refinery or checked it against causes the refineries documented.

awsedap.epa.govoversight.govapi.crossref.orgebi.ac.uk

Built with what exists

Monitor-level results from EPA's WebFIRE database and Benzene Fenceline Monitoring Dashboard; hourly winds from NOAA's Integrated Surface Database for the nearest airport; Python with pandas for wind-sector weighting; aerial photographs from the USDA's National Agriculture Imagery Program to mark process units and tank farms; pdfplumber to pull stated causes from published consent decrees and state reports.

  1. First step

    Download two years of tube results for one refinery that exceeded the action level, pair them with hourly winds from the nearest airport, and check whether the highest tube moves with the wind.

  2. Done looks like

    Every reporting refinery processed from 2018 to the latest quarter; for each refinery's highest 5% of two-week periods, a ranked compass sector with a confidence score; for at least 20 exceedances whose causes are documented in public records, the top-ranked sector contains the documented source in at least 70% of cases; code and tables published.

Open 02 · the brief

Ultrasonic listeners, measured

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A low-cost fixed ultrasonic leak listener, a microphone and small computer that listens for the hiss above human hearing that gas makes when it escapes a small hole under pressure, tested side by side with a commercial ultrasonic leak detector on a bench of calibrated compressed-air leaks. The result is a published curve, for each device, of how far away it detects each leak size against background noise.

Why it is still open

Hilditch and McNay found that about half of recorded UK offshore gas releases were apparently missed by fixed detectors. The closest low-cost work found is Senol Gulgonul's June 2025 detector built on an Arduino Nano 33 BLE Sense Rev2, whose abstract reports that it identified leaks on a pressurized pipe setup but gives no detection distances against measured leak rates, and a 2026 MEMS microphone array by Wang and colleagues that reports 94.67% detection accuracy on a laboratory platform, with no comparison to a commercial detector. No published, repeatable bench comparison of a low-cost listener with a commercial detector was found.

arxiv.orgelib.spbstu.ruebi.ac.uk

Built with what exists

The Arduino Nano 33 BLE Sense Rev2 board Gulgonul used, or a Raspberry Pi with a MEMS microphone that responds to ultrasound; a shop air compressor regulated to no more than 6 bar (87 psi); drilled precision orifices of known size; a thermal mass flow meter, such as Alicat's, to measure each leak; a rented handheld ultrasonic detector, such as UE Systems' Ultraprobe, as the reference; a loudspeaker playing recorded plant noise.

  1. First step

    Fit one orifice to a regulated compressor line and record ten minutes at five distances with the Arduino board and the rented detector.

  2. Done looks like

    Five orifice sizes at three pressures, at distances from 0.5 to 10 m and three noise levels, 20 trials each; for each device, the distance at which it detects 90% of leaks of each size, and its false alarms over 24 hours of noise with no leak; audio, code and data published.