BRENNEN

Industries / Industrial equipment and services / Industrial services / Industrial cleaning / Tank cleaning

Gas testing

Gas testing measures the air inside and around a tank before and during cleaning, for flammable vapor, oxygen and toxic gases such as hydrogen sulfide, and the job keeps that air in a safe range by venting it or by filling the tank with an inert gas that cannot support burning.

5

Milestones achieved, from Late 1920s to 2013.

0

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. Late 1920s

    A meter for vapor

    Oliver Johnson, working for the Standard Oil Company of California, developed a sensor to detect flammable mixtures in air in order to prevent explosions in fuel storage tanks. In such a catalytic sensor a heated platinum wire burns the gas, and the rise in the wire's temperature and resistance shows how close the air is to the lower explosive limit (LEL), the lowest concentration of a gas in air that can burn. By the account of the gas detector maker Interscan, a demonstration model was built in 1926 and a production model in 1927, and Johnson and Phil Williams founded Johnson-Williams Instruments in Palo Alto in 1928. The safety magazine ISHN also dates these sensors to the late 1920s and says they were developed against storage tank explosions. Johnson-Williams "Sniffer" gas indicators are held in the Science Museum Group collection in Britain.

    gasdetection.comishn.comcollection.sciencemuseumgroup.org.uk

    Credit

    Oliver Johnson, Standard Oil Company of California; Johnson-Williams Instruments

  2. 1969

    Washing jets wreck supertankers

    In December 1969 explosions in cargo tanks wrecked three supertankers. The Dutch-registered 207,000-ton Marpessa, whose crew was cleaning her empty tanks with jets of sea water, sank off Senegal on her maiden voyage; two weeks later a tank of the Mactra blew up in the Mozambique Channel, and the next day a blast tore the Kong Haakon VII off Liberia. Research into explosions during tank washing that followed, by Shell, the International Chamber of Shipping and university groups, pointed to static electricity: the spray breaks into a charged mist. John Hughes of the University of Southampton reported electric fields of up to 300 kV per meter in the center tanks of 200,000-ton tankers during normal washing with jets of up to 180 cubic meters per hour.

    LessonCleaning can create its own ignition source, so the air in the tank has to be kept from burning, not only kept free of sparks. 90022-4).

    content.time.comapi.parliament.uknature.comfiredoc.nist.govdoi.org

    Credit

    research by Shell, the International Chamber of Shipping, Edward Pierce and John F. Hughes, among others

  3. 1978

    Inert gas for washing

    The 1978 Protocol to the International Convention for the Safety of Life at Sea (SOLAS), the main treaty on ship safety, required an inert gas system, which fills cargo tanks with low-oxygen gas so that vapor cannot burn, on every tanker that cleans its tanks by crude oil washing and on tankers of 20,000 tonnes deadweight and above contracted after 1 June 1979 or delivered after 1 June 1982. The rule was tied directly to tank washing: older tankers of 20,000 to 40,000 tonnes could be exempted only if none of their washing machines pumped more than 60 cubic meters per hour. From 1 January 2016 the requirement was extended to new tankers of 8,000 tonnes and above. On land, API publishes separate guidance for work in inert confined spaces, Publication 2217A.

    traficom.filr.orgosha.gov

    Credit

    International Maritime Organization

  4. 2006

    A torch, not a meter

    At the Partridge-Raleigh oilfield in Raleigh, Mississippi, contract workers removed the crude oil residue from one production tank, flushed it with water and let it air for several days before welding a pipe to it. On 5 June 2006, to check that it was free of vapor, the welder put a lit oxy-acetylene torch into its openings, a practice called "flashing" that the US Chemical Safety Board (CSB) found was common. Sparks from the welding then lit vapor venting from an open pipe on the next tank, which had been neither cleaned nor isolated; the fire spread through a connecting pipe and a third tank exploded. Three contractors were killed and one was badly injured.

    LessonOnly an instrument can show that vapor is gone, and cleaning one tank does not clear the tanks piped to it.

    csb.govcsb.gov

    Credit

    investigated by the US Chemical Safety Board, report June 2007

  5. 2013

    One crew trusted another

    In March 2013, at a well site in British Columbia, Canada, a crew rinsed out with warm water a 400-barrel storage tank that had just been emptied of flowback fluid, the combustible waste that returns from a newly drilled gas well. No one tested the tank for combustible gas before the rinsing crew went in. Workers from a second firm, relying on a signal from the first, then lit a propane torch to melt ice at the tank's door; the vapor inside exploded, the tank top blew off and three workers were injured. WorkSafeBC, the province's workplace safety agency, found that the tank had not been ventilated and its vapor had not been kept below 20% of the lower explosive limit.

    LessonEach crew that works on a tank needs its own reading; a signal passed between crews is not a gas test.

    worksafebc.com

    Credit

    investigated by WorkSafeBC

Open

  1. 01

    Find the layer tests miss

    A scale model of a storage tank, such as a 2 m steel or plastic drum with a side manway and a roof vent, standing outdoors with no one inside, into which a harmless gas heavier than air, such as carbon dioxide at a few percent, is released at the floor and then ventilated the way tanks are degassed (a blower at the manway, an air mover on the roof). An array of 30 or more sensors records the concentration at every height and radius while common sampling plans are run against it: one reading at the manway, readings at manway and roof, and a probe lowered to the floor. The result is how often, and by how much, each plan misses a pocket of gas.

    The brief
    Credit · unclaimed
  2. 02

    Score meters on tank vapors

    An independent bench test of common portable gas detectors, with catalytic, infrared and photoionization sensors, under the conditions that tank cleaning brings together: heavy hydrocarbon vapor, low oxygen in a tank that was kept inert, and hydrogen sulfide, which can poison catalytic sensors. Each detector's reading is compared with the known concentration of a certified gas mixture.

    The brief
    Credit · unclaimed

Open 01 · the brief

Find the layer tests miss

Credit · unclaimed

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

What to build

A scale model of a storage tank, such as a 2 m steel or plastic drum with a side manway and a roof vent, standing outdoors with no one inside, into which a harmless gas heavier than air, such as carbon dioxide at a few percent, is released at the floor and then ventilated the way tanks are degassed (a blower at the manway, an air mover on the roof). An array of 30 or more sensors records the concentration at every height and radius while common sampling plans are run against it: one reading at the manway, readings at manway and roof, and a probe lowered to the floor. The result is how often, and by how much, each plan misses a pocket of gas.

Why it is still open

OSHA's tank cleaning guide says conditions around, on top of and below a tank may need monitoring and that test procedures should state their limitations, but it gives no figures for how often a reading at one point misses a layer. Inspection drones that fly into confined spaces, such as ScoutDI's Scout 137, carry a gas sensor that logs air quality during flight, but no published test compares such readings with a reference grid. A search found no measured comparison of sampling plans.

osha.govmeasur.ca

Built with what exists

Low-cost infrared carbon dioxide sensors, such as Sensirion's SCD41, on Raspberry Pi or Arduino loggers; a variable-speed blower; OpenFOAM, open-source software that simulates air flow, to compare the runs with simulation; Python for analysis.

  1. First step

    Build the drum with 12 sensors at four heights and run five tests with carbon dioxide released at the floor.

  2. Done looks like

    50 ventilation runs with at least three blower setups and two ways of releasing the tracer; for each sampling plan, the share of runs in which it reads less than half the highest concentration then present in the drum; data, drawings and code published.

Open 02 · the brief

Score meters on tank vapors

Credit · unclaimed

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

What to build

An independent bench test of common portable gas detectors, with catalytic, infrared and photoionization sensors, under the conditions that tank cleaning brings together: heavy hydrocarbon vapor, low oxygen in a tank that was kept inert, and hydrogen sulfide, which can poison catalytic sensors. Each detector's reading is compared with the known concentration of a certified gas mixture.

Why it is still open

Blackline Safety's guide to combustible gas detection says that catalytic sensors need at least 11% oxygen, are poisoned by sulfur compounds and lead, and are often fitted with filters that slow heavy hydrocarbons and lower their readings, while infrared sensors read accurately only for the gas they are calibrated to. The correction figures crews use come from guides that detector makers publish, and a search found no recent independent test of detectors under these conditions.

blacklinesafety.com

Built with what exists

Certified calibration gas mixtures held below 25% of the lower explosive limit, such as nonane in air for heavy hydrocarbons, methane in nitrogen for low oxygen, and hydrogen sulfide at a few parts per million; a sealed 20-liter test chamber with mass flow controllers inside a laboratory fume hood; five to ten detector models, bought or borrowed.

  1. First step

    Borrow three detectors and test them against one nonane mixture and one low-oxygen methane mixture.

  2. Done looks like

    At least three units of each model tested; each model's reading published as a share of the true value for each gas, at 21%, 15% and 10% oxygen, before and after a set hydrogen sulfide exposure; the method written so another laboratory can repeat it.