• Paul Kroes, B.Sc.Combustible Gas Detection: Don’t Believe Everything You See and Hear!

    I was at a tank repair facility when the client put the combustible gas monitor sensor into the tank prior to welding. Upon receiving a 0% LEL reading, they wondered how it could be, when they were sure that there should be combustible gases in there.

    Combustible gas detectors are generally catalytic bead sensors. They measure in the 0-100% LEL range. The LEL (lower explosive limit) is basically the concentration of gases that could go ‘boom’ when they hit 100%, if there were an ignition source. Generally they alarm at 10%, giving you a wide safety factor.

    Bead CatalystA catalytic bead combustible gas sensor has a flame arrestor shell, inside of which you would find two thin filaments, one with a Pt/Pd (platinum/palladium) catalyst bead and the other with no catalyst. When power is turned on, both beads have electrical current running through them and they’re hot, which is why there’s a flame arrestor. The temperature of the bead, the filament length and thickness, the catalyst, the algorithms, and even flame arrestor material are manufacturer specific, resulting in differences between different brands of LEL sensors.

    In general, when a combustible gas permeates through the flame arrestor, it comes into contact with the filaments. Nothing happens on one filament, but on the catalyst, the gas catalyses (burns) on the filament. This causes that wire to heat up, changing the resistance of that wire. An electrical circuit called a Wheatstone Bridge measures the change in the resistance between the two wires and expresses this as an electrical output, displayed as Percent LEL (%LEL).

    The sensor doesn’t actually know if it is propane, gasoline, or a complex brew of combustible chemicals; it just detects one wire getting hotter than the other and so it increases the reading on the display.

    Technically, we calibrate the display to a known concentration of gas, like methane, propane or pentane, among other common calibration gases. Each gas reacts differently on a particular LEL sensor because of the chemical properties of the gas; if we calibrate it to methane, it is accurate on methane and relatively close on everything else combustible.

    But back to the beginning question – why didn’t my LEL sensor read a gas concentration when we knew gas was there? Basically, it didn’t work.

    A couple of scenarios come to mind:

    • Sample Time: Maybe the sample time wasn’t long enough. If a sample hose and pump are involved, you have to be sure that the sample is getting to the pump, that there are no hose leaks, and that you allow time for the sample to come through the hose to the sensor. It takes at least a second or two per foot of hose for most pumps to get a sample to the sensor before it can even start detecting the gas.
    • Hose Blockage: If it has a pump, is the flow to the sensor adequate? Did you check to see if you have flow? You always have to verify that you have air flow to the sensors.
    • Temperature: If it is really cold, many combustible gases won’t evaporate, so you can’t measure them
    • Vapour Pressure: Sometimes a flammable liquid really isn’t that flammable. Diesel fuel doesn’t have much vapour pressure at 20°C. If it doesn’t evaporate into a gas, we can’t measure it.
    • Sensor Reaction Time: Ideally a new sensor has a very short T90. That means 90% of the response to the target gas happens in a few seconds. This changes with age, exposure to gas and other contaminants. It never gets better over time, only worse. If it takes 60 seconds to reach T90, you might miss the gas being measured.
    • Big Molecules: Some molecules (more than nine carbons) are so large that they do not permeate the flame arrestor or the catalyst properly. Maybe all the ‘light’ ends or small volatile molecules are gone, so only the heavy sticky stuff remains, resulting in no reading.
    • Heavier than Air Gases: Some gases like propane sink out to low areas; others like hydrogen tend to rise. If you don’t know what might be there you have to sample high, middle and low -otherwise it might be safe where you test, but not somewhere else!
    • Sensor Poisons: Si (WD40) or H2S have the effect of coating the Pt/Pd catalyst so that it becomes ineffective; if the wire doesn’t heat up, the Wheatstone bridge doesn’t see the current differential and it will happily read zero LEL in a high gas concentration
    • Lack of Oxygen: If there is no oxygen present, then there is no ‘burning’ of the gas on the filament, so no display. We need at least 10-12% oxygen concentration for a catalytic bead sensor to work properly, which is why an LEL sensor always comes with an oxygen sensor.

    So what happened to our instrument?

    As it turned out, our sensor was ‘poisoned’ and it would have never seen the combustible gas, even though it was there. The poison prevented the active filament from heating up, so no temperature resistance differential on the Wheatstone bridge, hence no output on the display.

    What caused it to be poisoned? Nobody knows, but something happened to the instrument prior to this usage. Once we bump checked it, we realized that it could not see a combustible gas and would happily read zero all day long.

    This is a classic ‘false negative’ and is potentially very dangerous. It is one of the reasons that it is critical to BUMP or calibrate each gas sensor prior to use. You never truly know where it’s been, what it has been exposed to and if it will work when you need it.

    It is also why even the most sophisticated instrument and standard operating procedure is only as good as the people running it. If you don’t check, you don’t know.

    For more reading you review these Application Notes: AP1007 and AP1018 at http://goodforgas.com/support/#appnotes, courtesy of GfG Instrumentation.


    Paul Kroes, B.Sc. | Instrumentation Specialist
    EHS Instruments Solutions

  • Kyle Kelly

    Yup thats why we always bump test the monitors before every shift. Great tips on why it may not pick up the LEL’s though.