Ask three engineers what the first alarm stage should be on a methane detector and you will get two answers: 10% LEL and 25% LEL. Both are defensible, and both appear in real installations. The confusion comes from treating them as alternatives when they belong to different layers of the same scheme.
Two stages, two purposes
A combustible gas detector normally carries at least two alarm thresholds. The lower one exists to make the plant aware of a trend. The upper one exists to trigger action: shut a valve, start extraction, stop a line.
The widely used figure of 10% LEL belongs to the first role. It is a pre-alarm or early warning, set well below the level at which anything must happen, chosen so that operators have time to investigate before the situation becomes serious.
The figure of 25% LEL belongs to the second role, and in China it is the statutory first-stage alarm under GB 50493, the design standard for combustible and toxic gas detection in petrochemical installations. The same standard sets the second stage at or below 50% LEL.
So an installation can legitimately show three thresholds: 10% as an internal early warning, 25% as the regulated first stage, and 50% as the second stage. Nothing is contradictory. The 10% value is an operating decision made by the plant; the 25% and 50% values are a compliance floor.
Where the 25% figure comes from
The lower explosive limit is the concentration at which a gas becomes flammable. Below it, there is not enough fuel. At 25% LEL, a methane atmosphere is at roughly 1.25% by volume, a quarter of the way to the point where an ignition source would matter.
The margin between 25% and 100% is the time and space in which a response can still change the outcome. Setting the alarm closer to the limit removes that margin. Setting it far below wastes it on events that were never going to develop.
The 10% figure has a similar logic at a different scale. It corresponds to a reading that is unusual but not yet actionable, which is exactly what an early warning should be.
Toxic gases follow a different rule
Combustible gases are measured against a physical threshold that is identical for every site. Toxic gases are measured against an occupational exposure limit, which is a health-based value and varies by jurisdiction.
The two families of limit in common use are the Chinese GBZ 2.1 occupational exposure limits and the ACGIH threshold limit values used widely outside China. They are derived differently and do not always agree, so a setpoint copied from a foreign project may not satisfy a domestic inspection.
The usual structure sets the first stage at the exposure limit and the second at twice that value. For gases with no established limit, the conventional fallback is a fraction of the immediately dangerous to life or health concentration: 5% IDLH for the first stage, 10% for the second.
Oxygen is a third case
Oxygen has neither an LEL nor an exposure limit in the usual sense. It has a normal range and two ways to leave it.
The conventional thresholds are 19.5% by volume for deficiency and 23.5% for enrichment. These numbers are consistent across OSHA, EU practice and Chinese standards, which makes oxygen the one channel where an imported setpoint is usually safe to reuse.
Enrichment matters as much as deficiency and is more often forgotten. An oxygen-enriched atmosphere makes clothing and hair burn violently, and a detector configured to alarm only on the low side will not report it.
Why setting alarms too low makes a plant less safe
This is the counter-intuitive part, and it is well documented in safety management literature.
A detector that alarms at every minor fluctuation trains operators to treat the alarm as noise. The response shifts from investigation to acknowledgement, then to bypassing the channel during shifts where it keeps going off. Once a channel is bypassed, a real event on that channel produces nothing at all.
The failure mode is not the alarm threshold itself. It is the loss of credibility that precedes the bypass.
This is the argument for the two-layer scheme rather than for lowering both thresholds. A pre-alarm that nobody acts on is still useful information in a trend log. A first-stage alarm that nobody believes is a safety system that has stopped working.
Setting the thresholds in practice
Four inputs determine the numbers:
- The applicable standard for the site, which sets the compliance floor for stage one and stage two.
- The gas in question, combustible, toxic or oxygen, because each follows a different derivation.
- The ventilation and layout, since a well-ventilated outdoor area develops a leak far more slowly than a confined pit.
- The response the alarm is meant to trigger, because an alarm that starts a fan needs different timing from one that evacuates a building.
Alarm values are configurable on every detector and module in the range, and they are set at commissioning rather than fixed at manufacture. Where a project specifies values that sit below the applicable standard, that belongs in the technical review rather than in a discovery at handover.
Threshold values differ by jurisdiction and by project specification. The figures here describe the common convention; the values that apply to your site come from the standard named in your project documents.