Testing and replacing flashback arrestors: the IS 11006 type tests annual checks and when a device is spent

A flashback arrestor is tested twice before it reaches the user, as a design and as part of a production batch, and then continually as an installed device. The design is proved by the type tests in IS 11006:2011, and each production batch by the factory tests the standard makes compulsory. Once it is fitted, nothing in the standard tells the user how often to check it, so the interval comes from regulators and manufacturers: inspect and test every 12 months, and replace after any flashback. The reason is simple. You cannot see inside a sintered element, and an arrestor that has stopped one flame is the arrestor most likely to fail on the next.

IS 11006:2011 cl 11 sets type tests on new devices with every safety function working. Tests are run at ambient pressure and 27 ± 5 °C, an Indian ambient rather than the 20 °C European convention, with air or nitrogen whose oil-vapour mass fraction is below 5 × 10⁻⁶ and suspended droplets below 1 mg/m³, dried to a dew point of 0 °C. Flow must be measured to ±10 % and pressure to ±3 %.

The non-return valve is tested in its most disadvantageous position, with gravity acting to open it, a detail that rules out testing in the easy orientation. The test mixtures differ by gas: LPG devices use 13–15 % propane in oxygen and hydrogen devices 40–50 % hydrogen.

IS 11006:2011 cl 12 is normative: a reverse-flow test on each device, and a pressure-resistance test and a flashback test on one sample from each lot of 100. This is stricter than ISO 5175-1:2017, whose production-test annex (Annex C) is informative only. A supplier claiming IS 11006 conformity should be able to show these records.

Because the conditions that defeat an arrestor develop after installation, and none of the type tests reproduce them. The failure modes reported by manufacturers and investigators are consistent: carbon from earlier flashbacks plugging the element; scale, rust and carbide dust; water in the element; oil in oxygen service; body corrosion; transport and handling damage; and a temperature cut-off that operated unnoticed. The worst is simple absence. In an Energy Institute case, a cutting torch lacked the required arrestors, trained operators did not notice, and a similar fire followed about seven weeks later. NTSB's 2009 special investigation of mobile acetylene trailer fires (SIR-09/01) recorded, among the equipment at one site, a flash arrestor on the compressors that had lost its packing, a device that looked fitted and protected nothing.

The C.P.I. made the same point in 1968 for plant devices: arresters "must be of approved type and tested periodically" (C.P.I., Safety in the Production and Use of Acetylene, Paris 1968), with at least one fitted on every cylinder charging rack between compressor and rack inlet.

A hydraulic arrestor is tested differently, because its quenching medium is the water level. The C.P.I. required the water in a low-pressure hydraulic valve to be checked every day and again after every oxygen blowback or flashback, with the inside examined after each flashback. IS 11006:2011 cl 6.3 requires that the liquid level can be filled, observed and adjusted without letting air into the circuit, and cl 9 that internal parts can be examined and maintained. EIGA Doc 239/23 names loss of water in the arrestor as a failure mode and sets a corrosion inspection every five years or after any shutdown of three months or more, because a water-seal vessel left standing corrodes. On KIGL generators the operation manual has the level checked at every hopper fill, the arrestors drained, flushed and refilled weekly, and drained and cleaned annually. The detail is in the wet arrestor guide.