Bursting discs vs flashback arrestors: why a relief disc cannot stop an acetylene detonation

A bursting disc and a flashback arrestor are both called safety devices, and both sit on acetylene lines, but they do different jobs and one cannot stand in for the other. A disc relieves pressure after it has risen. An arrestor stops a flame from travelling. Against an acetylene detonation, which moves faster than any disc can open, only the arrestor works. The test data show this plainly, and they also show where discs are genuinely useful.

The key measurements are by Miller and Penny, presented at the Institution of Chemical Engineers Symposium on Chemical Process Hazards, Manchester, 1960. A vessel 38 cm long and 11.5 cm in diameter was fitted at one end with a 5 cm bursting disc set to operate at 32 atm. That is a large disc for so small a vessel. Acetylene was exploded inside it, and the pressure near the disc was recorded.

In every case the pressure went well past the disc's 32 atm rating, to between 1.6 and 4.7 times the set pressure. The worst case was the one most like a real plant: an explosion arriving from a connecting pipe. As compiled by S. A. Miller (1965), the operation of a bursting disc had no retarding effect on a detonation, and its protective value was not great in deflagrations either.

A disc opens on a timescale set by its own inertia, the gas it must accelerate through the vent, and the vent area. A detonation front in acetylene travels at roughly 1,800–2,000 m/s in measured tests compiled by S. A. Miller (1965), with a sudden step in pressure at the front. By the time the disc has opened, the front has passed through the vessel and the peak has arrived.

The pressures involved are large. Deflagration raises pressure about 11 times; Miller and Penny measured 11.1 ± 0.6 in 16 runs in a 30 × 5 cm tube at 9–10 atm absolute. Detonation reaches 20–24 times in the front, two to two and a half times more on reflection, and 40–50 times the initial pressure as the destructive effect in a long tube. With cascading the rise can exceed 100 or 200 times, and Kogarko and colleagues (Khim. Prom., 1962) measured about 300 times at the end of 20 cm pipe. Carver, Smith and Webster (IChemE Symposium Series No. 33, 1972) measured more than 320 times in 12.7 mm tube, a lower bound, read from a disc that only just failed. The dead legs page explains how those extremes arise.

An arrestor works on the flame, not on the pressure. Fine passages present a large surface to the flame front and take heat out of it faster than the decomposition can supply it. For acetylene the passages must be very fine: the quenching diameter of the decomposition flame is about 0.6 cm at 2 atm absolute (Cummings, Hall and Straker, 1962), and detonation has been seen in tubes as narrow as 3 mm at 10 atm. The mechanism is set out on how a flashback arrestor works.

That changes the design case for everything downstream. A vessel that can only ever see a deflagration started inside it needs a test pressure of about 11 times its working pressure. A vessel that a detonation can enter through a pipe would need 40–50 times or more, which is not a practical design. The two conditions in the literature compiled by S. A. Miller (1965) are that no detonation can pass into a vessel through any connecting pipe, which requires arrestors, and that the vessel itself is too short for a deflagration inside it to become a detonation. IS 15190 (Part 1):2002 expresses the same thinking in its pipeline classes: Class 2 pipe is designed with a deflagration factor of 11, Class 3 with factors of 20 for undisturbed detonation and 35 where reflection must be resisted.

The same logic applies to the relief valve on a hydraulic arrestor or a back-pressure valve. The compilation by S. A. Miller (1965) describes the user-end hydraulic back-pressure valve and states that its relief valve is there to prevent pressure build-up from causes other than explosion and cannot relieve explosion pressures. The C.P.I. (Safety in the Production and Use of Acetylene, Paris 1968) adds that submerged non-return valves fitted with bursting discs must still be strong enough to withstand a flashback, allowing for the disc. In other words, the body must survive without help from the disc.

IS 11006:2011 follows the same principle for Indian arrestors. Clause 4.1 requires the casing to withstand ten times the maximum working pressure or 60 bar, whichever is higher, for at least 5 minutes, and clause 8 requires any safety valve or bursting disc that is fitted to be of full-flow design. The strength is in the casing; the relief device is supplementary. Testing is covered under flashback arrestor standards.