CO2 system: leaking cylinder valves release the whole bank (Eddystone, Red Eagle)
EL-0130 · Solved · rating 0 · occurred 08 Jun 2016
Source: /cases/el-0130
What happened#
Two UK ro-ro ships, same failure: nobody touched the CO2 release, but a bank emptied itself into the CO2 room.
Eddystone was in the Southern Red Sea, ambient 32 °C. The CO2 release alarm sounded with the remote release undisturbed; all 65 bank A cylinders were open. Red Eagle was 15 minutes out of East Cowes. All cylinders except the two masters released. On both ships the ER distribution valve held.
Diagnosis#
- Eddystone: Schmöle valve set screws must be at
10–11 Nm; no torque records. On a sister ship (June 2017 release),124of130set screws were not torqued to11 Nm. - CO2 room above the critical temperature (
31.1 °C): at32 °Ccylinder pressure would be over76 bar. 8check valves and the non-return valve to the trigger line were leaking.- Leakage alarm switch: range
0–10 bar, but set at12 bar. - Red Eagle: manifold connected directly to the trigger line, no vent valve. Actuators started opening valves at only
4.14 bar(maker's minimum:31 bar). - Nitrogen leak test of
26valves at50 bar:2leaked. One had a very slow leak, brass particles in the seat and wire-wheel pitting on the bonnet; the other was damaged by the release. The valve that started it was not identified. - Danfoss KPS 33 leakage switch (
0–3.5 bar): set at0 bar, diaphragm distorted.
Root cause#
One or more cylinder valves leaked into the manifold. On Eddystone the gas passed a leaking non-return valve into the trigger line. On Red Eagle the trigger line was part of the manifold. At trigger pressure the other cylinders opened one after another.
The leaks were most likely due to poor service work. Eddystone's set screws were probably not checked. Red Eagle's DAB valves had been refurbished by service suppliers with a wire wheel or similar, against the maker's instructions. On Eddystone, wet, oily service air (about 7 bar) used to blow through and test the system in 2015 likely helped the check valves fail early.
Timeline / response#
| Time | Event |
|---|---|
| 1625 (8 Jun 2016, UTC+3) | Eddystone: CO2 release alarm. Loud hissing in CO2 room. |
| — | C/E and C/O enter with BA: bank A open, bank B closed. |
| — | Pilot lines disconnected. Bank A leaks from failed hoses for 4 days. |
| ~0435 (17 Jul 2017, UTC+1) | Red Eagle: CO2 release alarm. Nobody allowed into CO2 room or ER. |
| 0515 | Berths; passengers landed. |
| 0630 | Technician in BA removes all actuation heads. |
Lessons#
- Cylinder valves are not "fit and forget". Check Schmöle set-screw torque (
10–11 Nm) yearly and record it. - Trace your system on the drawing. Can a manifold leak reach the trigger line? Check vent and non-return valves.
- Never blow through or test CO2 lines with service air. Use nitrogen or dry air, as MSC.1/Circ.1318 says.
- Check the leakage alarm switch against the drawing. It must alarm at the lowest possible manifold pressure, within its range.
- Heat raises the risk, maintenance removes it. Above
31.1 °Ccylinder pressure climbs fast and an under-torqued valve leaks more easily. But Red Eagle released at18 °C. - If a bank releases itself, stop the cascade. Disconnect pilot lines or remove actuation heads (BA only).
Frost on unions, hissing or a leakage alarm means CO2 in the room. Keep out:
10 %CO2 causes unconsciousness; at17 %you have under a minute.
Source: MAIB report 16/2018 — Unintentional release of carbon dioxide from fixed fire-extinguishing systems on ro-ro vessels Eddystone and Red Eagle. © Crown copyright 2018. Contains public sector information licensed under the Open Government Licence v3.0.
Comments · 0
No comments yet.