Safety Alerts 12

Relief

• RV: High discharge velocity/ ρv², led to tail pipe failure, gas release and local fire. In another case, led to auto-closing of discharge LO valve. Note: High velocity in tail pipes do not occur during group loads but when a RV/BDV alone opens with low backpressure. Support tail pipes properly

• RV: Quench column RV opened on Cooling Water failure. RV vibrations led to failure of bolted flanges. Local release and explosion, shutting steam to cracker furnace tubes. Loss of steam resulted in cracker furnace tubes rupture and release of quench oil. Quench oil pool fire damaged 4 cracker furnaces

• RV: Steam RV outlet had condensate accumulation. No drain or weep hole. When RV opened, water hit and damaged the tail pipe. WWW. Similar, where the weep hole was plugged / choked with rust. Note: In cold weather steam can freeze and block RV outlet BP

• RV: On ESD gas outlet SDV closed before inlet SDV, leading to blocked outlet. Large sluggish PCV responded slowly resulting in 3 RVs popping. PCV progressive opening reduced the flow to RV, resulted in chattering; chattering RV bellows failed. Local release

• RV: Cavern Gas Storage. MOC used good for screwed/ bolted application. Welding caused microscopic changes and corrosion cracks. Some RVs ruptured and making the caverns inoperable

• RD + RV: Leakage thru upstream RD led to higher backpressure on RD and hence over pressurization of the protected vessel. Monitor downstream pressure of RD regularly. Add PG/PAH

• RD: Can prematurely fail on fatigue and creep. Replace as recommended by manufacturer

• RD: on a 4" LPG pipeline to a storage tank ruptured during a ship to shore transfer. Fatigue failure. Not changed in service for 11 years Vs supplier recommendation to change every year

• RD: Failed RD on HX routed hot oil to flare, resulting in Flare KOD LAHH/PSD + gas release. High velocity relief gas impact on hot oil in the relief header led to water hammer damage of flare piping. Consider premature/ latent failure of aged RDs. Replace as recommended by manufacturer

• RD: Set at 2.1 bar, prematurely ruptured at 0.9 bar. Fatigue. Hot MDI chemical at 204°C was released and sprayed on workers as RD outlet was pointing down and ended 8" above ground. Discharge to safe location. CSB

• RD: Cooling Water (CW) RD failed on pressure surge when CW pump was started. Sent CW to flare

• SDV: Internal stop failed, shearing valve seats. Debris damaged downstream RV. Simultaneous failure of SDV and RV resulted in over pressurization. Avoid SDVs with internal stops; go for external stops

• PCV: Removed flare PCV without outlet isolation. Flare gases released locally. Fire - injury and death

Flare

• Header: Cold liquid and vapor release heated above MDMT of CS piping. Heater failure resulted in brittle fracture of CS piping. Local release and fire. Use SS piping

• Header: Prolonged flaring of Glycol Contactor HP gas and Slug Catcher wet gas led to hydrate blockage in HP flare header, pressurized it to 30 bar, twice 150# rating. Pockets in flare header accelerated blockage. Reverse rupture of Rupture Disks on seawater, used for cooling, sent gas to Seawater Caisson and Oil Production System. Bulk of the gas discharged overboard via Seawater Caisson, which bubbled to surface, formed a vapor cloud that set off gas detectors, triggering a full blowdown of the production train to the blocked flare. Fortunately, thanks to prevailing wind the vapor cloud didn’t ignite over the vessel that could have sunk it due to buoyancy loss. Do not mix wet and cold streams in the same flare header with pockets

• Header: Acid Gas. Sulphur and Iron Sulfide deposits can form as a result of SO2 + H2S slow Claus reactions. Blocked up to 80% flow area. Avoid routing SO2 streams to Acid Gas Header. Add RDs upstream of RVs in SO2 sources. Blind SO2 maintenance vents. Inspect and remove blockage

• Header: Liquid accumulation in flare header pocket poured out during maintenance. Flash/ pool/ jet fire. Avoid pockets in flare headers

• Acid Gas: 20 minutes flame out resulted in venting of stream with 16% H2S. Fatality. IChemE

• Seawater used for cooling. Seawater pump start-up surge burst RD and poured seawater into flare

• Header: Subsea header to remote flare tripod formed U pocket that got filled with carry over and condensed liquid that poured out of tip during flaring. Burning liquid floated on sea surface and drifted towards platform

• Water Seal Drum: Condensate accumulating on WSD was discharged to OWS with water. Large vapor cloud. Steam coil could have warmed, evaporated and sent it to flare

• KOD: Liquids carried at high velocity with gas dislodged internals - piping rupture/ local release/ fire

• KOD: Continuous HC liquid flow to a column with blocked outlet led to relief flooding unit Flare KOD. Entrained liquid in the flared gas led to liquid hammer that ruptured it's corroded outlet elbow. HC release, fire and explosion. Plant modification routing KOD liquids back to inlet instead of Slop Tanks as originally designed; Poor alarm rationalization and management

• KOD: Underground KOD floated up damaging attached piping after a violent storm and filled the underground basin. BP

• KOD: During an unscheduled shutdown of FCCU, a flammable mixture of air and residual tail gas reached the flare system. Explosion in the KOD. Probable source - pyrophoric scale. BP

• KOD: Plant start-up. Explosions in Water Seal Drum/KOD. Probable cause: Natural gas added to enrich flared gas + inadvertent routing of hot air resulted in combustible mix with a higher flash back velocity than stack tip velocity. First explosion damaged WSD hat and its effectiveness. On 2 restarts, ineffective WSD resulted in explosions in KOD. Inject enrichment gas after WSD. Avoid hot air

• KOD: Refinery start-up. Column overhead drum LT failure. Light HC liquid drained to flare. Flare KOD and WSD flooded blocking gas release to flare. PAH + sooty flame. Flare KOD LAH not functional and its electrical pump inadequate. Standby steam driven pumps not operational. Check liquid relief scenarios and size Flare KOD per API

• KOD: Ethylene plant start- up. Cold liquid hydrocarbon drained flooded the Flare KOD and flowed into the flare stack. Stack failed due to low temperature embrittlement. Both the independent LTs in the KOD failed as their top tappings were mistakenly isolated. BP

• Flare Tip: Hot tapping in a steam line used for steam-assist, resulted in hot-tap metal shavings blocking parts of steam injection ring. Steam injection affected and led to smoking flare

• Molecular Seals: have choked by carbon from incompletely burned gases. WWW

• Molecular Seals: Syn Gas Compressor seal oil accumulation in mol sieve erupted as fireballs, fell to ground and caused secondary fires in a Methanol Plant (Karthikeyan, CEP, Jan 20)

• Molecular Seals: A small corrosion hole on top of mol sieve allowed air leak into flare stack and mild deflagration, noticeable by a whoomph sound (Karthikeyan, CEP, Jan 20)

• Flare Gas Recovery: Incorrect suction pressure setting of compressor resulted in vacuum, air ingress and mild deflagration, noticeable by a whoomph sound (Karthikeyan, CEP, Jan 20)

• Stack: Elevated flare with lighter than air flared gas can result in air suck-in and internal explosion

• Stack & Structure: Excessive vibrations/shaking. Due to mixing of NH3 and CO2 rich gases in the presence of water. Partial blockage at stack bottom

• Stack: Internal explosion when purge gas failed. Good to monitor O2 level with O2 Analyzer. WWW

• Stack: Air Ingression. To decommission a stack, residual pressure in the WSD and header was brought down by reducing the seal level. Spade was inserted in the drum inlet and other piping. Explosion via one its opening, likely caused by the air pulled into WSD by stack chimney effect and possible ignition from an adjacent flare. Fill WSD with water and N2 purge WSD and stack to reduce HC inventory. Steam purging can help wet Iron Sulfide, a potential ignition source. But steam cooling and condensing in the stack can pull in vacuum and air. IChemE

• Stack: Air Ingression. Purged with steam for maintenance. Was difficult to turn a 35” blind on header with escaping steam. Switched to N2 purge. N2 cooled the stack allowing air to suck in when the blind joint was opened. Explosion. Use N2 only. WWW