The catalyst in a dehydrogenation reactor, which was usually operated under a hydrogen atmosphere, was changed while the reactor was isolated from the peripheral equipment by closing a 20-inch remotely controlled valve. The hydrogen pressure in the peripheral equipment was set at 20 KPaG, and the reactor was opened to the atmosphere. Anticipating some hydrogen leakage, suction from the piping was accomplished with a vacuum device and, nitrogen sealing was performed. When the piping connections were restored after changing the catalyst, flames spouted from the flange clearance and two workers were burned. One cause of the fire was poor management of the catalyst replacement process.

Incident Synopsis

A catalyst exchange was carried out in a dehydrogenation reactor at a linear alkylbenzene (LAB) manufacturing plant. To enable piping restoration after the catalyst exchange, a top flange with an elbow was mounted, and nitrogen purging of the reactor and piping was started. Leaked hydrogen gas ignited in the piping and flames spouted out through the flange during bolt tightening. The fire ignited the clothing of the two plumbers who were working near the flange.

The hydrogen flame spouted instantaneously, and no fire-extinguishing equipment was used. There was no damage to the plant. The clothing of the workers caught fire and they were burned. It was not recognized as a fire because there was no special damage, and it automatically extinguished itself. Fortunately, the workers were only slightly injured and their lungs were not burned.

Apparently there were two causes: 1) a leaking valve and 2) poor positioning of nitrogen injection. However, the work plan should have assumed that a valve might leak. When working with hydrogen gas, which easily leaks, using only one big gate valve with a 20-inch diameter to shut off hydrogen is not a good idea, since it can be expected to leak. Minimal countermeasures such as evacuating the leaked hydrogen with a steam ejector or sealing the system with a small quantity of nitrogen are inadequate.

The catalyst in this reactor is exchanged frequently, at about 55-day intervals. Therefore, to decrease the downtime for catalyst exchange operations, a new reactor isolation procedure with only one remote air-driven valve between the slightly pressurized peripheral section and the atmospheric pressure reactor vessel was proposed as a convenient method, and was approved by the plant manager. One cause of the incident might have been instituting a procedural change without a careful study, including an estimation of the potential hydrogen leakage, the nitrogen injection volume required for hydrogen evacuation, etc. The perfunctory study of potential safety issues that was executed for the revised reactor isolation procedure was inadequate to ensure worker safety.


At 05:30 on April 12, 1994, the LAB plant was shut down for a catalyst exchange.

At 13:10, the remotely controlled air-driven valves at the inlet and the outlet of the dehydrogenation reactor was closed.

The top and bottom piping of the reactor was kept at 20 KPaG with hydrogen gas.

On April 13, the reactor was left in a sealed condition at 40 KPaG with nitrogen gas.

At 09:00 on April 14, the reactor was depressurized, the top cover (the inlet side) was removed, and the blind flange (a partition board installed at the end of the piping) was set at the end of the inlet piping, and evacuation started with a steam ejector.

At 11:00, catalyst extraction was started.

At 14:00, catalyst extraction was finished.

At 16:00, new catalyst loading was finished.

At 16:10, the blind flange was detached.

At 16:30, the top cover of the reactor was mounted, and substitution of air in the reactor with nitrogen started.

At 16:40, Flames spouted out through the flange, and two plumbers were burned while tightening flange bolts.


Piping around the reactor was kept at 20 KPaG with hydrogen gas to protect against air contamination. Separation of the hydrogen-pressurized section and the atmospheric pressure reactor was executed by closing only one valve. A steam ejector was running to remove any leaked hydrogen. In the last stage of work to mount the upper piping to the reactor top flange after exchanging the catalyst, the blind flange was detached so the upper piping could be connected. At the location where the blind flange was set, removal with the steam ejector had already started for the hydrogen leakage through the valve. Air entered the piping from the flange during bolt tightening work. The air and the hydrogen that leaked from the remote control valve generated a combustible gas mixture in the piping. The gas mixture was ignited by a static electric spark, which might have been generated from solid material adhering to the internal wall of the piping.

According to the work plan, injecting nitrogen gas would purge the leaked hydrogen. However, the nitrogen was injected at the wrong point, so it did not provide a seal as planned.


The injured workers were taken to the hospital by ambulance. The flames automatically went out as the hydrogen gas was consumed.


The previous catalyst exchange procedure was reinstated, with nitrogen purging of the entire reactor and piping system.

Incident Date
Apr 14, 1994
  • Piping/Fittings/Valves
  • Piping
  • Piping/Fittings/Valves
  • Valve
  • Piping/Fittings/Valves
  • Flange
Damage and Injuries
When Incident Discovered
Lessons Learned

The plant tried to increase production and decrease workload, but failed to notice the potential negative outcomes of this change.
It is necessary to thoroughly simulate all potential negative outcomes of a change in procedures.
Although a valve stops a flow, it is necessary to consider potential leakage. A valve might leak due to fouling, solid material between seals, etc.