Skip to main content
Clean Hydrogen Partnership

Fire on a hydro-desulphurisation column

Event

Event ID
649
Quality
Description
The incident occurred in the hydrogen sulphide absorption tower of the kerosene desulphurisation unit of a refinery. The hydrogen leaked from the liquid level gauge and ignited causing a fire. The leak was due to a bolt which came loose and to a wrong insertion of the packing that seals the glass of the liquid level gauge and the flange of the main body.


Sequence of events:
The day before the leakage, the unit had been shut down in accordance with the scheduled repair plan. Only the hydrogen sulphide absorption tower affected continued to operate.
On the day of the incident, around 12:55, a construction worker who was nearby witnessed the leak by hearing a loud "bang" and notified the incident. The plant supervisor inspected the scene and ordered an emergency stop.
13:01, an emergency stop operation was started in the instrument room.
13:10 The affected tower could be disconnected from other units and the emergency release valve was opened to reduce the tower pressure from 50 bar to 1.5 bar. After decompression, cooling water was sprinkled and water was injected into the tower to raise the liquid level above the leak point.
13:25 The fire was extinguished.
Event Initiating system
Classification of the physical effects
Hydrogen Release and Ignition
Nature of the consequences
Fire (No additional details provided)
Macro-region
Asia
Country
Japan
Date
Main component involved?
Gauge
How was it involved?
Rupture & Formation Of A Flammable H2-Air Mixture
Root causes
Root CAUSE analysis
The INITIATING CAUSE of the leak was a wrongly assembled liquid level gauge.
The packing that seals the glass of the Klinger liquid level gauge to the main body was over tightened, resulting in deterioration. Because of this, during the two years of operation, a bolt without spring washer became loose, causing the oteh bolts to become overloaded. Probably the packing could not withstand the internal pressure and broke at a bolt next to the loosened one, causing the process gas to erupt. The ignition was attributed to static electricity generated during hydrogen ejection.

Therefore, the ROOT CAUSE could be attributed to wrong installation and lack of quality control by the manufacturer of the gauge. A factor could also be attributed to shortcoming in the inspection effectiveness of the plant operator. Morevoer, the absence of automatically shut-off valves to stop the flow in the event of a large gas leak hints atthe lack of a risk assessment related to the specific installation.

Facility

Application
Petrochemical Industry
Sub-application
Hydrodesulphurisation process
Hydrogen supply chain stage
All components affected
instrumentation, gauge
Location type
Open
Location description
Industrial Area
Operational condition
Pre-event occurrences
The component affected by the incident was left functioning as usual, despite the shutting down of the rest of the unit.
Repair work were ongoing on the rest of the unit.
Description of the facility/unit/process/substances
DESCRIPTIONOF THE PROCESS
The process used to remove hydrogen sulphide from kerosene consisted in hydrogen in in an aqueous solution of diisopropylamine.
The liquid level gauge was of the Klinger-type and was placed at the bottom of the tower.
Operating temperature 40°C, operating pressure 5 MPaG..

Emergency & Consequences

Number of injured persons
0
Number of fatalities
0

Lesson Learnt

Lesson Learnt

The following are three different return of experience based on this incident:
(1) The liquid level gauge had a life span of 10 years, therefore the manufacturer's responsibility in the failure of the product was relevant. However, the onsite inspection did not found any defect of the product during acceptance procedure. It could be that the defect was overlooked by an inspection which had become considered a simple routine, due to the good performance of previous products of the same type. It is rather difficult to identify a clear responsibility in this case.
(2) If the assumption done above on the routine inspection was correct, it strengthen the general statement that relying only on past experience can lead to misjudgement regarding the safety, reliability, and longevity of equipment and facilities.
(3) The liquid level gauge had a part made of glass. Breakage of the glass occurred often, especially at high temperature, due to the different thermal expansion of glass and metal. Alternative gauges were more suitable for high temperature applications such as the affected absorption tower. Indeed, following this incident, the plant installed another type of gauge. A general conclusion is then that operators should always consider improvement of instruments and components based on the technology progress and an accurate risk assessment.
(4) There were not automatically shut-off valves which could stop the flow in the event of a large leak. It signifies a lack of proper risk assessment related to the specific case.

Corrective Measures

1. The inspection procedure was modified and its frequency increased.
2. The liquid level gauge under severe temperature and pressure operating conditions was replaced by a differential pressure type liquid level gauge.
3. An automatic stop valve was installed on all liquid level gauges that were not changed.

Event Nature

Release type
gas
Involved substances (% vol)
H2 100%
Actual pressure (MPa)
6
Design pressure (MPa)
6
Presumed ignition source
Static electricity

References

Reference & weblink

JST failures database:<br />
https://www.shippai.org/fkd/include/fkd_showCase.php?id=CC0000066&text1… />
(accessed Dec 2024)

JRC assessment