Fire In the hydrodesulphurization unit of a refinery

The event occurred at the charge heater of a hydrodesulphurisation reactor.
The vertical cylindrical fired heater was used on the charge to a hydrodesulphurisation reactor. The flow was two-phase at the heater inlet, consisting in hydrogen and hydrocarbon.
At start-up, when the flow was lower than during normal operation, a radiant tube ruptured in one pass. This was thought to be caused by the interruption of the flow in the U tubes of that pass, due to the separation of the liquid-vapour mixture into liquid in the up legs and vapour in the down. Consequently, in absence of flow, the tube overheated and ruptured releasing the fuel, which ignited.

The INITIATING CAUSE was the overheating of a tube of the heat exchanger, due to low flow.

The ROOT CAUSE is probably related to a shortcoming in the start-up procedure and in the monitoring of the unit critical parameters such as temperature and mass flow. The distribution of flow to the multiple tube passes in this type of installation depends on symmetrical pipework and the pressure drop through the passes. There were no individual pass flow meters.
It may have a similar additional root cause components as event no. 656, where the absence of local temperature monitoring prevented the operators to identify deviation from normal operation.

reactors, reaction equipment and heater

The event happened at stat up

Only damage to equipment.

This is a case of a material failure in a technical system, due to shortcomings in operating procedure and lack of a monitoring system able to inform the operator with enough details on deviations from the expected operative parameters (local temperatures, mass flows).

The distribution of flow to the multiple tube passes in this type of installation depends on symmetrical pipework and the pressure drop through the passes. However, there were no individual pass flow meters able to monitor if the flow was really distributed as expected through individual passes.
It should not be assumed that two-phase flows distribute properly between multiple passes in all circumstances. This is particularly true of the furnaces with vertical tubes, and when re-starting operations after shutdowns.

(1) One improvement would be to fit separate flow meters on the gas and liquid streams to each individual pass.
(2) Where this would become too complex or impractical, multiple tube skin thermocouples can be fitted to each pass. In this case, a system able to detect failed thermocouples is required, because these tends to detach from the tubes surface.

It is unknown if the shortcomings identified by this accident and its related lesson learnt were transformed into improvements of procedures and monitoring systems.

Fires due to overheating and failure of tubes in a fired heater (' heat charger') occur regularly (see HIAD event 656 and for more recent cases events 1168, 1169 and 1178).