Fire in a refinery

This incident occurred at a reactor for the isomerization of wax in a lube oil plant. The high-pressure reactor ruptured during a catalyst activation operation. A hot (400 C) gaseous stream of hydrogen (70%) and light hydrocarbons (30%) was released and spontaneously ignited. The unit was immediately shutdown and depressurised. The fire was limited to the vicinity of the ruptured reactor and was extinguished within twenty minutes by onsite emergency services. There were no injuries but damage to equipment included the reactor and some piping, instrumentation and air-fin heat exchangers on an adjacent structure.
An investigation identified the following series of events preparing the failure of the reactor.
(1) The process consisted in a metal reduction operation in presence of hydrogen and a catalyst. Before the reaction could take place, it was necessary to activate the fresh catalyst. The target activation temperature was higher than the normal operating range of 300-380 degrees C, but still within the reactor design temperature. Since the reactor was a stacked bed reactor with entry at the top, the hydrogen rich gas had to pass over the fresh catalyst and then over the older catalyst beds.
(2) The gas was initially warmed-up via a furnace and passed through the reactor. As the inlet bed temperature target was approached, the furnace coil outlet temperature overshot its set point and three out of the four top bed temperature instruments went out of range, with the temperature at the bottom of that bed exceeding the vessel design temperature.
(3) In response, furnace firing was reduced and quench gas flows were increased to the reactor. Temperature control was poor as the furnace was tuned for normal process liquid/gas operation rather than gas-only activation. Two hours into the activation, the top bed temperatures had steadied out around the target activation temperature but the bottom temperature reading in that bed and all eight of the temperature readings in the two catalyst beds below remained offscale, beyond the design temperature of the reactor.
(4) The activation step was completed two hours later and the reactor cooled down. Three hours into the cool down phase, the reactor, which had a diameter of one metre and a wall thickness of 50 mm, ruptured at the base of the middle catalyst bed.

The INITIATING CAUSE was the release of a flammable mixture due to the rupture of the reactor, probably from thermo-mechanical stresses induced by too hight temperatures experienced.

This was only the second time that the activation had been performed on this catalyst system. The actual processes occurring within the reactor were not fully understood. The reactor was operated for ours outside the design specification and there were no high-temperature automatic shutdown systems.
Cosidering this and the fact that the furnace was tuned for normal process liquid/gas operation rather than gas-only activation, the job preparation should have been re-thought and re-designed.

The ROOT CAUSE can be attributed to lack of management intervention required by the change of the process, as well as failure in reacting to the deviation from the normal process values. The temperature readings were intentionally disregarded or misinterpreted. The operational procedures and training of managers, engineers and operators were inadequate.
(1) Failure to understand the processes occurring in the reactor and the effect of an increase in temperature on the rate of exothermic reactions.
(2) Inadequate cooling facilities.
(3) An inadequate control system and the absence of a high temperature trip device.
(4) Inadequate operational procedures and training of managers, engineers and operators.

de-wax isomerisation high pressure reactor

Incident started in the catalyst activation phase

DESCRIPTION OF THE FACILITY
The component involved was a wax isomerisation reactor, 23 metres high, 1 metre diameter and with thickness walls of 55 mm.
Operating conditions in the reactor were 64 barg and 430-480°C.

People at risk were those at work either in close proximity to reactor, e.g. on platform where fire occurred or in workshop and control room. Emergency personnel included plant operators who took first action to tackle fire and shut off gas supply to fire.

The fire damaged the reactor, the pipework, the instruments and adjacent vessels and structures. This caused 1 year lost sales of the product.

Establishment losses - material max euro: 5900000
Establishment losses - response max euro: 5300000

The event happened at 23:30. Rescue services interrupted road traffic, established a safety perimeter 400 m on either side of the truck, installed 2 protective hoses and evacuated 78 people to a gymnasium.

Technicians from the charter company operated to degas the tank until the leak stopped, then escorted the truck to a technical centre. At about 07:00, the emergency services carried out explosimetry measurements in the neighbouring buildings and allowed the residents to return to their homes.

This incident was a important source of lessons. The accident investigation identified several causes, mostly related to the inadequacy of the operational measures to early spot anomalies and to learn from them.
The findings can be cluster as follows (eMars report).

DESIGN SHORTCOMING
(1) Catalyst activation was not taken into account during the reactor design:
(2) There was no automatic trip in case of trespassing of the maximal reactor temperature.

NEAR MISSES are n important source of information to improve preventing measures. In this case, they were not taken into account:
(1) During initial plant start-up in 1991 and catalyst activation at 400 degrees Celsius an exothermic was experienced. Its cause was identified as a hydro cracking reaction of the paraffin in the treat gas, but no further action was taken;
(2) In 1994 during a preliminary hot de-waxed treatment at 430 degrees Celsius, temperature indicators were seen to be out of range. No investigation was made. The settings of the temperature indicators was 410 and 450 Celsius although the instruments had a potential maximum of 1000 Celsius. All alarms were set at 405 Celsius.

RISK ASSESSMENT SHORTCOMING
(1) The original risk assessment did not consider the catalyst activation stage and dismissed the possibility of a runaway exothermic reaction.
(2) The reactor installation was provided with an emergency shut-down control which was manually activated and required operators to respond correctly to the situation.
(3) In May 1994 a new catalyst had to be activated. A temperature 430 Celsius was planned, instead of the 400 C originally used, but there was no risk assessment on this change in operating conditions.
(4) During the activation stage 3 people went out onto the reactor ladders and platforms at various times, if they had been present when the reactor ruptured they would have been caught by the fire.

- ORGANISATION:
A. Near misses are an important indicator of the potential for accidents. The plant organisation has to record and assess them, and act upon.
B. Operation process parameters cannot be changed without performing a new (partial) risk assessment
C. Management of changes is a critical tool which should be adopted. It was not in the present case.

– TECHNICAL KNOWLEDGE:
At the time of the accidents, very little published information on runaway exothermic in continuous reactors was available. A case report in “Safety Digest of American Petroleum institute, Publication 758 Section 2 1979 Chapter 5” is similar to this incident. Also "Chemical Reaction Hazards - A Guide" Editors Barton and Rogers published by IChemEng 1993 and IChemSymposium Series No. 85 "Protection of Exothermic Reactors and Pressurised Storage Vessels" could have been used to identify the chemical process on going in the reactor.
Since then, hydro cracking and thermal cracking of paraffin at the temperatures experienced in this reactor are widely reported in standard text books.