Fire on the pre-heater of a desulphurisation unit

A tube ruptured in the charge heater ('fired heater') of a desulphurisation reactor due to a blockage caused by accelerated coking. Flames spread over an area of some 100 feet outward from the bottom of the heater and more than 50 feet out from the stack. Burner connections were blown loose adding to the fire at the base of the heater.

Within minutes operators reacted to shut down and block in feed, hydrogen make-up and product streams, to isolate fuel at battery limits and to initiate steps to depressurise the unit. Fire control forces were set up and began applying cooling water to surrounding equipment and on the furnace skin and stack. Special attention was given to the reactor and nearby exchangers. With all systems secured and isolated, the fire was allowed to burn itself out, which took approximately 30 minutes.

The failure was caused by accelerated coking in the furnace tubes because of insufficient hydrogen recycle coupled with severe flow imbalance. As the system (hydrogen supply was depleted and the heater continued to be fired harder, the coking risk problem was worsened. Heater pass temperatures became unstable, but the operators’ attention was focussed entirely on an attempt to improve mechanical flow balance using the pass inlet gate valves.

The INTIATING CAUSE was the rupture of one or more tubes of a fire heater due to over-heating, casued by los feed flow and accelerated coking.

SEQUENCE of CAUSES
On the day of the incident a recycle (and make-up) compressor was shut down due to hot valves. This resulted in a 3-folded reduction of the recycle gas rate. The operators did not make any attempt to compensate the feed rate, because (1) the then current operating practice did not call for maintaining a minimum hydrogen/hydrocarbons ratio and because (2) the operators knew from experience that one compressor fully loaded was sufficient. What they did not recognise, however, was that this compressor was operating at less than 50% of its normal capacity.

ROOT CAUSE: inadequate training and lack of process monitoring devices.

fire heater, feed tubes

On the day of the incident a recycle (and make-up) compressor was shut down due to hot valves.

DESCRIPTION OF THE PRE-HEATING FURNACE
In this unit, the feed stock, made of diesel and (recycled and make-up) hydrogen is preheated (1) by the reaction products via a heat exchanger and (2) via four passes in a "can type" furnace with burner placed at the bottom. Each heater pass consists of 10 rows of 9-chrome tubes around the circumference of the radiant section and 12 rows of extended surface 9-chrome tubes in the convection section. External inlet block valves on each pass were used for flow balancing. Combined heater outlet (reactor inlet) temperature regulates the fuel gas to four combination burners.

THE 'COKING' OF THE TUBES
Coking of furnace tubes is an unwanted buildup of solid carbon (coke) on the inner walls of hydrocarbon processing tubes, which is caused by thermal decomposition of the fluid. This fouling decreases heat transfer efficiency, increases tube metal temperature, and can lead to hot spots, equipment damage, and costly unplanned shutdowns. Factors influencing coking rate include feedstock type, high operating temperatures, high heat flux, and inadequate fluid flow.

Nobody was injured. The property loss was not provided: it can be assumed that the whole fired heater was destroyed, while the reactor did not incurred in heavy damages.

In this incident, the operators did not understand the real nature of the anomalous operative conditions and did not take the required mitigating actions. They must be trained to understand thoroughly the operation of complex plant with special attention to the significance of deviations from design operation.
As noted in other similar events, it could be added that a better understanding of the processes is facilitated by a good monitoring of the process parameters in the critical positions. It is highly probable that the facility was not equipped with flow meters and temperature gauges on individual passes, which could have alarmed the operators on low-flows and high-temperatures signals.
[See for example HIAD_1167 and HIAD_1168]