(213b) Flammable Liquid Process Tank Fire

In June of 1998 there was a fire incident at a coatings facility in the UK during a tank cleaning operation. It was reported that two contractors were injured, both had been transported to the hospital, one was not expected to live.

The investigation began at 0700 local time the next day. The facility was shut down pending the outcome of the investigation. Local police and Fire Service personnel were in charge of the official investigation. In addition, the government safety regulatory agency, The Health and Safety Executive (HSE), was involved.

Local management asked for assistance with their investigation and for a liaison with the various government agencies doing the official investigation. The Fire Service informed the company that, since one victim was not expected to live, this incident was being handled as a potential criminal investigation.

TANK CLEANING PROCEDURE

This incident occurred on a Monday afternoon as 2 contractor's employees were working at cleaning a tank that was used to manufacture coatings (paint). This system of tank cleaning, which had been used at this site for several years, was done by 2 people who would enter the tank wearing appropriate personal protective equipment, a supplied breathing air respirator and a rescue harness. The procedure was to scour the residual coating residue from the tank interior surfaces using a Class 1 (Flammable) solvent, in this case Xylene, and a section of a non-woven scrubber pad. After this residue was removed, the tank surface was wiped clean using paper towel-type wipers. This system resulted in a very clean tank interior with no residual material.

The tank was approximately 3 meters (9.6 ft.) high and slightly over 2 meters (7 ft.) in diameter. To reach the upper portions of the tank, a wooden ladder was placed in the tank. The solvent, Xylene, was available to the cleaner in a 4-liter metal can that had a plastic strap for a handle. This can was suspended from a rung on the ladder by means of a homemade ?S? hook. This 4-liter can was refilled as necessary from a 20-liter metal pail of xylene that was located outside of the tank, adjacent to the tank access port. The contractors alternated turns in the tank, working 20-30 minutes at a time.

INCIDENT DESCRIPTION

The first contractor worked his 30-minute turn in the tank, cleaning the dome. Shortly after the second contractor entered the tank, there was an ignition and the tank interior deflagrated. The bottom line of the incident was that the contractor who was outside of the tank received some 2nd degree burns on his hands and face while the contractor that was inside the tank at the time of ignition suffered 3rd degree burns to approximately 80% of his body surface. This man succumbed to complications from his injuries after about 2 weeks in the hospital. Upon his death, the incident became a criminal investigation under the jurisdiction of the local police.

As most fire investigators know, fires in closed tanks typically self extinguish. In a closed tank, since the combustion process consumes oxygen more rapidly than the fuel, the tank interior becomes oxygen deficient very quickly, typically in about a second or so, and the fire extinguishes itself. However, the atmosphere in the tank is now quite hot and expanding. This hot, fuel rich and oxygen deficient gas will exit the tank via the access port. As it exits, it mixes with ambient air and is re-ignited by the hot gas. The resulting event would most likely be described by an observer as ?the tank exploded?.

Tragically, if the contractor who was working in the tank had stayed in the tank during the event, he would have been unharmed. But, of course, the survival instinct is screaming, ?get out of here?, so he did.

Outside the tank, the second contractor was looking into the tank when the ignition took place. He (probably) reflexively threw himself backwards, away from the access port. In the process of doing this, he (probably) knocked over the 20-liter pail of Xylene sitting beside the access port. This spilled Xylene was almost immediately ignited. This was the scenario that the contractor inside the tank exited into. At this point, it is my opinion that this worker was covered with burning xylene. He walked out of the fire to the nearest safety shower, which was some distance away. The whole time his clothing was burning and smoldering. It was this extended period of time with the burning/smoldering clothes on that no doubt caused the extensive, and ultimately fatal, burns.

THE INVESTIGATION

The investigation consisted of determining - a. What was the fuel? b. What was the ignition source?

The answer to the first question seems to be very obvious, until one looks at the fire properties of Xylene and the ambient temperature at the time of the incident. Ambient air temperature in the vicinity of the tank at the time of ignition was approximately 16.5° C (55° F). When you look at the Flash Point (FP) of Xylene, 32° C (90° F), it becomes clear that something else ignited first. Gas Chromatograph (GC) analysis of liquid drained from the tank discharge piping, showed the presence of a significant amount of Methyl Ethyl Ketone (MEK). Subsequent investigation revealed that MEK was, on occasion, added to a dirty tank and left for a period of time to soften the paint residue and thus speed up the cleaning process. The fire properties of these 2 flammable liquids are:

Xylene MEK

Flash Point 32° C (90° F) -9° C (16° F)

Lower Flammability Limit 1.1% @ 32° C (90° F) 1.7% @ -9° C (16° F)

Upper Flammability Limit 7.0% @ 62° C (144° F) 11.4% @ 23° C (73° F)

At this point, it was fairly clear that MEK vapor was the fuel ignited inside of the tank.

The ignition source was somewhat less obvious. Since it was known what had burned and where it had burned, the list of possibilities was short. A discharge of static electricity was high on this list. The investigation showed that the 4-liter pail of Xylene, suspended by its plastic handle from the ladder was an ideal isolated conductor. This conductor was electrostatically charged by the action of ?sloshing? the section of non-woven scrubber pad around in the non-polar flammable liquid ? Xylene. The charged Xylene induced a charge on the metal can. When the can was jostled and contacted the grounded tank wall, there was a discharge and the MEK vapor was ignited.

To prove this theory, an independent expert in static electricity was retained to re-create the charging scenario and take measurements. This testing proved conclusively that this was a viable ignition scenario.

The local authorities accepted the company's report and conclusions as accurate, impartial and the most likely explanation of the incident. Fines were imposed but there were no criminal charges.

The key learning's from this incident investigation (and also from many other investigations) were:

1. Preserve the scene of the incident as much as possible until everything is documented - use barricades, guards, barbed wire, anything to isolate it.

2. Locate a secure area to store evidence, debris, etc. so that it will not disappear

3. Document the scene of the incident as soon as it is possible to re-enter the area using photographs, videos, hand made drawings, etc. Document where people, equipment, and debris was located using a plan of the area.

4. Get statements from everyone in the vicinity of the incident. Document what everyone saw, heard, smelled, felt, etc. This is must be done as soon as possible before people start to compare stories and/or forget critical information.

5. Get local officials on board as soon as possible if they are going to be involved.

6. Be as open and cooperative as possible with the authorities, involve them as much as possible in the investigation.

7. Involve workers in the investigation if possible. This will help to mitigate rumors that will no doubt start.

8. Get the best, most knowledgeable people you know of involved in the investigation. If you are over your head, get an expert. Your investigation and report must have credibility.