Pre-Incident Activity—
A severe storm on November 24 caused an electric power outage in the
refinery. The storm interrupted process operations and also stopped the
production of steam. At the delayed coking unit, the on-line drum had been
filling for about an hour and was approximately 7 percent full. The other
drum was full and was being cooled.
Although electric power was restored after 2 hours, an additional 10 hours
passed before steam production was re-established. During the interim, the
tarry oil in the piping between the furnace and the partially filled drum
cooled and started to solidify. Once steam was restored, the operators were
unsuccessful in attempting to inject it into the drum through the normal
route because of the plugged piping. (When normally injected, steam creates
passages in the tarry mass through which cooling water can later flow. It
also drives off remaining residual volatile petroleum and sulfur compounds
from the coke.)
A process interruption in 1996 had also resulted in a partially filled drum.
At that time, water was injected into the drum to cool the material inside.
However, when the drum was opened, a torrent of water, heavy oil, and coke
spewed out—which created a hazard and required a major cleanup. An internal
investigation team recommended that procedures be written for
cooling/emptying partially filled drums. However, this task was not
completed.
On the day of the fire, neither the process supervisor nor the operators had
any written procedures for handling partially filled drums. The process
supervisor was aware of the seriousness of the previous incident. He left
instructions directing the night shift not to add any water, and instead to
allow the drum and its contents to simply stand and cool overnight. On the
following morning, he met with the operators to determine how to empty the
partially filled drum. No engineers, who could have provided technical
support, were present at this meeting.
Preliminary Operations—
The supervisor and operators observed that the exposed part of the bottom
flange of the drum felt cool to the touch. They also noted that
temperature-sensing devices located beneath the insulation on the outside
surface of the drum indicated approximately 230 degrees Fahrenheit (°F), as
compared to the 800°F of a typically full drum.
One operator suggested adding 100 barrels of water to the drum. However, the
supervisor was concerned about such a course of action because of the
previous incident. Upon further discussion, they decided—because part of the
drum felt cool, and the temperature-sensing devices read only 230ºF—that it
was not very hot inside and it was safe to open the vessel as long as they
first injected some steam.
An operator connected a steam hose to the oil inlet piping at the bottom of
the drum. Several witnesses said that the steam warmed the top of the
piping, but the bottom remained cool. It is likely that steam flow had been
established, but the rate of flow was low.
Opening the Vessel—
Personnel expected a tarry mass to drain from the drum. The supervisor and
process operator directed that the drum be opened with a minimum number of
people present. Because they were also concerned that the limited flow of
steam might not sufficiently strip all the toxic compounds from the tar
inside the vessel, they required that the workers removing the bolts on the
drum heads wear self-contained breathing apparatus. The top head was
unbolted and lifted from the drum. The bottom head was also unbolted and
held in place by a hydraulic dolly. The operator then activated a release
mechanism to lower the dolly. Witnesses reported hearing a whooshing sound
and seeing a white cloud of vapor emanate from the bottom of the drum. The
hot petroleum vapor burst into flames. The process supervisor, an operator,
and the four contract personnel assisting were caught in the fire and did
not survive (Figure 2).
After the incident, Equilon relocated the controls for the hydraulic dolly
to allow workers to position themselves farther from a drum when opening it.
Followup Analysis—
The supervisor and operators analyzed the situation and devised process
changes to empty the drum. The relative coolness of the bottom flange
erroneously suggested to them that the temperature inside the drum was also
cool—when, in fact, only the material adjacent to the inside walls had
cooled. Unknown to the coker unit personnel present, the core of the mass
remained insulated from heat loss. Within the core, residual heat continued
to break down the petroleum, creating a pocket of hot pressurized volatile
oil. Had the limitations of temperature-sensing devices been better
understood, personnel may have realized that the low temperature readings
were not representative of the hot core. It was assumed that the entire drum
contents had cooled to safe levels during the 2 days since the power
failure. However, heat transfer calculations would have indicated that weeks
would be required for the drum contents to cool sufficiently via heat losses
to the ambient environment.
|
Fire control efforts
at Equilon refinery
Matt Wells, Skagit Valley Herald
The relative coolness of the bottom flange
erroneously suggested
that the temperature inside the drum was also cool—
when, in fact, only the material adjacent to the inside walls had cooled.
|
