FEATURES
FIRE REPORT
(Photo by Joseph Patrick)
Plan B
Prefire planning is designed to provide the proper strategies and—as in the case of this tank farm fire in Philadelphia—the proper alternatives.
In writing preincident plans for the city’s cluster of bulk oil storage facilities and petroleum refineries, Philadelphia Fire Department officers try to mentally prepare themselves with not only a Plan B for every strategy, but with Plans C, D, and E, as well.
On August 9 of last year, we used them all.
At 10:30 that Sunday evening, in the midst of a thunderstorm, fire dispatchers received a string of phone calls reporting an explosion in a tank farm owned by Chevron USA Inc. In the 32 hours that followed, one complication after another forced us to shift our strategy.
Alternate plans were available because some painful lessons from the past had caused us to plan and train even more carefully for such an incident. This one came 12 years almost to the day after a fire at the same tank farm—then owned by Gulf Oil Corp.—killed eight Philadelphia firefighters.
Since that time, fire company tours of the facility have increased from one to four per year (one each for each platoon), Chevron’s fire brigade is able to communicate directly with the fire department by radio, and face-to-face communications between tank farm and fire department officials have improved.
The 1975 fire destroyed three pumpers, but this time, we used “remote pumping” that kept the apparatus outside the tank’s collapse zone.
In addition, just 15 days before the more recent fire, city firefighters and Chevron employees trained together in a drill that simulated an airplane crash, a tank fire, and a hazardous-materials leak. [See “From Drill to Real in 15 Days” on page 32.]
Thus there was a sense of the familiar when the first officer on the scene at the real thing, a battalion chief, immediately requested a second alarm and began assembling critical size-up information by conferring with plant representatives who were already there.
Tank 58, a vertical storage tank 140 feet in diameter and 48 feet high, was ablaze, apparently as the result of a lightning strike. It was surrounded by tanks of similar size containing naphtha, gasoline, and products akin to gasoline. The exposures on side 1 also included overhead wires on wooden poles, carrying 23,000 volts of electricity, and a roadway lined with above-ground pipelines, known as pipe racks.
The fire tank contained one of the gasoline-like products to a depth of 8 feet, 9 inches. The 1 million gallons of product were burning beneath the somewhat crumpled cone roof; the seam where this roof met the shell had weakened, separating for 35 feet along side 1 and for 20 feet along side 3.
Beneath the cone roof, the tank had an internal floating roof—a sort of metal pan that rested on the surface of the product to help contain flammable vapors. It was unclear whether this internal floater had been damaged or even sunk. Nor was it clear whether the fire involved only the floater’s fabric seal or something more serious.
The first order given was to place Chevron’s internal foam system into service from the pump house. This operation, to be coordinated by the second-in battalion chief, would insert foam into the top of the tank from four stationary pipes, or foam chambers, on its shell. The pump house had two 2,400-gallon tanks of foam concentrate underground, connected to the tank via above-ground piping.
But with the first order came the first complication. A mechanical coupling in the pump house failed on pressurization, spewing raw foam concentrate throughout the building and crippling the system’s primary source of foam. It was going to take several foam units and a lot of foam to pressurize the system externally.
This was the situation as a deputy chief arrived on the scene. This officer struck the third alarm, at the same time requesting that another haz-mat task force be dispatched. One HMTF, consisting of an engine, a ladder, a foam pumper, and a unit carrying dry chemical, foam canisters, a cascade air supply system, and haz-mat suits, had responded on the first alarm.
With command moving up, the battalion chief who’d first taken control of the incident was put in charge of all HMTF operations and coordination with Chevron officials; the other battalion chief was switched from the pump house to the rear of Tank 58, with companies assigned to direct cooling lines on the sides of the burning tank and Tanks 57 and 59.
These lines, as well as many of the foam lines, had to be stretched by hand. After foam units had been placed in the immediate area, the narrow streets left no more room for apparatus amid the pipe racks.
By now it was clear that the explosion had damaged one of Tank 58’s foam chambers. This fire demanded an estimated 2,465 gpm for at least one hour, assuming the internal roof had sunk. Even if all four chambers were working, they’d fall short of that amount, since each could supply 400 gpm of finished, aerated foam. The haz-mat task force would have to supplement the supply. But that was possible only after the fire department isolated the tank’s foam lines from the rest of the yard foam system, which had been connected to the pump house.
With that done, we not only supplied the foam system by pumping into its gated outlets from mobile units, but also augmented the internal system with external foam lines operated from deluge sets that had aerated-foam nozzles. Chevron’s fire brigade assisted by using its foam unit to supply the system on side 2 and by placing hose lines and monitors in service for foam and water cooling lines.
The cooling with master streams had several purposes: protecting exposures, keeping the fire tank from collapsing, and cooling the steel to help the foam adhere better where the fabric seals contacted the tank.
Two fire department marine units supplemented the 125 psi of the plant’s yard hydrant system, for which a different pump house draws water directly from the Schuylkill River. This added 6,000 gpm to the system’s 5,000. A third marine unit was called and docked in case more 3½or 5-inch supply lines might be required.
A battalion chief was named water liaison officer to assign hydrants and work with the water department. Other battalion chiefs were eventually given responsibility for sides 1 and 2 and sides 1 and 4 of the burning tank. More sectors were added as the incident progressed, so that battalion and deputy chiefs took the roles of public information officer, personnel relief and rehabilitation officer, and safety officer.
The latter paid special attention to the condition of the dikes surrounding the tanks and monitored the amount of runoff water. The plant sewer pump kept the water level at a minimum.
Yet another officer acted as outside agency liaison, working with the police, who handled scene security and closed streets in the immediate area; Chubb National Foam Inc. representatives, who came to the scene to supply technical advice and facilitate deliveries from their plant 45 minutes away; Allied Chemical Corp. and Rohm & Haas Co., which responded with foam pumpers; and the Philadelphia Second Alarmers and Philadelphia Chapter of the Red Cross, which provided food and beverages for the fire forces.
The Philadelphia Emergency Operations Center, in the basement of the fire administration building, was also crucial in coordinating agencies. It began to be staffed at about the same time the fourth alarm was struck at 11:30 p.m. The center took over radio and land line communications with the fire scene and coordinated requests for assistance, equipment, and supplies—including the private-industry foam pumpers. The EOC became fully operational when outside agencies joined it the next morning.
Resources and relief would be a major element of this long siege. There were only small openings on side 1 of the fire tank, so foam application there proved difficult. We moved the deluge set to attempt better penetration, but after 2½ hours, the condition was slowly worsening.
Then Murphy’s Law visited again during the night. A second fixed foam unit got clogged and had to be shut down.
In their considerable discussions of alternatives, fire officers, Chevron officials, and National Foam representatives discounted the option of draining the tank, because that could shift the internal floater in a way that would worsen the fire. We also initially rejected the idea of subsurface foaming—the injection of foam through the tank’s supply lines. If this works as intended, the foam floats to the surface to smother the fire.
We turned to this option around 4 o’clock Monday morning. Firefighters set up a manifold to accept three 3-inch foam lines that would deliver 900 gpm into the tank. This would be augmented by the two remaining foam chambers and two portable monitors, for a total delivery of 2,700 gpm of foam solution. But enough foam had to be on location before the operation started, and the people from National Foam arranged emergency shipments.
At 7:10 a.m., the subsurface application began. An hour and 20 minutes later, the fire seemed to be getting worse. We still didn’t know for sure, but it was possible the internal floater had sunk in a way that interfered with the subsurface foaming.
The next choice was to add another four foam monitors to augment overhead delivery. The emergency operations center coordinated shipments of foam concentrate from different fire departmerit locations to the fireground.
During this period, more mechanical trouble caused a delay. Two fan belts broke on one of the fire department’s foam pumpers operating at the scene. A call went in to the department’s crash/fire rescue unit at Philadelphia International Airport, which had an auxiliary foam pumper in position within 20 minutes.
We would have used elevated aerial devices to get better access for our foam lines, but there were three problems: the narrowness of the only access road, which ran alongside the tank on side 1 and was just 12 feet wide; the closeness of the overhead wires, 30 feet from the immediate fire area; and the continual radiant heat.
By the time the additional ground monitors were in place atop the levees and the subsurface lines were prepared, one of the two remaining foam chambers had begun leaking severely. But at 3,300 gpm, the total foam application was well above the recommended 2,465 gpm.
Still, the frustration continued. At 10:30 a.m., there still had been no appreciable gain in fire control. Now we were quite sure the internal roof was blocking the subsurface foam. And the wind had kicked up to 20 mph, interfering with the overhead delivery.
We shut down the subsurface foam operation and repositioned the portable foam monitors atop the levees to take advantage of the wind. The pumper with the broken fan belts wouldn’t be back— the mechanics on the scene had uncovered additional problems with it. We had and used one foam pumper supplied by Allied Chemical and one from Rohm & Haas.
All the fire department’s supply pumpers were operating at capacity to provide the foam volume and large-caliber streams necessary. As the wind increased to 40 mph, it affected the reach of our streams. Increasing pressure to overcome these effects, our pump operators reported that they were operating near vacuum level on the compound gauges. Our water supply officer ordered more large-diameter supply lines laid from the fire boats to solve the problem.
(Photos by the Philadelphia Fire Department)
The yard hydrant system was operating at maximum capacity already. More lines were stretched for distances of 3,000 feet from outside the tank farm. This time, with the one remaining foam chamber Still in use, total application would be 3,500 gpm.
But after 75 minutes, it was the same story: Wind and distance again played havoc with the foam streams.
A 12:45 p.m. conference involving all concerned parties concluded that we should try transferring the product from Tank 58 to another tank within the tank farm. But even this attempt to move the fuel away from the fire was destined to fail. While the preparations were under way, the roof weakened noticeably near the joints where the fire was burning fiercely. Although cooling lines were in place, the entire cone roof began to collapse into the tank with a sudden, loud creak, and the sides of the shell folded inward.
The submerged roof appeared to rest on the internal floater, and the product above it produced a freeburning fire. The entire surface area was now fully involved. Flames and heavy, dense black smoke rose hundreds of feet into the air. All available cooling lines were placed on the pipe rack and tanks on side 1, which was severely threatened by the 40to 50-mph winds created by the fire storm. Two additional alarms were struck for companies to bring in supplemental water from city hydrants— a task which involved relay pumping and hose stretches of more than 5,000 feet.
The fire now involved the highvoltage power lines along side 1, so electrical service to that portion of the tank farm was discontinued. Thus ended the effort to transfer the product out of the tank. This also shut down power to the sewer pumping systems used to drain dikes of water accumulations. Our options were reduced to protecting the exposures with cooling water lines and continuing to apply foam to the burning surface. An eighth alarm was struck for additional personnel and apparatus to be maintained in staging.
The foam never did have an appreciable effect on the fire, and at 6 p.m., that operation was stopped and it became a waiting game until the fuel ran out. The foam lines were converted to water lines and turned to cooling the sides of the tank and the exposures.
We had to be ready in case the tank failed completely and released its contents into the dike area, spreading the fire and endangering Tanks 57 and 56, which shared the same area. Sector commanders were instructed that if that happened, all water lines used for cooling tanks in the concerned dike area would be shut down and foam lines would come into play. To assist the seven foam lines already in place, Rohm & Haas’s foam truck was repositioned next to Tank 56 and prepared to play foam into the dike area.
But the wind died down around 10 p.m. Around 3:30 Tuesday morning, the flames started to subside as the fuel ran out, and at 6 o’clock, fire officials declared the fire under control.
Lessons Reinforced
Familiarity increases efficiency. Inspections and on-site drills increased firefighters’ knowledge of layout, availability and use of auxiliary extinguishing and control systems, logistics needed and available, and organizational contacts.
Cooperation between the fire service and industry is a two-way street. The synergistic effects will benefit both sides. In this case, the exchange of information vital to handling the incident smoothly concerned plant operation, outside contacts, and foam supply.
Logical strategies for solving complex fire problems can be decided on and implemented only on the basis of preincident information and onscene size-up.
An incident command system is key to controlling an escalating emergency. The fireground must be sectored and staff officers assigned to auxiliary functions.
A central emergency operations center at a remote location enables a fire department to coordinate and communicate with other agencies. It also eliminates needless confusion at the command post.
For many of the firefighters at this incident, the underlying thoughts were flashbacks to the the fatal fire of August 17, 1975. There were both similarities and differences between the two fires.
Both occurred on a Sunday when Platoon C was working and was relieved by Platoon D. (The recent one went from 5 to 8 alarms, the earlier one from 6 to 12.) Both fires necessitated the shutting off of electricity, which in turn shut down the dike pumps.
The differences are more significant. The 1987 fire was started by lightning, the 1975 one by operator error. The initial explosion in 1975 damaged the piping going into the tank; when the electricity was shut off, the leaking product floated on the pooling water. The muffler on a foam pumper ignited the floating material, and it was this fire that killed eight firefighters and destroyed many apparatus and additional storage tanks.
The 1987 fire was contained to one tank and $2 million damage. There were two minor injuries; no apparatus were destroyed.
The lessons learned from 1975 and from the training exercise two weeks before made the difference. The interaction of planning for and performing the drill had created a mutual understanding and respect that carried through the emergency operation.
