Photos by Wayne Doolittle.
COVERED FLOATER IN JACKSONVILLE: STEUART PETROLEUM BULK STORAGE TANK FIRE
Editor’s note: This article originally appeared in the May/June 1993 issue of Industrial Fire Safety, a PennWell publication that we view as Fire Engineering‘s counterpart in the field of industrial sector fire protection and response. The vital lessons learned from this incident warrant a deviation from our standard editorial policy on reprints. These lessons must be shared with the fire service. Indeed, the firefight at the Steuart Petroleum tank farm will be remembered as a landmark response to what historically has been one of the most challenging fire problems ever faced.
Early on Saturday morning, January 2, 1993, the tanker Team Erviken was off-loading unleaded gasoline at the Steuart Agip Evergreen Terminal, located north of downtown Jacksonville, Florida, on the St.Johns River. At approximately 0315, a tremendous explosion, which could be felt for a one-mile radius, rocked the terminal, sending a fireball hundreds of feet into the air. The extended operation that followed yielded numerous important lessons for the municipal and industrial fire communities in the extinguishment of large flammable liquid storage tanks with covered internal floating roofs.
INITIAL RESPONSE
The explosion occurred from the ignition of an approximately 50,000gallon overfill of tank #22. The onduty terminal operator responsible for the filling operation, and the only Steuart employee on site, apparentlyfailed to direct the flow into an adjacent tank. He had either driven his company vehicle into the berm area, igniting the spill, or had parked the vehicle at the base of the tank to monitor the filling operation and was enveloped by the massive spill, igniting it in his bid to escape. Firefighting teams later located his body approximately 10 feet from the vehicle.
The Jacksonville (FL) Fire and Rescue Department (JFRD) communications center immediately received numerous 911 calls reporting the explosion and fire. The general area from which the reports were received is a combination residential/industrial area just north of the main downtow n port authority terminals. This location is dotted with heavy industry, railyards, and bulk petroleum storage sites. A maximum first-alarm haz-mat response, consisting of four engine companies, two truck companies, a battalion chief and district chief, a haz-mat unit, and two FMS-rescue units, was dispatched to the incident.
A plume of thick, black smoke was visible to responding units. Battalion Chief W. H. Burns requested two 1,000-gallon foam tankers to the scene. First-in engine companies arrived at 0321 and reported a large fire involving the bulk fuel storage facility. They immediately requested a second-alarm assignment.
Tank #22 (a 100-foot-diameter internal floating roof tank with a steel pan and a capacity of approximately 56,000 barrels, or 2.3 million gallons) was burning. Gasoline poured out of its eyebrow vents, and ground fires totally engulfed its base, impinging on tank #21 (a 3.2-million-gallon internal floating roof tank located 40 feet west of tank #22, which contained only residual gasoline product and vapor) as well as tank #23 (a 3.9million-gallon external floating roof tank, which was one-third full of gasoline and located 60 feet east of tank #22). The ground fire also raged around unprotected aboveground piping and manifolds. Many flange connections, stressed by the explosion and fire conditions, also were burning vigorously.
As firefighters established cooling lines to protect the adjacent tanks and exposed terminal piping, Chief Burns arrived and established command. He immediately consulted company representatives, who had arrived from an adjacent facility, to ensure that the fuel flow into the terminal was shut down. In response to the magnitude of the incident, he requested the assignment of additional water tankers and the JFRD’s largest fireboat. The water tankers each carried 75 gallons of foam concentrate and hydrant basins, which would be used as foam-induction points. The fireboat, a 65-foot marine firefighting vessel equipped with three 2,000-gpm fire pumps, would provide an ample water supply platform operating from the nearby St. Johns River. Two fiveinch supply lines were laid from the fireboat. (Later, another set of two five-inch lines would be laid from municipal hydrants outside the tank farm. These redundant lines were “gridded” to the fireboat supply lines, to be used as an emergency backup water supply.)
The ground fires and flange fires were suppressed in a relatively short period with foam lines and hand extinguishers. Success in this phase of the operation can be credited to the fact that all JFRD personnel had recently completed training in similar live flammable liquid fire evolutions. The experience of Denver area firefighters during a 1990 incident at the tank farm adjacent to Denver’s Stapleton Airport prompted extensive training in foam, dry chemical, and extinguishment techniques on three-dimensional fires. A third alarm was requested at 0353, approximately 30 minutes into the operation, as suppression activities were directed at the burning tank. This committed a total of 85 officers and firefighters, staffing 11 engine companies, five ladder companies, one squad company, three FMS-rescue units, one battalion chief, five district chiefs, and numerous special units, including the JFRD’s hazardous-materials response team. The suppression effort now accounted for nearly one-third of the available JFRD on-duty personnel and apparatus.
By 0417, Chief Burns received confirmation from company officials that all product flow into the terminal had been halted. Product flow continued, however, from the tank’s eyebrow vents, complicating ground fire suppression and threatening the haz-mat team, which was securing the nowextinguished flange leaks. Thermal heating and expansion of the product, plus the effect of incidental water flow into the tank from the cooling lines, were assumed to be the causes. As the flow of fuel continued, sector commanders raised the possibility of a rip or tear in the tank shell, resulting from the explosion. In response to this concern, all personnel, with the exception of those operating the cooling lines, were removed from the dike area while sector commanders, accompanied by the haz-mat team, reassessed the structural condition of the tank.
Tank #22 structurally sustained an enormous shock from the initial explosion. The sides of the tank revealed large visible ripples (which were “cold-formed”) and bulges at various weld sections. All access ports and vent covers of the external cone roof had been blown open, and fire was issuing out of most of these openings. The external cone roof had partially collapsed onto the internal floating steel pan, creating an irregular external roof line w ith numerous cavities between the roof and the floating pan.
FIRST ATTACK ON THE TANK
Assessment of the tank’s structural condition allowed fireground commanders to conclude that no immediate danger of shell collapse or failure was present. The order then was given to prepare for an attack on the burning tank.
Suppression operations required the placement of additional fixed water and foam supply lines inside the dike area. Efforts also were underw ay to transport all available JFRD foam supplies to the incident site. These efforts now were accelerated to ensure that an adequate supply of foam concentrate would be present prior to an attack.
Foam concentrate supplies, which were now arriving on the fireground, were a direct result of the JFRD’s interaction with various agencies and groups prior to the incident. An internal tank farm firefighting task force had been established to review the JFRD’s tank farm firefighting capabilities and foam concentrate supply parameters. Based on the task force’s recommendations, the JFRD had in excess of 8,000 gallons of foam concentrate on hand when the fire began. In addition, through the citywide hazardous materials interagency agreement, the JFRD had access to other local public and private foam concentrate supplies.
The U.S. Coast Guard’s (USCG) ability to access local military supplies also was a major factor in the supply equation. By late Saturday, local naval bases had contributed nearly 6,000 gallons of foam concentrate to the suppression effort, 500 gallons of which was being transported directly to the scene by helicopter from the Marine Corps air station in Beaufort, South Carolina.
The USCG also coordinated the effort to protect the river in the event of a massive gasoline spill. The containment area sloped away from the tanks and toward the St. Johns River. If a major breach of either the tank or the containment area occurred, the product would immediately flow into the river, threatening shipping lanes and adjacent industrial facilities. As a precaution, USCG representatives ordered the placement of protective floating booms surrounding the terminal property and restricted river traffic in the area of the terminal.
At 0439, a 1,250-gpm foam monitor, which had been placed on the north side of the tank, was directed into an eyebrow vent near the open manway on the northeast side. The foam attack continued for only 15 minutes when the tank began to overflow both foam and burning gasoline. This reignited ground fires in many places, and the attack was immediately suspended, as crew s suppressed the ground fires and reestablished a foam blanket.
The rapid overflow was attributed to the tank’s high product level, so company officials again were consulted and asked to remove product from the overfilled tank to provide sufficient outage in which to place a foam blanket. The transfer operation began at 0500 and continued for approximately two hours, when consultations between the incident commander, company officials, and the haz-mat team determined that sufficient product had been removed to attempt another over-the-top attack.
The second suppression attempt began at 0650, as the ground-mounted 1,250-gpm foam monitor nozzle on the north side of the tank again was activated. Finished foam solution now appeared to enter the side vents and blanket the external roof, flowing gently into the open roof access hatches.
The foam flow continued for an hour, confining the fire to the open manways on the northeastern and northwestern sides of the tank. It appeared that they could be extinguished easily with foam handlines or dry chemical extinguishers. To extinguish the remaining fire, members of Ladder 2 operated from an 85-foot aerial platform positioned adjacent to the tank. Its platform was used to direct foam from a 240-gpm handline into the still-burning manways. However, neither the foam line alone nor a combination of the foam line and hand-held dry chemical extinguishers proved successful. The attack had to be abandoned when the eductor to the 240-gpm foam nozzle operating from the platform became clogged, requiring maintenance.
MIXED FIRE CONDITIONS
The tank now had exhibited mixed fire conditions, at one point burning very vigorously while on other occasions almost totally subsiding, emitting a very “lazy flame” more indicative of a “rim-type” fire. This fire behavior led fireground commanders to conclude that the fire was burning in a number of different “compartments” formed by the collapse of the cone roof. Further, the possibility was raised that the force of the explosion tilted the floating pan to one side, wedging it in that position and forming voids under the pan.
The presence of a butane injection system in the tank also raised concerns. On numerous occasions the flames appeared to burn very efficiently, producing little or no smoke. This result seemed inconsistent with a gasoline-fed fire, so speculation began about the presence of recently injected butane or other additives. Company officials assured fireground commanders that no additives were present in the fuel; however, these flame conditions persisted throughout the fire.
From the failed attack, a second extinguishment plan was generated. It involved the coordinated use of larger quantities of dry chemical in combination with a foam attack. The last foam attack successfully reduced the fire conditions to what appeared to be purely “vapor fires” burning in the roof openings. If external roof damage was limiting foam access into all areas of the floating pan, sufficient quantities of Purple-K-Powder™ (PKP) might be able to access and inert these areas, allowing the foam to seal and extinguish the fire. To implement this plan, a request for the assistance of the Jacksonville Port Authority Fire Department’s two 450-pound PKP four-wheel-drive units was forwarded to the Jacksonville International Airport.
The break in suppression efforts allowed shift relief to be accomplished and fireground foam concentrate supplies to be restocked. Battalion Chief W. L. Mays reported to the command post after a helicopter survey of the scene and relieved Chief Burns as incident commander.
RENEWED ATTACK
By 1000, the preparations were complete for a renewed foam attack augmented by the PKP trucks. The 1,250-gpm foam monitor was relocated onto Ladder 2’s platform and operated as an elevated stream. Its role was to place foam onto the external tank roof in the general area of the large opening on the northeast side of the tank, where the external roof was deformed by the explosion and ensuing fire into a natural funnel. Finished foam solution deposited around the opening would gently flow into the burning opening. An additional aerial ladder truck, Ladder 1H, was placed on the western side of the tank. From this position. Ladder 1H operated a 240-gpm foam line directly into the side vent adjacent to the damaged access ladder. A PKP line was connected to each ladder unit. When the fire conditions seemed appropriate, the lines would be directed into the side vents in hopes of suppressing the fire.
Fire conditions seemed to worsen just prior to the beginning of the attack, but after approximately 20 minutes of operation, the foam application seemed to be achieving success. it was at this point the order was issued to discharge PKP into the side vents. The flames immediately turned purple and seemed to dissipate, with PKP clearly visible at all openings of the tank. Unfortunately, on completion of the PKP application, the fire stubbornly reappeared, and the foam lines were pressed back into service.
Suddenly, a five-inch hoseline supporting the foam attack ruptured, striking Lieutenant A. W. Bebernitz, Jr., the foreground safety officer, and Firefighter G. L. Pope, throwing them 10 to 15 feet. Lieutenant Bebernitz’s leg was fractured and firefighter Pope’s back was injured. Both required hospital treatment for their injuries. Fortunately, these were the most serious injuries firefighters sustained over the course of the entire incident.
The loss of the water line suspended the foam attack, and the PKP lines were removed from the ladders so the units could return to the airport and be refilled. Again, the failure renewed the examination of alternative ways of getting extinguishing agent into all areas of the tank. Fireground commanders now decided to attempt a subsurface attack in combination with the over-the-top attack.
SUBSURFACE AND OVER-THE-TOP ATTACK
It was thought that if the pan was wedged, forming void areas, perhaps the subsurface units would suppress any fires burning in these spaces. Earlier, consideration of a subsurface application, through the benzene injection system, was rejected because it was felt the injection port could not support the required flow rate. To correct this, company officials recommended injecting through the “highsuction” valve assembly. The highsuction port was chosen because it would allow foam solution to be injected above the water bottom, thereby not diluting the foam solution and rendering it ineffective. Two 240gpm, high back pressure-forcing foam makers were moved into the highsuction valve, and after some initial difficulties, a functioning subsurface flow was directed into the tank. It would be much later in the fire suppression effort before firefighters and facility personnel would discover that the high-suction line had been displaced by the explosion. It now rested within the water bottom that firefighters were trying to avoid.
The combined subsurface and overthe-top attack began just before noon. For most of the next hour, the haz-mat team struggled to maintain the subsurface flow. During this time, company employees also continued their efforts to transfer product from the tank. However, shortly after 1300, fire conditions rapidly intensified. This was quickly followed by reports from the sector commanders of raw gasoline flowing from the eyebrow vents. No ground ignition occurred, but foam operations immediately were shut down and foam safety lines were brought forward to protect the personnel and equipment inside the berm area.
The product flow from the eyebrow vents subsided with the cessation of the foam attack. Fireground commanders then met with terminal personnel to analyze the reported product transfer rates. It was concluded that only the subsurface attack could continue in relative safety. Therefore, at 1343, the subsurface attack was reactivated, and cooling lines were returned to the tank shell.
The subsurface attack continued until 1550 with no noticeable effect. By this time, the PKP trucks returned to the scene and were repositioned for a renewed attack. Chief Mays had decided, after staff consultations, to reattempt the nearly successful attack of the late morning.
For this attack, cooling lines first were applied directly to the external tank roof in an attempt to reduce the temperature of the surrounding metal. They were followed by the activation of prepositioned foam lines. Almost 2,000 gpm of finished foam solution was directed at the burning tank. After 20 minutes, the foam lines were shut down, and PKP again was applied into the eyebrow vents from Ladders 2 and 18.
The second PKP application was made at 1554. Once again, it appeared effective, but the application duration proved insufficient to achieve complete extinguishment. With this failure, cooling lines again were reactivated, and plans accelerated for what became the final foam attack of the day.
A SECOND EXPLOSION
At 1707, the 2,000-gpm foam application that preceded the PKP attack was reactivated. This flow continued for an hour with no apparent success. Then, at 1815, the report of foam and burning product flowing from the eyebrow vents was relayed to the command post. Two minutes later, the tank was shaken by what appeared to be an explosion in the area between the pan and the external roof. Burning gasoline was ejected from the eyebrow vents by the force of the explosion. This phenomenon was most evident on the south (leeward) side of the tank. Foam safety lines were immediately placed into service to protect the personnel operating within the berm area.
The cause of this event centers on three possible theories. One theory involves a combination of the weather conditions and the buildup of the foam blanket between the floating pan and the external roof. At the time of the explosion, the fireground was experiencing high wind gusts. It is possible that as the foam blanket filled the area between the pan and the external roof, the oxygen supply to this area was significantly reduced. Therefore, wind gusts of sufficient velocity through the eyebrow vents could reoxygenate this area, causing an explosion and overpressurizing this vapor-rich space. Another possible cause is the dropping of a large amount of unburned carbon deposits, or coke, created by the incomplete combustion of the unleaded gasoline from the underside of the external roof deck onto the floating pan, thereby displacing burning product. A final possibility is that the floating pan shifted under the weight of the foam attack, expelling burning product from the eyebrow vents in a “slopover”-type phenomenon. Whatever its cause, it became a “defining moment” of the first day’s efforts.
Concurrent with this event, the command post struck a fourth alarm and ordered the evacuation of personnel and equipment from the dike area. Within approximately 10 minutes, all ground fires had been suppressed, and the ladder apparatus were withdrawing from the berm area under the protection of the foam safety lines.
This last unsuccessful attack, culminating in tank #22’s unexpected explosion, frustrated firefighting personnel, who had spent the day battling this fire. As apparatus and personnel responding to the fourth alarm, along with the relief companies, approached the scene, they were greeted by a retreating and exhausted firefighting force. Within a 16-hour period, five separate yet unsuccessful attacks were mounted against the burning tank, utilizing thousands of gallons of water and foam concentrate, hundreds of pounds of dry chemical agent, and all other types of suppression technology at the JFRD’s disposal.
REASSESSING STRATEGY
Chief Mays now conferred with Fire Chief Charles D. Clark and other senior staff members who gathered at the scene during the operation to formulate a long-term strategy for the fireground. A number of factors had a direct bearing on the immediate suppression efforts and the JFRD’s ability to mount future offensive operations:
- The fireground foam supply had been seriously depleted by the day’s efforts. While it was adequate to achieve ground fire suppression, if necessary, major restocking would be required prior to renewed offensive operations.
- Company officials reported that product transfer operations were stymied. The transfer lines had become clogged, eliminating product transfer as an immediate valid alternative.
- The ability of company officials to predict the tank’s product level based
- on transfer and burnoff rates had proved ineffective and led to a number of potentially dangerous spillovers during the foam attacks.
- Now well past sundown, renewed offensive operations, even with fresh crews and portable lighting around the fire scene, were deemed to be too hazardous.
- The tank’s structural conditions favored long-term operations. The high product level and the failure of product transfer operations, combined with the relatively slow burnoff rate, made the continuation of longterm operations a plausible option. Further, the product level could be artificially maintained with the introduction of water, as necessary, to ensure the stability of the external shell.
In an effort to preserve the tank’s integrity, cooling lines were reapplied to the tank shell. Water lines also remained connected to the subsurface injection point so water could be introduced to maintain the product level. This way, the tank was slowly being burned off in a controlled fashion. For maximum safety, personnel inside the dike area were limited to those necessary for cooling and reconnaissance operations.
The water buildup in the dike containment area now became a primary concern of the command staff. The flow of thousands of gallons of water per minute from cooling and foam lines was rapidly filling up the dike area. This rising water level was pushing into the operational area on the northern side of the tank, and with the rising water came the risk of a catastrophic ground fire burning on top of this water cover. Another concern was its potential undermining of the dike wall, which would expose the river and other adjacent areas outside the dike. To avert this, command began interacting with local spill mitigation firms in an attempt to find pumps that could be set up to remove the water and any spilled product from the dike area. It was Sunday before the pumps could be placed and later before the water level was substantially reduced, but this action saved the lives of firefighters during a massive spill and ground fire on Monday.
Staff discussions also centered on solutions to the day’s failed suppression efforts. It generally was concluded that an inability to get sufficient finished foam solution into all areas of the tank, in combination with erroneous estimations of the true product level, led to the day’s failures. One avenue of foam application that had not been tried that might have provided the ability to access all areas of the tank was the use of foam wands. Foam wands, placed directly into the tank through the eyebrow vents, would allow all finished foam solution to enter the tank and gently coat the burning fuel. This would circumvent the major problem encountered with over-the-top attacks on closed roof tanks, where not all finished foam enters the tank and reaches the fuel surface. An over-the-top attack is limited by wind and weather conditions as well as the number and sizes of the tank openings, all of which can substantially reduce the volume of finished foam that actually reaches the burning product. To effectively pursue this option, fireground commanders realized that they would need outside technical assistance.
TANK FIRE SPECIALISTS
By midday Sunday, Williams Fire & Hazard Control, Inc., of Mauriceville, Texas, had been contacted and responded to the scene.
After an extensive scene survey, Dwight Williams, president of Williams Fire & Hazard Control, proposed another ground-based overthe-top attack, augmented by use of the foam wands he had brought and other wands that would be fabricated locally. Included among these fabricated wands were a four-inch “T” wand and another wand with a Bresnan distributor on the end. The proposed attack would expend 3,285 gpm of finished foam solution, consuming 99 gallons of foam concentrate each minute and a staggering 5,940 gallons of concentrate per hour.
A massive logistical buildup was required to prepare for the attack. The number of fireground personnel increased to well over 125, requiring complex planning for personnel rest and rehab. The foam, water supply, and tactical support sectors were probably the busiest sectors during these preparations. The tactical support sector interacted with local pipe fitters, who fabricated equipment to Williams’ specifications. This sector also fueled, serviced, and repaired all apparatus and equipment on the fireground. Throughout the operation, it also acquired additional vehicles necessary for each sector’s operational needs. The foam sector, predominately staffed by JFRD volunteers, shuttled foam with forklifts, front-end loaders, wreckers, and pickup trucks as well as by hand from the central storage location established on site to all preselected foam-eduction positions.
The operation’s demand for apparatus and personnel placed a severe load on the JFRD’s resources. To counter this, alarm response assignments were reduced on a citywide basis. This left more of the remaining units available to cover other emergencies should they arise.
Preparations for the attack were not complete until almost midday Monday, January 4. Firefighters had labored throughout the morning to place the foam wands into the tank at points designated by Williams. By 1100, the equipment had been placed, and the attack was initiated. The foam application continued for approximately one hour, when burning product and foam once more began flowing from the tank’s eyebrow vents. As the severity of the overflow conditions increased, the operation was abandoned, all lines were shut down, and all unnecessary personnel were withdrawn from the dike area.
GASOLINE BACKFLOW
At this point, a nearly disastrous chain of events occurred. When the foam flow into the four-inch-diameter “T” foam wand was cut off, the product level in the tank rose above the bottom of the wand. This created a reverse siphon flow of raw gasoline directly from the top of the tank into the wand and its five-inch supply line, exiting at the aerator ports of the 500gpm high back pressure-forcing foam maker well back in the dike area. Since this eventuality was not foreseen, no backcheck devices or control valves were installed in the wand or its supply line. To compound the problem, the reverse flow occurred as Ladder 2 was being withdrawn from the dike area, and it was initially assumed that the flow was only water resulting from the ladder’s undercarriage severing a supply line that rested under the surface of the dike area’s water and foam layers. Approximately five to 10 minutes passed before the problem was identified by teams operating in react mode. Williams capped the hose fitting and stopped the flow of gasoline.
Minutes later, burning product flowing from the tank ignited the gasoline spilled from the foam wand, which now had drifted into the impounding area. Fireground forces immediately assumed a defensive posture and protected the adjacent tanks, as a ground fire burned at one end of the impounding area for approximately 15 minutes. The dike fire was allowed to burn itself out, removing the danger of the spilled product. Thankfully, no one was injured in the spill or ignition, due in large measure to the foresighted actions of the fireground commanders, although one of the industrial pumps extracting excess water from the impounding area was destroyed by the fire.
Fireground foam supplies were seriously depleted by this latest effort, removing the possibility of any renewed offensive operations. Once more, fireground operations settled into a holding action, as cooling lines were reapplied to the tank shell.
Attention now turned to the process of transferring product from the burning tank. Transfer operations had been underway earlier in the day, reportedly achieving a rate of 1,100 barrels per hour. However, a mechanical failure around noon substantially curtailed this operation. Fireground commanders felt that the attack had failed because the product level was still too high, creating insufficient space between the internal pan and the external roof cone. More product had to be removed to provide for an effective period of offensive operations.
In an effort to overcome these problems, crews from a local petroleum firm attempted to replace the pump, located at the foot of the tank. The pump had been destroyed by ground fires, and it was hoped that a replacement would solve the transfer dilemma. Unfortunately, it proved to be impossible under the dangerous circumstances; and while crews were able to reroute some piping to achieve a gravity flow, a restriction quickly developed in the line, resulting in a very low transfer rate. Another means of product transfer had to be found.
The accidental suction provided an idea for an alternate method of product transfer. If a large pipe similar to the wand that leaked could be fabricated and placed into the tank through the large opening on the northeast side, suction then could be mechanically created and the product drawn off from the top of the tank. The tank then could be “burned off” in a controlled fashion, by injecting water into the bottom of the tank to maintain a high product level and siphoning it off at the same time. This idea provided the added advantage of ensuring the tank’s structural integrity.
The pipe fabrication required for this six-inch siphon was not completed until late Tuesday afternoon, and placement was attempted at 2000. First, the tank was foamed in an attempt to reduce the flame intensity. It was hoped that this would provide some protection for those individuals in the tower ladder platform placing the suction device. How ever, as members attempted to insert the siphoning unit, a severe, fast-moving thunderstorm struck the foreground. Wind gusts quickly made the operation extremely hazardous, and it was aborted. During the withdrawal, the main control valve was broken off the siphon, rendering it useless. Repairs were initiated on the piping, and late Wednesday morning a second effort resulted in the suction device’s successful placement.
After the latest setback in the product transfer operations, a meeting was held with the foreground commanders and the JFRD staff to again address overall strategy. Sufficient foam supplies now had been stockpiled (12,000 gallons) to allow for consideration of another attack. In addition, the Port Everglades Fire Department loaned the JFRD a 2,000-gpm portable foam monitor. After a careful assessment of the available options, Fire Chief Clark authorized one final foam extinguishment effort and allocated 6,000 gallons of foam concentrate from the storage stockpile. If this failed, the backup plan was to use the six-inch siphon to transfer as much product as possible, allowing the remaining few inches to burn off.
THE FINAL ATTACK
The attack would employ a layered approach, beginning with using only the subsurface lines and the foam wands. The subsurface application would employ a rate of 500 gpm, while the five foam wands would distribute a total of 1,025 gpm. Calculations showed that this combined flow would be adequate to “sandwich” the fire and extinguish it; if it was not, other options could be employed. In total, this attack would distribute 1,525 gpm of finished foam into the tank.
If the wands failed or if extra punch was needed, the attack would be augmented with the use of prepositioned portable foam monitors. A 1,250-gpm foam monitor was positioned just northwest of the tank, while a 1,000-gpm foam monitor was set up on the tank’s southeast side. The 2,000-gpm foam monitor was set up on the terminal’s western dike wall; from there it was capable of dropping foam or water onto the tank’s roof. This monitor would only be employed, if necessary, in the final stages of the attack, estimated to be less than 15 minutes, to overwhelm the remaining pockets of fire and pound the external roof to loosen built-up coke deposits.
For this attack, the fireground was configured to provide a total finished foam flow of more than 7,000 gpm, including the safety handlines. If all resources were employed, the perminute demand for foam concentrate would exceed 200 gpm. The water supply and foam sectors immediately began coordinating efforts to accommodate the fire flows required. A total fire flow capacity in excess of 9,000 gpm would be required to support the planned offensive operations. For this attack, all foam would be educted from points outside the dike wall. This would allow for a safer operation and a more efficient resupply effort. Much of this day’s efforts were concentrated on supply and attack line positioning, along with the replenishment of foam stockpiles to support this effort.
To provide sufficient space in the tank for the foam attack, the product level, which had been held constant, again was lowered by terminal officials. The transfer rate was very slow, but by the late afternoon a noticeable reduction in the product level of the tank was evident.
It was well after dark before the outage was acceptable to commence attack. As it neared completion, a significant deterioration of the upper tank shell was noted. To combat the worsening condition of the upper tank shell, the decision was made to accelerate the attack schedule and proceed as soon as practically possible. By 2107, all sectors reported that they were staffed and ready; at 2112, air horns heralded the start of the attack.
Almost immediately, the attack ran into difficulty. The distributor nozzle on the three-inch foam wand seized up from the heat and did not operate to full effect. To compensate, command ordered the activation of both the 1,000-gpm and the 1,250-gpm foam monitors.
Since the attack proceeded effectively and the fire conditions were being reduced, the 2,000-gpm foam monitor was activated earlier than anticipated and flowed for longer than the initial projection. As a result, foam stockpiles now were rapidly being depicted by a foam flow in excess of 170 gallons of concentrate per minute. Conditions became so critical that firefighters were tilting the hydrant basins to allow the remaining foam concentrate to be educted. At this moment, two tractor trailer loads of foam concentrate, each arriving from a different embarkation point, arrived at the terminal gate. Their foam supplies were quickly shuttled to the foam-eduction positions just as the on-scene foam supply was exhausted.
At the 50-minute mark in the operation, foam and burning product again began flowing from the eyebrow vents. It was concluded that the fuel level could not have reached the eyebrow vents in the short time that the attack had been in operation. Therefore, the foam had to be filling the space between the wedged floating pan and the external roof. At the 55-minute mark, knockdown was achieved.
Foam application continued for another hour because of the reignition potential from the coke buildup. Foam and raw fuel poured from the eyebrow vents. Only a few visible flames continued to burn at the southern “fishmouth” opening. A few feet of foam also coated the entire dike impounding area, a combination of overflow from the extinguishment effort and the foam handlines attempting to suppress vapors from overflowing gasoline. As parts of this foam blanket drained, patches of the foam were lifted by the wind and carried over the pumps on the dike wall, where the pumps’ exhausts ignited the vapors entrapped in the foam.
The timely delivery of the newly arrived foam concentrate supplies had allowed the foam operations to continue without interruption. This allowed the fireground forces to completely overwhelm the remaining flames, and at 2309 the fire was declared extinguished.
POSTEXTINGUISHMENT
Foaming operations continued for seven more hours, at 15-minute intervals, to cool the tank and to prevent coke reignitions. One reignition occurred around 0300 on Thursday morning and was quickly suppressed. This precautionary foaming operation remained in effect at ever-increasing intervals for the next 32 hours.
On Friday afternoon, a group was contracted to recover the remaining gasoline from the tank. Transfer was accomplished with the use of a nitrogen-inerting system, which maintained an inert gas within the vapor space of the damaged tank, allowing for the safe removal of the unburned gasoline. Transfer operations, including the use of the six-inch siphon, continued until Monday, January 11. Firefighters were on standby during this time for the “safing” of the tank. Company officials estimate that almost two million gallons of product were recovered from the damaged vessel during the various transfer operations.
An interesting event occurred late Friday afternoon just after the inerting system was established. Without warning, product suddenly was ejected from tlie eyebrow vents. When this occurred, no control operations of any kind were being applied to the tank. This incident never reoccurred, and those who witnessed the event believed the floating lid, now cooled, had sunk to the bottom of the tank.
The tank itself was dismantled twoand-a-half months after the incident, allowing investigators an unrestricted view of internal tank conditions. They found the tank’s interior walls had been coated with a fiberglass liner, which now flaked off in large sheets, draping onto the floating pan, and in all probability caused the clogging problems encountered while attempting the off-loading of product from the tank. Wicking of the fiberglass also may have occurred—a Class A hazard inside the tank. This coating had not been noted on the tank drawings, and company officials were not aware of its existence.
The roof trusses were twisted and distorted, having been seriously warped by long-term exposure to intensive fire conditions. In contrast, the internal H-columns supporting the roof trusses and serving as guides for the internal floating pan held up amazingly well, protected as they were for most of the incident by the high product level.
The internal floating pan was found on the tank bottom, collapsed on the lock-down legs and tank piping. It had sustained extensive warping as a result of the fire conditions. The pan’s warped areas were grossly distorted—some areas or sections by as much as eight feet—in comparison with the normal pan line. The most seriously warped area of the floating lid was directly under the large opening on the northeast side of the tank. More significantly, approximately one-third of the floating roof’s seal remained undamaged.
These results supported the supposition made during the fire that the floating lid was indeed tilted and jammed into the roof trusses by a combination of the overfill and explosion, creating the difficult extinguishment conditions experienced during the fire. Its resting position also was compatible with the observations of those who associated the sudden product overflow on day seven with the sinking of the floating roof. This phenomenon was probably the result of the cooling and contracting of the tank and its supports, which allowed the pan to be released.
The results of research and discussions with numerous individuals associated with the petrochemical industry have led many to conclude that this incident represents the first successful extinguishment, utilizing portable foam equipment, of an internal floating roof tank of this size containing unleaded gasoline. While this claim appears impossible to confidently verify, it does appear that this represents a unique accomplishment, and one—given the number of similar storage containers throughout the industry-worthy of examination.
This fire provided the JFRD with a number of valuable lessons in terms of firefighting tools and techniques and refinements to its incident command policies, procedures, and practices. Many of these improvements currently are being implemented. However, the safe and successful resolution of this very dangerous incident stands as a testimony to the courage, dedication, and expertise of all those affiliated with the operation.
LESSONS LEARNED AND REINFORCED
Editor’s note: These lessons were compiled from interviews with Dwight Williams, president of Williams Fire & Hazard Control, and from the Jacksonville Fire and Rescue Department incident critique. We are grateful to all those who have devoted much time and energy to sharing these learnings with firefighters around the country, especially Fire Chief Charles D. Clark and Division Chief Randall W. Napoli, JFRD, and Dwight Williams.
- This fire operation is a good example of an experienced and welltrained fire department working together with a private firefighting company. The beneficial synergistic effect of the interaction between these two organizations was the key to successfully extinguishing this fire.
- By maintaining the structural integrity of the tank’s shell, controlled long-term operations were possible. Suppression operations could be planned and properly prepared prior to implementation. Shell integrity was maintained through a variety of factors: the use of cooling lines, the relatively slow burnoff rate (a result of the overflow and the failure of the product transfer operation), and the
- addition of water into the tank through subsurface injection lines, creating a large, controlled burnoff. Cooling lines on the portion of the tank wall subject to fire attack must be maintained for the duration of the fire.
- Covered floating roof tank fires are a great firefighting challenge. The combination of the tank design and damage caused by the overfill and explosion contributed to the limitation of access points for foam application and the creation of void areas that were difficult to penetrate with extinguishing agents. As a result of this dilemma, several extinguishment efforts were required, with foam flows and durations significantly in excess of those anticipated in NFPA 11. Note, however, that the final extinguishment itself required approximately 7,200 gallons of foam.
- Firefighters used ingenuity, imagination, and innovation to overcome the numerous problems they faced. Final extinguishment came through a combination of attack methodologies:
- subsurface injection, foam wands, and an over-the-top attack with portable foam monitors. The fire literally was sandwiched and extinguished between a massive foam application from the bottom and on top of the floating pan. The ability to adapt to a changing incident is critical to the successful extinguishment of these fires. Strategies and tactics must constantly be reevaluated and new methods built on the lessons learned from unsuccessful attempts at extinguishment. For example, the concept of the large mechanical siphon for product removal was borne out of the backsiphon of product through the highpressure foam maker.
- Information concerning the amount of product in the tank proved to be unreliable. Although well-intentioned, such misinformation can prove to be a major problem for forces extinguishing a tank fire, in which case product level is critical. Firefighters subsequently used threequarter-inch sample valves adjacent to the tank ladder to determine the
- levels of product and foam in the tank.
- After the manifold fires had been extinguished, flanges at the base of the tank required constant tightening to avoid any additional three-dimensional fires.
- Safety is a primary concern. The importance of having “react teams” operating portable monitors and hoselines ready to protect personnel operating near the tank in a variety of capacities, including hardware positioning and adjustments, cannot be overemphasized. These teams must be prepared to act immediately to protect personnel and contain the fire. Numerous occurrences at this fire, including the backflow of product through the high-pressure foam maker and the unanticipated explosion, illustrate the need for react teams.
- In the Steuart incident, the tank’s floating pan was thrust up to the tank cover and wedged there in the explosion. In situations where the pan does not sink to the bottom of the tank, the possibility exists that the pan can shift
- during firefighting operations. Such a shift can cause product to be expelled through the eyebrow vents. The outcomes of a potential pan shift must be considered, particularly for ground forces around the base of the tank.
- Moving significant quantities of foam concentrate supplies must be initiated early, as supplies may come from long distances.
- Three foam-induction points eventually were established. Keeping the portable folding tanks full was an essential, labor-intensive task.
- Eyebrow vent screens are difficult to remove. A maul, a long section of pipe, and considerable physical effort were necessary to push them out of the way for the foam wands.
- Holes put at the top of wands prevent back-siphoning should the wands become submerged in product.
- Relief companies should be staggered to allow for operation/attack continuity.
- Automatic high-level alarms, automatic diversion, and automatic differential chamber, opening or closing the main valve of the assembly.
To alleviate confusion, the following definitions are reprinted from the 1993 edition of NFPA 14, Installation of Standpipe and Hose Systems.
Pressure-control valve. A pilot-operated valve designed for the purpose of reducing the downstream water pressure to a specific value under both flowing (residual) and nonflowing (static) conditions.
Pressure-reducing valve. A valve designed for the purpose of reducing the downstream water pressure under both flowing (residual) and nonflowing (static) conditions.
Pressure-regulating device. A device designed for the purpose of reducing, regulating, controlling, or restricting water pressure. Examples include pressure-reducing valves, pressure-control valves, and pressure-restricting devices.
Pressure-restricting device. A valve or device designed for the purpose of reducing the downstream water pressure under flowing (residual) conditions only.
Pilot-operated valves control pressure to a portion of the system —i.e., a lower zone in a high-rise building. These valves, which are rated for 175psi or 300-psi inlet pressure, are not practical for hose-valve applications. They may be useful for sprinkler systems on high-rise floors where system demand —400 or more gpm —requires a connection to a 2 1/2-inch or larger standpipe. Often, a single valve of this type is used to regulate pressure to a specific riser serving the lower floors of a building where multiple risers (zones) are used with a single fire pump.
Fire hose valves and sprinkler system floor control valves are pressurereducing valves of the direct-action type. They are installed on standpipes where static pressure is greater than 175 psi and less than 400 psi. The valve reduces outlet pressure to a predetermined static and residual pressure at a given flow.
The basic valve design consists of a piston, a valve stem, and a valve disc, which is free to float in the valve after the valve handwheel is turned to its fully open position.
In some pressure-reducing valve designs, the outlet pressure vs. inlet pressure is based on the diameter of the piston relative to the valve seat. Other valve designs include a spring whose tension has been factory-adjusted or field-set to reduce inlet pressure to the desired outlet pressure for a given flow.
APPLYING PRESSURE-REGULATING HOSE VALVES
Factory-set valves currently are manufactured by Elkhart Brass, Grinnell, and Powhatan Brass. When valves are ordered, the contractor should specify maximum inlet pressures, desired outlet pressure, and desired flow for the various valves. The contractor also must ensure that each valve is installed on the correct floor, based on inlet pressure conditions. Pressure-reducing valves may be applied over a short range of inlet pressure conditions. Depending on the manufacturer’s design, each valve may be applied to only four to nine different pressure ranges. A given valve’s outlet pressure will be lowest at the highest applied point and highest at the lowest applied point on the riser. If a valve is tested and outlet pressures are too high or too low, the valve must be moved. These valves are not field-adjustable.
Field-adjustable valves are produced by Giacomini and Zurn Industries. A contractor adjusts valves by referring to the manufacturer’s instructions and charts, which are shipped with each valve. To adjust the valve properly, the contractor must know the static and residual pressure available at the valve inlet and the static and residual pressure and desired flow at the valve outlet. The valves may be adjusted before or after installation.
The Giacomini valve is adjusted by rotating a cylinder in the valve bonnet, using the ⅛-inch-diameter steel rod provided with the valve. The cylinder is rotated until the number shown on the appropriate chart lines up with the setting arrow.
Note: The numbers on the valve and those listed in the charts are dimensionless. They do not indicate valve outlet pressure.
Hie Zurn valve is adjusted by removing the handwheel assembly. The 1 ⅛-inch-deep socket provided by the contractor is used to adjust a nut on the “floating” valve stem. The distance from the top of the valve stem to the top of the adjustment nut must be equal to the distance shown on the valve manufacturer’s charts. The handwheel assembly is reinstalled, and the valve itself is ready to be reinstalled and operated.
TESTING VALVES
How do your know if a valve will work as specified? Test it! Pressure regulating devices fail for several reasons, including the following:
- They are incorrectly applied or adjusted.
- Mineral deposits or other waterborne debris has collected inside the valve.
- Their components have worn.
- Water supplies to the system have deteriorated.
As stated in NFPA 13 and NFPA 14, new valves should be tested before the system is accepted. At a minimum, the test must indicate inlet static/residual pressure and outlet static/residual pressure at a specified flow. Static pressure should not exceed, and residual pressure should not be less than, the limits specified.
NFPA 25, Inspection, Testing, and Maintenance of Water Based Fire Protection Systems, requires sprinkler-system valves to be flow-tested annually. The standard requires standpipe hose valves to be flow-tested every five years. My personal opinion is that hose valves’ outlet static pressure also should be tested on an annual basis. This can be done when the sprinkler floor control valves are tested, at little additional cost.
(Photos by author.) Flowmeter by Gerand Engineering Co. Patent pending.
Valves may operate improperly for several reasons, including the following:
- improper installation or adjustment;
- debris in the valve;
- worn piston rings;
- leaking valve seats;
- corrosion of valve parts, including springs; and
- problems in other parts of the system, such as “worn” fire pumps, closed valves, and plugged piping.
Other items can cause pressurereducing/control valves to deliver inadequate pressure and flow, including
- diminished public water supply and
- the addition of backflow preventors to the building’s supply piping, installed after the building was built.
TEST PROCEDURES
Tests of pressure-regulating hose valves can be done by installing a pressure gauge and flow-measuring device at the valve outlet. Ensure that sufficient hose is available to reach from the flow-measuring device outlet to a place where approximately 250 gpm can be safely discharged The 1993 NFPA 14 standard requires a three-inch drain to be installed for testing purposes. For older systems, apparatus have been developed to test these valves, using existing standpipe riser equipment (see “NFPA 25 and the High-Rise Building,” Industrial Fire Safety, July/August 1993, p ⅜2 >.
Static tests on hose valves require installing a valve cap with a pressure gauge (zero to 400 psi) and a small bleeder valve on the hose valve outlet. Fully open the handwheel, note the pressure, and close the valve. Open the bleeder valve to discharge trapped pressurized water into the bucket. Remove the test cap, replace the valve cap, and move on to the next valve.
All components of the extinguishing systems need to be tested initially and periodically. Ask yourself the following questions:
- Are all parts properly applied and functional?
- Is there wear to any part of the system or another change that may result in inadequate operation?
- Do the responding fire companies know how the system works?
Responding departments’ familiarity with the system is as important as thorough testing and inspections. Inspections and testing are precautionary measures, but knowing how pressure-regulating devices work is essential if problems are to be directly addressed in the case of an actual event.