When it comes to fighting tank fires, Sunoco Fire Chief William C. Kelly freely admits to cheating. Last July, when a lightning strike ignited a 100-foot-diameter xylene storage tank at Sunoco's Eagle Point refinery in Westville, N.J., Kelly exceeded the NFPA guidelines for foam application by nearly 40 percent.

"It's good to have those guidelines, but if we can put the fire out faster by cheating, I'm okay with that," Kelly said. "The sooner the fire goes away and the smoke plume leaves, the sooner the community gets happy again.

"Cheating is okay."

Using a one-point application, firefighters attacked the fire with a non aspirating nozzle capable of 3,000 gpm maximum. The NFPA guideline only called for 1,200 gpm under the circumstances. Instead, Kelly ordered that the nozzle be set for a 2,000 gpm delivery.

The 20-minute foam attack — "shooting juice" as Kelly calls it — ended a 3½-hour tank fire producing enough smoke to be visible from Philadelphia almost 15 miles north. Using a three percent application, firefighters achieved flame collapse within the first 10 minutes of the attack. Unofficially, that application rate may have gotten as high as 12 percent, Kelly said.

"The folks in charge (of the nozzle), when they saw that flame being knocked down, got really impatient and started turning the dials to the right," Kelly said. "I don't know what we were shooting by that time. I don't care. I don't think anybody cares as long as the fire went out."

Kelly serves as fire and emergency services supervisor for three Sunoco refineries in Philadelphia, Marcus Hook, N.J. and Westville, N.J., all within 20 miles of each other. He operates from an office in Marcus Hook. Altogether, the three refineries have an emergency response team of 33 full-time firefighters and more than 140 volunteers.

By coincidence, the ERT spent the entire month of June training on storage tank firefighting, Kelly said.

"We had just done four consecutive weeks of doing everything we did on fire day in preparation for a tank fire," he said.

Chemicals produced at Eagle Point include benzene, toluene, and xylene, a chemical that occurs naturally in crude oil but is supplemented by additional amounts made in the refining process. It is used in many common products, such as paints, rubber, adhesives, plastic bottles and clothing.

Xylene is also highly flammable. On July 11, 2007, the Eagle Point refinery was put on alert regarding approaching electrical storms in the area. All employees were warned to leave the towers and move indoors. The strike that triggered the fire hit about 4:45 p.m.

"We had one operator who was heading toward the tank that was hit to gauge it," Kelly said. "Because of the weather alert he came back to shut off the transfer pump. That was fortunate for him."


The storage tank stood 38 feet tall, with five sheets from top to bottom in its shell. Filled to within 10 feet of the top, the tank contained 1.5 million gallons of xylene. It utilized an older style cone roof still common in northeastern U.S. refineries.

American Petroleum Institute approved designs for cone roof tank include a weak seam around the top as an emergency precaution. In this case, the roof did exactly what it was designed to do in an explosion, Kelly said.

"It's designed so that an overpressure lifts the roof up and gets it out of the way so that you can actually get to the fire," Kelly said. "The roof came completely off the tank and hung off the outside."

With the tank's contents now exposed, the fire went full surface. Wind from the southwest blew smoke over inland areas of New Jersey. Emergency responders from all three of Sunoco's northeast refineries aligned as one task force, reacting immediately. All available 7¼-inch hose and foam concentrate was rushed to the scene.

The burning tank was located in its own containment dike, although product escaping the tank was never a problem. With regard to surrounding tanks, there was no immanent threat of the fire spreading through the refinery, Kelly said.

"We have really good spacing at the Eagle Point tank farms, so there wasn't a lot of initiative for cooling surrounding tank exteriors," he said.

A monitor was positioned to protect a nearby tank that would be vulnerable to radiant heat only if the wind shifted. Other than to test that monitor, little time or effort was wasted on protecting other tanks.

Likewise, Kelly did no work to cool the exterior of the burning tank itself.

"From what I have learned, unless you can cool the exterior a complete 360 degrees around you just increase the chance of a shell failure," he said. "We don't like to cool the shell."

As rehearsed in June, the game plan was to make a one-point application to the burning tank. With the roof thrown back at almost the 12 o'clock position, the application would be made from the six o'clock position, Kelly said.

"We use a single point of delivery," he said. "We don't like to circle the wagons."

In preparation for the attack, Kelly utilized a pair of distance measuring devices similar to what golfers and hunters use to calculate the best location for the pickup truck mounted monitor.


"We're familiar with the tools and know what its throw is," Kelly said. "We actually stood up at the correct distance from the tank, measured it off and then put an X in the dirt where the monitor would go. When it arrived, we were able to park it in the right spot."

To operate the pump proportioning system, Kelly had a 6,000 gpm pump on a trailer set up about 700 feet away from the monitor. Dual 7¼-inch hoses linked the pump to the one-point application device. Rather than using totes, foam concentrate was fed to the manifold at the proportioning site from four tanker trucks. Additional tankers were available if needed through mutual aid.

"The good part is that nobody is really in harm's way," Kelly said. "The only folks who are close to the fire are the four or five manning the single-point application device. As part of our safety precautions, we had a pickup truck about 60 feet away with the motor running and facing away from the fire. If anything went wrong, the evacuation plan is for the designated operators to jump in the pickup and go."

One major hiccup temporarily delayed extinguishment. The first foam attack was prematurely curtailed 10 minutes into the operation when a pickup truck drove over and damaged the charged supply line feeding the pump.

"We were all getting cocky and ready to get our bonuses," Kelly said. "Then we lost water. It took about 25 minutes to get water reestablished and try a second shot."

Twenty minutes into the second foam attack the fire was out with only two feet of product lost from the tank. Damage was limited to the top two sheets of the tank shell. However, a foam blanket had to be maintained over the surface for the next three days.

"For various reasons we weren't able to pump it out right away," Kelly said. "We still had storm activity, so we kept a good blanket on it. We used more foam in the maintenance than we did in putting the fire out."

Going by the NFPA guidelines, the entire operation should have been accomplished with a little less than 3,000 gallons of foam concentrate. Kelly said he has no problem with having actually used almost 1,000 gallons more than that.

The key to dealing with tank fires is practice and pre-planning. Every tank at Eagle Point has its own pre-plan, Kelly said.

"When most of our industry measures tank fires in days, achieving extinguishment in 3½ hours was quite an achievement for us."