JACK FAIRCHILD GREW UP IN BROWN MILLS, New Jersey, located just south of the swath of acreage that includes Fort Dix and McGuire Air Force Base. Just to the north of the bases was Jackson Township, home to Six Flags Adventure Park, where a fire in the Haunted Castle attraction killed eight teenagers on May 11, 1984.
The blaze occurred the day before Fairchild’s tenth birthday, and he recalls the adults around him buzzing about the nearby fire and loss of life. “We were planning a big party, and most of the discussion was about the appropriateness of having a celebration,” he says. “We’d had a home fire a year or two before, which I can still vividly picture in my head. The combination of the two really unnerved me.”
They also fascinated him. When the time came to look at colleges and figure out what he might want to study, he chose the fire protection engineering technology program at Delaware Technical Community College. “It just struck a chord,” he says.
Fairchild wasn’t alone—the Haunted Castle fire struck a chord with a lot of people. The interest was due in part to some of the fundamental questions that arose in its aftermath—primarily as part of an ensuing criminal trial—and to the surprising answers generated by those questions. For example, Haunted Castle was not sprinklered, and the question arose over whether lives would have been saved had a sprinkler system been installed. One expert told the court that, in his opinion, sprinklers would have made no difference in the loss of life—an otherwise predictable observation on behalf of the defense, except that it was made by the chair of the NFPA Technical Committee on Automatic Sprinklers, who at the time also happened to be the chair of NFPA’s Board of Directors. Conversely, an analysis of potential sprinkler performance in the Haunted Castle conducted a year after the fire found that sprinklers could have prevented the fatalities.
As the thirtieth anniversary of the Haunted Castle fire approached, Fairchild, now manager and senior associate of fire protection and life safety at Ballinger, an architectural/engineering firm in Philadelphia, decided to take another look at the event. He wanted to use state-of-the-art computer modeling to analyze the growth rate, potential sprinkler response times, water delivery times, maximum heat release rates, and tenability criteria of the fire to see if he could reach a definitive answer on the sprinkler question. He was joined on the project by Brad Casterline, a project manager at FSC, Inc., an engineering firm in Overland Park, Kansas, who contributed experience in fire and panic egress modeling.
Fairchild and Casterline will present the results of their work in an education session ("Revisiting Analysis of Potential Fire Sprinkler Performance in the Great Adventure Fire of May 11, 1984") at the 2014 NFPA Conference + Expo in Las Vegas in June.
“We thought it would be interesting to look at this again, considering the improvements in modeling over the past 30 years,” says Casterline. “The place was just a disaster waiting to happen.”
|
 |
A postcard of Haunted Castle, a popular attraction at Six Flags Adventure Park in New Jersey.
|
|
The fire and its aftermath
The work by Fairchild and Casterline is the latest chapter in a saga that began with the tragic deaths of eight young people on a spring evening in 1984.
The Haunted Castle was a popular attraction at Six Flags. It consisted of a series of darkened, convoluted passages meant to disorient visitors and was populated with displays—rat lady, hunchback, the spider room, and so on—intended to scare them. The Castle’s wood-frame façade featured arches and turrets, covered with sprayed and painted urethane foam, where visitors entered and exited. Most of the attraction was contained in 16 commercial truck trailers, connected by plywood and wood framing. The trailers housed two separate Haunted Castle attractions, mirror images of each other, which operated simultaneously during peak times. (Only one side was operating the night of the fire.) An additional trailer housed the control room, which included the electrical panels for the attraction and two pressurized-water fire extinguishers.
The interior of the attraction was constructed of plywood and wood stud partitions, with plywood ceilings suspended below the trailer roofs. The corridors were covered variously with wood, paper, and fabric, and one was covered with polyurethane foam. There were two emergency exits equipped with panic hardware and illuminated exit signs. The attraction also included emergency lighting that would illuminate the interior if the power failed. There were no smoke alarm or sprinkler systems.
The fire broke out shortly before 6:30 p.m., when a 14-year-old boy who was using a lighter to find his way through the attraction accidentally walked into the polyurethane foam padding covering a wall and set it on fire. He tried to put the flames out, but failed. According to an investigation report prepared by John Bouchard, then assistant division director of NFPA’s Engineering Services Division, the boy “apparently continued through the Castle” without notifying anyone of the problem.
|
 |
Each of the two Haunted Castle attractions consisted of eight commercial truck trailers, connected by wood and plywood and containing a maze of passageways.
|
|
|
 |
The walls of the "strobe room," where the fire began, were covered with a polyurethane foam padding.
|
|
|
 |
An additional trailer housed the control room, which accessed both sides of the attraction.
|
|
Roughly 30 people were in the attraction when the fire began. Some encountered smoke and immediately headed for exits, while others thought it was a Haunted Castle effect; one group of visitors apparently saw the fire in its incipient stage and “remarked how real the illusion seemed,” according to Bouchard’s report. The foam-covered walls ignited, and the fire, fueled by the plywood and other materials, spread rapidly. Flashover occurred three and a half minutes after the fire began, according to the investigation report, and soon the entire attraction was engulfed in flames.
A Castle employee notified the Great Adventure Fire Brigade at 6:35 p.m. En route, the fire captain noticed the heavy smoke and told park security to notify the Jackson Township Police Department and start implementing the park’s mutual aid plan. Heavy smoke kept the park’s firefighters from advancing into the attraction, and a defensive attack was begun at the rear.
Two volunteer fire companies arrived shortly before 7 p.m. and were also unable to advance into the structure, which was rapidly being consumed. “In 20 minutes, the building looked burned down,” one park visitor told reporters. “There was nothing there.” When the fire was declared under control, at 7:45 p.m., 15 mutual aid companies had responded with 300 firefighters. Eight teenagers, ranging in age from 15 to 19, were found dead, victims of smoke inhalation and carbon monoxide poisoning. Seven were found in the same corridor.
Four months after the fire, Great Adventure, Inc. and its parent company, Six Flags Corp., were charged by the state of New Jersey with aggravated manslaughter, as were two park managers. According to The New York Times, the indictment “charged the companies with ‘reckless’ conduct amounting to ‘extreme indifference to human life’ for failing to take adequate fire prevention measures.”
At the trial, Chet Schirmer, president of Schirmer Engineering Corporation and chair of both the NFPA Technical Committee on Automatic Sprinklers and the NFPA Board of Directors, along with Rolf Jensen, president of Rolf Jensen and Associates, appeared as defense witnesses. They set off a storm of controversy in the fire protection industry when they told the court that, in their opinions, sprinklers would have saved some of the property, but they would not have saved the lives of the eight teenagers who died in the fire. Schirmer also told the court that smoke and heat detectors would “not have been of any particular value in this building.” (For more on Schirmer’s testimony and NFPA’s reaction, visit the story online at nfpa.org/hauntedcastle.)
Schirmer based his opinion on assumptions about the way the fire developed and the results of previous fire testing. He also assumed that any sprinkler system installed in the attraction would have been a dry-pipe system, which takes slightly longer to discharge water once activated than a wet-pipe system.
“If the sprinkler system does not operate quickly enough or if the water does not hit high enough on the walls…you cannot control the survivability of people in the space of origin,” he told the court. “In other words, it is necessary that water hit high enough on the walls to prevent fire spread, and it is also necessary that the sprinkler operate fast enough to knock the fire down and maintain tenability limits in that space of origin.”
|
 |
Haunted Castle actually consisted of two separate attractions, mirror images of each other (depicted here in blue and red), connected by a control room. Only the red side was operating the night of the fire. Circles indicate people in and around the attraction when the fire began. Crosses indicate where fatalities were discovered.
|
|
NFPA’s Bouchard, testifying on behalf of the prosecution, said that sprinklers or smoke alarms might have allowed the fire to be detected earlier, perhaps allowing the victims, none of whom was found in the room of fire origin, to escape. “The fire would have been detected earlier and perhaps controlled to a tenable situation,” he told the court.
After an eight-week trial and 13 hours of deliberation, a jury found Great Adventure Inc. and Six Flags Corp. not guilty.
The Haunted Castle fire prompted several important changes to NFPA 101®, Life Safety Code®. A new section titled “Special Provisions for Special Amusement Buildings” was added to the 1988 edition, requiring every special amusement building, such as the Haunted Castle, to be protected throughout by an approved—and properly installed and maintained— automatic sprinkler system. If the building was portable or movable, the “sprinkler water supply may be by an approved temporary means.”
If the amusement building had low lighting levels, the code required that it be equipped with a smoke detection system, the activation of which would sound an alarm at “a constantly attended location on the premises.” The actuation of either the smoke detection system or the sprinkler system would “cause the illumination in the means of egress to increase” and silence any “conflicting or confusing sounds and visuals.”
These requirements have been refined and expanded over the years, but their intent remains the same as when they first entered the code. They can be found in Chapter 12, “New Assembly Occupancies,” and Chapter 13, “Existing Assembly Occupancies,” in the current edition of the Life Safety Code. (For more on code changes brought about by the Haunted Castle fire, visit nfpa.org/hauntedcastle.)
The first sprinkler analysis
In response to the verdict, and finding it difficult to believe that a sprinkler system would have made no difference, the board of the National Fire Sprinkler Association (NFSA) asked Russell P. Fleming to prepare an analysis of expected sprinkler performance to see if it would support the Schirmer opinion.
Fleming—who would later serve on NFPA’s Standards Council and its Board of Directors and write a long-standing column for this magazine—began reviewing testimony, and found what he considered three significant errors made by the expert witnesses called by the defense in their assumptions about fire spread and the potential performance of sprinklers during the incident.
First, Fleming noted that they assumed an exhaust fan would have taken the heat out of the compartment of origin for the first minute after the fire started until it failed. The expert witnesses then restarted the fire as a small fire instead of acknowledging that, by then, it would have grown large enough to activate sprinklers. He also questioned the defense’s assumption that it would take a full minute for water to reach the sprinklers in the dry-pipe system once they opened. He noted that this assumed that the delay was based on the traditional maximum time of water delivery in a test simulating a single sprinkler opening, not that the time of water delivery would have been reduced considerably when multiple sprinklers open, as they would have given the rapid growth of the Haunted Castle fire. Finally, Fleming questioned the assumption that standard sprinklers would have responded too slowly and would not have put water high enough on the wall to stop the fire from spreading. Based on work that had gone into the development of residential sprinklers, Fleming believed that sprinklers had successfully proven their abilities in spaces with similar geometries.
Using the defense assumptions in combination with the quantitative methods available to him at the time, Fleming analyzed the likely performance of automatic sprinklers in the Haunted Castle fire. In his final report, “Analysis of Potential Fire Sprinkler Performance in the Great Adventure Fire,” Fleming concluded that “although hazardous conditions would likely have developed within the [room of origin] itself prior to water delivery, hazardous conditions would not have developed outside the [room of origin], and safe egress from the facility would likely have been maintained.” In short, “it is likely that the loss of life would have been prevented.”
Fleming’s report appeared in Sprinkler Quarterly magazine, and the plaintiffs’ attorneys later contacted NFSA to ask if they could use it in support of their civil suits. Most of the suits were settled out of court, however, and Fleming does not know if the report played any role in negotiations.
The new sprinkler analysis
Years later, while reading about fire modeling, Fairchild stumbled across Fleming’s report. The contents fascinated him. “Fleming performed an exquisite analysis based on the tools available at the time,” says Fairchild. “I realized he laid the groundwork for a great comparison—he beat the path, and we retraced his steps.”
For their analysis, Fairchild and Casterline used computer models, courtroom testimony, and any other information they could find. Along with computing capabilities immensely more powerful than Fleming’s, the pair was able to utilize a far more advanced modeling tool known as the fire dynamics simulator (FDS), software that could account for the interaction of numerous fire phenomena. FDS is known as a “field model,” where all physical objects, including obstructions, vents, fans, and persons, are entered into the model and given appropriate properties. The simulation is run, and the algorithms in the model account for the various fire effects—hydrodynamic, combustion, heat conduction, suppression, detection, and more—that influence each other. Fleming’s models, by comparison, could accommodate only one fire phenomena variable at a time. Fairchild describes FDS, along with an accompanying evacuation model that contains tenability algorithms, as “a tool Russ could only have dreamed of.”
FDS does not include a fluid delivery model, however, so Fairchild enlisted the help of Mike Mahomet of Tyco, who holds the patent on Tyco’s fluid delivery time software, SprinkFDT, currently the only UL-listed software on the market for calculating dry-pipe sprinkler system activation times.
Fleming began by determining that a 6-megawatt fire was the minimum required to reach flashover in the area of fire origin. He combined this information with trial testimony to conclude that an ultra-fast fire growth curve was appropriate. Fairchild and Casterline accepted Fleming’s prescribed burning rate of the model fire because they felt there were too many unknowns to try to predict that rate. Using FDS, they were better able to predict when the design fire would become oxygen limited. Fleming then compared residential test data to confirm that the combustible walls could create a 6-megawatt fire. Fairchild and Casterline agreed.
Fleming was unable to model fire suppression, extinguishment, or control, but he did have a tool called DETACT-QS—the acronym refers to detector activation—which computes heat detector (sprinkler) activation time based on the sprinkler response time index, temperature rating, and spacing, in concert with the heat release rate (HRR) and ceiling height. “It’s great for a rough approximation of initial detector activation,” Fairchild says, “but beyond that other effects come into play . . . A fire is a complex reaction where one variable can have a great influence on another.” Because the fire was located on a wall, DETACT increased the heat release rate by a factor of two. To account for the exhaust fan in the area of origin, the heat release rate curve Fleming entered into DETACT assumed all of the heat was removed up to 100 seconds into the fire. Fairchild and Casterline modeled the exhaust fan directly into FDS.
Using the previously developed data and expert witness testimony to locate the sprinklers, Fleming calculated the dry-pipe sprinkler operating times. Fleming could only calculate operating times before water was delivered—DETACT could not model suppression—but Fairchild and Casterline could model sprinklers directly into FDS. In FDS, the water delivery time included a programmed delay based on the times produced by SprinkFDT. Fleming used a computer program developed by Factory Mutual to determine dry-pipe valve trip time, which Fairchild and Casterline also accomplished with SprinkFDT. Fleming determined sprinkler discharge densities, which Fairchild and Casterline confirmed.
Fleming reviewed his sprinkler application rate and water distribution pattern data, compared it to full-scale testing of a mobile home that had been allowed to burn to near flashover, and concluded that sprinkler discharge would have suppressed the fire. Fairchild and Casterline modeled the sprinkler application rates and water distribution patterns directly into FDS, and concurred that sprinklers would have applied water at a sufficient rate and in a distribution pattern that would have suppressed the fire.
On the topic of sprinkler suppression and its impact on life safety, however, Fairchild and Casterline hit a snag. To determine the impact of sprinkler control on life safety, Fleming again referred to test data and concluded that, if flashover had been prevented, life-threatening conditions would not have developed in areas remote from the fire, such as those where the eight fatalities had been discovered.
Using FDS, Fairchild and Casterline measured carbon monoxide and oxygen depletion values. But while they agree with Fleming that sprinkler control would have been effective, they found that the activation time was much more difficult to predict. In their analysis, activation time depended in part on the properties of the foam and plywood, such as their density, hydration levels, the surface area of the foam, the wood species of the plywood, and more. “Predicting sprinkler activation times is highly dependent on fire growth rate, which DETACT has no ability to model,” Fairchild says. “We were hoping to find a smoking gun, but there doesn’t appear to be one. Instead, we can offer probabilities.”
He stresses that it’s not game over, though. “We’re still studying it,” he says. “I’m not giving a defined conclusion until June 10. But maybe not even then.”
Bottom line: So far, at least, even with all their whiz-bang technology, Fairchild and Casterline cannot provide a definitive answer to the question, “Would sprinklers have prevented eight teens from dying in the Haunted Castle in 1984?” It’s just as difficult to say if Schirmer’s opinion was wrong. “We’ll never be able to replicate the exact conditions of that night 30 years ago,” Casterline says. The answers lie somewhere in shades of gray.
“Sprinklers are sometimes thought of as a magic bullet that can solve nearly any fire protection problem,” Fairchild says. “They’re really just another piece of the overall building safety puzzle, and they’re dependent on the other pieces they’re attached to. Just as seatbelts and airbags are critical safety components in a car, they would not perform as well if connected to a soap box racer.”
That’s a sentiment Chet Schirmer would probably agree with.
Kathleen Robinson is editorial operations manager for NFPA Journal.