The
investigation into the causes of a line-of-duty death is one of the
most painful and difficult tasks a fire department will ever have to
face. Determining the exact sequence of events and contributing factors
that led to the death of a firefighter can take months. Even then, it’s
an uncertain science—a blending of witness testimony, evidence collected
at the fire scene and educated “best guesses” about what happened.
But an exciting new tool is helping departments hit by LODDs see a whole
new picture of these incidents using the science of computer modeling.
Given specific data from the scene, it creates three-dimensional and
animated pictures of what may have happened at LODD incidents based
on the laws of physics.
Developed by the Building and Fire Research Laboratory of the National
Institute of Standards and Technology (NIST) Fire Safety Engineering
Division, it’s called the Fire Dynamics Simulator (FDS). The FDS simulator
is already offering valuable new insight into the sequence of events
that led to recent LODDs, but this is only the beginning. NIST researchers
hope the simulator will lead to realistic computer-based training to
better prepare firefighters for the hazards they face.
NIST has been using computer modeling for years, but only recently has
it reached a point where the simulator can be run on a desktop computer.
The latest version has a feature called “Smokeview” that translates
physical data into animation and 3-D graphics—making it much easier
for the average fire investigator to use.
Dan Madrzykowski is leader of large fire research for the Building Fire
Research Laboratory at NIST. “We noticed that a picture is worth a thousand
words (in computer modeling). With the Smokeview aspect, instead of
graphs and numbers, we can now generate a picture that will give a time
and history,” he says.
NIST’s first opportunity to use FDS to help a department piece together
events leading to LODD occurred last spring, when it assisted the District
of Columbia Fire and Emergency Services Department (DCFEMS) in its investigation
of the Cherry Road Fire of May 30, 1999. The fire took the lives of
two Washington firefighters and left investigators with several baffling
questions. DCFEMS contacted NIST and asked if they would be able to
assist.
Since the fire occurred close to NIST’s facilities in Maryland, NIST
was able to respond to the scene, evaluate the structure where the fire
occurred and take the precise measurements that the computer model would
need.
According to Captain Tim Gearheart, the safety officer for DCFEMS, “The
only information we had to provide to NIST was the basic timeline.”
They did not provide them with any timeline information regarding fire
ground operations, because one of the questions they hoped the simulation
would answer was specifically when did ventilation occur and what role
did it play in the incident.
“As a committee, we were trying to explore every avenue and not leave
any stone unturned,” says Gearheart.
SEE IT ONLINE
The result of the collaboration between NIST and DCFEMS is a simulation
of the Cherry Road incident that is contained on a CD ROM. It can also
be seen on the web at http://fire.nist.gov/6510.
Robert Duval, senior fire investigator for the National Fire Protection
Association, assisted in the reconstruction and says FDS helped investigators
take a large step forward. “The incident was confusing because of the
timeline and the type of injuries. The committee was having difficulty
visualizing what had happened,” Duval says. “Was it a backdraft or a
flashover?”
By translating the physical data into graphics, the information suddenly
takes on a new dimension that clarifies the significant contributing
factors in a way that was not previously possible. “The CD shows the
heat flow and the temperatures, by transposing that with the location
of the firefighters…it cleared it all up and helped make sense of their
injuries.”
One of the perplexing questions involved the timing of when a door was
opened and how it affected the fire. By using the model, the investigators
were able to run different scenarios and see what the computer would
predict as the outcome. They could then match what the simulator showed
with information they had collected from the scene and from witnesses.
“What we looked at in DC was ‘What would happen if we took this window
out a little earlier, what would happened if we had ventilated on the
roof, or opened this particular door,’” says Frank Washentiz, safety
and occupational health specialist with the National Institute of Occupational
Safety and Health (NIOSH). Washenitz works in NIOSH’s Division of Safety
Research, and is responsible for conducting firefighter fatality investigations
across the country. “This tool gives us an opportunity to answer a lot
of questions that we could not ask without the fire model,” he says.
(NIOSH also participated in the investigation of this incident and published
a report on its findings.)
PUTTING IT TOGETHER
How is a simulation put together?
One of the first steps is to obtain a good set of measurements of the
building where the fire occurred. This includes the room dimensions,
the location of openings such as doors and windows, stairways and other
features.
Next, it is critically important
to identify what is referred to as the fuel package or fuel load that
was involved in the fire, the total quantity of combustible contents
of the space. NIST’s simulator is plugged into a database of the heat
release rates of different types of furniture and furnishings, expressed
as British Thermal Units (BTUs) or Kilowatts (kW) per second. The model
divides the space involved in the fire into thousands of “cells.” In
the DC simulations, the cells measured 8 inches by 8 inches by 4 inches
high.
Once the physical data is entered into the computer, it is time to run
the simulation. The computer will model the conditions for each cell,
and then combine all of them together to provide the user with an overall
prediction of the fire.
On a desktop computer with a 600 MHz processor it takes about 20 hours
to run a single simulation, according to Madrzykowski. “That’s using
76,000 cells. If you increase the number of cells, you increase the
computational size.”
So what did the Cherry Hill simulation show?
The fire occurred in a two-story townhouse, with a basement that was
accessible from the exterior. Shortly after midnight, the occupants
on the second floor were wakened when a smoke detector activated. They
went down to the first floor and exited the building through the front
door, leaving it open.
When firefighters arrived on the scene, they reported that heavy smoke
was coming out of the open front door. Crews made entry, and removed
a front window on the first floor from the interior to provide ventilation.
Crews also removed windows on the second floor, front side.
Within approximately two minutes, another crew opened the sliding glass
door on the basement and made entry. Within a short period, there was
a rapid buildup of heat inside of the building. Firefighters on the
first floor were able to escape, as were the firefighters in the basement.
According to Madrzykowski, the firefighters in the basement reported
that they saw a “tunnel” in the smoke that gave them a clear avenue
of escape out the sliding glass door through which they had entered.
However, there were three firefighters operating inside of the building
on the first floor near the stairway to the basement were trapped by
the rapid heat buildup. Two firefighters were killed, while the third
one, who was located in between the other two, suffered severe burns.
A puzzling aspect to the fire, reports Madrzykowski, was the extremely
rapid heat buildup. “There were two experienced teams, with two charged
hoselines. How come no one used a hoseline to defend themselves?” In
addition, it was difficult to explain why it appeared that the middle
fire fighter suffered severe burns, while the outer two were killed
in the fire. One of the fatalities was burned in a directional fashion
up the front of his body, while the other fire fighter was burned uniformly
all over his body. What would account for the difference in burn patterns?
The fire had started in the basement. The NIST report described the
following damage:
“The post fire investigation determined that the fire started near an
electrical fixture in the ceiling of the basement. The basement had
severe fire damage throughout, indicating a well-mixed, post-flashover
fire environment. The stairway from the basement to the first floor
also showed signs of flame impingement on the ceiling and walls. The
door at the top of the basement stairs was open during the fire and
had been partially burned away. The basement stairway opened into the
living room on the first floor. The living room had significant deposits
of soot throughout, with limited thermal damage. Most of the paper on
the gypsum board walls and ceiling remained intact and sofas in the
room only showed signs of pyrolization or limited burning on the upper
portions of the back cushions and top surfaces of the seat cushions.
Areas in the living room away from the basement door opening had less
thermal damage.”
This unusual scenario puzzled the investigators from DCFEMS. By using
the FDS, they were able to replicate the conditions that the firefighters
were encountering and determine what the theoretical temperatures were
inside of the building. According to the NIST summary…
"
…the venting of the sliding glass doors in the basement increased
the heat release rate of the fire very rapidly. The FDS calculation
indicates that the opening of the basement sliding glass doors provided
outside air (oxygen) to a pre-heated, under-ventilated fire compartment,
which then developed into a post-flashover fire within 60 s. The fire
filling the basement forced high temperature gases (approximately
820 °C (1500 °F)) up the basement stairwell at velocities in excess
of 8 m/s (18 mph). The high velocity gas stream flowed into a pre-heated,
oxygen depleted first floor living room. The FDS predictions show
the hot gas flow moving across the living room ceiling and banking
down the back wall of the townhouse.
Between the doorway to the basement and the sofa on the back wall
of the townhouse, the temperatures from approximately 0.5 m (1.6 ft)
above the floor, to floor level are in the range of 180 °C to 260
°C (350 °F to 500 °F). These thermal conditions developed within seconds
of the rapid fire growth in the basement."
The
model showed that superheated gases moved up the stairs at approximately
18 miles per hour. This townhouse was only 33 feet deep, which meant
that the gases moved through the townhouse in less than two seconds.
One firefighter was in the direct path of this fast-moving body of
super-heated air.
Another fire fighter was found near a wall. Based on the simulations,
investigators believe that when the heat came blasting up the stairway,
it encountered the wall, which served as a barrier and caused the
heat to immediately move downward to where the firefighter was located.
As to why other firefighters were able to escape with only minor burns,
Madrzykowski likened the interior conditions to that of a hair dryer.
“The basement door acted as a high-velocity nozzle, but if you were
off to the side, you were not in the direct path of the hot gas. Like
a hairdryer, if you place your finger in front of it, your finger
is hot, but if you move your finger an inch or two to the side, you’ll
feel no heat.” This is why firefighters that were within just a few
feet of the victims were able to feel the heat, yet still survive.
While the current model provides invaluable assistance, it has limitations,
concedes Madrzykowski. It takes a certain level of expertise and knowledge,
as well as a high-end PC to run the FDS. But as NIST continues to
refine and further develop the program, Madrzykowski is hoping that
they will be able to provide a tool that any firefighter can use on
a desktop PC.
Another limiting factor (in the model) is the materials database.
Not every material involved in a fire scene has been input into the
materials database. NIST is continuing to input that data into the
model.
“The science is fairly young,” he continues, “but a lot of progress
has been made in the past generation. However, there has not been
a lot of technology transfer…it needs to get out to the fire service.”
It’s not yet available to assist in every LODD, but the FDS simulator
is being used to assist in investigations of a high-rise fire in New
York City that killed three firefighters and a lightweight wood-truss
roof structure fire at a fast-food restaurant in Houston that killed
two firefighters.
TRAINING APPLICATIONS
Investigating these real-world LODDS in a variety of different scenarios
is helping NIST to lay the foundation for training applications in
the future. Madrzykowski hopes the FDS will soon evolve into a tool
that any firefighter can use on a desktop PC.
The long-range goal is to develop a desktop simulator that can be
made available to fire departments across the nation. Unlike other
“games,” this one will be based on real-incidents and will abide by
the laws of physics.
“Its value is to use it as a training tool,” he says. “The one thing
that we hear from firefighters is that they are getting less and less
on-the-job live-fire training because there are fewer fires and fewer
training opportunities.”
Gearheart reinforced this point. “If NIST continues to get the funding
they need to make it user-friendly…they are working in that direction,
they just haven’t gotten to it yet. We hope that this incident helps
to push it in that direction. As a training tool, it would be invaluable.”
COMMERCIAL APPLICATIONS
With the sponsorship of the U.S. Fire Administration, NIST is inputting
data on the thermal properties of various materials used in firefighter
turnout gear into FDS. Once that data in incorporated into the model,
it can be used to assess the effectiveness of firefighter PPE.
“You will be able to assemble an ensemble using various materials
such as Nomex or PBI, and then place this ‘target’ inside of the scenario,”
says Madrzykowski. The user will then be able to evaluate the impact
upon a firefighter when opening a particular window or door at a given
time in the fire changes the environment. Madrzykowski is hoping that
this will be available within several months.
Other plans include incorporating the physical properties of various
hose streams into the FDS model, enabling users to evaluate the effectiveness
of straight-stream versus fog nozzles, or see what impact compressed
air foam might have on fire. “This is at least five years away,” says
Madrzykowski.
One day in the not too distant future, he says, “The firefighter will
be able to draw on a catalog of situations and see the impact of specific
actions—like opening a door or ventilating the roof. “This will allow
the user to see the impact in real time on various scenarios. By doing
this on the desktop, it will hopefully become an ingrained response
on the fireground when they are faced with a similar situation.”
FDS may facilitate better communications with firefighters as well.
By using its animations, trainers may be able to demonstrate the concepts
of heat flow and the other conditions encountered during firefighting
that could only be previously demonstrated by equations and charts
and graphs--not the most user-friendly method.
Madrzykowski notes this improved communication goes both ways. Scientists
and engineers involved with FDS development are learning more from
firefighters about the “real world” of firefighting than they ever
could from inside a lab.
Copies of the FDS CD Rom of the Cherry Road Simulation can be obtained
by contacting Madrzykowski at madrzy@nist.gov with your name and address.
You can download the full report on the Cherry Road simulation by
clicking on this link: http://fire.nist.gov/6510/
You
can also contact NIST for more information regarding Simulation of the
Dynamics of the Fire at 3146 Cherry Road NE, Washington D.C., May 30,
1999 (Reference NISTIR 6510).
