Wind-Driven Fire Research: Hazards and Tactics




Fires in high-rise buildings create unique safety challenges for building occupants and firefighters. Smoke and heat spreading through the corridors and the stairs of a building during a fire can limit building occupants’ ability to escape and can also limit firefighters’ ability to rescue them. In 2002, there were 7,300 reported fires in high-rise structures (buildings of seven stories or higher). Most of these high-rise fires occurred in residential occupancies such as apartment buildings. In fires that originated in apartments, 92 percent of the civilian fatalities occurred in incidents in which the fire spread beyond the room of origin.1


Changes in the building’s ventilation resulting from opening doors or windows can increase the growth of the fire and allow it to spread beyond the room of origin and also increase the spread of fire gases throughout the building. In some cases, such as the October 2003 Cook County Administration Building fire, the fire flow into the corridors and the stairway prevented firefighters from suppressing the fire from inside the structure. This fire resulted in six building occupant fatalities and several firefighter injuries in the stairway.2


The failure of a window in the fire apartment in the presence of an external wind can create significant and rapid increases in the fire’s heat production. Combined with open doors to corridors, stairs, or downwind apartments, wind-driven fire incidents have resulted in firefighter fatalities and injuries.3,4


The objective of the study discussed below was to improve firefighter and building occupant safety through a better understanding of wind-driven fire conditions and firefighting tactics that could be used to mitigate the fire hazard. The study included conducting two series of wind-driven fire experiments to demonstrate the hazard and examine the potential firefighting tactics. The first set of experiments was conducted in a laboratory; the second set was conducted in a seven-story acquired structure in New York City.

Both sets of experiments were designed to expose a public corridor area to a wind-driven, post-flashover apartment fire. The door from the apartment to the corridor was open for each of the experiments. The conditions in the corridor were critically important because firefighters would use the building’s corridor to approach the fire apartment, and occupants from an adjoining apartment would use it to exit the building.




The National Institute of Standards and Technology (NIST) and the Fire Protection Research Foundation (FPRF), with the support of the Department of Homeland Security (DHS)/Federal Emergency Management Agency (FEMA) Assistance to Firefighters Research and Development Grant Program and the United States Fire Administration (USFA), conducted a series of wind-driven fire experiments in the NIST Large Fire Facility in Gaithersburg, Maryland.

The experiments were conducted in NIST’s Large Fire Facility, which permitted researchers the best control over the experiments. Additionally, the facility permitted accurate measurement of the heat release rates and gas concentrations, which would have been difficult and cost-prohibitive in an acquired structure.

The test structure was composed of three rooms: a bedroom, a living room, and a target room, which were all connected by a hallway. The target room represented another bedroom but with its door closed. Conditions in this room were monitored to see how much heat and smoke would be forced around the edges of the door under wind-driven conditions. A door from the living room led to a corridor that extended 7.3 meters (24 feet) in each direction when measured along the inside of the exterior wall. The south side of the corridor was closed with no exit and no flow path. The north side of the corridor had an exit vent in the ceiling, which provided a flow path. The only other opening to the facility was the bedroom window, when it vented during the fire experiments. The window served as the wind inlet during the experiments (Figure 1).

Figure 1. Floor Plan of Experimental Structure with Flow Paths
Click to Enlarge
The bedroom is the room of fire origin. Source: The National Institute of Standards and Technology.

To ensure control and repeatability, the experiments employed an airboat as a mechanical wind source for all of the wind-driven fire experiments (photo 1). Operating the fan at between 800 and 3,000 rpm resulted in air speeds of between 2.2 and 11.4 meters per second (m/s) (five to 25 mph) at the structure’s window opening. Full details of the experimental setup, instrumentation, and complete experimental results are documented in NIST Technical Note 1618.5

Click to Enlarge
(1) Photos courtesy of the National Institute of Standards and Technology.

The bedroom furnishings had a combustible mass of 19.6 kilograms per square meter (kg/m2) or 4 pounds per square foot (lbs/ft2). The living room furnishings had a combustible fuel load of 12.2 kg/m2 (2.5 lbs/ft2) (photos 2, 3). Although the mass of the furnishings was different for the two rooms, the measured peak heat release rates of both the furnishings in the bedroom and in the living room were approximately eight megawatts (MW). These heat release rates did not account for the heat release rates from the interior finish of the rooms, such as carpeting and padding. Each of the experiments had a similar fuel load.

Click to Enlarge
Click to Enlarge

Researchers conducted eight experiments. The fires were ignited in the apartment’s bedroom. Prior to the failure of or venting of the bedroom window, which was on the upwind side of the experimental apartment, the fire’s heat release rate was approximately one MW. The heat release rates from the post-flashover structure fires were typically between 15 and 20 MW. When the door from the apartment to the corridor was open, temperatures in the corridor area near the open doorway, 1.52 m (five ft) below the ceiling, were more than 600°C (1,112°F) for each of the experiments. The heat fluxes measured in the same location, during the same experiments, were more than 70 kW/m². These extreme thermal conditions are not tenable, even for a firefighter in full protective gear. These conditions occurred within 30 seconds of the window failure.

Experiment 1 findings. Experiment 1, conducted without any external wind, provided valuable baseline data and demonstrated several important points relevant to firefighting.

Smoke is fuel. A ventilation-limited (i.e., fuel-rich) condition had developed before the window failed. Oxygen-depleted combustion products, containing carbon dioxide, carbon monoxide, and unburned hydrocarbons, filled the rooms of the structure. Once the window failed, the fresh air, entrained by natural ventilation, provided the oxygen needed to sustain the transition through flashover, which caused a significant increase in heat release rate. Less than a minute after the window was vented, the heat release rate increased from approximately 1.5 MW to more than 14 MW.

Venting does not always mean cooling. In this experiment, the post-ventilation temperatures and heat fluxes all increased because of the ventilation-induced flashover.

The natural ventilation caused fire-induced flows.Velocities within the structure exceeded five m/s (11 mph), just as a result of the fire growth and the flow path that was set up between the window opening and the northwest corridor vent.

The fire gas flow direction. Thermal conditions (temperature and heat flux) were twice as high in the “flow” portion (north) of the corridor as they were in the “static” portion (south) of the corridor in Experiment 1. Thermal conditions in the flow path were not consistent with firefighter survival.

Figure 2 shows the temperatures along the flow path at an elevation of 1.52 m (5 ft) below the ceiling, or 0.91 m (3 ft) above the floor. Note the difference between the temperature in the flow path from the bedroom through the living room and into the north portion of the corridor. The south portion of the corridor was not in the flow path.

Figure 2. Experiment 1: Temperature vs. Time, with No Imposed Wind
Click to Enlarge

Experiments 2-5 focused on the effect of wind-control devices (WCDs). These experiments employed two WCDs, which function by covering the window opening and blocking or reducing the flow of air into the room. Both devices used in these experiments were made from a proprietary high-temperature textile material that is flexible, resists abrasion, and can withstand temperatures of approximately 1,100ºC (2,000ºF).

The main differences between the two devices are size, weight, and stiffness. The smaller WCD measured 1.8 × 2.4 m (6 × 8 ft), weighed approximately 12.3 kg (27.1 lbs.), was reinforced with metal rods, and was secured by a rope fastened at each corner. A single firefighter could deploy a device of this size and shape from the floor above the fire (photo 4).

Click to Enlarge

The second WCD measured 2.95 × 3.66 m (9.66 × 12 ft), weighed approximately 20.5 kg (45.2 lbs), had a chain sewn into the bottom to assist with deployment, and had tether straps attached at each corner. This device would typically require two or more firefighters to deploy and secure in place (photo 5).

Click to Enlarge

Experiments 2 through 8 all used a mechanically generated wind, ranging from 3 m/s to 9 m/s (7 to 20 mph). The fuel load in the structure was the same for all of the experiments. Each of these experiments demonstrated a rapid transition to untenable conditions in the corridor [even for a firefighter in full personal protective equipment (PPE)] after the window failed.

Experiments 2 through 5 focused on the impact of WCDs. In these experiments, the WCDs reduced the temperatures in the corridor outside the doorway by more than 50 percent within 60 seconds of deployment (Figure 3). The heat fluxes were reduced by at least 70 percent during this same time period (Figure 4). The WCDs also completely mitigated any gas velocity resulting from the external wind.

Figure 3. Experiments 2-5: Temperature vs. Time
Click to Enlarge
Figure 4. Experiments 2-5: Heat Flux vs. Time
Click to Enlarge

Experiments 6-8: Externally applied water. Experiments 6 through 8 focused on the impact of externally applied water. In these experiments, the externally applied water streams were implemented in three different ways: a fog stream applied across the face of the window opening; a fog stream directed into the window opening; and a solid water stream directed into the window opening. The fog stream applied across the window was not effective at reducing the thermal conditions in the corridor. The fog stream directed into the window decreased the corridor temperature by at least 20 percent and the corresponding heat flux measured by at least 30 percent. The solid streams directed into the window resulted in corridor temperature and heat flux reductions of at least 40 percent within 60 seconds of application. None of the water applications reduced the gas velocities in the structure. In some cases, the gas velocity increased during water application, because of the momentum imparted from the water.

Door control. In Experiment 7, the fire was started with the door from the living room to the corridor in the closed position. The window failed at approximately 300 seconds. The door was opened 377 seconds after ignition; this point is designated as time “zero” (T = 0) in Figure 5. This figure clearly shows how the door acted as a WCD and a thermal barrier to protect the corridor from extreme thermal conditions. Temperatures along the flow path (corridor north position) exceeded 600°C (1,112°F) within 20 seconds of the door being opened. The temperatures in the south portions of the corridor, which were not in the flow path, increased at a much slower rate. These temperatures were measured at 1.52 m (5 ft) below the ceiling. These data demonstrate the importance of door control and the importance of keeping firefighters out of the flow path of fire gases.

Figure 5. Experiment 7, Temperature vs. Time
Click to Enlarge

These experiments clearly demonstrated the extreme thermal conditions that a “simple” room-and-contents fire can generate and how these conditions can extend along a flow path within a structure when wind and an open vent are present. Two possible tactics that could be implemented from either the floor above the fire (i.e., WCD) or from the floor below the fire (i.e., external water application) were demonstrably effective in reducing the thermal hazard in the corridor. However, these experimental results also indicate that the post-deployment thermal conditions for any single tactic were still of a level that could pose a hazard to firefighters in full PPE.

The experiments also provided potential guidance for firefighters as a part of a fire size-up and approach to the room of fire origin: note wind conditions in the area of the fire, look for “pulsing flames,” examine smoke conditions around closed doors in the potential flow path, and maintain control of doors in the flow path.




Further research in an actual building was required to fully understand the ability of firefighters to implement these tactics; to examine the thermal conditions throughout the structure, such as in stairways; and to examine how these tactics interacted with natural and positive-pressure building ventilation strategies.

When a wind-driven condition exists in a structure or the potential for one exists, the normal direct frontal fire attack tactic of stretching an attack line down the hallway and into the fire apartment cannot be executed without burning or severely injuring the attack team. Crawling under or through these extreme conditions or even flowing a large-caliber line and trying to move down the hallway is not a safe or even possible option—hence, the need to examine alternate strategies in these structure fire experiments.

NIST, the Fire Department of New York (FDNY), and the Polytechnic Institute of New York University, with the support of the DHS/FEMA Assistance to Firefighters Research and Development Grant Program and the USFA, conducted a series of wind-driven fire experiments in a seven-story building on Governors Island, New York. Fourteen experiments were conducted to evaluate the ability of positive-pressure ventilation fans (PPV), WCDs, and exterior water application using floor-below nozzles (FBN, also known as high-rise nozzles) to mitigate the hazards of a wind-driven fire in a structure. In addition, the experiments considered the fire department’s ability to safely implement these tactics.

Positive-pressure ventilation (PPV) fans. The experiments used a single type of fan, a 0.7 m (27-inch) diameter PPV fan powered by a nine-hp gasoline engine to ventilate and pressurize the structure’s stairwells. Fans were positioned about 1.8 m (6 ft) from the doorways into which they were flowing air and tilted back 10° from vertical. Fans were located at the front lobby doors (photo 6), at the base of the stairwells on the first floor, and on two floors below the fire floor, depending on the experimental configuration.

Click to Enlarge

WCDs.  These experiments used two different WCDs. The first device, made of a silica fabric aluminized with a foil material, measured 3.0 × 3.7 m (10 ×12 ft), weighed approximately 14.0 kg (30.9 lbs), and had a strap at each corner and one at the top center to secure it in the desired location. The blanket necessitated that a team of at least two firefighters be positioned above and below the fire apartment to deploy and secure the blanket. The bottom 3.7-m side of the large WCD was weighted with a chain to assist with deployment (photo 7).

Click to Enlarge

The second device was the smaller WCD (measuring 1.8 × 2.4 m) used in the laboratory Experiments 2 through 5 above. A single firefighter above the fire apartment could deploy this WCD; additional firefighters could be used to secure the bottom of the WCD from below (photo 8).

Click to Enlarge

Floor-below nozzles. Two different types of floor-below nozzles (FBNs, also called high-rise nozzles) were used. The first, FBN 1, was a bent pipe to which a nozzle of choice could be attached. FBN 1 was designed to be positioned straight out of the window of the floor below the fire; by adjusting the distance, the crew could direct the stream into the floor above (photo 9).

Click to Enlarge

The second type, FBN 2, also featured the bent-pipe design but was also designed to hook onto the windowsill of the floor above. There were a couple of FBN 2 prototypes, with different configurations to get the tip of the nozzle into the above apartment, but for the purposes of this analysis, they both will be grouped together (photo 10).

Click to Enlarge

Both FBN types could accommodate any tip or water distribution device. These experiments used a 11⁄8-inch smooth bore tip and three different fog tips: a navy nozzle; a sprinkler head-type tip; and a common fog tip. The fog tip was considered any tip that created a broken water stream.




Each experiment started with a fire in a furnished room. Twelve of the 14 experiments used a natural or mechanical wind to intensify the air flow. Each tool was evaluated individually as well as in conjunction with each other to assess the benefit to firefighters and to structure occupants. The data collected—temperature, differential pressure, and gas velocity inside the structure—were used to evaluate the impact of the PPV fans, the WCDs, and the exterior water application tactics. Each experiment was documented using video and thermal imaging cameras, which also captured specific fire phenomena that are not typically observable on the fireground.

During these experiments, a public corridor and stairwell area was exposed to a wind-driven thermal flow from a post-flashover apartment fire. The door from the apartment to the corridor was open for each experiment.

Because of the excess fuel pyrolysis/generation (lack of ventilation), the room of fire origin (i.e., the bedroom) could not transition to flashover until windows self-vented and introduced additional fresh air with oxygen to burn. If no wind was imposed on the vented window, the fire did not spread from the room of origin, and it never left the apartment of origin. Even with no externally applied wind, the creation of a flow path from the outside, through the fire apartment, into the corridor, and up the stairs to the open bulkhead on the roof increased the temperatures and velocities in the corridors and in the stairwell, resulting in hazardous conditions for firefighters and untenable conditions for occupants on the fire floor and above in the stairwell.

With an imposed wind of nine to 11 m/s (20 to 25 mph) and a flow path through the fire floor that exited out of the bulkhead door on the roof, temperatures higher than 400ºC (752ºF) and velocities of approximately 10 m/s (22 mph) were measured in the corridor and stairwell above the fire floor. These extreme thermal conditions are not tenable, even for a firefighter in full protective gear. Complete details and data from the experiments can be found in NIST Technical Note 1629.6

Size up the potential for wind-driven fire conditions.Initial incident size-up should include considering wind conditions. Wind conditions can vary widely in an urban environment; because of wind flows around buildings, shielding by buildings may give you the perception on the ground that no significant wind is present. However, another side of the building or a higher elevation in the building may be exposed to wind conditions. Wind speeds between 10 and 20 mph are high enough to create a wind-driven fire condition in the structure with an uncontrolled flow path.

If the fire has vented a window, observing the flames’ behavior provides important information. If the fire apartment has a high pressure relative to the outside because of an imposed wind, the flame will “pulse” out of the window to balance the overpressure. If the flames are being forced out of the window, it indicates that a flow path has been established through the building and the flow direction may be favorable to interior firefighting. If the flames are pulsing or being forced into the window, conditions may not be favorable to interior firefighting, and you should use caution when approaching the fire floor. Even if flames are being forced out of adjacent windows in the fire apartment with a high amount of energy, there could still be sufficient energy flows on the fire floor to create a hazard for firefighters.

No flow path, no wind-driven conditions. A wind-driven condition exists when hot gases or flames are flowing horizontally out of the room of fire origin. Firefighters have described the wind-driven fire condition as a “blowtorch.” For our purposes, a wind-driven fire condition existed when the fire gases were well mixed and of equally high temperature from the floor to the ceiling, of at least 400ºC (752ºF). For this condition to occur inside a structure, the fire must be in a flow path. In these experiments, the inlet to the flow path was the upwind window in the room of fire origin. The flow path then went through the apartment, went into the corridor, and exited out of the bulkhead door on the roof via the stairwell. Without a flow path, the wind-driven fire condition inside the structure cannot occur.

Control the door.Door control is the most basic means for interrupting or controlling the flow path in the building. Check the fire floor stair door for heat or hot gases flowing around the edges. At first, open the door only a few inches to check for rapid changes in smoke volume or velocity and thermal conditions. If the thermal environment changes quickly, close the door to interrupt the flow path. In a smoke-filled environment, visual changes to conditions may not be apparent without a thermal imaging camera. A similar approach would be used on the door to the fire apartment.

PPVs.PPV fans alone could not overcome the effects of a wind-driven condition. However, when used in conjunction with door control, WCDs, and FBNs, the PPV fans were able to maintain tenable and clear conditions in the stairwell. The key to successful use of PPV fans was to mitigate the wind-driven fire condition via door control or other tactics. Then the PPVs can be used to clear the stair and then pressurize the stairwell to provide a safe working environment. Although the PPV fans, when used alone, could not reverse the flow of a wind-driven fire, PPV fans always improved conditions in the stairwell.

WCDs.In all of the building experiments, the WCDs reduced the temperatures in the corridor and the stairwell by more than 50 percent within 120 seconds of deployment. The WCDs also completely mitigated any velocity caused by the external wind. The WCDs were exposed to a variety of extended thermal conditions without failure.

Externally applied water. The externally applied water streams were implemented in different ways: a fog stream inserted into the fire room window, a fog stream flowed from the floor below into the fire room window opening, and a solid water stream flowed from the floor below into the fire room window opening. In all cases, the water flows suppressed the fires, thereby causing reductions of at least 50 percent in temperature in the corridor and the stairwell. The water flow rates used in these experiments were between 125 and 200 gpm.

Stored energy.Wind-driven fire conditions can generate and transfer energy throughout the flow path. If doors or WCDs are used to stop the wind-driven fire conditions, energy and fuel may be trapped on the fire floor. These experimental results indicate that the thermal conditions resulting from the residual heat on the fire floor were still at a level that could pose a hazard to firefighters in full PPE. However, when used in combination with PPV fans to force cool air into the stairwell and out through the fire floor, and/or with the cooling effect from water streams, the fire floor temperatures were reduced to tenable conditions for firefighters in full PPE in minutes.

The findings from the experiments conducted in the laboratory and in the seven-story building experiments were in agreement and supported the use of the three tactics.




The data from this research will help provide the science to identify methods and promulgation of improved standard operating guidelines (SOGs) for the fire service to enhance firefighter safety, fireground operations, and equipment use. If the demonstrated technologies continue to prove effective in fire department field trials and pilot programs, the next step may be to examine the need for standards and standardized test methods to define a minimum level of acceptable performance of these devices.

It is also very important to understand that wind-driven fires are not just phenomena that occur in high-rise buildings. The same hazards can occur in single-family dwellings and other low-rise buildings. It is just as important to attack the fire from an upwind position and to use alternate strategies to ensure that the crews go home safely.

Training will also be required to implement any of these tactics safely and effectively. NIST, with the support of the USFA, has developed a DVD set that includes an overview presentation and reports and videos from both sets of experiments. These DVDs can be used as part of your department’s training package. To request a Wind-Driven Fire DVD, e-mail your name and mailing address to Daniel Madrzykowski at For information on other NIST research for the fire service, visit




The two-year wind-driven research project demonstrates what partnerships and coordinated research can accomplish in addressing the needs of the fire service in a well-supported, focused two-year effort.




1. Hall, John R., “High-rise Building Fires.” National Fire Protection Association, Fire Analysis and Research Division. Quincy, MA., August 2005.

2. Madrzykowski, D., and W.D. Walton. “Cook County Administration Building Fire, 69 West Washington, Chicago, Illinois, October 17, 2003: Heat Release Rate Experiments and FDS Simulations.” Gaithersburg, MD: National Institute of Standards and Technology, NIST Special Publication 1021, July 2004,

3. NIOSH F99-01, “Three Firefighters Die in a 10-Story High-Rise Apartment Building—New York.” NIOSH Firefighter Fatality Investigation and Prevention Program, Morgantown, WV., August 1999,

4. NIOSH F2001-33, “High-Rise Apartment Fire Claims the Life of One Career Firefighter (Captain) and Injures Another Career Firefighter (Captain)—Texas.” NIOSH Firefighter Fatality Investigation and Prevention Program, Morgantown, WV., October 2002.

5. Madrzykowski, D., and S. Kerber. “Firefighting Tactics Under Wind-driven Conditions: Laboratory Experiments.” Gaithersburg, MD: National Institute of Standards and Technology, NIST Technical Note 1618, January 2009. Download at

6. Kerber, S., and D. Madrzykowski. “Firefighting Tactics Under Wind-driven Conditions: 7-Story Building Experiments.” Gaithersburg, MD: National Institute of Standards and Technology, NIST Technical Note 1629, April 2009. Download at

Daniel Madrzykowki will present the classroom session “Wind-Driven Fires in Structures” at FDIC 2010 in Indianapolis on Friday, April 23, 8:30 a.m.-10:15 a.m.

DANIEL MADRZYKOWSKI is a fire protection engineer with the Building and Fire Research Laboratory at the National Institute of Standards and Technology (NIST) in Gaithersburg, Maryland. He has conducted research on fire suppression, large fire measurements, fire investigation, and firefighter safety. Madrzykowski has a master of science degree in fire protection engineering from the University of Maryland. A member of the National Fire Protection Association and the International Association of Arson Investigators, he is also a registered engineer and a fellow in the Society of Fire Protection Engineers (SFPE).

STEPHEN KERBER is a fire research engineer in the Corporate Research Division at Underwriters Laboratories (UL). His areas of research include improving firefighter safety, fire service ventilation, sprinkler spray measurements, and smoke management fire modeling. Prior to joining UL, he was a fire protection engineer at the National Institute of Standards and Technology. A 12-year veteran of the fire service, he received his bachelor’s and master’s degrees in fire protection engineering from the University of Maryland and is also a registered professional engineer.



More Fire Engineering Issue Articles


Fire Engineering Archives


No posts to display