The Governors Island Experiment
Wind is always a potential issue when fighting fires, but it is particularly dangerous in high rise fires, as the incident at Vandalia Avenue made clear. Firefighters had started testing ways to counter wind using portable blankets, pressure fans, and high-rise nozzles that could be deployed from the floor below the fire. However, firefighters had no way of measuring the effectiveness of these systems since they had never been evaluated by scientists and engineers. This was the basis for the Governors Island experiments (Madrzykowski).
The partnership between Brooklyn Poly, the Fire Department, and the NIST was formally initiated in 2007. Earlier experiments in Chicago, Toledo, and Ottawa (which the FDNY had participated in) had produced some preliminary data and the Fire Department hoped that a full burn test would validate that data in a way that would confirm their understanding of what had happened at Vandalia Avenue and give firefighters better information on how to fight high-rise fires.
The NIST began by conducting small scale tests at their headquarters in Gaithersburg, MD. The purposes of these tests was to quantify the level of hazard in conducting a full scale experiment and to provide a sound starting point for Brooklyn Poly researchers (Madrzykowski).
Ultimately, the researchers at Brooklyn Poly, led by Dr. Sunil Kumar settled on a scope for the experiment, summarized as follows by Bugliarello: "Given the complexity of the problem, we will pick only a subset of issues, specifically determination of the flow fields in and around the test building for different building external and internal configurations, wind conditions and pressurization patterns, and demonstration of their impacts on firefighting approaches such as warning signs of whether to open windows or not. We will also look limitedly, because they are not standardized, at containment in fixed systems..." (Bugliarello 2007, 2).
It is worth highlighting Bugliarello's comment about impacts on firefighting approaches. In every fire, but especially in high-rise fires, firefighters need to make decisions about how to balance the competing needs to vent smoke and provide access to the fire with the needs to shield the fire from high pressure air from wind and other sources and to keep smoke from spreading. These needs are always in tension, and every window and door deliberately opened or left closed can be a determining factor in how a fire scenario plays out. For example, during the 2022 Bronx apartment fire at a Twin Parks high-rise, the Fire Department was forced to make critical decisions about how best to give firefighters access to the fire floor while still being able to evacuate trapped residents whose apartments were rapidly filling with smoke. Although the building's management company claimed that the Fire Department opened the doors on the third floor too often, an investigation by the New York Times concluded that the floor had already completely filled with smoke before the FDNY began opening doors to fight the fire (Singhvi et al.).
The Governors Island experiment would consist of 14 independent burns. The critical challenge for these burn tests was to find a way to reliably simulate wind using fans while also taking actual wind into account. The fans would need to react in real time to changes in the actual wind flow in order for the net wind effect to remain constant during the experiment.
Full scale wind simulation had proven to be a difficult challenge at the prior tests as well. The NIST experiments at Gaithersburg used a rescue boat with a large propulsion fan as the wind source, but this proved unreliable for generating consistent wind patterns (Madrzykowski). Wind tunnels do allow for precise wind control, but as Cresci discovered in the 1970s CUES experiments, these tests are often better in theory than in practice, as pressure effects are extremely sensitive to minute changes, and are therefore difficult to scale down reliably.
For the Governors Island test, Brooklyn Poly researches installed a remotely operated weather station on the roof of nearby Building 844 to monitor real time wind conditions as well as to measure the fire's impact on external temperature (Kumar).
A crucial part of the experiment was the fire blanket deployment tests, such as Experiment 7K, named for the location of the test (apartment K on the 7th floor). As can be seen in the accompanying NIST video — which appears to be sped up although the provided video description does not indicate that it is — firefighters were able to effectively deploy the blanket to reduce the effects of the fire.
Not shown in the video is the procedure for attaching and securing the blanket. The video does show that the blanket was connected by straps to the roof of the test building. In a real world fire, the roof would only be useable as a connection point for fires near the highest floors.
A second video showing a blanket used during an actual fire displays several key differences. The blanket in the second video is connected to attachment points on the fire floor rather than the roof, and it appears that firefighters are using a ladder to service it. As in the previous video, the procedure used to set up and deploy the blanket is not shown.
In addition to fire blankets, researchers also analyzed positive pressure fans (similar to the ones tested during the 30 Church Street experiment) and nozzles that could be used from floors below the fire to spray water externally through open windows to reduce temperature.
The final report from the NIST made three conclusions:
- Positive pressure fans cannot overcome the effects of wind on their own, but are still useful when used together with fire blankets and floor-below nozzles.
- Fire blankets were able to reduce hallway temperatures by 50% in just 120 seconds and were able to withstand intense thermal conditions without failure.
- Floor-below nozzles were also effective in reducing hallway temperature, and importantly, could do so with minimal amounts of water sprayed as a fog stream through an open window (Kerber and Madrzykowski 425).
