Firefighters are often forced to rush into burning buildings without much information on what it looks like inside. It is a mad scramble to save lives and avoid getting caught in a collapsing inferno.
Firefighters have to do all of this on the fly, while also worrying about the possibility of a flashover, a deadly phenomenon when flammable materials ignite simultaneously, producing an intense blaze only limited in size by available oxygen.
It is difficult to predict when a flashover might occur, so any type of warning would be critical to saving lives—which is where artificial intelligence comes into play.
NIST develops tool to save firefighters
The U.S. National Institute of Standards and Technology (NIST) has developed P-Flash, or the Prediction Model for Flashovers, an AI-powered tool which could provide firefighters with a warning of when a flashover will occur.
The predictions are based on temperature data from a building's heat detectors, and it is even designed to operate after those detectors fail, using the other remaining devices.
NIST has tested the tool's ability to predict flashovers in over 1,000 simulated fires and more than a dozen real-world fires.
Extensive research suggests the model shows promise in anticipating flashovers, and with more development it could "enhance the ability of firefighters to hone their real-time tactics, helping them save building occupants as well as themselves."
NIST researcher Christopher Brown, who also serves as a volunteer firefighter, had this to say about the need for P-Flash:
"I don't think the fire service has many tools technology-wise that predict flashover at the scene. Our biggest tool is just observation, and that can be very deceiving. Things look one way on the outside, and when you get inside, it could be quite different."
NIST uses machine learning to protect firefighters
NIST notes that computer models predicting flashovers based on temperature are not new, but they have always relied on a constant stream of temperature data. In a lab, it is easy to obtain this data, but in real life it is a different story.
Heat detectors typically operate at temperatures up to 150 degrees Celsius (302 degrees Fahrenheit), but that is much lower than the temperature flashovers typically occur at, which is 600 degrees Celsius (1,100 degrees Fahrenheit).
To address this gap of lost data, NIST has applied machine learning to the model.
"Machine-learning algorithms uncover patterns in large datasets and build models based on their findings. These models can be useful for predicting certain outcomes, such as how much time will pass before a room is engulfed in flames."
NIST chemical engineer and co-author of the study, Thomas Cleary, explains it like this:
"You lose the data, but you've got the trend up to where the heat detector fails, and you've got other detectors. With machine learning, you could use that data as a jumping-off point to extrapolate whether flashover is going to occur or already occurred."
Using big data to protect firefighters
A tremendous amount of research and testing went into creating P-Flash. To start, researchers gave the algorithm temperature data from heat detectors in a three-bedroom, one-story ranch style home, the most common home type in most states.
Machine learning algorithms require immense quantities of data, so burning down hundreds of buildings and homes was not very practical. Instead, researchers burned the virtual home repeatedly using NIST's Consolidated Model of Fire and Smoke Transport, or CFAST. This is a fire modeling program validated by real fire experiments.
Here is how NIST tested P-Flash's accuracy:
"The authors ran 5,041 simulations, with slight but critical variations between each. Different pieces of furniture throughout the house ignited with every run. Windows and bedroom doors were randomly configured to be open or closed. And the front door, which always started closed, opened up at some point to represent evacuating occupants. Heat detectors placed in the rooms produced temperature data until they were inevitably disabled by the intense heat.
To learn about P-Flash's ability to predict flashovers after heat detectors fail, the researchers split up the simulated temperature recordings, allowing the algorithm to learn from a set of 4,033 while keeping the others out of sight. Once P-Flash had wrapped up a study session, the team quizzed it on a set of 504 simulations, fine-tuned the model based on its grade and repeated the process. After attaining a desired performance, the researchers put P-Flash up against a final set of 504."
And here is what was concluded from the tests:
"The researchers found that the model correctly predicted flashovers one minute beforehand for about 86% of the simulated fires. Another important aspect of P-Flash's performance was that even when it missed the mark, it mostly did so by producing false positives—predictions that an event would happen earlier than it actually did—which is better than the alternative of giving firefighters a false sense of security."
After all of this, there was still one more important aspect to the P-Flash development. NIST mechanical engineer and corresponding author Wai Cheong Tam sums it up like this:
"One very important question remained, which was, can our model be trusted if we only train our model using synthetic data?"
While the tool performs very well in a controlled setting, it can be much more difficult to predict flashovers in the real world.
The NIST team notes that there are a few weak spots with P-Flash, including what they are calling the enclosure effect, where a fire starts in a small area, that affects the accuracy of P-Flash's ability to predict flashovers.
However, NIST researchers are still confident about the AI tool's future abilities, envisioning that it can be embedded in hand-held devices that are able to communicate with detectors in a building through the cloud.
When fully developed, it will become an important tool for firefighters, allowing them to identify danger spots and adjust tactics before arriving at the scene—maximizing their chance of saving lives and property.
