Wildfire Simulation Technology – Part 3

Beneath every wildfire model there is a set of assumptions, ideally based on a theory of how fire ignites and spreads. The model, however, is only as good as the theory that drives it and the data that goes into it. As computers have gotten smaller and...


Beneath every wildfire model there is a set of assumptions, ideally based on a theory of how fire ignites and spreads. The model, however, is only as good as the theory that drives it and the data that goes into it. As computers have gotten smaller and faster, models have become more powerful. However, the fire simulations are heavily dependent on the assumptions made by a particular model. This last installment on wildfire simulation software takes a hard look at the current thinking on this aspect of fire science in the U.S.

Mark Finney, a research forester at U.S. Forest Service (USFS) in Missoula, MT, and one of the foremost authorities on how assumptions on wildfire apply to models, provides some background material on how we arrived at our current thinking in fire science. “There were a lot of basic studies done 50 years ago trying to understand wildland fire from an engineering perspective, but since wildland fire is an applied discipline, there’s a lot of interest in coming up with practical products that you can use,” he explained. “Whereas people would normally try to understand what’s going on and then model it once they had a theory, there actually is no theory to fire spread.”

The Science of Combustion and Spread

Despite appearances, fire combustion and spread are not simple phenomena. “There are a lot of variables and a number of things that cannot be known very well when you try to model something. But when you look at very simple experiments in laboratory conditions, we still can’t explain exactly the physics going on there,” said Finney. “It’s been recognized for a long time that wildland fire spread is a process of repeated ignitions. If the sequence of ignitions fails, then you don’t have spread.”

Finney contends that people have jumped right into determining the rate of spread rather than trying to better understand the process of ignition due to the need for operational kinds of tools (i.e. models). “They only modeled the conditions under which spread occurred, which is only part of the equation because most of the interesting stuff is ‘When is it going to spread, when is it not going to spread and what are the various factors that contribute to that threshold?’” he explained.

“You have all kinds of things (from moisture content to spacing between the fuel elements), so there are so many parameters that can contribute in different ways in different combinations (like weather, wind and turbulence) to produce an ignition that it is complicated, but you can still understand what the requirements are for ignition,” Finney continued. “Our approach to this is not to model fire, but to understand the physics of fire spread. There are some very interesting things that come out of this that are exactly the opposite of what most models assume, which raises some questions.”

And there are some things that simply may not be possible. “It is hard to extract the process you’re interested in without making it so out of context that it isn’t relevant to what you were originally studying, for instance keeping out how radiation or convection work in a wildland fire is tough because they’re always there,” said Finney.

Running experiments on combustion and spread is not only time consuming, it’s a bit pricey. “At the moment, running an experiment with fire in a laboratory is expensive, uncertain, and you can find almost no one to back it,” said Finney. “We have pretty good facilities here [at USFS], but it’s very expensive to have the kinds of technical people here – all kinds of engineers, machinists, and technicians – to run an experimental program.”

Look Before You Leap

Proceeding with a simulation without having a good understanding of all the parameters of a fire can be risky, especially with lives and property at stake. “What we’ve found is that people jump straight into modeling without having a full understanding of what’s going on,” Finney continued. “It seems like models were pursued, not as a means of understanding the phenomena so much, but in order to produce something that could be implemented in an operational way and that kind of shortcut the understanding part.”

This content continues onto the next page...