If you want to understand how fire works, then you have to burn stuff. That’s where Ruddy Mell comes in. He’s a research combustion engineer and physicist at the U.S. Forest Service’s Pacific Wildland Fire Sciences Lab. Mell’s job is to work with teams of fire experts to create controlled burns, collect all the data they can, and then build physics-based models that can predict what could happen when seriously dangerous fires burn out of control. Mell talked with us about why our current wildfire models are so insufficient and how they go about trying to control the world’s most unpredictable element out in the field.
Why do we need to study wildfires?
At least three reasons. Two of them are kind of combined. They have to do with fires in the wildland and urban interface, where wildland vegetation is adjacent to where people live and fire causes damage to homes and roads and power lines and cell towers -- anything that people have built that causes enough damage that the consequences need to be addressed.
The other problem is smoke. That’s a significant problem. Even if it doesn’t burn buildings the smoke is a problem if people are downwind. The health effect has been shown to cause increased hospital visits for respiratory problems. In some parts of the country, the southeast in particular where there are a lot of old people that are retired, it can be a big problem.
Also in the southeast US the vegetation tends to grow back very quickly, so they have to deal with this smoke issue because the vegetation is there to burn. One of the ways they deal with fires there is to do fuel treatments, where fuel is vegetation. They’ll burn it periodically just to keep it down so it will be easier to contain if there’s a wildfire. They’re limited in doing prescribed fires because of all the people around. They want to do this to keep it safe, but it’s hard to do.
So the wildland fire problem is a fire problem, a vegetation problem, and smoke problem. To address the problem you have to think about all that. When modelling comes in, you need models for fire and better models for smoke.
The purpose of these research burns is to provide data sets for model testing and validation and development. The best example of a model that’s used by people everyday are weather models. Imagine the world if we couldn’t look up the forecast. You can’t use experiments alone to help with weather predictions. Suppose you go out and measure temperature and wind at some site, there’s no guarantee it will be like that a year from now. You need models to help predict out into the future.
Where we are now with fire modelling is probably where the weather community was 30 years ago. We’re really behind.
You can’t do it just from measurements and you can’t really do enough experiments or research burns to be able to say what’s going to happen in the future. You still need a number of models. Where we are now with fire modelling is probably where the weather community was 30 years ago. We’re really behind. I’m guessing because fire doesn’t influence as many people as weather does.
But as the population grows, with the smoke issue, that’s going to be changing. That’s why we’re doing this, because the fire is affecting communities. If it wasn’t for that, if the fires where all happening in the middle of nowhere there wouldn’t be the interest.
What’s wrong with our current fire models?
The experiments that have been done to date for the models we do have aren’t sufficient for the new models. They just didn’t collect enough and the right information. The models they were developing were simpler. The weather models work so well because they have models of different parts of weather. They can focus models on different parts of weather dynamic and the main thing is they’re focusing on the physics. So on the fire side we don’t have any models beyond the basic beginning research stage.
The models we do have were developed using experiments, but they weren’t focused on the physics, they were focused on the spread rate. They didn’t care how they spread or why, just how fast the leading edge of the fire would spread given the wind conditions and terrain.
They focused on the spread rate because if you are an incident commander you’re concerned about how the head fire is going to be in six hours. You want suppression efforts in the most effective way and you want to know how long you have to do that. It’s related to firefighter safety. But there’s no physics here.
The current models are used operationally. They were based on models developed largely in the 1960s. Remember Sputnik? A lot of science in the US was energized after Sputnik. A lab was created in Montana partly focused on fire models. They did a lot of experiments and came up with the model that’s used to today called the Rothermel model (PDF).
When an object burns it’s because the heat flux is sufficiently high. The old models didn’t consider this at all.
Dick Rothermel and a few other engineers came up with a better understanding of fire behavior. They didn’t focus on a model that captures the physics because they just didn’t have the computers. Even if they wrote down the equations they couldn’t solve them.
So now we want the physics. To address that problem things like spread rate and flame length aren’t good enough. You need to predict heat flux. When you go out in the sun and you feel the heat on your skin, that’s the sun’s heat flux. It’s a transfer of heat to some object. When an object burns it’s because the heat flux is sufficiently high. The old models didn’t consider this at all. The physics-based models naturally solve the heat flux because the physics is driven by the heat flux, so it’s naturally included.
How do you measure these things in a controlled environment?
There’s a lot that’s known about the general behavior of fire. People know at what wind speeds and vegetation conditions you’ll have a damaging fire. You wouldn’t want to do a research burn under those conditions. You don’t want the fire to get out of control. When you’re doing research, you’re limited to conditions that won’t cause damaging fires.
With science and a model you always want to do it as simple as you can, so if something happens you can’t explain there aren’t too many confounding factors. You want to really be simple and isolate the different factors as much as you can.
So that’s why we simplify the terrain by making it flat and level and simplify the vegetation by making it grass as opposed to, say, grass and trees
We developed a relationship a with the Texas Fire Service. They were doing burns in military bases because they want to keep vegetation down for fire safety and to facilitate training exercises. So we came in modified what they were doing to meet our research needs. They had the simplest terrain and simplest vegetation and we just waited for the wind.
The current models aren’t really well founded for higher winds. There aren’t very many measurements and there’s a bit of controversy about what happens in grass fires when the winds get stronger. For us the big thing is, what would you have to do if your community was surrounded by grass? Could you just remove the grass and replace it? Or just water it really well to create a buffer between you and the wildland? What would it take to have a buffer? That depends on this wind. How fast does the wind have to be so the fire can cross that?
It’s a very simple question but there has been no systematic research to address that.
What kind of tools do you use to measure the fire?
Fire behavior is driven by three characteristics of the environment: vegetation, weather, terrain. It’s called the fire behavior triangle.
We want to measure things that help us test the physics-based models. They have everything in them you can think of -- vegetation and it’s properties, wind, and rain. We need to measure things that put inputs into these models. We measure moisture and how much vegetation there is in a given area in pounds per square foot.
Then we have to measure the wind. As the fire proceeds across the experimental plot it can change direction because of the wind and you can’t predict a wind shift. Anything that may affect the fire behavior and physics we have to measure.
So we have anemometers, which measure the wind speed and direction. The main issue with all this instrumentation is some of it can be expensive so you can’t put it in the fire. We put those around the edges of the plot at different heights. They’re just a propeller that spin and swivel in the wind. That gives us the wind speed and direction. We have 12 around the plot that’s 100 metres by 100 meters. Then we have other wind instrumentation anemometers on 30-foot towers.
Then we have three SODARs. Those are instruments that measure the wind speed but can go up to 200 meters. They send a sound wave up and measure winds speed based on that. You can hear them pinging.
Then we also have UAV’s with wind measurements on them. They fly at about 1,500 feet in a circle of a 1,000 feet radius. They also measure air temperature and take movies and photos of the ground with infrared cameras. The fire emits radiation at frequencies that are in what’s called the infrared region. Unlike visible light, infrared radiation is not blocked by smoke. With infrared cameras we can see the fire through the smoke.
We also use Fire Behavior Packages. They have instrumentation for temperature, relative humidity, and heat flux. Those can go inside the plot because they’re hardened to be resistant to fire damage. They sit on a tripod about four feet off the ground and they are directional. You have to face them in the direction you think the fire is going to come from. They have a field of view and outside that they won’t sense anything. That’s the tricky thing. We wait until just before we light the fire to place them because only then we know the wind direction.
You must have to run!
The firefighters go out into the plot, aim it them in the right direction, and then get out of there.
The firefighters go out into the plot, aim it them in the right direction, and then get out of there.
We also put a hardened video camera inside a box with ice packs to keep them cool. They face the Fire Behavior Packages so we can see how tall the flames are and how fast they’re moving when they go past the packages. We want to be able to relate the heat flux to fire size.
It’s primitive what we’re doing in many ways. The development of instrumentation for fire hasn’t received enough attention. A fire outside compared to a furnace or combustion in a power plant generator is so uncontrolled. People want to minimize the effects of, it’s not something you can control, and it hasn’t received enough attention.
For example, one other instrument in the plot is called a thermocouple rake. The thermocouple is just a wire that measures resistance. When it heats up, the electrical properties of the wire change. So you can measure voltage changes and relate that to temperature. You can measure the wind with them too. It’s very simple but the constraint with fire is you have to have something that measures the fire but it doesn’t get destroyed or go out of calibration during large changes in heat.
So that’s why we have a video camera out there. It gives us an understanding of what the conditions are that the instrumentation is being subjected to.
How do you start the fire?
The procedure is: you have to pick your plot site and pick it so the expected winds are oriented in a way that’s going to match the expected wind direction.
Then what the firefighters do is they have to go around the edges of the plot so the fire won’t go out into surrounding vegetation. So they go and mow all the vegetation down and sometimes do what’s called black lining. They use what are called drip torches -- think of them looking like a watering can -- they have a handle and fluid comes out of a spout. It’s metal and holds a combination of gasoline and diesel. You ignite the end of it and drop the liquid on the ground. You’re putting liquid on fire on the ground. This is also the most commonly used device for starting controlled fire.
If you really want to be sure the fire stays in the plot, you go around the outside of your plot with these drip torches and people go along with you using different tools to put out the fire. They do that by using these things that look like a mop. They only do this if the vegetation is sufficiently short and they just whack the fire. They walk along and create this black line around the plot you want to burn so the fire can’t leave the plot.
Now you’re ready and created a plot and it comes to the burn day. You have to wait for good temperature and wind conditions. If you do it too early there might be too much moisture, but if you wait too late the winds might die down.
We decide on how long the ignition line should be. If a fire starts small, it creates an extended U shape, like an ellipse, and the leading edge of the fire will accelerate and gain speed as the fire gets bigger. That’s a really good test of a physics based model to be able to predict the acceleration. So we choose the length of the fire line depending on if we want to test that.
A lot of this was based on work on experimental fires in Australia. If you’re standing up wind and looking at fire going away from you, directly down wind is the biggest part and on each side are the flank fires. Once they get far enough apart, roughly 60 to 100 meters, there’s no longer an acceleration of the head fire. If everything stays consistent, the fuels are the same on the ground, and the wind speed doesn't change, it stays as an ellipse. But of course nothing ever stays the same.
It’s critical the ignition is as good as it can be. If you mess it up you might not have the U shape. Everything has to be as clean as it can so when you test the model and something’s different you have much better ability to say what it is.
As long as you know the conditions, you can control the fire. You can do many things with it. The largest thing that’s hard to control is the wind. You can do it in the lab, but then you’re not at the full scale.
How many people do you need to do a controlled burn?
If there was variation in terrain we’d need somebody to measure that. There’s a crew that does vegetation measurements and a crew that does wind measurements. And then there’s a fire measurement crew and crew doing measurements from the plane.
And then you need the firefighters to light the fire and make sure everything’s safe -- do the logistics of where everyone is.
At this latest burn, Colorado State University had one of the UAV pilots. San Diego State University was doing the wind measurements. From the Forest Service there was a group from Missoula doing the low wind measurements and the Fire Behavior Packages (that’s their specialty, they have a trailer and travel around the country doing this).
Then we had a local Texas UAV pilot we contracted. And another forest service group from Seattle was doing vegetation measurement. And of course NIST also owns the UAVs.
Then the Texas Forest Service were the ones running the show -- starting the fire and preparing the plot, doing the safety briefs and working with the base commander to allow us access.
There are something between 40 and 50 people.
It’s quite a production! Have you always been interested in studying fire?
I started out more interested in geology and geophysics. So I actually went into the oil industry first doing work a fuel wells. Then I went back into the field and did diamond exploration for the upper peninsula of Michigan. Some farmer guy in the Lower Peninsula found a diamond in his field deposited by glaciers. Kimberlite was then found to the north in the Upper Peninsula. This is a type of rock that shoots up from the mantle in pipes and can contain diamonds. Then I went back and got a degree in applied math. For a while I was doing underwater acoustics, like submarine detection.
You’re a jack of many, many trades.
What happens is, in graduate school you’re a baby bird with it’s mouth open wondering who is going to feed you. If you hook up with a professor that has money you don’t have to pay for anything. The money ran out in underwater acoustics and so I ended up in combustion. The classes I was taking were all in physics, so it didn’t make a lot of difference. I ended up at NIST for my postdoc, studying fire inside buildings. Then we realized this fire problem of wildland and structure fires is the problem of the future and that’s how it started.
I’ve been part of about six burns now. There’s another one happening next month in the Pine Barrens in New Jersey. That’s a much more complicated vegetation and the fuel guys are there right now taking vegetation samples.
Photos and videos courtesy Ruddy Mell
Not all science is done in a lab by guys in white coats staring into microscopes. Lots of discoveries require brave men and women to put their boots on the ground and get down and dirty in dangerous environments. Every month we’ll profile one of these field scientists, tell you how they do their job, and explain the science behind what they do. If there’s a scientist or field of science you’re dying to hear more about shoot us an email or a tweet: erin at erinbiba dot com, @erinbiba