Can We Predict Fire Movement?

[MevsEinstein]

TL;DR

Flames in a fire whips in "random" ways. Can they be predicted?

I was staring at this barbecue fire. While my family was enjoying the heat, I was wondering why the flames all of a sudden shrink and then resize, and also take different shapes. Is there a way to predict this movement? Just curious

 

[Drakkith]

I was wondering why the flames all of a sudden shrink and then resize, and also take different shapes.

It's a complex interplay between local air temperatures, densities, flame source geometry, and probably a dozen more variables, but mostly it's because of the chaotic way that air masses of different temperatures and densities interact that gives you the random 'whip' effect. A mass of rising hot air doesn't rise smoothly, it tends to make lots of 'whirls' and other instabilities, such as Rayleigh-Taylor and Kelvin-Helmholtz instabilities. Throw in some turbulence (not the same as instability) and you get the complicated hydrodynamic system that is a fire.

Is there a way to predict this movement? Just curious

Certainly. It's just very complicated and requires computer simulations.

 

[MevsEinstein]

Certainly. It's just very complicated and requires computer simulations.

So what are the formulas that determine its movement?

 

[DaveC426913]

Google thinks flame propagation is explained by heat conduction and diffusion.

 

[Baluncore]

So what are the formulas that determine its movement?

You need a finite element model. Start by defining an element with fuel and air inputs.

Then run the model so air is drawn in from below, to the fuel, where the gas is heated and expands as oxygen is removed, and exhaust gasses are added, then the gas rises due to lower density. As the flame rises it radiates energy and cools to become smoke, water vapour and CO2 etc.

 

[Drakkith]

So what are the formulas that determine its movement?

Here's somewhere to start. It's a chapter on Governing Equations of Fluid Dynamics (air is a fluid) from the engineering department at Auburn University. Good luck.

 

[MevsEinstein]

Here's somewhere to start.

Thank you very much! I will try to understand

 

[jbriggs444]

Certainly. It's just very complicated and requires computer simulations.

If the goal is to predict the behavior of a fire in real time then you are out of luck. By the time you get the required inputs to an acceptable accuracy and process them, it is tomorrow, the fire is out and your predictions were only good for a few seconds anyway.

https://en.wikipedia.org/wiki/Butterfly_effect

 

[DrClaude]

Here's somewhere to start. It's a chapter on Governing Equations of Fluid Dynamics (air is a fluid) from the engineering department at Auburn University. Good luck.

And that's just the beginning, because combustion adds its own complexity, as @Baluncore noted, you have to account for the chemistry going on and the release of heat.

 

[Vanadium 50]

Combustion is a very difficult problem. It is solved numerically using very large computers. It is of enormous commercial importance, and millions of dollars a year are spent studying it,

 

[MevsEinstein]

If the goal is to predict the behavior of a fire in real time then you are out of luck. By the time you get the required inputs to an acceptable accuracy and process them, it is tomorrow, the fire is out and your predictions were only good for a few seconds anyway.

https://en.wikipedia.org/wiki/Butterfly_effect

After reading the wikipedia article, my brain blew up. Before posting this, I knew that if there was a way to predict the movement of fire, it would be complex. But I didn't think of all the variables needed.

 

[Baluncore]

After reading the wikipedia article, my brain blew up. Before posting this, I knew that if there was a way to predict the movement of fire, it would be complex. But I didn't think of all the variables needed.

I reached the point where I was considering a scattering matrix for the full EM spectrum. Then I had to stop thinking about the finite element equations because my head exploded.

I think a 1:1 scale model needs to be built. Then the experiment can be run, while it is observed for a few hours, through a glass of liquor.

It should be possible to observe reality, then extract the image parameters and typical variations over time. The statistical caricature of fire can then be generated as an extrapolation from any random image.