The development of flame simulation solver?

[ZeusYang]

Hello, @doyubkim Flame is a kind of pretty significant and attractive natural phenomenon. Essentially, it's also a sort of fluid. However, it seems that combustion simulation is always ignored in fluid simulation. I was wondering whether you had been planning to develop fire simulation solver in the future?

[doyubkim]

To simulate “flame” directly, I think the most classical approach is this: http://physbam.stanford.edu/~fedkiw/papers/stanford2002-02.pdf

Adding more complex combustion models are also available from number of papers including: http://physbam.stanford.edu/~fedkiw/papers/stanford2007-02.pdf

A bit of cheating can be done by rendering temperature field from regular smoke sim with color map as shown below: https://doyub.com/baroclinic-turbulence-with-varying-density-and-temperature/

Similar to the multi-phase flow sim, flame sim always been in the wish list for a while. But I don’t have concrete plan to implement it as I’m more focusing on v2 at the moment. Still, let’s keep this issue open so that it’s in the backlog :)

[cesss]

In the mean time, I think it would be pretty useful if you implement the "cheating" technique from your paper on top of the smoke simulator, because that video is pretty impressive, and would be realistic enough for many uses.

[doyubkim]

That's actually possible already since the smoke solver exposes temperature array (which then can be mapped with a color ramp/map). However, the current cubic solver is not "turbulent" enough to look like a flame.