I am trying to simulate an premixed ethylene fuel rich stagnation flame in Fluent. With all the tests I've done I've struggled to get convergence, even using 1st order equations and relaxation factors. Included are pictures of the settings I used. I have set some boundaries to pressure outlets to simulate open air. I am having reversed flow in those pressure outlets but I know that's not the whole issue. I am setting the nozzle as a velocity inlet with the velocity and species mixture defined in the boundary condition. I patch in the temperature and product species and then run the calculation. The results I do get look somewhat fine, however the temperature is too low, and continues to drop as I do more iterations(the flame seems to be going out). If anyone could give any pointers or options for me to make my simulation work I would be grateful. I have tried many tips given from this community and while some of them have helped, my simulation is still having problems.
Problems with flame simulation
- 119 Views
- Last Post 21 December 2018
Can you repost the images so we can view them? Please also list what you've tried: I assume there's some mechanism to stabilise the flame and the combustion zone is within the flammable limits?
Hey, sorry I took so long to reply. The images should be re-uploaded. I have tried: changing using transient instead of steady, changing the viscous model to laminar, changing scheme to Coupled, using pseudo transient method with time larger time lengths, changing boundary condition simulating open atmosphere to outlet rather than pressure outlet (with pressure outlet I'm getting reversed flow but it shouldn't be a huge issue), changing relaxation factors, changing order of equations used, running the simulation cold for a few hundred iterations and then igniting. None of these have worked, however i haven't tried all combinations of these. The combustion zone is within flammable limits as this flame has been tested in another simulation and through experimentation. I'm not quite sure what you mean by "mechanism to stabilize the flame"; the chemkin imported mechanisms are for the ethylene flames I'm modeling and for stabilization I was assuming since I'm modeling it using the steady model in fluent it would stabilize it.
No worries. We're not allowed to open attachments, so whilst I can see you've uploaded I can't view them!
By flame stabilisation I mean how is the flame staying in the burner: most systems use swirl or a bluff body to keep the flame on the burner. Now, with back flow, does this penetrate anywhere near the flame?
Oh ok, i see. This should work then for the pictures: https://imgur.com/a/aL2S36I
As for the stabilization, I don't think I use either of those considering I don't even know how to use them. The back flow isn't really close to the flame and is pretty minimal so it shouldn't be an issue I don't think.
Flame stabilisation is a part of the burner/combustor design: ie it's a way to keep the flame anchored where you want it. Otherwise you may finish up with a pulse jet system, or (on a rig I used on secondment) with the flame stabilised on the corner of the control room..... H&S were not happy about that one!
Links also count as opening attachments as it's relating to some legal definitions that CFD numerics look like kindergarden maths...... Inline images are good!
Ok i think i got the pictures in. As for flame stabilization, in the real tests I'm sure we have it; if I needed to somehow incorporate it into the model I probably haven't, the model is very simple as can be seen in the pictures.
Those look sensible (I'm not a combustion specialist) although k-e is known to be weak for jet flows.
Looking at the residuals you'd need to run much longer: 150 iterations in no where near sufficient. The shape of the residual plot implies some transients, I don't know whether this is due to flame instability or something else. However, 1m/s over about a 2cm gap may not give you a stable flame. Check with the experiment but it's probably also worth trying the model using the transient solver: initial time step will be 1m/s/ cell size / 10
