I'm modeling a tubular combustor with swirler& vane passage fuel injection as part of my PhD and I'm having big problems with the combustion model.

For these swirling combustion flows the finite rate/eddy dissipation model is essential but when I use it the flame goes out eventually after around 25 iterations even when having the outer wall at 1500K and patching in the temp of 1300K.

The boundary conditions were worked out by my supervisor and previous theses confirm them. The aerodynamics are spot on but when adding combustion (methane) the flame keeps going out for this model. The eddy dissipation model can be used but combustion occurs where the flow velocity is very high so flame should be quenched. I wrote a UDF to halt reactions > 80,100,120m/s but results are still not great but more sensible.

have you read the book computational f.d. in industrial combustion by baukal the chapter "gas turbine". it may help you. if you agree we could be good coworkersbecause i also work on it in the sharif university.

Not read that book but was planning to get it soon to read it for that exact reason! Sorted out the last problem- needed the hot wall to be alot hotter.

I did read Theoretical and Numerical Combustion and thought it was good- theres a new book coming out just now called An introduction to the finite volume method that has a section on GT's. Meant to be really good.

Phil

the reasons of the flame quenching are various. it may be from your mixing model.one is the stability criteria that relates with the wall temperature.can i know what is your solver and models?

I've sorted that problem out by having the wall at high temp and patching the temp in at the beginning. The problem now is that the wall temp is unrealistically high but at least the results are good. My supervisor says I may need to write a sub-routine for an adiabatic wall but not sure yet if that is correct. I don't think its worth that until I'm using the proper detailed reaction mechanism.

model results now seem to be good compared to previous 30000 node 10 years ago model.

Solver: Fluent, 3d, segregated, steady state

Turbulence: standard k-epsilon as some previous work that I am comparing my model to for initial validation used this model

Combustion: species transport (same reason as for turbulence) finite rate/eddy dissipation energy equation methane/air 300/600Kelvin 146/4.62m/s 2.5/250mm diam

Phil

In most industrial combustor cases, you tend to run with a combined combustion model using an appropriate finite rate reduced chemistry. Alternatively some people use a strain rate limiter instead.

These typically quench zones, i.e. boundary layers, that can light up if you only have turbulence based model.

You seem to be having the opposite problem. Starting the flame at 1300K seems cool as the head ends of most industrial gas turbines are running nearer 1500C (not K). These systems will typically have 1-200C of turndown in premix before they have to stage to a higher NOx more stable mode.

I am also surprised that the wall is having such a big influence. It may well be that the problem lies elsewhere (i.e. in the flame holding zone) and that not having a very hot wall is simply the last straw. Typical maximum realistic wall temperatures would be in the 1000-1100C zone for coated parts.