mesh independence study for non-premixed oxy-coal combustion

[ll00023]

I am running a 2-d RANS case for non-premixed oxy-coal combustion in an elevated pressure environment.

the models used: kw-sst for turbulence, species transport with finite rate/eddy dissipation model for combustion, DPM model for particle, DO model for radiation.

3 meshes are generated in ICEM; quality check is good.

the temperature, TKE, volatile concentration are quite close for these mesh. but heat of reaction profile shows big differences between coarse and medium/fine mesh.

is this caused by finite rate/eddy dissipation model, because this model use a piece-wise function to get the reaction rate?

Thank you.

 

 

[Rob]

It may also be the coarse mesh not being sufficiently resolved in the flame region. I'm assuming the particles are much smaller than the cells in all cases as you're using DPM. 

[ll00023]

Particle average size is 65 micron. they are smaller than cell size.

I think you are right that coarse mesh is not being sufficient to resolve the flame region of the non-premixed flame at the flame front.

the coarse mesh can capture the location of flame front, the features of turbulent flow. so now it is a cost-benefit analysis. as you can see from the table, upstream (x=0.1) temperature is similar but downstream (outlet) shows differences. I assume the downstream part does not converge well for the fine mesh.

What's your suggestion to deal with this situation?

Thank you for your reply, Rob. I appreciate it.

 

[Rob]

I'd look at contours etc to understand the flow field. A single point/surface result isn't always overly diagnostic. Why assume something when you can look at the results?

[ll00023]

The flow field looks reasonable and consistent. but I can't post it here.

the issue here is: the inconsistence of heat of reaction at the flame front seems shaking the whole simulation's credibility.

I assume the simulation capture the main feature of the flow and combustion. but if the goal of the simulation is to find the flame dynamics, then the mesh needs to be much finer than the current case.

[Rob]

And I can't comment on what I can't see. As an aside, if you're using Academic licences you do need to be publishing at some point. 

 

[ll00023]

Hi Rob, thank you for this reminder. these are the result of coarse mesh case. 

it predicts the flame well. that's why I said the result, generally speaking, is reasonable.

 

temp

vel

 

[ll00023]

the data extraction location is indicated by the red line.

[Rob]

I checked with a colleague. As you're monitoring at a position, can you check the change in mesh hasn't move the reaction zone slightly? Also check point monitors, residuals and fluxes. 

[ll00023]

these results from coarse mesh.

temperature at upstream. it fluctuates a little bit but I assume this is acceptable because the turbulence is expected to be strong, even if this is a RANS case.

this is another monitor point in downstream.

Temperature profile of three meshes: 

 

[ll00023]

mass flux shows that fine mesh case has much lower dpm mass flow rate. it is quite strange.

in order to keep the same setting, I used file/write setting from medium mesh case and then read it to coarse and fine mesh.