[Weiqiang Liu]

Hi all,

I am modelling methane catalytic combustion in micro-channel with fluent. Both gas phase and surface mechanism are included.

Also effective diffusion model is incorporated into CFD. When channel diameter is 2.6 mm and 2.1 mm, reasonable results can be obtained. However, when channel diameter decreases to 1.8 mm, divergence happens. I tried to modify under relaxation factor or mesh quality and nothing worked. Divergence happened after several hundreds of iterations.

I am wondering why divergence will happen when only geometry is changed. Can anybody give me some suggestions?

Best regards

Weiqiang

[Kummi]

Have you tried decreasing time step size?
What time step size you use?

[Weiqiang Liu]

Originally Posted by Kummi

Have you tried decreasing time step size?
What time step size you use?

it’s a steady state case. I am very confused about this. I thought convergence would never be a problem if I only decrease the channel diameter

Best regards

Weiqiang

[Kummi]

What are all the factors u relaxed? And their values plz..
How u initiated your problem ~ like initialization with velocity or?

Can u post the error here plz

[Weiqiang Liu]

Originally Posted by Kummi

What are all the factors u relaxed? And their values plz..
How u initiated your problem ~ like initialization with velocity or?

Can u post the error here plz

Hi Kummi,

The error messages are like:
Divergence detected in AMG solver: species-1
Divergence detected in AMG solver: temperature

I used 0.7 URF for species and 0.4 for energy. Actually I tried very small URF and it did not work.

I used standard initialization and initialized from inlet. In order to ignite the mixture, I patched a 1600k temperature to the whole computation domain.

I don't know why I can not upload picture here. Thanks very much for your concern

Best

Weiqiang

[Kummi]

In order to ignite the mixture, u have patched 1600K?
Basically, due to certain optimum temperature the surface ignition takes place. May I know, why did u patched? Hav u patched at inlet or along whole surface?

In this kind of work, long ago I haven't patched anywhere. The patch makes the region maintains the same temperature irrespective of changes due to surface temperature.. Right??

With minimal radius in your previous cases, have you validated the work ? The validation might give u enough coincidence in proceed further

[Kummi]

I personally think that the divergence is due to temperature patches you made on the domain.

[Weiqiang Liu]

Originally Posted by Kummi

I personally think that the divergence is due to temperature patches you made on the domain.

Hi Kummi,

I validated my results with larger diameter of channel , in which 1600k temperature was patched to the whole computation domain.

I am wondering why divergence will happen when diameter is decreased. When I further decreased diameter to 1.5 mm, then divergence disappeared.

This really confuses me

Best

Weiqiang

[rampada]

Dear
Did you check the maximum aspect ratio of the cell due to reduction in diameter? sometimes solution gets diverged due to this and can use coupled as solution scheme as starting point...may work.

[Weiqiang Liu]

Originally Posted by rampada

Dear
Did you check the maximum aspect ratio of the cell due to reduction in diameter? sometimes solution gets diverged due to this and can use coupled as solution scheme as starting point...may work.

Hi,

The maximum aspect ratio is even smaller for diverged case.

Yes, I used coupled scheme in my model. I found an interesting thing that when diameter decreased from 2.6 mm to 1.8 mm, divergence happened with 1.8 mm. However, when diameter further decreased to 1.5 mm, it converged very well.

I am wondering is it because my UDF of defining effective diffusion in porous zone is not right. However, I can get converged results with this UDF when channel diameter is 2.6 mm or 1.5 mm. I am really confused.

Best regards

[Weiqiang Liu]

Originally Posted by Kummi

I personally think that the divergence is due to temperature patches you made on the domain.

Hi Kummi,

I solved the problem by step by step convergence strategy. I increased the surface reaction rate step by step and then divergence disappeared.

Best.