[C72]

Hi all.

I am currently modelling a scaled down building model comprised of 3 regions - the outer field, the building structure, which is a porous region with 0.5m wall/roof thickness and a Pv of 25714kg/m^3.s and the inner building which is another fluid region.

I have a core mesh base size of 1m with 20 prism layers round the outside of the building, a PL total thickness of 300%ob and stretching 1.2.

My velocity at inlet is a field function that varies with height to emulate atmospheric boundary layer conditions.

When i run my simulation past the point of acceptable convergence, my pressure drop monitor plots through each respective wall face and roof seem to produce erratic unsteady oscillations. Now ive read that this can be caused by incorrect meshing of high gradient regions or boundary layers.

I am no expert on meshing, so i am just wondering if anyone has any meshing tops for me to achieve a better convergence for a model of this type -which is effectively just a solid block at the outermost level with no curve features. I've attached a picture of the current mesh. Hope someone can help.

Many thanks!

[LuckyTran]

I would have used the trimmed cell mesher for something like this, the geometry suits it.

What you can do is make a field variance monitor and do iteration averaging. Then plot the variance. It will show you exactly which cells are experiencing oscillations rather than you taking a random guess at where the problem might be.

[C72]

Hi LuckyTran, thank you for your response. Unfortunately can't seem to use the trimmed cell mesher as this doesn't accomodate multiple input parts to an automesh - those being my building inside, porous walls/roof and the outer field. That's the only reason i've had to use the less appropriate polyhedral mesher.

I will have a go at trying this field variance monitor as you suggest. How exactly would i go about this? Do you just input all parts of the flow domain, change trigger to iteration and then select my chosen report as show in the picture?

[C72]

Also if i could ask, how do you use this plotted data to see exactly where oscillations are occurring in the mesh for say the pressure drop variance through a face? I've just not done something like this before so just want to be sure. Thanks again