converging large space with k-e model

CFD system for detailed predictions of airflow and heat transfer processes in and around building space.
Post Reply
firemansam
VE Newbie
VE Newbie
Posts: 3
Joined: Mon Apr 07, 2014 4:00 am

converging large space with k-e model

Post by firemansam »

Hi,

I'm working on a very large space in microflo.

When I use the Constant Effective Viscosity method, my solution converges, and I seem to be obtaining mesh independence.

However, when I apply k-e to the model, it cannot reach convergence, even after adjusting time steps etc.

attached is my microflo monitor. The solution reaches convergence with the constant viscosity method. At approx 800 iterations, the k-e model is applied, and you can see the result.

http://s26.postimg.org/3ydryye3d/kemodel.jpg

Is my result still valid using a constant viscosity method?

How can I reach convergence with a k-e model?

Unfortunately, my mesh cannot be refined much further because the program runs out of allocated memory (when will a 64 bit version be released?).

Thanks
harshadj
IES Staff
IES Staff
Posts: 54
Joined: Wed Dec 22, 2010 4:34 am
Location: Glasgow
Contact:

Re: converging large space with k-e model

Post by harshadj »

There is not much that can be done in a large space due to the limitations of grid size. What you can also do is put up some cell monitors. Convergence does not necessarily mean the reduction in residuals. It can also be gauged from the values of variables stop changing after a the number of iterations. If the cell monitors also oscillate like the residuals only then will solution be totally useless.

Another thing to try out is changing the discretization scheme to hybrid or power law and re-run the simulation.
IES Worldwide Technical Support
firemansam
VE Newbie
VE Newbie
Posts: 3
Joined: Mon Apr 07, 2014 4:00 am

Re: converging large space with k-e model

Post by firemansam »

Is my result still valid using the constant viscosity method?

It is my understanding that constant viscosity method will introduce an artificially high level of diffusion into my model.

Since I am modeling distribution of CO2 in my large space, is the cosntant viscosity method appropriate?
rubixcube101
VE Newbie
VE Newbie
Posts: 1
Joined: Tue Dec 20, 2016 6:59 am

Re: converging large space with k-e model

Post by rubixcube101 »

Hi All

Sorry to bring up an old thread.

Has the OP or IES representative found a solution to this problem? I am running a simulation of a similar nature and encountering the same problems regarding oscillation of the residuals (all residuals). If I reduce the relaxation factor this does help to reduce the oscillations, however the solution does not seem converge even after 5000-10000 Iterations.
Post Reply