Help:Finding sources of error
 

Hi,

I've written a NS FVM solver. I'm using Ghia's lid driven cavity result to test for accuracy.

The u values for the same grid size typically have difference of less than 2%, except one which is different by 20% (0.003 vs 0.004).

However, the v values typically have difference of 2-5%. I'm wondering if that is acceptable...

I'm trying to find out the source of error. I was thinking of wrong implementation of BC but then my u values will not be so close then. Hence I suspect that there are some careless mistakes in my v,y BC. Is that logical since my v values differs by 2-5%?

Thank you very much

Re: Help:Finding sources of error
 

Firstly, if You would like to get help here, please give description of Your solver.

NS FVM solver rather wide formulation

When I had the same problem, I fixed it having found a mistake in high-order discretization schemes of convective terms (there was a bug).

With BC there have not to be problems

U and V all zero except upper wall where U=1.

As a rule on the fine grid (300x300) with HLPA scheme You should obtain excellent results

All that from my experience with SIMPLE solver on staggered as well as colocated grids

Re: Help:Finding sources of error
 

By the way, how do You get Ghia's velosity distribution?

Re: Help:Finding sources of error
 

what do you mean by how to get Ghia's velosity distribution?

the velocities are given in his paper in journal of computational physics 1982 48 p387