grid convergence problem
 

Hi
My channel is 50 X 50 micrometer square channel having length 1000 micrometer.I am using the workbench for the analysis. I am giving different mesh size by changing the value in
I am getting different number of elements after each change.
I tried Number of elements ranging from 7000 to 3 lakhs.
My inlet boundary condition is mass flow rate 8.333e-7 kg/sec
Outlet boundary condition is outflow
Fluid water
No turbulence.Simple flow
viscous laminar.No energy terms
But when i find out the average inlet and outlet total pressure from post processor..and to find out pressure drop through the channel.When i am doing this the pressure drop value also increases from 3670 pa to 3980 pa as the number of elements increases from 7000 to 3 lakhs. So now i cant get what is the optimum number of elements required for the further proceedings.grid convegence is not happened.



Thanks

when you change mesh from 7000 to 3 lacks you have change in pressure 3670 to 3980 pa which is 8% increase from initital value at 7000 nodes.

What is the difference between mesh size of 150,000 and 300,000 nodes?

the number of elements pressure drop in Pa
312500 3982.393
214896 3931.363
160000 3939.024
96237 3890.693
20000 3759.658
10688 3670.525


A linear increase as the number of elements increase. what to do which elemnt size is better..or any problem of meshing....

thanks for your reply

So in your last two cases (two fine meshes) the change is 1.3%!. If it is acceptable to you then you have achieved grid convergence. Even with lower mesh density you have grid convergence if error is acceptable to you.

Plot your all cases in excel and see if you have almost straight line after some mesh density.

i tried in excel but it is linearly increasing trend not at all a constant trend...then what should i do...any problem with meshing..

If error is below your requirements then you can assume solution to be mesh independent...

I use the Grid Convergence Index proposed by Roache to perform grid convergence studies
You may find the description of the procedure here:
http://www.google.com/url?sa=t&rct=j...45645796,d.eWU

Through this, you can estimate grid uncertainty and if your grid solutions lie within the asymptotic convergence range

regards