CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   2nd order interpolation for NS solver (

Quarkz July 13, 2005 02:11

2nd order interpolation for NS solver

I'm now trying to modify my cartesian grid to structured grids. My main objective is to model airfoil.

I'm using 2nd order projection mtds to solve the NS eqn and hence at some pt, I need to evaluate the diffusive and convective flux.

According to Peric's book on CFD, it uses linear interpolation between the 2 nodes on either side of the face to evaluate the face values. Moreover, it is stated that this is 2nd order.

However, a paper by Kim & Choi (A 2nd order time accurate FVM for unsteady incompressible flow on hybrid unstructured grids) uses bilinear interpolation with 6 nodes to get the face value and it is also a 2nd order.

Hence, can I just use Peric's linear interpolation or must I use Kim's bilinear interpolation to achieve 2nd order accuracy?

Thanks alot

ramp July 13, 2005 02:46

Re: 2nd order interpolation for NS solver
Are u modifying ur cartesian grid to structured or unstructured grid ?

Quarkz July 13, 2005 02:48

Re: 2nd order interpolation for NS solver
To structured grid for airfoil (o or c grids)

coolblizz July 13, 2005 13:03

Re: 2nd order interpolation for NS solver
Hi Quarkz....

The linear interpolation in Peric's book is 2nd-order only if the grids are orthogonal. If the non-orthogonality is severe, the interpolation becomes 1st-order.

The interpolation by Kim and Choi will be more robust in this case since it uses more cells.

Are you modifying your cartesian grid to body-fitted structured grid?

Quarkz July 13, 2005 19:49

Re: 2nd order interpolation for NS solver

thanks. yes, i'll modifying my grid to body-fitted structured one for airfoils. in that case, is it recommended to use bilinear interpolation at the airfoil's leading and trailing since non-orthogonality is more pronounced at those regions?

coolblizz July 14, 2005 14:06

Re: 2nd order interpolation for NS solver

Yes, sure. You will definitely want to minimize the loss of information at those regions. Just take care that if you are doing supersonic flows, the face values will have to be managed using some slope limiters for shock treatment.

Just a side thought, is the name "bilinear interpolation" used in the paper by Kim and Choi? The word bilinear sounds to me it's an interpolation done on orthogonal grids.

Quarkz July 15, 2005 21:50

Re: 2nd order interpolation for NS solver
yes it is written that "...and is equivalent to the bilinear interpolation in rectangular grids, ...."

All times are GMT -4. The time now is 06:42.