Interpolation of pressure from one mesh to another mesh

[kajalschopra]

I have performed interpoaltion of applied pressure from one mesh (mesh of tetra elements) to another mesh (mesh of brick elements).

I have done the interpolation.

I want to introduce some checks if my interpoaltion is rightly done..

Please can anyone advise what could be the checks that may be acrried out?

Please help..

Kajal

 

[olivermsun]

Some ideas:
Integrate the pressure over the domain.
Interpolate some random points from both meshes and see if you get the same results.
Interpolate back to the original mesh to see if you get something remotely similar.

 

[kajalschopra]

I did not get you here;

1) You say integrate pressure over the domain

Whay you mean exactly heer? does it mean that I should add the pressure on each of the nodes of each eleemnt?

Total pressure over the whole domain = Sum of pressures on every mode of every eleemnt?

Please can you eleaborate?

2) You said: Interpolate some random points from both meshes and see if you get the same results.

Is this continuation of 1)

Please he;lp

 

[kajalschopra]

Also, I was thinking of:

1) Get the total pressure acting on every element face before interpolation (i.e on on elements of source mesh)- I can do this as I have some intrinsic functions available in the utility I'm using...

2) Get force on every element of 1 = total pressure * area of eleement face

3) Get total force summing up forces on every element face

1,2,3 were with the source mesh

Now, 4,5,6 on destination mesh

4) Get the total pressure acting on every element after interpolation (i.e on on elements of destination mesh)- agan using intrinsic functions available in the utility I'm using...

5) Get force on every element of 4 = total pressure * area of eleement face

6) Get total force summing up forces on every element face

Compare 3 and 6 which would give an ided of quality of interpolation done.

Please advise

 

[olivermsun]

1) Yes. Your second post explains it.

2) If the procedure you describe in the second post is expensive then you can just pick a few points in the domain and see whether the values on the two meshes agree. If you're doing the interpolation just once or you want to check your algorithm then the whole procedure is worth doing.