Questions overset interpolation with DyM very slow for large 3D cases

hxaxtma · November 29, 2017, 14:40

Hi guys,
At the moment I am playing around with the new overset feature of OF 1706+. Everything is fine so far, I observed just for large 3D cases and moving mesh cabalities (for example a flapping 3D airfoil) a huge decrease in speed up (overPimpleDyMFoam). This is due to the cutting hole algorithm of the overset analysis itself.

There are three kinds of interpolations possible:
inverseDistance
trackingInverseDistance
cellVolumWeight.

So which one would you recommend in order to speed up the cutting hole algorithm. Mesh is changing here every time. At the moment I am using inverseDistance approach by defining a searchBox with subsequent Divisions.

Thanks in advance,

Yann · December 29, 2017, 07:25

I'm working with overPimpleDyMFoam on a moving object in 3D meshes and I tried the 3 algorithms on the same case.

On my case, inverseDistance and trackingInverseDistance have pretty much the same speed, while cellVolumeWeight is way much longer (about 5 times longer than inverseDistance)

Those tests have been done only on one case so I'm not sure this results is valid as a general rule.

Furthermore, I used the default setup for the 3 algorithms.
For inverseDistance, I've seen it's possible to specify 3 parameters : layerRelax, searchBox and searchBoxDivisions.

How do you use the searchBox feature and did you notice a speed up with your setup compared to the default setup?

November 29, 2017, 14:40	Questions overset interpolation with DyM very slow for large 3D cases	#1
hxaxtma Senior Member Join Date: Jan 2014 Posts: 171 Rep Power: 8	Hi guys, At the moment I am playing around with the new overset feature of OF 1706+. Everything is fine so far, I observed just for large 3D cases and moving mesh cabalities (for example a flapping 3D airfoil) a huge decrease in speed up (overPimpleDyMFoam). This is due to the cutting hole algorithm of the overset analysis itself. There are three kinds of interpolations possible: inverseDistance trackingInverseDistance cellVolumWeight. So which one would you recommend in order to speed up the cutting hole algorithm. Mesh is changing here every time. At the moment I am using inverseDistance approach by defining a searchBox with subsequent Divisions. Thanks in advance, bennn, christoph45 and Ansh like this.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
enabling multiprocessor for sliding mesh (DyM) cases	qjh888	OpenFOAM Running, Solving & CFD	5	September 29, 2016 03:33
Slow Cases using HPC and PBS job scheduler	Alhasan	OpenFOAM Running, Solving & CFD	5	February 1, 2015 12:07
Questions on the overset method of unstructured grid.	Sheldon Lee	Main CFD Forum	1	June 5, 2014 14:54
Slightly embarassing questions about Rhie-Chow interpolation	mercator	Main CFD Forum	16	June 11, 2013 05:35

December 29, 2017, 07:25		#2
Yann Senior Member Yann Join Date: Apr 2012 Location: France Posts: 181 Rep Power: 12	I'm working with overPimpleDyMFoam on a moving object in 3D meshes and I tried the 3 algorithms on the same case. On my case, inverseDistance and trackingInverseDistance have pretty much the same speed, while cellVolumeWeight is way much longer (about 5 times longer than inverseDistance) Those tests have been done only on one case so I'm not sure this results is valid as a general rule. Furthermore, I used the default setup for the 3 algorithms. For inverseDistance, I've seen it's possible to specify 3 parameters : layerRelax, searchBox and searchBoxDivisions. How do you use the searchBox feature and did you notice a speed up with your setup compared to the default setup?