XANSYS Message: 34231 [Go back to message list]
No rating yet
Rate item:

Subject: Re: *MOPER with MAP
Author: Paris Altidis
Date: 2002-02-21 13:40:00

Rick, From my archives it appears that it was you who asked for the same
question.
Here is the thread as I have it.

============ THREAD BEGIN
=== R2 (Rick Fischer)
I read John Swanson's comment about the "default value of field 8", so I
never counted fields for myself. I'm using Dr. John's recommendation of
1/sqrt(no. input nodes) for ratio, so I guess until it goes bad, I'm not
gonna worry. Still interested about what it is doing with nodes outside the
region of input nodes. I am using this to map nodal thickness from one mesh
to another. Mesh one (InXYZ) comes from another solver that works off an
STL file from ProE. Mesh two is generated in Prep7 on an IGES model, from
the same ProE part file. Sometimes on nominally coincident edges, mesh two
(ResXYZ) nodes fall off the edge of mesh one by a smidgen, apparently due to
modeling error. Mesh two is generally finer and better shaped than mesh
one, so my first choice would be to extrapolate to these slightly off nodes.
If it is taking a nearest node value to a mesh two node, I could live with
that. If it is taking a value from a nearest mesh one node, it could give
me a problem. I suppose I could live with a zero value from the *moper
operation and go back and check for zeros and extrapolate them myself, but I
was originally doing that to all the nodes, looping thru one at a time. It
was slow, and that's I why I'm goint to the map option in the first place.

=== R1
This is from John Swanson's last note on the subject:

c ***** Map option (one set of nodes to another) ****
c *MOPER,ResVal(1,1),ResXYZ(1,1),MAP,InVal(1,1),InXYZ(1,1),
c kDim,Ratio,kOut
c kDim = 2, 2d 3, 3d
c Ratio = MaxElSize/MaxGridDim of input data
(autodefault) (was 0.10 default)
c kOut = 0, use closest point values 1, zero if outside
ARG8, by my count should be 'kOut' which will allow the user to control what
happens at the boundary. Right now the behavior is, I believe, to use the
nearest node value. This can make for quite a slow process. Per the prior
discussions, ANSYS is implementing a toggle to simply (and quickly) report
back a zero if it is out of bounds. Also, the beta MAP feature did not
support the *VMASK command. This is supposed to be fixed in the next
go-round, so that the zero valued interpolation points could be used in a
mask for subsequent calculations. I'm not sure what the 'Ratio' does
exactly (I haven't observed that much of a difference), but John had said
that:

"I have found that the default value of 0.10 in field 8 is much too high for
large problems. I suggest you supply a value in this field. A good value
is:

Value = 1.0 / (sqrt(nIn)), where nin is the number of nodes in the input
(known) set. (for kDim = 2)

Value = 1.0 / (cuberoot(nin)), where nin is the number of nodes in the input
set (for kDim = 3)"

I'm guessing 'Ratio' is just a characteristic length kind of value for
identifying possible elements to interpolate into. My guess is that this
would result in a speed issue rather than an accuracy issue, although I
didn't note that much of a difference in what I was doing. I did attempt to
interpolate at a few 'known' positions and the interpolation was as accurate
as my calculator could predict.

=== Q
I have a couple of questions regarding the map function on the *moper
command:
1. What is the significance of the ratio value in field 8? I've played
with different settings and can't see any effect. Is this an accuracy or
speed issue?
2. What is currently being done with nodes outside the region? It sounded
like from previous posts that they were assigned values from nearest nodes.
Nearest in the original set or nearest of the mapped nodes?
Rick Fischer
============ THREAD END

Paris Altidis
Staff Engineer
Borg Warner Automotive
708-547-2719
paltidis@a...

> -----Original Message-----
> From: rick.fischer@m... [SMTP:rick.fischer@m...]
> Sent: Thursday, February 21, 2002 12:03 PM
> To: xansys
> Subject: Re: [xansys] *MOPER with MAP

> I dont know anything more than you about the ratio field, but I seem to
> remember a post by John Swanson sometime back where he gave a relation for
> determining the ratio. I think it was based on some function of the
> number
> of points. Perhaps you can find that post in the archives.

> "Christian Semler" on 02/21/2002 11:59:24 AM

> To:
> cc:

> Subject: [xansys] *MOPER with MAP

> Hello:

> I have been playing recently with the *MOPER command to do some CFD
> mapping
> on a thermal model. It works really great. The only point I don't fully
> understand is the argument "Ratio" which is used to limit the search for
> the
> nearest points. I changed its value with no much effect, so I was
> wondering
> how it works exactly.

> The help says the following: Relative pinball search radius (pinball size
> to
> total size based on min and max coordinates contained in Par2). Defaults
> to
> 0.1. This argument is used to limit the search for the nearest points.

> Is that distance related to the distance of the exterior nodes to the
> surface (i.e. the distance perpendicular to the surface) or is it the
> actual
> distance of the exterior nodes to the different nodes at the surface.

> Also, if the model is quite big (which is often the case), isn't 10% of
> the
> model rather big? Even with the default equal to .1, the *MOPER sets the
> results to zero if Kout=1, which means that it is not able to find a
> corresponding nodes with a large search radius.

> Thanks in advance for any suggestion.

> -Christian

> Christian Semler, Ph.D.
> MECHANICAL DYNAMICS LTD.
> Montreal
> Tel: (514) 761-3333

Posts possibly associated with message #34231AuthorDateScore
34227*MOPER with MAPChristian Semler2002/02/21 
34228Re: *MOPER with MAPRick Fischer2002/02/21 
34231Re: *MOPER with MAPParis Altidis2002/02/21 
34232Re: *MOPER with MAPAntonio L. Negron2002/02/21 
34234Re: *MOPER with MAPChristian Semler2002/02/21 
35997*MOPER, with MAP optionsmarqueansys sebastien.marque@2002/04/10