XANSYS Message: 42310 [Go back to message list]
No rating yet
Rate item:

Subject: Re: APDL - selecting nodes
Author: Glailson Mattos
Date: 2002-09-30 14:55:00

The mesh was refined near the line by the command LREFINE and there
are wide gaps in the nodes numbering.
I think that *GET the node number and use the NNEAR function is the
best approach.

Thank you Don, for your suggestion.

Glailson Mattos
Metallurgist Engineer
USIMINAS - BRAZIL

--- In xansys@y..., Shaffer Don wrote:
> I have not looked in the documentation on this
> but you may be able to sequence thru your nodes
> using one of the *get commands for lines (perhaps.)
> Typically the mesh sequence for a line with only the
> two endpoints is to place a node at each end then
> fill in between with the number based upon
> a.) user specified number of edge divisions
> b.) specified edge length
> c.) user specified global element size
> d.) default element edge length

> I am uncertain if that is the same sequence that is
> used for splines. You may want to do testing to see
> if that is the case.

> Another approach is to *get the number of nodes on the
> line, or alternatively use the line inquiry functions
> and then select only those nodes attached to the line
> from the full model, next *get the node number of the
> node on one end of your model. Next use the nodenearest
> functions to get the next closest node. Unselect the
> first node from your current group of selected nodes.
> now get the node nearest the second node in the group.
> continue through this until you reach the end of the
> number of nodes associated with this line. Get the next
> line and attached nodes, continue the identification
> process starting with the last node identified in the
> first line. After identification create the groups
> of the node numbers you have saved in an array.

> Regards
> Don Shaffer
> Siemens Westinghouse Power Corp.
> Orlando, Florida

> -----Original Message-----
> From: glailson_mattos [mailto:gmaraujo@u...]
> Sent: September 30, 2002 12:36 PM
> To: xansys@y...
> Subject: [xansys] Re: APDL - selecting nodes

> Hi, Rene!

> The line created by BSPLIN is sinuous and has more than 300ø from
> start to end (almost close). Hence I guess this approach would not
> run because the lack of a logical orderind rule.

> Thanks, any way.

> Glailson Mattos
> Metallurgist Engineer
> USIMINAS - BRAZIL

> --- In xansys@y..., "Fiege, Rene" wrote:
> > Glailson

> > you can use vector/matrix commands.
> > 1) create a matrix with the node numbers and the coordinates
> > matrix(NodeNumber,x,y,z) -> including all coordinates you have
> to filter
> > 2) sort matrix using the direction you are interested in (*moper)
> > 3) compress to the nodes of interest
> > 4) use the *do loop on the vector, which is faster than the
> selection

> > I suppose you select the nodes by a location in a specific
> coordinate
> > system, which is faster, because you do not need a *do loop at
all.
> You can
> > directly create a masking vector for the matrix comparing the
> coordinate
> > data to a specific location information using *voper (in fact in
> this case
> > the matrix could be a vector holding the coordinates only).

> > Hope this helps,

> > Ren‚ Fiege
> > Meyer Werft
> > Germany

> > -----Original Message-----
> > From: glailson_mattos [mailto:gmaraujo@u...]
> > Sent: Monday, September 30, 2002 2:27 PM
> > To: xansys@y...
> > Subject: [xansys] APDL - selecting nodes

> > Hi, XANSYSers!

> > My model has a sinuous line build by the BSPLIN command. Attached
> to
> > that line there are a lot of nodes from the mesh operation.
> > I need separate those nodes in two groups from each ten
consecutive
> > nodes on the line (the nodes' numbering is randomic and there
isn't
> > any geometric rule). I mean create two components like the
> following
> > scheme

111111111100000000001111111111000000000011111111110000000000 .......

> > I am able to count the select nodes and separate the nodes's
number
> > in two arrays using a *DO loop and after that issue a NSEL,s,....
> and
> > NSEL,a,... for each array index, but this solution is slow and
not
> > very smart.

> > Does anyone know a smart and fast method to get it?
> > Any help is welcome.

> > Glailson Mattos
> > Metallurgist Engineer
> > USIMINAS - BRAZIL
Posts possibly associated with message #42310AuthorDateScore
42292APDL - selecting nodesGlailson Mattos2002/09/30 
42297Re: APDL - selecting nodesRene Fiege2002/09/30 
42304Re: APDL - selecting nodesGlailson Mattos2002/09/30 
42308Re: APDL - selecting nodesDon Shaffer2002/09/30 
42310Re: APDL - selecting nodesGlailson Mattos2002/09/30 
42311Re: APDL - selecting nodesOtto, Mike J MikeJOtto@2002/09/30