CFD Online URL
[Sponsors]
Home > Forums > Pointwise & Gridgen

[Pointwise] why does joining takes so much time?

Register Blogs Members List Search Today's Posts Mark Forums Read

Like Tree1Likes
  • 1 Post By cnsidero

Reply
 
LinkBack Thread Tools Display Modes
Old   November 22, 2011, 09:33
Default [Pointwise] why does joining takes so much time?
  #1
Member
 
Join Date: May 2010
Posts: 44
Rep Power: 6
la7low is on a distinguished road
Hi,
I try to join a number of (coons patch) surfaces which were created from curves. The joining of 128 such surfaces lasted for ~6000s(=1.5h) and the resulting file was ~260MB. Why does it last so long and why is the file so big? What other options do I have in Pointwise to join quite a few number (hundredish) of surfaces in a fair amount of time (max some minutes)?

Thanks in case someone can help or give ideas! Id appreciate it!
la7low is offline   Reply With Quote

Old   November 22, 2011, 13:59
Default
  #2
Senior Member
 
Chris Sideroff
Join Date: Mar 2009
Location: Ottawa, ON, CAN
Posts: 358
Rep Power: 11
cnsidero is on a distinguished road
la7low,

I wouldn't recommend joining surfaces, particularly since it's not necessary in Pointwise. There are couple of ways to approach this.

First, if you want to reduce the number of "meshable" surfaces, I suggest you use the solid modeling cabaility. Solid modeling allows you to "join" surfaces not a physical sense - which is very computationally expensive as you've found out - but in a topological sense. The way I think of solid modeling is that it's creating a "glue" that describes how adjacent surfaces should be connected to each. I won't describe how to use solid modeling here, you'll want to refer to tutorial #3 of the tutorial workbook (found in Help > About). Since you have a lot of surfaces you want to "join", pay particular attention to the section on quilting.

Check out this video for a quick overview of solid modeling:

http://www.pointwise.com/videos/soli..._modeling.html

There other approach is to not join or use solid modeling, instead leaving the surfaces as is, create surface meshes on all the surfaces and join the surface meshes. This technique is called fault-tolerant meshing.

Check out this video for a quick overview of fault-tolerant:

http://www.pointwise.com/videos/faul..._tolerant.html

Let me know if this is what you're after.

Enjoy, Chris

Quote:
Originally Posted by la7low View Post
Hi,
I try to join a number of (coons patch) surfaces which were created from curves. The joining of 128 such surfaces lasted for ~6000s(=1.5h) and the resulting file was ~260MB. Why does it last so long and why is the file so big? What other options do I have in Pointwise to join quite a few number (hundredish) of surfaces in a fair amount of time (max some minutes)?

Thanks in case someone can help or give ideas! Id appreciate it!
cnsidero is offline   Reply With Quote

Old   November 22, 2011, 18:16
Default thx, why?
  #3
Member
 
Join Date: May 2010
Posts: 44
Rep Power: 6
la7low is on a distinguished road
Dear Chris,
firstly thanks for the quick reply!
Secondly, I admit I should have explained, but I am about to create a special surface made out of several bits. Thus, I am not dealing with a dirty geometry with tiny little unwanted surfaces, but all of my surface elements are needed. Those surface elements will create the whole surface.
Still, your recommendations are good although not breaking the ice:

The 1st way: (I tried this)

I assemble the model, but then I have lots of quilts (same number as the number of surface elements). But then I will have lots of connectors when I create connectors on database entities. So then I have to delete the unwanted connectors and join lots of others...

The 2nd way (was about to continue this way):
I leave surfaces as is the create the connectors on them, then delete and join as before (only I dont have many useless quilts and a model).

Lastly, back to the beginning:
Could you explain me why is it so computationally expensive to join the surfaces?

Thanks again!
la7low is offline   Reply With Quote

Old   November 22, 2011, 18:20
Default s is not there
  #4
Member
 
Join Date: May 2010
Posts: 44
Rep Power: 6
la7low is on a distinguished road
By the way, I made a grammatical mistake in the topic name...
la7low is offline   Reply With Quote

Old   November 22, 2011, 19:07
Default
  #5
Senior Member
 
Chris Sideroff
Join Date: Mar 2009
Location: Ottawa, ON, CAN
Posts: 358
Rep Power: 11
cnsidero is on a distinguished road
Quote:
Originally Posted by la7low View Post
Secondly, I admit I should have explained, but I am about to create a special surface made out of several bits. Thus, I am not dealing with a dirty geometry with tiny little unwanted surfaces, but all of my surface elements are needed. Those surface elements will create the whole surface.
None of the techniques I described will remove or modify surfaces - they only reconstruct the proverbially "glue" I described earlier. They can be used for dirty CAD but also, as in your case, can be used to help simplify the resulting mesh.

Quote:
Originally Posted by la7low View Post
The 1st way: (I tried this)

I assemble the model, but then I have lots of quilts (same number as the number of surface elements). But then I will have lots of connectors when I create connectors on database entities. So then I have to delete the unwanted connectors and join lots of others...
Exactly. So now you have a model with a lot of quilts - start joining the quilts to reduce the number of domains and connectors that will be created. The purpose of quilting is to try and recreate the engineering intent in the surface definition which was never there to begin with in the CAD system. Tip: don't join quilts across hard edges (sharp corners) otherwise you will lose the definition there.

Remember, when you mesh a model, Pointwise will create one domain for each quilt. So reducing the number of quilts will reduce the total number of entities.

Quote:
Originally Posted by la7low View Post
The 2nd way (was about to continue this way):
I leave surfaces as is the create the connectors on them, then delete and join as before (only I dont have many useless quilts and a model).
Yes but in the process of joining Pointwise will automatically remove the connectors at the seam where the domains were joined so clean up should of extra entities be painless.

Quote:
Originally Posted by la7low View Post
Lastly, back to the beginning:
Could you explain me why is it so computationally expensive to join the surfaces?
That's a good question. I would have to ask a developer as to why this might be. My limited knowledge on the subject of computational geometry would lead me to believe that the joining of many surfaces into one surface is a non-trivial operation. At their core, almost all CAD formats use the non-uniform rational B-spline (NURB) surface as the fundamental geometric surface entity (our Coons surface is similar in form to a NURB surface). There are analytical equations equations that define a NURB surface. Therefore reconstructing a new "joined" NURB from a collection of smaller surfaces would seem to be a non-trivial operation. I am just speculating of course.

Now that I think about, this is precisely how and why CAD packages construct "solids" from a collection of surfaces. It's much easer (and likely more robust) to create a complex object from simple fundamental entities and tie them together with topological "glue" than one large entity. Our solid modeling capability, uses the same principles that a full feature CAD system does for assembling solids. BTW, there is no requirement that a solid model be closed - it's pefectly valid for them to be open.

Phew ... I digress.

Getting back to your original inquiry, let me know if I'm still pointing you in the right direction or I'm getting further off base.
la7low likes this.
cnsidero is offline   Reply With Quote

Old   November 23, 2011, 09:23
Default solution
  #6
Member
 
Join Date: May 2010
Posts: 44
Rep Power: 6
la7low is on a distinguished road
Dear Chris,

okey, due to your suggestions I think I found the best way finally:

1. create assemble models on the series of surfaces
2. join the quilts with create-assemble quilts (just found this out today, before I would have tried edit-join (as it greyed out)
3. Create connectors on the joined surfacepatch-quilt, then I just join the connectors on two sides of the assembled surface (lets say along u direction of the surface).
Nice, thanks a lot again, you pointed me to the good direction!

About the analytical join: I can grasp it a bit, thanks for the info.
Still, it would be nice to make a faster analytical reapproximator. I think Pointwise tries to reapproximate the surface elements with the given tolerance but to reach this prescribed level of tolerance it has to use lots of terms in the equation of the NURB describing the joined surface. This might be the reason for huge .pw file size and longlasting effort...but not sure though...
la7low is offline   Reply With Quote

Old   November 29, 2011, 17:52
Default domain creation
  #7
Member
 
Join Date: May 2010
Posts: 44
Rep Power: 6
la7low is on a distinguished road
Yet another question:
Based on what Pointwise decides on whether a domain can be created on existing databases? I have assembled my surfaces to a model and I can't create domain on it (it is a squre-like surface), whereas Pointwise allows me to do that in case of the prepared geometry avalable for download from here. What should I do with model databases in order to be avalable for creating domains on them?
Thnaks in advance!
la7low is offline   Reply With Quote

Reply

Tags
coons patch, join, pointwise

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
TimeVaryingMappedFixedValue irishdave OpenFOAM Running, Solving & CFD 26 July 24, 2014 05:30
Time step size and max iterations per time step pUl| FLUENT 24 January 16, 2014 07:35
directMapped problem panda60 OpenFOAM Bugs 4 July 8, 2010 11:23
Time step in transient simulation shib FLUENT 0 June 17, 2010 14:07
unsteady calcs in FLUENT Sanjay Padhiar Main CFD Forum 1 March 31, 1999 12:32


All times are GMT -4. The time now is 13:39.