FEAP User Forum

FEAP => nurbFEAP => Topic started by: alina.i on June 02, 2018, 02:56:03 AM

Title: Knot insertion in order to split patch
Post by: alina.i on June 02, 2018, 02:56:03 AM

Dear FEAP Team,

I was testing knot insertion command in order to split my patch into separated patches (obtain C-1 continuity). So, for second-order functions, I insert knots 3 times. However, after the last insertion, I obtain wrong values of coordinates and weights.  Is there a way to solve this problem? The body is small, see input and two meshes attached.

I am using Version 8.4. License belongs to TU Dresden.
I would really appreciate any hint.

Best regards,
Alina Israfilova

Title: Re: Knot insertion in order to split patch
Post by: Prof. R.L. Taylor on June 02, 2018, 09:31:49 AM

The knot insertion algorithm only makes sense to C_0.  After that there are no control points at the ends.  Once you are at C_0, to split the patch would be done the same way you would split standard C_0 isoparametric elements.  To understand this think of what you would have for control points for the knot vectors (0,1) and (0,0,1,1).

Title: Re: Knot insertion in order to split patch
Post by: alina.i on June 18, 2018, 04:19:23 AM

I see. Thank you for your help, Professor!

Best regards,
Alina Israfilova

Title: Re: Knot insertion in order to split patch
Post by: Yaakov on July 12, 2018, 07:50:43 AM

 In pre-process, knot insertion ('not local') can be done very easily in NURBS Toolbox ('h,p,k', fully 'free', Octave), which is still being developed by Prof. Alessandro Reali's group at the University of  Pavia.

https://octave.sourceforge.io/nurbs/

If the code works in Octave, the code could also work in Matlab. ('openoffice <-> microsoft office')

But NURBS Toolbox is only used for pre-processing, the main job (99,999999%) must be done in FEAP85/84.

Title: Re: Knot insertion in order to split patch
Post by: Yaakov on July 22, 2018, 08:26:38 AM

@Alina Israfilova, compare with IgA in FEAP84, IgA in FEAP85 includes many new features, more robust, much more fast, etc.

Title: Re: Knot insertion in order to split patch
Post by: alina.i on July 26, 2018, 06:58:51 AM

Quote from: Yaakov on July 12, 2018, 07:50:43 AM

In pre-process, knot insertion ('not local') can be done very easily in NURBS Toolbox ('h,p,k', fully 'free', Octave), which is still being developed by Prof. Alessandro Reali's group at the University of  Pavia.

https://octave.sourceforge.io/nurbs/

If the code works in Octave, the code could also work in Matlab. ('openoffice <-> microsoft office')

But NURBS Toolbox is only used for pre-processing, the main job (99,999999%) must be done in FEAP85/84.

Thank you!! I will take a look!

Title: Re: Knot insertion in order to split patch
Post by: Yaakov on July 30, 2018, 08:03:22 AM

You are welcome!

NURBS Toolbox contains very limit number of NURBS GEOs. But for h-p-k global refinement, NURBS Toolbox can be a very good assist for pre-processing in  IGA85/FEAP85. 

P.S. if you need k refinement in  NURBS Toolbox , first step:  p refinement,  then next step h refinement.

Quote from: alina.i on July 26, 2018, 06:58:51 AM

Quote from: Yaakov on July 12, 2018, 07:50:43 AM

In pre-process, knot insertion ('not local') can be done very easily in NURBS Toolbox ('h,p,k', fully 'free', Octave), which is still being developed by Prof. Alessandro Reali's group at the University of  Pavia.

https://octave.sourceforge.io/nurbs/

If the code works in Octave, the code could also work in Matlab. ('openoffice <-> microsoft office')

But NURBS Toolbox is only used for pre-processing, the main job (99,999999%) must be done in FEAP85/84.

Thank you!! I will take a look!