Traction load on NURBs

[T Niemann]

Dear Prof. Taylor, dear Forum-Admins,

when I run a 3D IGA problem with loading normal to the surface of a patch, it works fine. But if I try to apply a traction (tangential to the surface, here: coordinate y-direction) with:

NLOAd 
   TRAC 1 1 5 2 1      !patch, face, traction, direction, prop-num

the force is still applied in the normal direction (x-direction). Is this a bug or did I make a mistake in the input? All the examples I found are dealing with normal forces only. My version is 8.5.2i.

Thank you in advance!

[Prof. R.L. Taylor]

It may be a bug.  Can you post the input files so I can test to see what may be wrong.

[Prof. R.L. Taylor]

I looked further and it appears that the traction is only code for 2-d problems at present.  I will look at it some more later and see if we can add the other cases for the next release or a patch.

[Prof. R.L. Taylor]

I made a stab at adding the traction case.  I wonder if you can check if it is o.k.  (or send me your input file and I can check too).

the routine goes in ./igafeap/program for ver8.5

[T Niemann]

Dear Professor Taylor,

thank you very much for the quick answer! I checked your "stab" and it seems to work, i.e. the load is applied in the desired direction.

During checking I found out that in line 94 et seqq. of plblock.f you attached still the case for ndm.eq.3 is still missing (max. integration points 'lval' that seem to work are 5). I assume the definition of 'lval' is not correct, as in the code it is meant to be the number of integration points per direction (maybe also for normal forces). According to IGAManual8.5 a traction is applied for example by:

LOAD PROP xyz
NLOAd
     TRACtion patch face traction direction prop-num
LOAD END

A mistake seems to be that xyz should be assigned to 'ldprp' (i.e. the prop-num!) and what is called prop-num is in the code 'lval' (the value of the integration points per direction), is this correct?

These are just hints, as mentioned, the possibility to apply traction forces in 3D helped me already! 

Best regards

[Prof. R.L. Taylor]

The manual is wrong.  The 'prop-num' should be 'quad-pt' , the number of points/direction (default in 2-d is 7 and in 3-d is 4).

The other values should come from the load table - but may not have been tested.
If you have a problem set up I would like to test to make sure they are not needed or are properly placed in the table.

[T Niemann]

Attached you find the used input-file.