Hi
I have modeled pipes and elbow (picture below) using *ELEMENT_BEAM_ELBOW using LSDyna. My model converged but somehow the results are not making sense. The von misses stresses are way larger than what is expected (I am getting 10 times higher stress). When I check the d3plot, I found that the elbow elements are separating badly (picture below). It happens for the second elbow as well. My *ELEMENT_BEAM_ELBOW definition is as below:
*ELEMENT_BEAM_ELBOW
1 1 1 4 3 (element ID PID N1 N2 N5 and the row below is midpoint N3)
2
2 1 4 8 3
6
.............................................
*INTEGRATION_BEAM
1 0 0.0 9 0
0.161925 0.159125 0.0 0.0 0.0 0.0 0.0 0.0 (outer radius as 0.161925 m and inne radius as 0.159125 m)
*SECTION_BEAM
1 14 1.0 -1 2 (
SECID, ELFORM, SHRF, QR/IRID, CST
)
0.0 0.0 0.0 (
PR, IOVPR, IPRSTR
)
*DATABASE_EXTENT_BINARY
0 0 36
0 0 36 (36 integration points as
BEAMIP


This is the first application of the *ELEMENT_BEAM_ELBOW keyword I have seen, so below are my suggestions based only on reading the User Manual.
Perhaps your problem is you are using the same middle node N3, i.e. 3, for the two elbow beams you listed? This node is used, along with N1 and N2, to define the curvature if the beam is curved. N3 should have a unique node number and be placed as shown in Figure 19-2 of the User Manual (see below).

Also, the so called “orientation node” N4 is only a place holder as its coordinates will be internally generated in the case of a curved beam:
“If a curved beam is defined initially, the orientation node is automatically calculated as the center of the beam curvature. However, an orientation node is still required at the input.”
So this node number can be used for multiple curved beams; adjacent circular arc beams should share the same center of curvature in any case.
Perhaps these suggestions will be helpful, --len

--
You received this message because you are subscribed to the Google Groups "LS-DYNA2" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ls-dyna2+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ls-dyna2/6502ec58-4687-4a52-9b12-b7c838cfbc7bn%40googlegroups.com.
1\ I would not use *ELEMENT_BEAM_ELBOW for straight beams, only in the elbow regions. Rather use the “nominal” *ELEMENT_BEAM with *SECTION_BEAM ELFORM=1 and your tubular cross section user defined integration.
2\ This sentence in the User Manual
“If a straight beam is defined initially, the orientation node must be defined and should be on the convex side of the beam.”
Only makes sense if you are trying to represent a curved elbow using straight beam segments, as nominal straight beams have no convex/concave side.
This is what I think you might have inadvertently done in your three element elbow definition. --len
To view this discussion on the web visit https://groups.google.com/d/msgid/ls-dyna2/fd1598d0-a385-4876-997b-0e73f0ec8dben%40googlegroups.com.
I ran your model using
ls-dyna_smp_d_R14_165-ge650b57117_winx64_ifort190.exe
and the model ran to completion, i.e. termination time 1.0
There is something “odd” about your elbow elements as shown in the bottom graphic– the same geometry appears in the d3plot file at time=0
For example, I think your elbow element 11 should be numbered
11 1 24 25 26
33
Node 25 is the center node.

To view this discussion on the web visit https://groups.google.com/d/msgid/ls-dyna2/fbf5c66a-a1fc-4fc4-8806-bb2262e9af92n%40googlegroups.com.
11 1 24 26 33
25
However I don’t understand about this, “There is something “odd” about your elbow elements as shown in the bottom graphic– the same geometry appears in the d3plot file at time=0”
Thanks again.
Have you seen this example in the User Manual?
n1-n3-n2: 1 3 2
ovalization nodes: 5 7 6
8 10 9
Associated with this diagram

The N3 node goes between the two end nodes.
I think your incorrect node numbering is why the initial d3plot geometry looks ”odd” at time=0
Thanks Manik.
I renumbered some of your elbow elements using my interpretation of the node order from the User Manual. It was then obvious to me this was NOT the correct ordering.
So your elbow elements are input correctly and I ran your model to completion using a relatively new version of LS-DYNA. I cannot say anything about the results.
However, it seems the “elbow separations” you noted in your original email are a function of the graphics between LSPP using the input file to display the initial geometry and LSPP using the d3plot file to display the initial geometry, see images below.
I note that when LSPP displays of the input file the elbows are curved and when LSPP displays the initial state from the d3plot file the elbows are straight. --len


As a follow up, I asked one of the LSPP developers about this graphics discrepancy and this was the response:
“The problem is in the d3plot file, there is no information to tell LSPP that the beams are Elbow beam. To solve the problem, after you load the d3plot files, also open the keyword input file, and tell LSPP that belongs to the same model, once you do that, the display will be correct.”
--len
To view this discussion on the web visit https://groups.google.com/d/msgid/ls-dyna2/000a01da1bc7%247f099ff0%247d1cdfd0%24%40schwer.net.
On Sun, Nov 19, 2023 at 3:10 PM <l...@schwer.net> wrote:
Thanks for the update.
I urge you to report your observation about the elbow beams to ANSYS Support. --len
To view this discussion on the web visit https://groups.google.com/d/msgid/ls-dyna2/CAJxaeX9xKtFpoPGXxSnsQP9xeiCiZjtJ4Pe8ucseB32nOzovgw%40mail.gmail.com.
Thanks for the update.
I urge you to report your observation about the elbow beams to ANSYS Support. --len
From: ls-d...@googlegroups.com <ls-d...@googlegroups.com> On Behalf Of Manik Mia
Sent: Monday, November 20, 2023 5:52 PM
To: L...@schwer.net
Cc: LS-DYNA2 <ls-d...@googlegroups.com>
Subject: Re: [LS-DYNA2] Need some help in *ELEMENT_BEAM_ELBOW
Thank you for the valuable information. The model with elbow elements seems not working. As per the suggestion of one of my friends, I used regular beam elements (ELFORM=1) for both straight pipes and elbow ones and the results are pretty close! Thanks again.
Regards,
Md Manik Mia
Graduate Research Assistant
Louisiana State University
Baton Rouge, Louisiana, USA
Contact Number: +1 225 778 9181
On Mon, Nov 20, 2023 at 12:12 PM <l...@schwer.net> wrote:
Of course straight beams can be used to “approximate” a curved elbow. The question to be asked is how accurate are straight beams in such a situation?
I would think to best answer this question of accuracy you should build a 3D shell element model of an elbow and compare results with your straight beam model approximation. --len
PS – I assume increasing the number of straight beams in an elbow geometry would improve their accuracy.