Using HEX20 elements in conjunction with BEAM3 elements

21 views
Skip to first unread message

Mriganabh

unread,
Jun 17, 2020, 5:43:24 PM6/17/20
to mastodon-users
Hi all,

I am trying to use HEX20 elements in conjunction with BEAM3 elements. The model contains 4 HEX20 elements and 4 BEAM3 elements (as shown in figure-1). The beam elements are sitting on top of the solid elements. I merged the nodes where the beam and the solid meet.


Figure-1
The simulation aborts due to segmentation fault. I have shared my input files. Let me know where I am going wrong.

P.S. - The simulation runs fine when I use HEX8 elements.

Thanks,
Mrignaabh
cantilever_solidbeam_2element.i
solidbeam_cantilever_2element.e
accel_x.csv
accel_y.csv
accel_z.csv

Christopher Wong

unread,
Jun 17, 2020, 5:51:22 PM6/17/20
to mastodon-users
Have you successfully used HEX20 elements in MOOSE before? I'm not sure if there's support for this element type. 

Mriganabh

unread,
Jun 17, 2020, 6:11:50 PM6/17/20
to mastodon-users
Hi Chris,

Yes. I have used HEX20 in Mastodon.

Around 6-7 months back, I had tried using HEX20 in conjunction with HEX8 elements but Dr. Sweta advised me against it since, at that time in MOOSE, the quadrature rule cannot be set per block. Hence, the entire model would be solved at a higher quadrature. 

Thanks,
Mriganabh

Chandrakanth Bolisetti

unread,
Jun 17, 2020, 6:20:17 PM6/17/20
to mastodon-users

Hi Mriganabh,

 

Couple of things:

  1. When you connect beams to solids, you need to make sure that the beam moments are transferred into the solids. We do this by modeling some stiff ‘connector’ beams that connect the beam elements and solid elements. Please refer to example 9 for this.
  2. While the connection may not be the source for the seg fault, we prefer that users don’t see seg faults at all. If you change your solids to HEX8, do you still see a seg fault? If so, can you share your model?

 

Chandu

 

-- 

Chandu Bolisetti, Ph.D.

Facility Risk Group Lead

Idaho National Laboratory

716-352-5107 (M)

208-526-8161 (O)

 

 

From: mastodon-users <mastodo...@googlegroups.com> on behalf of Mriganabh <mrigan...@gmail.com>
Date: Wednesday, June 17, 2020 at 4:14 PM
To: mastodon-users <mastodo...@googlegroups.com>
Subject: [EXTERNAL] [mastodon-users] Re: Using HEX20 elements in conjunction with BEAM3 elements

 

Hi Chris,

 

Yes. I have used HEX20 in Mastodon.

 

Around 6-7 months back, I had tried using HEX20 in conjunction with HEX8 elements but Dr. Sweta advised me against it since, at that time in MOOSE, the quadrature rule cannot be set per block. Hence, the entire model would be solved at a higher quadrature. 

 

Thanks,

Mriganabh

On Wednesday, June 17, 2020 at 5:51:22 PM UTC-4, Christopher Wong wrote:

Have you successfully used HEX20 elements in MOOSE before? I'm not sure if there's support for this element type. 

On Wednesday, June 17, 2020 at 3:43:24 PM UTC-6, Mriganabh wrote:

Hi all,

 

I am trying to use HEX20 elements in conjunction with BEAM3 elements. The model contains 4 HEX20 elements and 4 BEAM3 elements (as shown in figure-1). The beam elements are sitting on top of the solid elements. I merged the nodes where the beam and the solid meet.

 

Image removed by sender.

Figure-1

The simulation aborts due to segmentation fault. I have shared my input files. Let me know where I am going wrong.

 

P.S. - The simulation runs fine when I use HEX8 elements.

 

Thanks,

Mrignaabh

--
https://mooseframework.org/mastodon
https://github.com/idaholab/mastodon
---
You received this message because you are subscribed to the Google Groups "mastodon-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mastodon-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mastodon-users/15b55bc8-4624-4cbd-8e74-fbcd6189f873o%40googlegroups.com.

Mriganabh

unread,
Jun 17, 2020, 6:37:14 PM6/17/20
to mastodon-users
Hi Chandu,

I will keep that in mind for the future. 

And no, the simulation runs fine when I use HEX8 elements in conjunction with BEAM2 elements. The reason I switched to BEAM3 when using HEX20 is cause I was getting an error that the "EDGE2" element type is bad for second-order approximation. Upon changing to a 3 noded beam element, I was able to get rid of that error but am facing the segmentation fault.

Thanks,
Mriganabh

To unsubscribe from this group and stop receiving emails from it, send an email to mastodo...@googlegroups.com.

Som Dhulipala

unread,
Jun 17, 2020, 6:38:01 PM6/17/20
to Chandrakanth Bolisetti, mastodon-users
Hi Mrignabh,

I ran your input file using MASTODON DBG and this is the error message I am getting:

Assertion `_local_re.size() == 2' failed

StressDivergenceBeam: Beam element must have two nodes only.

at /Users/dhulls/projects/moose/modules/tensor_mechanics/src/kernels/StressDivergenceBeam.C, line 106

[3] /Users/dhulls/projects/moose/modules/tensor_mechanics/src/kernels/StressDivergenceBeam.C, line 106, compiled Jun 17 2020 at 16:30:15


It seems there is an issue with your beam connectivity, but I need to dig a little deeper.


Thank you.


To view this discussion on the web visit https://groups.google.com/d/msgid/mastodon-users/3E25A8A8-3DC2-4E1F-9001-251B1D001A8A%40inl.gov.


--
Som Dhulipala, Ph.D.
Postdoctoral Research Associate,
Facility Risk Group,
Idaho National Laboratory

Chandrakanth Bolisetti

unread,
Jun 17, 2020, 6:47:45 PM6/17/20
to mastodon-users

Hi Mriganabh,

 

Looks like Som just debugged it. You can use HEX20 but you cannot use 3-noded beams in MOOSE/MASTODON. The beams are currently limited to 2-noded elements.

 

Even when you switch to 2-noded elements, you might run into convergence issues if you don’t model the connection properly.  

Error! Filename not specified.

Figure-1

The simulation aborts due to segmentation fault. I have shared my input files. Let me know where I am going wrong.

 

P.S. - The simulation runs fine when I use HEX8 elements.

 

Thanks,

Mrignaabh

--
https://mooseframework.org/mastodon
https://github.com/idaholab/mastodon
---
You received this message because you are subscribed to the Google Groups "mastodon-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mastodo...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mastodon-users/15b55bc8-4624-4cbd-8e74-fbcd6189f873o%40googlegroups.com.

--
https://mooseframework.org/mastodon
https://github.com/idaholab/mastodon
---
You received this message because you are subscribed to the Google Groups "mastodon-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to mastodon-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mastodon-users/7b5a833c-eea3-4c03-872e-2fffb96e610ao%40googlegroups.com.

Mriganabh

unread,
Jun 17, 2020, 6:50:13 PM6/17/20
to mastodon-users
Hi Som,

1. What does the Beam element must have two nodes only mean? BEAM3 element in Trelis is a 3-noded element with the mid-point node having 0 DOFs. Hence, I thought the mid-point node might act as a dummy node. Correct me if I am wrong.

2. The bottom-most node of the beam and the center node at the top face of the solid has the same node number (Node ID - 19). I request you to go through the mesh file and double-check that. I am not sure if that's what is causing the error.

Thanks,
Mriganabh

To unsubscribe from this group and stop receiving emails from it, send an email to mastodo...@googlegroups.com.

--
https://mooseframework.org/mastodon
https://github.com/idaholab/mastodon
---
You received this message because you are subscribed to the Google Groups "mastodon-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mastodo...@googlegroups.com.

Mriganabh

unread,
Jun 17, 2020, 6:53:44 PM6/17/20
to mastodon-users
Hi Chandu,

Then how should I go about using HEX20 in conjunction with BEAM elements?

I had initially tried using BEAM2 elements and I get the following error.
--------------
*** ERROR ***
ERROR: Bad ElemType = EDGE2 for SECOND order approximation!
--------------

Thanks,
Mriganabh 

Som Dhulipala

unread,
Jun 17, 2020, 6:55:54 PM6/17/20
to Mriganabh, mastodon-users
Hi Mriganabh,
  • Please see Chandu's latest response. He says three noded beam elements cannot be used with MASTODON/MOOSE.
  • As Chandu mentioned previously, it's recommended you connect the vertical beam to horizontal rigid beams. Since the vertical beam has rotational DOF and the solid element does not, connecting the vertical beam to rigid horizontal beams is recommended. Please see example 9 on MASTODON website.
  • If you get a segmentation error in the future and want to find out what's wrong, please do this: run 'export METHOD=dbg' and then compile MASTODON using 'make'. THen run your input using 'mastodon-dbg' instead of 'mastodon-opt.' 
Thank you.

To unsubscribe from this group and stop receiving emails from it, send an email to mastodon-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mastodon-users/8c5411b0-181e-4c68-bd2d-f1dd1ca6fdebo%40googlegroups.com.

Christopher Wong

unread,
Jun 17, 2020, 7:03:22 PM6/17/20
to mastodon-users
Maybe use BEAM2's with HEX20's, but specify your displacement variables separately. For the beam block, use the default first order approximation, as MOOSE's Timoshenko beam elements may only use linear Lagrange interpolations. For the solid block, keep your second order interpolations. This might work.

Mriganabh

unread,
Jun 17, 2020, 8:13:07 PM6/17/20
to mastodon-users
Hi Som, Chris & Chandu,

I got the simulation to run but I had to make a few modifications. I have attached a working version of the input file. The following steps should be followed to run simulations with HEX20 elements & BEAM elements:
1. Specify the element type to be BEAM3 while modeling. 
2. Specify the displacement variables separately for BEAM3 & HEX20 elements.


Som: Thank you for the heads-up regarding dbg version of MASTODON. I will get that compiled on my local system.

Chris: I followed your instructions and created separate variables for beam & solid elements. 

Chandu: I had to specify the element type to be BEAM3. Using BEAM2 with HEX20 resulted in the following error. 
--------------
*** ERROR ***
ERROR: Bad ElemType = EDGE2 for SECOND order approximation!
--------------
Like I mentioned earlier, the BEAM3 elements only introduced a midpoint node in the BEAM2 elements without any DOFs. Hence, BEAM2 & BEAM3 had same number of DOFs. I am guessing the new node works as a dummy node and is required to use HEX20 & BEAM elements together. 

Best,
Mriganabh
cantilever_solidbeam_2element.i

Christopher Wong

unread,
Jun 17, 2020, 8:19:54 PM6/17/20
to mastodon-users
Make sure you really understand what that mid-point node is doing. Verify your results carefully with ParaView or something. It looks like all you needed to do was use linear interpolations for your beam displacements. If you don’t need that third node on your beam, then don’t use it - it’d be a waste of memory and unnecessary complexity.

Mriganabh

unread,
Jun 17, 2020, 8:53:03 PM6/17/20
to mastodon-users
Hi Chris,

I do not need the third node on the beam. However, using only 2 nodes on the beam leads to the error I had mentioned earlier " Bad ElemType = EDGE2 for SECOND order approximation!". Hence, I am not sure how to get rid of that.

Thanks,
Mriganabh

Christopher Wong

unread,
Jun 17, 2020, 9:04:17 PM6/17/20
to mastodon-users
I understand this issue. But, you are no longer using a second order approximation, so you should try to use edge2 elements now. Although, your newmark auxvars still are, so maybe you need to do the same for those.

I suppose if there are no further issues, your results are adequate, and you’re ok with extra nodes, than you’re good to go. Personally, I’m a bit disturbed by them hehe. Eventually, you will have to explain your modeling decisions in your research publications.

Mriganabh

unread,
Jun 17, 2020, 9:08:54 PM6/17/20
to mastodon-users
Yes. Ideally, I would not want to use them either. Let me see if I can figure out a way to get the simulation to work using BEAM2. I appreciate the detailed discussion.

Thanks,
Mriganabh

Christopher Wong

unread,
Jun 17, 2020, 9:10:56 PM6/17/20
to mastodon-users
I appreciate them too, Mriganabh. Good luck!
Reply all
Reply to author
Forward
0 new messages