Discussion Groups

Simulation Mechanical and Multiphysics

Reply
Active Contributor
tibor121774
Posts: 42
Registered: ‎08-18-2013

Study on Reinforced Branch Connection on a Header Pipe

235 Views, 6 Replies
08-20-2013 08:54 PM

Hello to all!!!

 

I'm doing a study for a reinforced branch connection using (one FEA software) and to (another FEA Sofware).I've attached a pdf  file with filenames Sample 5 and Sample 6 for everybody's reference. The geometry that I had can be considered thin shell (with D/t ratio greater than 20). Please correct me if this is a correct assumption.

 

In Sample 5, I've made a two run in (another FEA sotware) using:

1) Solid Brick Element (one with mid node and the other without mid node) and

2) Plate and Shell Elements (using Linear Plate Element, Thin-Shell Element, and General Shell Element).

Please bear with me for I'm still new in the FEA. In this sample 5, the Solid Brick Element gives a good comparison result to (one FEA sotware) results. But the Plate and Shell element results are around 2 to 2.8 times greater than the (one FEA sotware) results. I was thinking that the Plate and Shell element will give good comparison result since this is a thin shell. But maybe there is something wrong with my input. I've highlighted the figures in yellow with discrepancies. I was also also thinking, that there's another formulation using a 3-d shell like a solid but considering additional 3 degrees rotation other than 3 degrees translation. And I was thinking I was comparing a two different formulation. Your help is greatly appreciated on this.

 

In Sample 6, I've made a run using Plate and Shell Element  in (another FEA software). This time I've made the header pipe and branch pipe very thin with a D/t ratio of greater than 200. In this example I've run 3 cases:

1) Using Linear Plate Element

2) Thin Shell Element, and 

3) General Shell Element .

At this example 6, I'm getting a good comparison result for the bending moments (inplane, outplane, and torsion) and axial force except for thin shell element which gives low result for outplane and axial as highlighted in yellow.

The major discrepancy that I've got for this example 6 is on the pressure stress. For thin shell element run it gives a good comparison result. But the Linear Plate and General Shell Element are almost 3.9 greater as highlighted in yellow. 

 

Actually, I've also tried the Solid Brick Element for this example 6, but I'm getting a warning and error. But I haven’t tried to resolve it since I was thinking this is a very thin element, and a solid brick with 3 translations ( and no rotational dof's) will really not work for this example.

 

As a disclaimer note: "I'm really new to FEA so the results presented here maybe wrong. Maybe there are some error in my input, and assumptions." Please correct some of the theoretical statements that are not right in the above statements.

 

To summarize my questions:

1)     I would really appreciate from the experts if someone can verify the study that I've made above (Sample 5         and 6).

2)     Please give me pointers on how to properly evaluate the two given system. Please inform me if                additional information is required.

3)     Is there really a 3-d shell element like solid element but with 3 additional rotational dof’s (other than 3       translational dof’s) in the FEA world? I mean, other than the 2D plane element   formulated to a 3-d shell element with mid nodes?

4)     If the answer in number 3 question is true, is this feature available in Autodesk Simulation?

5)    And lastly, I would appreciate if someone can give me their own file for this study. I was thinking my                   modelling for applying pressure ,bending and torsion moments and axial force in sample 5 and 6 are      not correct.

 

As a final note, the loadings that I’ve run considers acting one  at a time for each type of load (inplane moment, outplane moment, torsion, axial force and pressure). It’s only in sample 5 that I’ve made a combined run.

 

Thanks all for your time. Greatly appreciate your help on this study.

 

Cheers to all!!!!!!!!

Please use plain text.
Active Contributor
tibor121774
Posts: 42
Registered: ‎08-18-2013

Re: Study on Reinforced Branch Connection on a Header Pipe

08-21-2013 02:25 AM in reply to: tibor121774

Please note that the header has a hole in this example. Not the trunnion support type.

 

Thanks all!!!!

Please use plain text.
Distinguished Contributor
ilyas
Posts: 139
Registered: ‎06-13-2012

Re: Study on Reinforced Branch Connection on a Header Pipe

08-21-2013 09:47 PM in reply to: tibor121774

Hi,

 

I may not have the answers to all your problems, but here's my take on the DoF questions you're facing.

 

As you already know, brick element only support 3 translational DoF. But we learn in Engineering school that there are 3 forces and 3 moment acting on a structure, so why not the brick element support rotational DoF as well? The answer will require you to understand that moments are what I would describe as idealizations used to represent a combination of eccentric loads. The idealization is required because the model itself is geometrically simple, such is the case of plate/shell elements and beam elements. But in the real world, there are only forces acting from element to element, from an atom to an atom. It is either push or pull. There are no moments. Moments would be the result of a combination of some of these forces. This is the basis for brick elements. Because it takes into account the full 3D geometry of the object, it only needs to transfer forces to approximate the real world. By having a large enough number of elements, the brick elements are able to capture the "moment" as felt by plate or beam elements. So the answer to the question is, Brick elements do not need the rotational DoF.

 

That is all I have for you. If anyone finds anything wrong with how I understand this, let me know.

 

Regards

Ilyas

Please use plain text.
Active Contributor
tibor121774
Posts: 42
Registered: ‎08-18-2013

Re: Study on Reinforced Branch Connection on a Header Pipe

08-22-2013 06:55 AM in reply to: tibor121774

Thanks Ilyas for your reply...Your explanation of the theory make sense to me...Actually on Sample 6, I have re-tried to use solid brick element. What I've done is to refine the mesh and adjust aspect ratio to 5 (I've just learned aspect ratio yesterday from searching the net)....The error is gone and the moments and axial forces now matches between two software within reasonable limit...I'm struggling now is on the internal pressure which does not match (up to 3 times difference) ....I've tried to make the mesh finer which takes some time but still gives same result....Maybe, I need more reading in FEA to unlock its mysteries.  Your views on this topic are greatly appreciated....

 

 

Cheers to all!!!!

Please use plain text.
Distinguished Contributor
ilyas
Posts: 139
Registered: ‎06-13-2012

Re: Study on Reinforced Branch Connection on a Header Pipe

08-22-2013 09:34 AM in reply to: tibor121774

Well, if you've reached the point of mesh convergence, then there is little point in refining it further. I would suggest you examine the results in more detail. Where is the location of maximum stress? Does the stress contours share any similar patterns? 1000MPa is huge, though it is not impossible to get such a result with Linear Static Stress. Is the applied load excessive or well within the elastic limit of the material? You could also check the applied load whether they tally up the same for brick and plate elements. If you know some hand calculations, now would be a good time to get your hands dirty again. Find out which of the results are bogus.

 

Regards

Ilyas

Please use plain text.
Active Contributor
tibor121774
Posts: 42
Registered: ‎08-18-2013

Re: Study on Reinforced Branch Connection on a Header Pipe

08-22-2013 06:10 PM in reply to: tibor121774

Thanks Ilyas...I've check in the plate and shell element (which has a feature to separate membrane and bending components on the stress results) and I found out that it's the bending component which has a high stress...The membrane component for re-pad is around 219 MPA and 156 MPa for branch (compared to one FEA software of 147 Mpa)...The bending component for re-pad gives 1390Mpa, for branch 119Mpa (compare to one FEA software of 82 MPa.) I'll be concentrating my study why re-pad is getting this high bending component for an internal pressure of 0.25MPa....Maybe I need to check which stress tensors has the highest stress component and investigate from there...Anybody's advice is greatly appreciated....

 

 

Cheers to all!!!!!!!!!!!!!!

Please use plain text.
Valued Mentor
AstroJohnPE
Posts: 494
Registered: ‎08-30-2012

Re: Study on Reinforced Branch Connection on a Header Pipe

08-23-2013 10:05 AM in reply to: tibor121774

My suggestion is "Create, Post, or Provide an Archive of your model" and maybe someone can look at the model and find the problem. It is hard, at least for me, to understand the discussion when all we have to work with are a few pages of numbers.

John Holtz, PE
Mechanical Engineer
Pittsburgh, PA

16 years experience with Simulation Mechanical
Please use plain text.