• dimtz
  • New member
  • Offline
  • Registered: 2013-06-14
  • Posts: 2

Topic: Erroneous results modeling shear transfer

Hi,
In my master's thesis I am investigating the shear transfer mechanisms in concrete interfaces, i.e. the roughness/aggregate interlock and the dowel action of the reinforcement bars, using typical push-off specimens (two opposite-superimposed L-shaped concrete specimens, with discrete reinforcement bars crossing the interface between them). As a first step I modeled a couple of experiments (Hofbeck(1969); Walraven(1981)) of monolithically cast specimens.
Following as much as possible all the specifications of the experiments published in the aforementioned papers, I achieved to get very close results regarding the ultimate load. However, I keep getting different response in terms of ultimate slips/displacements (results appear to be 1.5 to 2 times more "elastic" than reported) and the resulting failure "snapshot" differs from that described in the papers, as the reported diagonal cracks (<20~45°) do not show up as expected. Any suggestions?
Moreover, as I intend to model the dowel action of the reinforcement, is there any other way to model that in the vicinity of the shear plane than that described in §2.1.16/Troubleshooting manual, so as not make the models extremely complex?
And, finally, is there an actual difference between modeling monolithically cast specimen using a single macroelement (meshed appropriately with tetra- and hexahedral FE, in turn), and multiple 6-edge macroelements (meshed with brick FE)? I do notice some differences in the cracking patterns near the edges of the neighboring macroelements.
Many thanks and regards,
Dimitris

Quote Post 1

  • dpryl
  • CC software developer
  • Offline
  • Registered: 2007-07-13
  • Posts: 1,107

Re: Erroneous results modeling shear transfer

Hello Dimitris,
not all options mentioned in 2.1.16 make the model complex. I would ssuggest to first try the smeared reinforcement approach for the shear contribution of bars.

If you see something strange near the macroelement borders (surfaces connecting neighbouring MEs), it usually means something is wrong with your mesh. The safest way usually is to have mesh compatibility at the internal interfaces (apply this option to the contacts through the FE Mesh menu). When Master-Slave connection is used between 2 MEs, make sure the side with the larger finite elements is the Master (and the side with the finer mesh the Slave).

Finally, if you like to get advice with your model (the recommendations in 2.1.20 My analysis results do not match the experiment/expectations. How can I improve my model? did not solve the differences), please send us your model along with a description+sketches of what you are modelling, the graph comparing the measured and calculated Load-Displacement curve, and your ATENA User ID (WR).

dpryl's Website

Quote Post 2

  • dimtz
  • New member
  • Offline
  • Registered: 2013-06-14
  • Posts: 2

Re: Erroneous results modeling shear transfer

Thank you for your prompt reply.
First of all, as you will discover in the models I've already sent to you, I think that its scale and nature implies using only discrete reinforcement (adopting smeared approach for such big variations of reinforcement in the same element seems questionable). I will try to model the reinforcement using shell elements in the neighborhood of the shear plane, as advised in §2.1.16.2/Troubleshooting manual. If I understand it correctly, I should also create an opening in the surrounding concrete ME, isn't it?
Secondly, the neighboring MEs' contacts I was talking about are already set up to be compatible. What concerns me is that albeit in the single-ME model the cracks appear to cross the shear plane, in the multi-ME one they just appear aligned parallel to the shear plane (which is also a contact).

And a maybe-simple question: Should I be concerned, if running the example manual's NGAP(3D) model I get a totally different-looking l-d curve and if running the tutorial manual's Leonhard's beam I get ultimate load 5% lower than the reported?

Regards,
Dimitris

Quote Post 3

  • dpryl
  • CC software developer
  • Offline
  • Registered: 2007-07-13
  • Posts: 1,107

Re: Erroneous results modeling shear transfer

You have an email about the shear model.

With respect to the differences in the examples, thanks for the report, we will check them. AFAIK, both the models and the example manual are being updated to reflect small changes in the program and material models.

The 5% in the Shear Beam might be due to some meshing changes and/or the Aggregate Interlock. In the Interface material model, there were some fixes and changes since the previous update of the Example Manual, but "totally different" is not what I would expect...

dpryl's Website

Quote Post 4