Innovative bamboo truss modeling problems

Hello!

I am a student trying to replicate a rather special bamboo truss (found in the paper Villegas et al. (2019) Combined culm-slat Guadua bamboo trusses), where the upper and lower chords are bamboo culm and the web is mad of bamboo slats connected to the culms on either side of the attachment points. To make things clearer (extracted from the above paper):

The authors of the paper did a numerical beam-model in ANSYS to see if they could replicate their theoretical results, and it is fairly correct (some differences due to the non-linear behavior of the semi-ring joints that they use, they remain of an order of 5-10% difference in maximum deflection).

My problem is that I am not able to obtain the same deflection values that the authors did (11.4 mm for a total service load of 7.063 kN applied punctually on the upper chord where the truss elements meet, or even replicate the load/deflection curve). Using an orthotropic material, with the engineering constants taken from (ResearchGate - A bamboo Beam-Column Connection Capable to Transmit ...), my model is too flexible and I get a maximum deflection of 24.28 mm.
I tried other orthotropic mechanical characterizations from other sources (Kathry & Mishra, 2012, Finite element analysis of bamboo and jo... and Chand , Shukla & Sharma, 2008, Analysis of Mechanical Behaviour...), to no avail.

Of course, the problem could be with the material properties I inputted but I am trying to contact the research team to see directly with them. In the meantime, I am looking to make sure the model itself is not flawed.

It also seems to me that gravity was not accounted for in the numerical of the paper, but it seemed to much of an oversight to be possible (still, the deflection curve of their paper goes through 0).

There are several points I am not quite sure about: after all I am still fairly new to Karamba3D and may still have some things to learn about the inner mechanics of the plugin.

The very first is: should I put eccentricities of the slat-elements of the truss in the definition of their cross-section (directly with the Cross Section box) or as an offset of the beam element (with the ModifyElem box)? I tried both approaches and they seem to yield similar results (max. deflection change by 0.65mm in my latest model).

Second is: is it good practice to subdivide the beam elements in more than one element (and connecting the pieces rigidly) in order to get better results? I imagine some meshing or subdivision is performed when the analysis is run but there is no way of visualizing it (that I found in any case). Subdividing the chord elements seems to give smoother deformation results (though I did not check stress I have to admit).
My issue on this topic is that the subdivision of the slat-elements of the web is problematic. On the screenshot below, where the elements are divided in two, lets take the example of node 18. It seems to me that all elements of the diagonal element (28, 29, 34 & 35) are all rigidly connected to the node 18. 28 & 29 are not connected together, independently from 34 & 35. The added rigidity may not be a bad thing for my model, but it is not correct I think? Is there a way of solving the problem?

Element tags: