Seismosoft | Earthquake Engineering Solutions

Dear seismosoft,
I’m trying to understand how does the software work, so I started to use a very simple model (a single element which represent a cantilever wall) and compare the moment-curvature capacity that I obtain from pushover analysis with my hand calculation. I used a reinforced concrete rectangular wall section, I switched off the “automatically transform masses into gravity load” option, and I didn’t put any axial force on the element.

First I modeled the wall as “infrmDB”. I don’t understand why the moment capacity at the base off the wall that I obtain is different when I take as applied loading an incremental force instead an incremental moment at the top. When I apply an incremental load the top I obtain the same result that I got from my hand calculation, when I apply a force, the moment values are two time higher. I think that if I use the same section, the capacity of the wall should be the same.

Since modeling the wall as single element with an incremental moment at the top gave me good results, I tried also to compare different element classes and what I obtain is very strange: when I use “infrmDB” or “infrmFB” or “infrmFBPH” element whit a plastic hinge length which is 50%, I obtain the same results, but when I change the length of the plastic hinge I didn’t reach the same values of the moment capacity (they are much more smaller) and when I try to use “infrmDBPH” elements I obtain completely different results in term of moment-curvature relationship even with the plastic hinge length as 50%. With “infrmDBPH” elements the results are also very irregular, even if I change convergence criterion.

I tried to read the reference paper related to the lumped plasticity approach (Scott M.H., Fenves G.L. [2006] "Plastic hinge integration methods for force-based beam–column elements," ASCE Journal of Structural Engineering, Vol. 132, No. 2, pp. 244-252.) but I didn’t understand why I don’t get the same results.

Do you think that I got something wrong conceptually?
Do you think that a possible alternative to represent lumped plasticity approach is to use an elastic element combined with a link element at the base?
I’m sorry for the long message, I just would like to be more clear as possible.
Thank you very much if you can help me.
Sabrina

Dear Sabrina,

You raise many issues in your post, which are difficult to address in a complete but succinct manner, adequate for this Forum.

We can reassure you that there are indeed rational explanations for the results your are observing: e.g. if one uses a single 'infrmDB' subjected to a transverse load, one will not get a proper moment distribution throughout the element's length, since, unlike with 'infrmFB' element, internal force "exactness" is not imposed.

As mentioned above, it is unfeasible to dwell on these theoretical issues herein, however, we suspect that if you read the following publication, you may find the answer to many, if not all your queries and doubts:

Calabrese, A., Almeida, J.P., Pinho, R. 2010. Numerical Issues in Distributed Inelasticity Modeling of RC Frame Elements for Seismic Analysis. Journal of Earthquake Engineering 14(S1), 38-68.

Seismosoft Support

Dear Seismosoft,
I read the paper and I think I understand more or less what is the point.
Could you tell me which integration method is implemented in seismostruct for "infrmDB" elements? Is it the Gauss-Lobatto integration method or the Gauss-Radau one?
Thank you,
Sabrina.

It varies. Please do read the Help System. Seismosoft Support

Dear Sabrina,

Let me try to give you some additional hints.

Elements “infrmFBPH” can be used if you observe a softening response of your cantilever wall (that is, a post-peak range of behavior that is of interest), as it will allow you to define the length over which the inelasticity will concentrate (after the peak). If the response of your member is of the hardening-type, then I suggest that you use “infrmFB” elements (one per member is enough if the section is the same throughout the length) with a significant number of integration sections (e.g., 6 or 7). If the wall is tall, then you can even consider one shorter element near the cantilever base (where nonlinearity will be larger) and another for the remaining part of the structural member.

Finally, “infrmDB” elements use Gauss-Legendre integration (and not Gauss-Lobatto or Gauss-Radau, as you query).

Rgds