Seismosoft | Earthquake Engineering Solutions

Hello!

I am modelling a 4 storey before and after it was retrofitted with a shear wall as a way of increasing the building's stiffness.

I am running a pushover analysis for the +X/-X and +Y/-Y directions for both the building before and after the retrofit.

My first question is how to connect the shear wall with the beams? I am supposed to use rigid links, but which degrees of freedom should be restrained?

Then, I have a problem reducing the members' stiffness, as per the Greek Retrofit Code. The reduced stiffness is calculated as K = (My*Ls)/(3*èy). As I am using the model con_ma, I can only change the compressive strength of concrete (right?). I should reduce it from 16000 (C16/20) to about 1500 as per my calculations. The pre-retrofit model is running perfectly before reducing the compressive strength of concrete, but does not run with the reduced on.

Finally, in many cases an analysis that "worked", as in it runs without any errors, the next morning the same analysis stops at a very early step (~25 or 180 to 190) with this error message "Unable to apply the entire phase load. Phase terminated" In a few cases it run up to step 77 and then got the same message.

Any ideas will be greatly appreciated.

Thank you in advance
Giannis

Hi,

- the publication mentioned in http://www.seismosoft.com/forum/topic.asp?TOPIC_ID=92 (search this Forum for 'wall modelling') discusses extensively the issue of which DOFs to restrain when modelling concrete walls/cores. In any case, clearly it all depends also on your actual structural configuration - e.g. are your beams pinned or clamped to the walls? Sensitivity studies will also be helpful in making a decision on this modelling issue

- if you are using fibre modelling with a concrete material model of the 'con_ma' type, concrete cracking is already implicitly considered. In any case, if you prefer to manually reduce the stiffness of concrete members, whilst still using fibre elements, then you should select a different concrete model - e.g. con_tl

- from your description, your model seems to be very sensitive to numerical precision issues, which might be a symptom a modelling error somewhere (check eigenmode shapes and/or deformed shapes), or of exaggeratedly relaxed convergence criteria (do "play" with these).

Seismosoft Support