Seismosoft | Earthquake Engineering Solutions

Hi, I've been having difficulty on simulating my experimental columns with lap splice deficinecy at base, where starter bars and longitudinal flexural bars are lapped under lateral cyclic displacements.

In the forum, use of reduced modulus for bond-slip had been recommended, however, such that technique didn't work out for me accurately and can't get the post peak degrading response due to lap splice failure.

Any help or suggestion on modeling the lap splice degradation is greatly appreciated in cantilever columns under single curvature!

What value are you using for the strain-hardening parameter of the rebar steel material? Try to use something between 0 and 0.01. Further you should have geometric non-linearities activated from the program settings.

SeismoSoft Support

Yes, thanks. That totally makes sense to use such small strain-hardening parameter, where bond loss governs. However, I recall geometric non-linearity is chosen as the program default.

Hi,

Reducing the Young Modulus can indeed be a good way of modelling the slip part of the phenomena, but not the subsequent pullout (when it occurs).

In order to try to capture the pullout, you obviously need to reduce the yield strength, following the recommendations of e.g. FEMA-356 (lap splice provisions).

Alternatively, you may check out the proposals of e.g. Cho&Pincheira (2004), Melek&Wallace (2004), or the series of papers by Elwood&Moehle. (if you do a Google search on "modelling lap splices elwood cho melek" you will readily find the full references)

Rui

Dear fakharifar
it is the one of the most important case in accuracy of modeling.seismostruct doesn't have it.
i used bond effect in sharifi paper that indicate well relation which E changed to Eeq and the result gets much better.
http://dx.doi.org/10.1061/(ASCE)0733-94 ... 34:6(1061)
i modeled the frame by ABAQUS and seismostruct to check the Eeq(sharifi's paper)and it get well accuracy in my experimental models.
you can send me, your file to let you know.

Thank you really so much Dr. RuiPinho. The Elwood document really helped.

Thanks dear Salar for sharing your expertise on this topic with me. I will contact you to further refine the model upon updating it. Appreciate it.

Dear Fakharifar,
Would you please suggest me how did you model the lap splice.
Thanks in advance for your help.

Regards
Muntasir

Dear Muntasir,

FEMA 356 provides and equation as the maximum steel stress attainable in lap spliced bars.

Work done by Cho and Pincheira (2006) and its modified version in ASCE/SEI 41 also provides the steel stress.

The equations from above mentioned references made my results much closer.

However the below documents are more helpful for further elaborate calibration of the model:

(1) Ken. Elwood (it is a presentation including all the FEMA and Cho equations, link below):

http://tinyurl.com/oorhu3v

(2) Orakcal, K., & Chowdhury, S. R. Bond Slip Modeling of Reinforced Concrete Columns with Deficient Lap Splices.

(3) Melek, Murat, and John W. Wallace. "Cyclic behavior of columns with short lap splices." ACI Structural Journal 101.6 (2004).

Hope it hels,
Good Luck.

Also, Zero length section element is a method in which bond loss could be accounted, which is already brought up in the forum. Unfortunately, it is not included in Seismostruct. Opensees has this capability though.

Is there any way to assign a stress-DISPLACEMENT relation instead of stress-strain relation to a really small length element to capture the bond loss component of degrading response? The element's rotation, should be close to section curvature (has some error though since it is not zero length).

Is implementation of stress-DISPLACEMENT feasible simulating bond in SeismoStruct?

Thanks for help in advance.

Hi fakharifar.mostafa,

A zero-length section element (or a stress-displacement relationship) may indeed constitute an interesting feature for SeismoStruct to implement some time in the future, especially for those using the program for advanced research.

Personally, however, I still do believe that for the vast majority of situations where bond-slippage needs to be modelled (and these in themselves tend to be relatively few), the workarounds discussed above are quite appropriate (and relatively simple to implement, also by non-researchers).

Anyway, I am sure that Seismosoft will take note of your suggestion, and follow up on it should they ever find the time and resources to do it.

Best,

Rui