Base isolated bridge

[mmoradian]

Hello everyone.
I’m about to model a base isolated bridge. I use the “bl_kin” link element for representing the behavior of rubber isolator.¸
I have two questions about base isolator modelling in Seismostruct.
1- Is the software considering the bidirectional behavior of the isolator? If yes it is based on which theory or reference?
2- The inserts parameters for link element are K0, fy and r. Which thory is used in software for calculate Keff of the isolator for modal analysis,

I’m so thankful of your help.

[huffte]

Hi mmoradian. For your questions:

1. The degrees-of-freedom in the link elements are uncoupled. See User's Manual, Page 301.
2. The initial stiffness is used when an eigenvalue analysis is performed, which may be verified by a simple test model.

The link element you have selected works rather well, I have found, for both Lead-Rubber-Bearing (LRB) and possibly Friction-Pendulum-Systems (FPS). Best of luck.

[mmoradian]

Thank you so much dear Tim for your response,

I want to knwo how seismostruct calculate Keff ? while displacement (D) of the link element is not defined by user.

And also which method do you propose for coupling the shear deformation in two direction (longitudinal and transversal) of the bridge ?

Thank you again.

[huffte]

I do not believe that SeismoStruct calculates a 'Keff'. This is what is meant by uncoupled - the displacement in the x-direction has no effect on the stiffness in the y-direction. I believe that the stiffness in each direction depends solely upon the displacement in that degree of freedom. So, you see, this is an approximation. However, if you work out a little algebra you'll find that the resultant stiffness is the same as the stiffness specified in each direction. The approximation comes in because when the isolator is yielded in the x-direction, it can actually withstand less deflection in the y-direction to yield, but the uncoupled treatment does not account for this.

For example, suppose the displacement in the x-direction is 0.90Dy and the displacement in the y-direction is 0.9Dy as well. Then, since the element is uncoupled, nothing has yielded. However, in reality, the resultant displacement is SQRT[(0.9Dy)^2=(0.9Dy)^2]=1.27Dy. So yielding has actually occurred.

I have no clear cut answer for you, unfortunately, on how to best treat the effect. I would say as an initial advise to track the displacement sin each direction and compare them by post-processing the results in a spreadsheet. Check the ratio of x-deformation to y-deformation throughout your loading history. If the loading is truly along an approximate diagonal with demands about the same in each direction, then, for example, one could infer that the yield force used for input into the link element in each direction should be 0.707 (cos45) times the real yield force of the physical isolator.

I hope this helps a little, mmoradian. I'll mull it over some more tonight over dinner.

Best of luck with your interesting work.

[mmoradian]

Thank you again dear Tim.

Your response for my first question was perfect.
In fact when I compare between Sap2000 and SStruct, it makes me confused, because in Sap2000 Keff is used for modal analysis and K0 and Fy are used for nonlinear analysis.
But in SStruct we have just K0 for both type of analysis (modal and nonlinear).Beside, I can't find in user manual (initial stiffness is used when an eigenvalue analysis is performed).

[huffte]

I seem to have read it somewhere in the Manual or Help System that the initial stiffness is used for eigenvalue. But, like you, I could not find it this time so I ran a simple model to verity it. And logically, it does make sense that K0 would be used since the only other choice, unlike SAP2000 where Keff is explicitly entered by the user, in SeismoStruct would be the post-yield value.

SAP2000 does offer the advantage of incorporating coupled link type elements if desired, yet I still believe there is a great deal of value in the SeismoSoft link elements.