P-Delta Preset Analysis in ETABS

[Checkmann]

Hi everyone,

I have been doing a lot of research but cannot find a solid answer on how to account for P-Delta effects properly in ETABS (big P-Delta, let’s ignore little P-delta in this post). I know there is also a method to add a load case specifically for P-Delta, but I want to look at the Preset P-Delta Options in ETABS, as seen in the attachment.

There are 3 methods:

1) None
2) Non-iterative – Base on Mass
3) Iterative – Based on Loads

I would like to know when you use each of these methods. Method 1 is to simply ignore P-Delta effects in the model. As far as I understand if you do this you have to check Equation 12.8-16 of ASCE 7. If your Theta is under 0.10 then you can totally ignore P-Delta. But if it is greater than 0.10 then you have to increase your member forces and deflections by a factor of 1/(1-Theta). This is always an option.

But if you know you already will have to consider P-delta - say for a flexible moment frame building - it seems easier to let the software do the calcs and adjust the forces automatically. So that gets me to which method do I use, iterative or non-iterative? I have seen posts suggesting both ways, and the CSI wiki is really confusing.

https://wiki.csiamerica.com/display/etabs/P-Delta+analysis+parameters

It says that “when gravity load is specified, we recommend using Iterative.” But what does that mean? When is gravity load not specified? When is it appropriate to use non-iterative based on mass procedure?

So then if we assume I should use Iterative based on loads approach what loads do I add and which factors do I apply? The CSI Wiki link above seems to suggest that I add Dead and Live loads with the factors they are assigned using the lateral load analysis combinations. For example, if I have 1.32D + L + EQ then I would add 1.32 to the DL and 1.0 to the LL for P-Delta. But I have read other publications that suggest using 1.2 for DL and 1.6 for LL. Does anyone have any clarity on this?

Thanks for your help!

 

[Trenno]

From what I understand, you would use iterative P-Delta if you have a susceptible structure or doing the detailed design. Iterative P-Delta will increase model run times.

Non-iterative is an approximation. It places very soft springs at each joint in the model - you can see it when you turn on reactions. I queried this last year with CSI and got the following answer:

These are reactions at internally assigned springs (with negligible stiffness) used for applying Pdelta effect based on mass. See attached for Non-iterative Pdelta based on mass which is an approximate procedure that can be used for lateral only models. Iterative Pdelta is always recommended especially for complete models (gravity+lateral).

 

[JoshPlumSE]

My thoughts:

1) Ignore P-Delta: You might use this early on in design. As you're working on setting member sizes. The reason for this, is that if you don't have your drift under control, you might get your analysis to diverge when using option 3.
2) Non-Iterative based on Mass: Honestly, I don't see a reason to use this very often. The help file suggests cases where you have rigid diaphragms, but you don't have loads defined for each level, but where masses are defined. I can't picture many structural models where this would be true.
3) Iterative based on loads: Always use this method. Unless you are in the early stages of design where you don't have all your member sizes defined yet or where you haven't properly defined your loading yet.

 

[Checkmann]

Just to follow up on this. I have had some talks with the engineers at CSI over the last few weeks and they recommend using Iterative Based on Loads and applying 1.2D + 1.0L in the load cases. Or if your live load factor is not 1.0 for your seismic or wind case then apply whatever the load factor is in the lateral load case.

 

[IDS]

I don't use ETABS, but I would be interested to know how they do the non-iterative analysis, if there is an open document available.

But with computer speeds these days I agree that the iterative analysis is the way to go.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[johnie134]

first check the second order to firsts order moment ratio. if it is >10% then go for an iterative non-linear analysis for pi-delta consideration.