Include Load Factors in P-Delta analysis?

[shadva]

I am designing a slender column, and have a question about whether or not to include factored loads when doing a p-delta analysis (with Staad). I was told that it would be like factoring it twice, since the 'delta' would be from factored loads, and I should use unfactored loads to obtain 'delta', which makes sense, but I feel like the factors are there for the uncertainty in actual load conditions, and thus the factored 'delta' could very well occur. Does anyone have any opinion on this, or better yet, point to the location in a code for better direction?
Thanks in advance.

 

[phuduhudu]

The load factors are there to create a statistical certainty of compliance. Therefore if there is a 0.1% chance of the load exceeding that factored value and you take that as an acceptable limit then I agree that all the consequences of the load exceeding that limit ought to be taken into account including P-delta effects. Although normally you would not use factored loads to calculate deflections because you don't require as much statistical certainty of a deflection being exceeded, in this case your deflections become in themselves a load case.
Carl Bauer

 

[Riz]

The load factors need to be in place. The factoring twice issue is incorrect. PDelta analysis has been well explained in the Staad manual.
I was wondering as to why u need to carry out Pd rather than a normal perform analysis.
Maybe u let us know more about your slender columns and structure it has to support.

 

[shadva]

The structure is a concrete pile supported pier with some heavy vertical and lateral loads on it. The pile bents will act as an unbraced frame. Some of the piles are quite long and will pick up some extra moment from p-delta effects. I understand that we would use unfactored (service) loads to obtain regular deflections, but when those deflections are contributing to additional moment in the pile, I feel the loads should be factored. If they aren't, then you are saying the pile will never go to the ultimate load condition. The code requires you to design for some ultimate (factored) load condition, and if you aren't taking into account all the consequences of that load condition (e.g. amplified moment from p-delta effects) then it isn't a correct design. What do you think?

 

[Riz]

I agree with your feel of a ULS approach.

I would be inclined to carry out the analysis & design as follows:

1. Factor all the loads and carry out a normal perform analysis command.
2. Member sizing based on the results from above.
3. 2nd Order analysis with factored loads imposed on the columns. Invoke a PDelta analysis command.
4. Revisit your design with deflections/displacements obtained from 3 above to check for additional moments.

How r your columns connected?
Do u have any bracing in place?
How high r your columns?, size? Slender??