Direct Analysis 2

[msquared48]

Josh:

Can I use a response spectrum load case in a DIRECT ANALYSIS in RISA 3D?

In a response spectrum analysis, you are essentially finding out the maximum responses at various nodes/members. So the various responses are not at the same time instant. One cannot carry out a second order analysis without knowing the actual real time values for the displacements at all nodes. Also due to the modal combination, the results lose their signs which makes it impossible to find the true nature of the force ( tension/compression) or the direction of displacement. In addition, in direct analysis, the structural stiffness changes between iterations depending on the forces in the members which makes the analysis type nonlinear in nature.

The steel frame system is for the support of a Coker furnace and other petrochemical equipment - non-building structure with an R=2.

Mike McCann, PE, SE (WA)

 

[JoshPlumSE]

The combination of response spectra analysis and Direct Analysis Method is currently a topic of debate at AISC and ASCE.

If you ask the folks from the seismic committee, they will tell you that you should ignore the Direct Analysis Method stiffness adjustments for the Response Spectra Analysis (which is what RISA does). The idea here is that you don't want the Direct Analysis Method to affect the loading applied to the structure.

Also, there is inherent non-linearity in the seismic design codes (with R, Cd, Omega, et cetera). So, using an increased 2nd order effect to mimic some non-linearity could be considered double dipping.


If you ask the folks in the stability committee, they are more likely to say that you should run the analysis twice. Once with the reduced stiffness included in the response spectra analysis and once with it excluded.


The question that every software program has to ask is, "Is it more important to use the correct response spectra resutls (i.e. natural period, mode shapes and such), or is it more important to amplify the PDelta effect". Since RISA does NOT include the P-Delta effect in its response spectra results, the answer was easy for us. But, as always it requires that the user look at the Theta factor from ASCE or such to see if they need to include some extra P-Delta effects for this load combination or what.

 

[JoshPlumSE]

A lot of people have told me that it's not possible to BOTH amplify the P-Delta effect for Response Spectra analysis and use the proper dynamic results (mode shapes, natural periods and such). I don't believe this is true.

The way everyone currently does response spectra analysis, it is true. We typically solve for spectral displacements for each mode and then back out everything else from there. However, you could generate a different set of "spectral forces" for each mode. It would then be possible to apply these spectral forces to your structure and solve it with a static analysis. Then you combine the results using SRSS or CQC or such. If you used the regular stiffness matrix to generate the spectral forces, then reduced the stiffness matrix per the DA method then, each 'modal force solution' would both start with correct spectral forces and amplify the 2nd order effects per the DA method.

Now, I don't believe anyone actually does it this way. I just know that it's possible. If someone where to do it this way, I'd want to say to them, "Your using your R factor to reduce your forces (and displacements) by a factor of 2.0 (for your case) up to 8.5. And, you're ignoring the amplification that those displacements would have on your structure's P-Delta effect. But, you're going to look at the 20% increase that you see from DA method and saying that is somehow more significant and must dominate the behavior of the structure. Why?"

 

[IDS]

Can I use a response spectrum load case in a DIRECT ANALYSIS in RISA 3D?

The answer seems to be that you can, but that doesn't mean it is a good idea.

Current seismic design methods currently use grossly simplified analysis methods combined with empirical adjustment factors, calibrated over many years. If a more detailed analysis method (but still grossly simplified) is used, that results in reduced design forces and ductility demand, it is likely that this will result in more building collapses under extreme events, unless the empirical factors are re-adjusted.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/