ASCE 7-16 Seismic Design for Skid With Nonstructural Components

[m_ridzon]

I'm on a team of mechanical engineers and we are a little bit out of our wheelhouse. We have taken training for ASCE 7-16 seismic design, so we know the basics. We have been tasked with designing a process piping skid to ASCE 7-16 seismic criteria. We are designing the skid, as well as the vessels and piping that are on the skid. We are brainstorming a reasonable approach to take...

We contemplated first analyzing the skid using Modal Response Spectrum analysis and the design response spectrum from ASCE 7. I believe this would be a nonbuilding structure. All of the major components on the skid would be represented with masses. In post-processing, we would extract accelerations for all of the major components and run detailed submodels of each component using the Equivalent Lateral Force method with said accelerations.

I'm looking for feedback to see if this approach seems reasonable and am open to other ideas or considerations we overlooked.

 

[JoshPlumSE]

Maybe I have something different in my mind. But, I think of equipment skids as basically a steel platform that is directly connected to the concrete foundation. The main purpose is to provide something that you can bolt your equipment onto.

If that's what you're dealing with, then it seems like major overkill to use the Response Spectrum method to design it. I would tend to use the "non-structural components on rigid supports" type of anchorage demand requirements. Like ASCE sections 13.3 and 13.4..... probably with Rp and ap values per Table 13.6-1.

 

[m_ridzon]

Yes, your understanding of the purpose is correct. I found the following image in the back of ASCE 7-16 that is a good example of my situation.


Your idea sounds reasonable. But my understanding is that 13.3 and 13.4 are for the mounted components. What about the skid itself? I think it would classify as a "nonbuilding structure not similar to buildings" and be governed by section 15, correct?

 

[HTURKAK]

our idea sounds reasonable. But my understanding is that 13.3 and 13.4 are for the mounted components. What about the skid itself? I think it would classify as a "nonbuilding structure not similar to buildings" and be governed by section 15, correct?

Your description implies Nonstructural Components rather than non-building structure.

Pls look to the worked examples of the following doc. sections 17 and 18 Nonbuilding Structure Design , and Nonstructural Components.

If you need more feedback, pls provide more info.

 

[m_ridzon]

Your description implies Nonstructural Components rather than non-building structure.

Pls look to the worked examples of the following doc. sections 17 and 18 Nonbuilding Structure Design , and Nonstructural Components.

If you need more feedback, pls provide more info.

Thank you! Your attachment was very helpful!

 

[m_ridzon]

When designing nonstructural components, is there ever a case where a response spectrum can be applied directly to the nonstructural components in modeling software? Let me elaborate...

From my understanding of Chapter 13 of ASCE 7-16, to do an RSA approach, you have to run the RSA on the building. Then with the acceleration of each story level of the building, you use Eqn 13.3-4 to determine a static force for the nonstructural component. If desired, the engineer can utilize the floor response spectra per §13.3.1.4.1, but still ends up calculating an applied force for the nonstructural component? I don't think directly running an RSA on a nonstructural component is permitted ASCE 7-16. Can someone clarify?

 

[driftLimiter]

No if the structure is mounted in the building, then the response spectrum at its base is different than the building's response spectrum at its base.

In order do a response spectrum analysis of the non structural component you need to generate the so called 'in-structure response spectrum'

 

[m_ridzon]

Okay, that makes sense so far. In the case you generate the "in-structure response spectrum," would you use Eqn 13.3-4 to generate the seismic force for the component? Or can you directly apply the spectrum to the nonstructural component?

 

[TLHS]

There's basically two ways to approach this.

(1) Simplify as much as possible and use a large base shear factor for everything. Do very high level static calculations, get yourself a seismic factor as a percent of gravity (or a couple for different categories of equipment) and run with it. Make sure you have a qualitative understanding of the stiffnesses of the various items and overbuild things that are stiffer and will absorb load.

(2) Model everything as a combined model and do a static or modal analysis for everything together

Honestly, even in item 2, for something of this scale a modal analysis is likely not all that helpful. The periods for these things are generally super low and rigid and will basically always end up at peak spectral.

The other issue is transportation loads. In a lot of cases transportation load will significantly govern unless there are specific transportation bracing considerations being used, or in the case of the shipping weight being significantly less than the operating weight (tankage that will get filled at site or items that have significant site installed components).

Depending on what stuff specifically you're responsible for, you may make slightly different conservative assumptions for different parts. You don't necessarily have to use the same loads to design the attachment of legs to a tank as you use for the legs being attached to the skid. The book-keeping becomes silly and overly complicated if you let it and it means future revisions involve touching everything. Try to keep the paperwork for each component separate and not feeding into the others as much as you reasonably can and don't try to optimize aggressively on the steel, these things get kicked around in ways you aren't expecting and the warranty claims aren't worth being cute.

Also, don't fall into the trap of thinking that fancier analysis is going to be more accurate. The amount of interactions between components in these things are so significant that the dynamic responses you're calculating are pretty far off no matter what you do.

 

[m_ridzon]

Okay, thank you for that feedback again. The main debate I'm having with a colleague is whether ASCE 7-16, Chapter 13 allows an engineer to analyze a nonstructural component by directly applying a response spectrum to said component in the modeling software. He believe it's permissible. However, my understanding of Chapter13 is that it's not permitted.

 

[driftLimiter]

Have you reviewed section 13.3.1.4 in detail?

This is where it allows the use of dynamic analysis and details the requirements associated. You cannot do a dynamic analysis of the component without first knowing the floor accelerations of the building.

 

[m_ridzon]

Yes, I have reviewed 13.3.1.4. It says dynamic analyses are allowed for the building and/or the floor, but you then take the acceleration that was determined from that RSA model and plug it into Eqn 13.3-4 to determine the seismic force for the nonstructural component that resides on that floor. So no matter what, it seems like I have to model the nonstructural component with an applied seismic force. That insinuates I am not permitted to apply the response spectrum directly to the nonstructural component.

 

[TLHS]

None of the ASCE-7 sections fully cover this type of design. It's sort of equipment supported by a structure, but it's also sort of one structure with subcomponents, and it's also sort of components on a foundation system. You can draw the box in a variety of valid ways. To some degree you're going to have to use first principles and judgement on how to apply loading. I have seen many many calculation packages on this sort of thing and people do it all sorts of different ways.

You can do a response spectrum analysis if you want if you treat the components more like structures not similar to buildings. This is a valid way to look at it. It also seems overly complex for the scale of equipment you seem to be talking about though. At the end of the day, structures not similar to buildings stuff gives you similar numbers to the non-structural components stuff except in some of the edge cases surrounding rigid structures or in a couple of cases where the R values don't really align.

 

[TLHS]

Like I had said above, though, you can also design the tank using software with a response spectrum analysis and then do holddown and skid design using simplified numbers from a non-structural component style static analysis.

 

[m_ridzon]

Okay thanks for this additional feedback. I can tell that it's a grey area that can be perceived different ways. But I think I have an idea of what's involved now.

Thanks!