NRC Structural code 3

[WARose]

I have a client who want some pipe supports (and some other small stuff) done not only by a pass through IBC/ASCE but through the nuke code as well. (I think the reason(s) are: he spent years in the nuke business (before he went to the Chem biz; so he is falling back on what he knows) and (possibly also) they want pretty much zero chance of any of this failing.)

I haven't looked at that stuff in years. Can someone remind me what NRC doc governs structural design and what its name is?

Thanks in advance.

 

[Insert_Smurf_Here]

The code of record depends on the class of safety/seismic category of the structure you want to design.

Typically, the critical structural elements in containment are governed by AISC N690 (steel design) & ACI 349 (Reinforced Concrete design). Piping design codes typically falls under ASME B31.1 "Power Piping Code".

There should be a Design Control Document (DCD) that is typically found on the NRC's website that outlines the general design criteria for nuke plants in USA.

 

[WARose]

There should be a Design Control Document (DCD) that is typically found on the NRC's website that outlines the general design criteria for nuke plants in USA.

Thanks. Are you referring to this:

https://www.nrc.gov/reactors/new-reactors/design-cert/apwr/dcd.html#dcd

Seems like something a private company generated. Do they have any go-by standard to generate this?

 

[Insert_Smurf_Here]

I am not sure if the NRC provides a generic template. The steps for licensing of a nuclear power plant are outlined per 10 CFR Part 50:

https://www.nrc.gov/reading-rm/doc-collections/cfr/part050/full-text.html

You can also google a NRC approved design as an example ("AP1000 DCD").

Can the client provide you with a design criteria document?

 

[WARose]

Can the client provide you with a design criteria document?

I'm not sure....but thank you for all this info.

 

[bones206]

The basis of design varies from project to project with nukes. I would just go with the latest versions of the codes for each material like ACI 349 and AISC N690.

I actually worked on AP1000 design criteria docs back in the day and trying to envelope the Nuke codes and IBC was no small task. Gets real fun when you have to start enveloping the China codes too!

I still have most of the nuke codes if you need any specific info WARose.

 

[WARose]

thanks Bones

 

[WARose]

Bones/smurf.....a couple follow up questions here:

1. Apparently seismically qualifying electrical & mechanical equipment in the nuclear world is a big deal. I think I have an idea as to how the process works....but one thing I am not sure about is: what is the structural engineer's role in it (if any)? Maybe give them criteria?

2. How do they figure seismic loads (typically) in the nuclear world? A lot of what I am reading in material codes (like ACI 349) they are similar to their counter parts in regular structural design. Is that how it works with seismic? They fall back on ASCE 7-10....but with modifications? (I ask because it's not readily apparent to me in some of the Design Control Documents I've looked at.)

3. With the first 2 questions in mind.....how do they get the anchorage forces for equipment skids? Through the qualifying process? Or something similar to what ASCE 7-10 calls for (for mechanical/electrical equipment)?

Thanks.

 

[le99]

The design approaches fall in line with AISC and ACI standards, but depending on the level of safety, the load factors and load combinations can vary very much from those standards.

 

[CLT49er]

That US APWR DCD brings back some memories. I haven't worked in the nuclear industry in years and its been over a decade since I worked on the US AWPR but if I remember correctly all of the seismic loads were determined using a site specific dynamic analysis including soil structure interaction (the SSI was done on the newer plants but I am not sure about the older ones). The older plants still had a dynamic analysis performed but the models were much simpler lumped mass and stick models. The output of that dynamic analysis would have been used to generate response spectra at various elevations and other points of interest in the plant. Those response spectra would then be used for the design of piping and electrical supports and for the seismic qualification of equipment (most likely on a shake table for new equipment). I wasn't involved in the generation of the seismic loads (there were a couple of PhDs doing that) but I remember using the response spectra for designing items within the plant. I think the analysis to generate the response spectra for the US APWR took a couple of years. You might be able to find some of the response spectra for various existing plants if you dig hard enough on the NRC website but they will most likely be useless as they are site (and even building) specific. ASCE 7 was not used for seismic loads on the nuclear island but may have been used for the turbine building and other non nuclear structures at the plant. ASCE 4 and ASCE 43 would have been used on the US APWR but not on older plants. Some of the methodology in ASCE 7 was used for calculating wind pressures but the wind speeds used were much higher than those shown on the maps in ASCE 7. I am pretty sure I remember seeing 300 psf wind pressures but those may have been for tornado loads.

Chapter 3 of that US APWR DCD outlines the criteria used for developing the seismic loads and there are some figures showing the models, time histories, and response spectra inputs around page 3.7-83 (buildings), 3.9-256 (reactor internals). The DCD is still a high level document and you would need to dig through various NUREG and other regulatory documents for other criteria as well as look at ASCE 4 and ASCE 43. There was also a lot of proprietary documentation used as back up for the DCD that is most likely not available on the NRC website. I didn't see any of the response spectra that would have been used for the equipment qualification in the DCD. Typically those response spectra would have been too detailed for the DCD and would have been in another document that I cannot remember the name of and that is most likely not publicly available. It was thousands of pages long though.

Using AISC N690 or ACI 349 wouldn't be that difficult. Determining seismic loads in accordance with the nuclear regulations is serious overkill. Its going to require a very advanced understanding of structural dynamics. The equivalent lateral force method is not permitted for nuclear plants. The company I worked for had PhDs working on it. Granted your pipe supports are simpler structures than an entire nuclear plant so maybe it could be done by a single person in a shorter amount of time but you will end up with seismic loads several orders of magnitude above what you would get with ASCE 7. If I remember correctly the peak vertical accelerations on the polar crane for the US APWR were in the 20 to 30g range. I remember designing equipment supports for some other existing plants that had seismic loads in the 8 to 12g range pretty regularly. Some of the supports holding up electrical conduit would not have looked out of place as columns in a 6 story building.

We used to joke that if the design basis earthquake ever happened on the US APWR they would have to scrape the operators off the ceiling of the control room. Good thing the reactor would trip on its own after an earthquake.

 

[CLT49er]

Instructure Response Spectra is what you would be looking for if you were going to qualify equipment or design pipe supports inside a nuclear plant for seismic loads. These are building and floor specific at each nuclear plant in the USA. Like I said in my previous post generating these for a non nuclear pipe support structure would be overkill.

 

[WARose]

The equivalent lateral force method is not permitted for nuclear plants.

I think something similar to it must be (according to one of the DCDs linked above):

3.9.2.2.2 Seismic System Analysis Methods

The seismic system and subsystem analysis methods (including response spectrum
analysis, time history analysis, and equivalent static load analysis) are discussed in
Subsections 3.7.2 and 3.7.3.

Maybe that's only for the lesser category of seismic loads. (A lot seems to depend on that.)

 

[CLT49er]

Equivalent static still uses the instructure response spectra. Basically you do a response spectrum or time history analysis of the entire building to generate instructure response spectra. Those instructure response spectra become the input for designing the substructure (piping or other equipment) in the plant. You could then run a time history analysis or response spectrum analysis of your substructure (pipe support for example) or you could design it using equivalent static loads calculated from the instructure response spectra. Its been awhile since I have done it so I can't remember if it depends on the subsystem for which method is permitted (the DCD may not state this but other documents may). In practice I think most piping systems and electrical systems other than maybe the reactor and reactor coolant piping would then be designed using equivalent static loads calculated from the instructure response spectra. I always used equivalent static loads for designing substructures but I never designed piping supports for anything on the primary coolant loop. Usually there were two ways of calculating the equivalent static loads. You either take 1.5 times the peak value from the instructure response spectra (this can be a very high acceleration - see post above) or you can tune the structure past the frequency at which the peak acceleration occurs and use that acceleration. I don't remember if you have to tune it all the way to the zero period acceleration or not. This can still result in really large members because you may have to get the structure to a very high natural frequency (each plant will have specific rules) but I seem to remember values of around 25 Hz. This of course still results in very large members but much lower seismic loads. This helps with anchorage, especially when using post-installed anchors.

Section 3.9.2.2.2 is in the section for mechanical systems not the primary buildings which is why equivalent static loads would be permitted for those systems (compare 3.7.2 versus 3.7.3). Part of the problem is the DCD is a very high level document that is describing how the applicant is going to meet the NRC regulations. The DCD references a lot of other NRC documents that contain the actual regulations. One of those may explain how the equivalent static loads are calculated but I can't remember which one because it has been 10 years. You might want to look through the references in the DCD. There could also be something in an IEEE or ASME standard referenced by the DCD or another NRC document. There are NUREGS, Reg guides, the SRP, and nuclear specific industry standards such as ASCE 4,43, IEEE 344.

In the case of the US-AWPR I don't think the DCD was completed before Mitsubishi suspended and eventually terminated the application. It may look complete but the NRC may have still be sending in more RAI/RFI triggering changes because they weren't happy with something. The US-APWR stops on revision 4 whereas the AP1000 is on revision 19 and it is still under construction. There are other design criteria documents for more specific systems in the plant but these are typically not publicly available because they don't require review from the NRC (they can't be less restrictive than the DCD or UFSAR). Those may introduce more criteria or limitations for specific structures or systems for a particular nuclear plant. The plants I worked at had additional design criteria documents and procedures for things like cable tray and conduit supports and pipe supports that introduce more rules. I think that is where the 25 hz number I am remembering above comes from. Another plant might have a document that says it has to be 30 hz.

Usually new electrical components had to be shake table tested but maybe a valve would be permitted to be designed using an equivalent static load calculated from the instructure response spectra. It depended on how complex the equipment was and plant specific rules and procedures (these had to conform to whatever the utility told the NRC they were going to do in the DCD or UFSAR). There is a whole other thing called SQUG, which I am not even going to get into.

Nuclear power is like an onion. There are layers upon layers of safety systems and layers upon layers of documentation and rules. I hope I am explaining this well but it has been 10 years and there are documents that reference other documents that reference documents that you aren't going to have access too. I think it took me a year of working in the industry before I had all of this even halfway figured out.

 

[bones206]

WARose I think what you may be gleaning from these replies is that there is a whole ecosystem of regulations, codes, design criteria, supporting calcs etc, that go into designing nuclear structures. Without that entire ecosystem in place providing inputs to your design, you are really just cherry-picking. But that’s ok, in your specific case.

My experience is similar to CLT49er with regard to seismic. Lots of fancy dynamic analyses and response spectrum methods, which in turn were validated with simple lumped mass stick models. Not sure exactly how the piping folks did their seismic analysis as it wasn’t really my world, but I know it was a dynamic analysis. May have been response spectra analysis of specific anchor nodes on a pipe run to generate the design loads for the pipe supports.

 

[CLT49er]

Check out section 3.7.2.5 of the US APWR DCD.

 

[CLT49er]

Based on the 5 or so plants I have worked at I don't think I ever saw a peak acceleration over 30g and that was for the US APWR on the polar crane, which was a newer design. They were all east coast plants. If you are just trying to approximate nuclear seismic loads then 1.5*30*weight should get you conservatively in the ball park. It really is overkill though for a non nuclear application.

 

[WARose]

My experience is similar to CLT49er with regard to seismic. Lots of fancy dynamic analyses and response spectrum methods, which in turn were validated with simple lumped mass stick models. Not sure exactly how the piping folks did their seismic analysis as it wasn’t really my world, but I know it was a dynamic analysis. May have been response spectra analysis of specific anchor nodes on a pipe run to generate the design loads for the pipe supports.

Thanks for getting back to me Bones. Reading through the regs....it made me think you would ultimately get the anchor loads for equipment directly from electrical & mechanical. Mainly because if everything needs a dynamic analysis.....not too many structural engineers are going to be doing a FEA model of a transformer or pressure vessel itself.

 

[CLT49er]

The large transformers in the switch yard wouldn't have been required to be seismically tested because they weren't safety related (needed to prevent a meltdown). Smaller transformers inside the plant would have been shake table tested. Half of the engineers I worked with at the time had PhDs and they did do a dynamic analysis of the pressure vessels. It came down to whether there were moving parts that couldn't be realistically modeled in an FEA program. I think the containment vessel model took two to three weeks to run on a $15k-$20k computer and that might have been just the static model with all of the construction stages for the post tensioning.

One of the pump vendors did a dynamic FEA analysis of the pump but then shake table tested the electric motor. We got the anchorage loads for the pump from their dynamic analysis. I think we usually got equipment anchorage loads directly from the vendors. I don't remember what percentage was from analysis versus testing.