Moment Frame Baseplate Design - High Seismic 3

[sticksandtriangles]

I am looking for some guidance on the FEM design of high seismic moment frame baseplates.

A little background:

Design criteria, fixed base column:

Column = W14x730​

Moment = 2200kip*ft, Omega Moment = 6600 kip*ft​

Axial = 140 kips (T) Omega Axial = 420kip (T)​

I have sized up an unstiffened baseplate... 9" thick grade 32 steel baseplate (massive!!). Uses (5) 3 1/4"Ø grade 105 anchor bolts. This is based on a purely rigid BP assumption and based on DG#1 examples.

I want to explore what the thickness of the baseplate needs to be using a stiffened design and do a little FEM modelling to confirm the software rigid BP assumptions.
Using RAM connections stiffened baseplate design module, it looks like a 7 1/2" thick baseplate can get the job done.

Ideally I would like to get into the 5"-6" range of BP thickness with stiffeners as required and have started to dive into a little FEM modelling to see if I can not get into this range.

FEM Model Development:

I have developed a SAP2000 model with compression only elements (gap elements) for the grout bearing, tension only elements (hooks) at the anchor bolt positions and modelled the baseplate as a non-linear shell. This is my first hurrah into heavy non-linear modelling so I am somewhat intimated by trying interrupt my results. I have started with the unstiffened design to get a feel for the more simple approach.

Questions I have:
[ul]
[li]How do i accurately model the stiffness of the grout bed/concrete foundation below?

My thoughts, E of concrete is readily defined, but what do I assume for deflection? What I did in my model was take E*trib area/deflection = stiffness (utilized 8600 kip/in for a 1 square inch finite element)​

[/li]

Utilizing this approach with a linear compression element yields very high concrete stresses. I have stresses in the range of 13ksi. my allowable for 5ksi concrete is around 5.5ksi​

I was thinking maybe I could model some non-linearity into the compression only element to allow for stress resdistrubtion within the bearing area to limit stress to the 5.5ksi range, but I am not sure if that is truly how the grout bed below would behave. I would assume that it locally crushes and fails? My DG#1 original design utilizes a rectangular bearing strength assumption though so I feel ok with this.​

[/ul]
[ul]
[li]Anchor bolt stiffness?[/li]

I did A*E/L for the anchor bolts. Yields a stiffness of 8020 kip/in. I have a long embedment of 60" on these 3 1/4"Ø ABs​

[/ul]

Image of high concrete compressive stress (each square is 1"x1"):

Images of the SAP model for reference under Axial and Moment Load listed above (shows signs of local yielding at edge of flange):

Thanks!




S&T

 

[JLNJ]

How is your base plate attached to the column?

It seems like you will need a big BIG bracket to ge the forces from the column into the bolts and one factor of that bracket size is that it is big enough to reduce the bearing stresses to something manageable.

 

[KootK]

Wow!! Fun day at the office. I can't answer most of your question as you'd want them answered but I'll throw a few things out there for conversation starters.

1) I've done some elastic FEA on normal, gravity column base plates and pretty much all of those will also overstress the concrete locally. Distribution is the only rational story to tell.

2) I don't think that I'd even bother trying to model the concrete. In addition to the issues that you've raised, you've also got the flexibility of the soil, the effect of any bolt prestress, creep in the concrete, the flexural flexibility of whatever the heck your foundation element is... on and on. It's a rabbit hole of uncertainty with practically no end to it. Instead, I'd generate a plastic force distribution like you mentioned above and just apply that tot he base plate as a load directly. For criteria, maybe just keep things from yielding or deforming a ridiculous amount.

3) At this scale, I sure would like to:

a) Embed your column deep into a massive grade beam or;

b) Take the bolts up to an elevated bearing seat to increase stretch length and simplify load transfer.

 

[WARose]

Questions I have:
How do i accurately model the stiffness of the grout bed/concrete foundation below?

Not sure I have ever done that. In such cases, I have typically modeled the [perimeter of the] bolt holes as (pinned) support points and added the uplift/downloads (delivered by the column flange/web) where required. Typically the uplift controls the design anyway.....so (for a moment) the inaccuracy on the other side doesn't really matter.

By the way, looks like you are using solids (if I read that pic right). Is that really necessary? I've always modeled it with plates.

 

[Rabbit12]

I have to ask. What are you designing with forces like those?!

 

[r13]

How is the uplift force to be sustained? As an approximate, T = 2200*12/22.4 = 1178 kips, quite large, how about shear? Suggest to increase the foot print (with stiffeners) of the base plate to spread the force to a larger area. Another thinking is to provide bracings in the framing to take out some lateral load, and design the column as pinned, or semi-rigid connection.

 

[KootK]

I'm going to guess... hospital, interior frame.

 

[r13]

Industry building, or equipment support, I guess.

 

[sticksandtriangles]

Thanks for all the replies so far. A few responses

How is your base plate attached to the column?

Full pen welds everywhere.

I have to ask. What are you designing with forces like those?!

Interior hospital frame as KootK had suspected.

By the way, looks like you are using solids (if I read that pic right). Is that really necessary? I've always modeled it with plates.

They are modeled as plates, just shown in the extruded view.

Suggest to increase the foot print (with stiffeners) of the base plate.

I am going to move onto that after I feel comfortable with the results from the unstiffened case, i will keep you posted.

2) I don't think that I'd even bother trying to model the concrete. In addition to the issues that you've raised, you've also got the flexibility of the soil, the effect of any bolt prestress, creep in the concrete, the flexural flexibility of whatever the heck your foundation element is... on and on. It's a rabbit hole of uncertainty with practically no end to it. Instead, I'd generate a plastic force distribution like you mentioned above and just apply that tot he base plate as a load directly. For criteria, maybe just keep things from yielding or deforming a ridiculous amount.

I will play with defining a spring that goes linear elastic up to a point then plastic and see what the results are. Coming up with a rationale stiffness does seem to be quite the rabbit hole, but does greatly affect the bearing results.

3) At this scale, I sure would like to:
b) Take the bolts up to an elevated bearing seat to increase stretch length and simplify load transfer.

I've heard and seen this done before, but I can not see what the benefit is, I must be missing something. Care to explain?
















S&T

 

[WARose]

I will play with defining a spring that goes linear elastic up to a point then plastic and see what the results are. Coming up with a rationale stiffness does seem to be quite the rabbit hole, but does greatly affect the bearing results.

You are using FEA results to check the bearing stresses on concrete? Seems a little odd to me. I.e. something that could be resolved much more quickly with hand calcs. (In fact, take away the stiffeners in the plate and it all could be done with hand calcs.)

 

[KootK]

Interior hospital frame as KootK had suspected.

Bitchin'! Being right's better than sex... lasts longer. Much longer.

I've heard and seen this done before, but I can not see what the benefit is, I must be missing something. Care to explain?

The benefit is twofold, depending on what you're up to. In order of awesomeness:

1) Directness of load path. You steer your anchor bolt tension directly into your column flanges, bypassing the base plate entirely. One could even do this in compression, conceivably, but it's trickier and I've never seen it done in the wild

2) If you're pre-stressing your anchor bolts, this gives you a chunk of stretch length that doesn't have to be sleeved into the concrete.

That's right, feel Omer Blodgett judging your FEM models from beyond the grave...

 

[PROYECTOR]

Hello sticksandtriangles,

Design criteria, fixed base column:

Column = W14x730
Moment = 2200kip*ft, Omega Moment = 6600 kip*ft
Axial = 140 kips (T) Omega Axial = 420kip (T)

The Omega factor only need be applied to the seismic load component in your design load combinations (not the result of the load combination). So, I think you might be overestimating the axial load (That could make an important difference in this case).

I want to explore what the thickness of the baseplate needs to be using a stiffened design and do a little FEM modelling to confirm the software rigid BP assumptions. Using RAM connections stiffened baseplate design module, it looks like a 7 1/2" thick baseplate can get the job done.

I would not use RAM Connection to design a stiffened base plate, the method used by RAM is too conservative. In such cases, I prefer to use yield line theory to find the base plate thickness.

As KootK suggests, a good option to transfer large loads is to use a base chair connection. However, the architect and/or the owner may not want an exposed base plate. The base chair and the anchor rods can be covered with dray wall or something, but the available space will be reduced anyway. If the architectural aspect is not a problem and you decide to go for that solution, I recommend you consult the Bo Dowswell's paper "Bending of Top Plates in Base Chair Connections".

Another option to transfer large moments is to use an embedded column base connection. That way, the base plate and anchor rods are required only for erection and for transferring the axial load. The moment and shear are transferred by direct bearing of the column into the concrete. A grade beam is often used to do the job. You can find some guidance in the AISC Seismic Design Manual (2nd or 3rd edition).

Full pen welds everywhere.

Don't forget to consider the need of weld access hole, which can be quite large in this case. This will greatly reduce the available shear strength of the column web.

Good luck!

 

[JLNJ]

A couple of questions:

In the frame analysis did you model any flexibility into the base anchorage?

What limitations are there when CJP welding such thick sections? It seems a little iffy.

 

[sticksandtriangles]

Warose, I would not say that I am using the model for bearing stress calcs, but it is something that I would like to get into the ballpark.

I have "solved" my bearing stress issue by definig a non-linear link element in SAP:

Units of displacement = in, force = kip.

New bearing stress plots look like this with the gap element bahaving as expected (hits 5.5 kips and then goes plastic with a lot more bearing area now).

The benefit is twofold, depending on what you're up to. In order of awesomeness:

1) Directness of load path. You steer your anchor bolt tension directly into your column flanges, bypassing the base plate entirely. One could even do this in compression, conceivably, but it's trickier and I've never seen it done in the wild

2) If you're pre-stressing your anchor bolts, this gives you a chunk of stretch length that doesn't have to be sleeved into the concrete.

Part 1 makes a ton of sense and would be great to incorporate. Part 2, given that I've never designed prestressed anchor bolts, i assume you have to "sleeve" in the concrete so the AB doesnt stress the concrete along the length of rod and potentially transfer load to the concrete in lieu of the bolt itself? I also need to read a little about the benefits of prestressing ABs. Not sure of the benefit or purpose.

I have been referencing that Blodgett book as I go

The Omega factor only need be applied to the seismic load component in your design load combinations (not the result of the load combination). So, I think you might be overestimating the axial load (That could make an important difference in this case).

These loads are from only applying the omega to the seismic component of the force... I think i've got this part correct.

Also good point on the weld access hole for the web.

We might also explore embedding the column into the concrete for direct load xfer, but this would be an oddity in typ construction for sure (but I guess a 9" thick baseplate is as well!)

The SAP model I am developing also works well to take out the conservative nature of the baseplate thickness sizing, you can spot areas of local yielding a make a call on wether you feel OK with that part yielding (kinda of like yield line theory)

In the frame analysis did you model any flexibility into the base anchorage?

What limitations are there when CJP welding such thick sections? It seems a little iffy.

Frame analysis did not consider any flexibility of anchorage or foundations below. Might be worth exploring to see if this can reduce the load.
With regards to the CJP, I have the same concerns. We are meeting with our steel contractor shortly and I will be sure to ask about such a massive weld.




S&T

 

[sandman21]

We might also explore embedding the column into the concrete for direct load xfer, but this would be an oddity in typ construction for sure (but I guess a 9" thick baseplate is as well!)

Attempting to provide a baseplate would be the oddity for supporting the moment frame of this size. Assuming this is a OSHPD project they will require modeling of the foundation and connection for drift, CBC 1616A.1.16. Good luck

 

[KootK]

Part 2, given that I've never designed prestressed anchor bolts, i assume you have to "sleeve" in the concrete so the AB doesnt stress the concrete along the length of rod and potentially transfer load to the concrete in lieu of the bolt itself? I also need to read a little about the benefits of prestressing ABs. Not sure of the benefit or purpose.

1) My understanding is that the primary benefit arises in situations with frequent load reversals that would cause fatigue issues. You get that non-intuitive thing happening where the stress in the bolts doesn't actually change until you overcome the prestress in them. This has the effect of narrowing the band of stress variation in the bolts and, thus, improving their fatigue response.

2) Frequent stress reversals? Surely that's a perfect fit for building lateral systems under wind & seismic? Not so. As an industry, we seem to have decided that neither wind nor seismic produces enough lifetime reversals at high enough stress levels for anchor bolt fatigue to be an issue. I see most anchor bolt pre-stresssing in the industrial world and none at all in the ICI building world. It is a difficult thing to control the amount of prestress in an anchor bolt accurately over it's lifetime given the reality of concrete creep etc.

3) You may or may not need to sleeve a chaired anchor bolt. You need a stretch length that will allow you prestress the bolt as you desire without fracturing it. In some cases, the chair height alone may be enough for this. In other cases, it might need to be a combination of the chair height and some sleeving into the concrete. You don't really need to worry about the the stress in the concrete because the designing the concrete/bolt interface to deal with those stresses would have been part of your anchor bolt design regardless of whether or not the bolts were prestressed. Obviously, if your prestress is more than your actual, external load generated bolt tension, the prestress may govern the anchor bolt design.

 

[Aesur]

I am curious what cost difference you expect in these plate thicknesses and are you being compensated for spending a good portion of your budget fine turning these connections, is the cost to benefit ratio worth spending many hours (typically at your cost) to design these to save the contractor a few dollars? I just want to understand why you are going this far with these connections without currently knowing the scale of the project or how many connections there are. I have seen this done in large bridges, where a few engineers spent months to remove one bolt from connections, but it saved money in the long run (of which they got a cut of the savings) because it removed enough bolts, but I have never seen this setup done in buildings.

 

[sticksandtriangles]

Sandman12, this is not an OSHPD project. Interesting to hear baseplates being an oddity.

Aesur, there are a number of these columns that would require this large baseplate (about 8). As of right now, I have not sunk too much time into this study... I hope to not do too much more. I will say I have been pleasantly surprised at "how easy" it has been to do non-linear modelling. I think this maybe more of a case of me not knowing what I do not know though .

This is all anticipation of the contractor screaming about a 9" thick baseplate. No kickbacks anticipated on getting the baseplate size down... that would be nice though if it worked that way.

S&T

 

[PROYECTOR]

Hi sticksandtriangles,

I have developed a SAP2000 model with compression only elements (gap elements) for the grout bearing, tension only elements (hooks) at the anchor bolt positions and modelled the baseplate as a non-linear shell. This is my first hurrah into heavy non-linear modelling so I am somewhat intimated by trying interrupt my results. I have started with the unstiffened design to get a feel for the more simple approach.

I have also used SAP2000 to evaluate the nonlinear response of base plates as part of a research program. I compared the results with a series of experimentally validated FEA models in ANSYS to calibrate a simplified modeling approach in SAP2000 (less computationally expensive). The results did not match perfectly, but are good enough for design purposes. I think our modeling approaches are quite similar.

These are some of the modeling approaches adopted.

1) The anchor rods where modeled using two-node link elements using a bilinear force-deformation relationship.

2) The column, base plate and stiffeners where modeled using layered shell elements considering the plasticity of those elements.

3) The concrete was considered elastic and was modeled using a Winkler-Pasternak subsoil model. The elastic stiffness of the concrete was determined by theory of elasticity considering the concrete as a semi-infinite solid subject to a uniform distributed load. The solution can be found in Timoshenko & Goodier “Theory of Elasticity”.

4) For load application, a body constraint was assigned to the nodes at the top of the column.

I have "solved" my bearing stress issue by definig a non-linear link element in SAP:

I think it's a very clever solution. What model did you use to find the Force-Deformation relationship of the concrete?.

The SAP model I am developing also works well to take out the conservative nature of the baseplate thickness sizing, you can spot areas of local yielding a make a call on wether you feel OK with that part yielding (kinda of like yield line theory)

A good criterion to establish what level of yielding is acceptable, is setting a limit on the plastic deformation. For example, the Eurocode establishes a maximum of 5% on plastic deformations when designing connections by advanced analysis methods (nonlinear FEA). In fact, this is the same approach behind the software Idea Statica Connection.

 

[ATSE]

"...This is all anticipation of the contractor screaming about a 9" thick baseplate." Step back from the design (which looks like fun) and consider actual steel properties, and the realities of modern fabrication.
Thick steel is prone to cracking, and can have inclusions. What we SEs love about structural steel - ductility, toughness, weldability - are all but lost when fabricating very thick sections.
Even 4" thick plate steel is challenging to weld (with good results) for an experienced fabricator. If you can't make this design work with steel less than 6" thick, I would take a different approach.