Analysis Discussion: HSS Baseplate, usually assumed as pinned. Yet in my FE analysis it seems Fixed. 2

[IngDod]

Greetings, attached is a summary of my FEA analysis.

I am currently designing a two story OMF building. My initial assumption was to assumed the columns pinned at the base, I did this because it seems to be the common practice. But now that the time has come to design the base plates I'm having problems (cant sleep) simply assuming its pinned and designing it solely for axial force and shear (using AISC's DG1), it troubles me that if this connection is actually fixed (or at least with high rigidity) the baseplate will fail.

Now, I've always had problems grasping the concept of steel connection design; some people say that since you assume is pinned and you designed for axial loads only then it will behave as such. I proceeded to do a FE model, if the connection is indeed pinned then I should see little (no connection is truly pinned of course) moment being transferred to the baseplate. What I get is that the base-plate bends considerably due to the moment, now i would expect this since it is precisely this bending which allows rotation; however the stresses in the plate are much higher than what it can resist... So while it rotates, it seems to me that it would fail too. Now i proceed to increase the baseplate size so that it can resist the stresses, now the rotation is much much smaller... giving me the impression that by increasing the thickness I have made the connection rigid.

Please see the attached summary as it is much easier to understand. The initial baseplate designed for axial load only is 1cm thick, while the one designed based on the FEA is 2.3cm thick. I added results for more thicknesses to study the behavior, with 5cm thickness the rotation seems very low.

I don't really know how to judge if the connection rotation is high or low, or whether it approaches rigid or pinned behavior. I'm getting rotations of 2 degrees for the 1cm thick plate and around 0.5 degrees for others... is this a lot? is this rigid?. How much should a connection rotate in order to be considered as pinned?

Thanks and I hope you can help me with this dilemma.

 

[AELLC]

I think we referred to that nonsense as Ultimate Strength Design, back in the day. I went to many a seminar where they championed it. We should start a separate thread of what nonsense it has created, and how many engineers are confused.

I was hi-tech when I went to graduate school, however. My thesis was Progressive Collapse of Steel Frames.

 

[CELinOttawa]

Lol... Well AELLC, you're certainly an American.

 

[Archie264]

>>>I have read Russell Fling's (Designer of 30,000 concrete beams by ASD elastic method) essay of why USD design is nonsense. That was good read. (But I lost my copy)<<<

I would love to read that as well; it sounds interesting. Do you remember if it was a stand-alone article or part of a book?

 

[AELLC]

I always have had this perception of engineers in the UK, Europe, China, NZ, Australia, Canada, etc as being more advanced than here in the USA, but we make better softwares, computers, and airplanes.

 

[CELinOttawa]

I don't think it is at all fair to say that the Engineers are more advanced outside the US, but I do think that the rigour with which most countries train their engineers is no longer "standard" in the US.

From school to school the quality of graduates varies much more greatly out of the US than out of most other countries with which I am failiar... As a result your codes tend to me even more "recipe book" than anywhere else. That holds back true understanding and lends itself to the comoditization of Engineering.

VERY good engineers (and quite sh*te engineers) exist in all jurisdictions.

 

[AELLC]

I think the biggest problem here in the USA is:

We test and categorize the heck out of our High School students, but we don't know how to teach them math and science properly.

BTW - my HS counselor recommended to me that I don't go to college. Still trying to figure that one out.

 

[CELinOttawa]

I'd say you've proven your BS Guidance ClownSeller wrong with aplumb(bob). *evil smiles*

 

[AELLC]

I wonder if IngDod got his project finished without any more fretting. We all spent a lot of time helping him.

 

[IngDod]

@AELLC: Sorry I had not posted... I had plenty of non-engineering things to do today; but i have been following the thread all day from my phone.
The baseplate is A36, the HSS tube is a different steel (im not in the US, I dont know if HSS in the US is also A36). I use HSS because thats produced in my country, W shapes are imported so the price is very high. Here we use HSS for both beams and columns.

@Ron: I agree with you on structures not actually seeing the loads for which they have been designed. For example I would say that the 1.6 amplification on the live load pretty much guarantees that the live load will not be exceeded, unless of course the owner decides to change the use of the structure... say turn a second floor of a commercial building into a storage area.. that would be a big problem and indeed I have seen it happen; however I would say this is not a design error but rather a user error.

I checked the model and qualitatively it seems to behave as expected, I recently took a SAP2000 class that covered modeling baseplates and I based my model on this.

@Sail3: Thanks for such a thorough answer, I understand what you say about the redistribution of stress; my intention with the FEA model is to determine the connection stiffness at the onset of yielding in the plate; My understanding is that as the plate further yields and redistribution occurs the rotation of the connection would increase; thus a connection that behaves as pinned in the FEA model should approach such condition in real life. However I do feel uncomfortable not being able to model or predict this redistribution.. While I know that redistribution occurs I do not know how to calculate or estimate it. If anyone knows of a way to do this I would be most grateful.

@CELinOttawa: I agree with you, I also do not design solely by black box programs. In this case I am trying to better understand the behavior of the connection, in the past I have relied on AISC DG1 for calculations. I use FEA as a tool, I do not blindly accept the results; if I did we would not be having this conversation. If you have a "back of the envelope" method to quickly estimate connection stiffness and rotation I would be most grateful if you could share it.

@Gumpmaster: Thanks for your reply, I would pose the question how can I determine if there is too much yielding? There must be a method to ascertain this or at least to help in making an engineering call.

@rowingengineer: To be honest I have never designed a semirigid connection, my understanding is that you would need the moment-rotation curves for such a connection. In the past I have chosen connections that are widely assumed to be pinned or fixed. But I can see how such analysis and design would be better, specially since no connection is truly pinned or rigid.

@johnbridge231: I added springs to the area elements that form the baseplate, this springs are compression only and are very stiff to simulate concrete pedestal. The bolts are modeled as a pinned restraint; the loads are non-linear and the analysis is as follow: apply axial load, then use the matrix resulting from this load to apply the moment load.

 

[AELLC]

IngDod,

We don't see A36 but rarely in the USA now. Most steel is Fy = 50ksi.

HSS means Hollow Steel Section, not High Strength Steel. Fy can be from 36 to 50ksi, for rectangular tube section, the 2 most prevalent here are 46 and 50 depending on the exact steel spec.

I am surprised you use so much HSS instead of W-section. Here, the W is cheaper and much easier to make connections. Plus, with HSS and you have a non-thru shear tab, the question is, does the tab buckle the wall of the HSS, or does it punch thru? Sounds like a can of worms.

 

[CELinOttawa]

Here's a quick and reliable trick for checking and working with stiffness of a connection in steel: %(EI)

I'm not trying to be smart... Most software will allow you to enter this value as a stiffness at a connection, and in hand calcs you can apply a moment based on the partial stiffness.

Whatever moment you develop at that point and are happy with, you design your system to sustain without failure (both SLS & ULS).

Thus is exactly the type of situation Wooten was envisioning: You will never know the exact state of stress, even with FEA, so pick a possible state of stress and design against that. Plastic materials WILL conform.

 

[IngDod]

Here we use A36 mainly for baseplates and such. Yes for some reason the hot-rolled W-Sections produced locally only go up to 14cm depth.. everything else is imported. There are companies that manufacture W-shapes by welding three plates, but even that is not very common. The HSS beam and column situation is truly terrible.. For gravitational loads i would say is not that bad.. but for lateral loads there is of course the risk of punching trough the wall of the columns when using moment frames. Dont start me on the connections... its awful.. directly welding steel tubes is a recipe for disaster if you intent to develop any moment, I usually detail all the beams with a haunch in both ends to decrease rotation in the connection and avoid plastification and/or punching of the column wall.

I base most of my calculations in the design guides provided by CIDECT, which is an european agency that promotes and research the use of rectangular and circular hollow sections. The top of the cherry is that the hss sections produced here are not compact and 90% of the country is under the risk of a seismic event. Dont try to convince anyone here to use a single bolt on their structure... absolutely everything is welded... the baseplate.. they weld rebars to the plate instead of using proper anchor bolts... The rebar manufacturer clearly states that they rebars are not weldable. Thank god that most structures in my city don't go over two floors, I studied my M.S in the US.. imagine my shock after coming back here and facing all this steel tubes beams and columns... even in universities here they teach steel design with only W-Sections... Its absurd.

 

[AELLC]

Very interesting...what country?

 

[CELinOttawa]

Sounds like Indonesia, or maybe Sri Lanka.

Vietnam still uses Bassemer steel for EVERYTHING.

 

[IngDod]

After reading all your helpful posts I am inclined to design the baseplate as pinned. Following SAIL3's advice to design the bolts for a fixed condition seems like a very good practice and on the safe side.

As I see it right now a column-baseplate connection fixity depends primarily on the thickness of the plate (supposing no stiffeners are used), a thin plate behaves as pinned because it bends considerably, causing the column to rotate and thus not gaining moment. Is this correct?

My dilemma starts here: Since I assume the baseplate to be pinned I size it solely for axial load, which results in a thin plate; But in reality the plate has to bend so it must have stress... If I assume there is no axial load how can I verify that the bending of the plate is not "too much"?

@CELinOttawa: I have done this before.. But how would this allow you to calculate the connection stiffness?. Or are you saying that the connection stiffness will be conditioned by the moment I design it with?.. say i model my frame for only 10% of the columns moment capacity to be transmitted to the base... I design the baseplate for this moment.. And since I designed it for this moment this results in a stiffness that will attract approximately only this moment?

 

[IngDod]

The country is Venezuela.. I will try to take some pictures of connections tomorrow... I'm pretty sure this things don't exist anywhere else.

 

[AELLC]

IngDod

Make the baseplate 3/4 to 7/8" thick. That is common for 2-story steel building.

 

[CELinOttawa]

IndDod: You've got it. After the capacity is reached, rotation and yielding transfers the load to other parts of the structure, making the structure conform to your model.

So long as you have a stable state that provides sufficient Strength, Stability, and Stiffness, plastic structures will comply. It is effectively a close-cousin of the Hardy Cross Method (though I have no idea if you've learnt it, it is DEFINITELY worth learning about at a minimum).

 

[IngDod]

@CELinOttawa:
Thanks, you've truly opened my eyes on this subject... And I can totally see how a FEA model cannot capture this behavior. I was taught the moment distribution.. which I believe is the hardy cross method... So I can see what you mean, however I never thought of applying a similar concept to the connection - member interface. I think my error is thinking of connections as a separate part of the structure.. a habit probably enforced by the use of "stick frames" software where connections are idealized as nodes.

Having established that FEA does not capture stress redistribution due to yielding, it would seem to me (feel free to call me an idiot if I am wrong) that it would still give a close to reality result for connections where no yielding is desired.. For example a fixed connection where i wanted to reduce rotation to the minimum.

@AELLC:
Thanks, I will aim for this during my calcs.

 

[AELLC]

IngDod,

You are still too FEA-oriented...others have designed countless low to high-rise steel buildings without any FEA at all.

How many years since you got your degree at University?