Question regarding modeling of truss portal frames... 2

[IngDod]

Greetings and thanks for reading my message,

I am modeling a two story building in SAP2000, this building has frames spanning 18m so the beams are trusses. My question is how should i model the connection between truss chords and frame columns.. currently i set the connections not to transmit moments, only shear and axial force... this causes a pair of forces in the columns that generate a bending moment; however i am not sure if this is the expected behavior of trussed beams supporting gravity loads. Reading structure books and examples trusses are always modeled as being simply supported, but to obtain a similar behavior in my structure i would have to release the axial transfer of forces between column and truss. I am at a loss here.. thanks.

 

[paddingtongreen]

Oh dear, It worries me that you have to ask this question, do you not have a mentor? You clearly need one.

If it is to act as a portal, you must have a moment connection at the top of the column, that is what the chord couple is supplying. If it were a beam, your model would have a moment connection. I don't understand why you would lose the ability to transfer axial loads if you fix the chord connections. If you make them fixed, you will reduce the chord axial forces slightly but when you add the chord moments you would have about the same total moment as before.

The trusses you have been looking at are not used in portals, they are simply supported on columns and girders with the floors/roofs acting as diaphragms and with bracing in the walls to carry the horizontal loads. No portals to be seen.

Really, you should understand, qualitatively, the behavior of the structures before you model them, otherwise, if you put any wrong information in, you may not notice its effect in the strange output.

Michael.
"Science adjusts its views based on what's observed. Faith is the denial of observation so that belief can be preserved." ~ Tim Minchin

 

[IngDod]

Thanks for your answer, I know this is a rather basic question... one I should know. But its the first time I am facing this kind of structure and i had no basis for comparison, my first assumption was to model it with pins (No moment transfer, only axial and shear). But as I began to research on the subject I found only examples of roof trusses (which are simply supported)and nothing on trussed frames, and hence my doubts.

After checking the moment capacity of the connection I am going to use for chord to column I realized that this connection has more capacity than what is transferred by the chord (bending of the chord due to self weight) to the column if a fixed connection is assumed, so my conclusion is that the connection would act as fixed when only gravity loads are acting; for lateral loads where I expect moment to be greater in the columns i am yet to perform this check.

It is my understanding that the moment transfer from truss to column (Gravity Loads) or column to truss(Lateral Load) would occur by transfer of axial loads through the chords and truss members to and from the column (The truss resisting the displacement of the columns due to moment in the column or the columns resisting the displacements of the chords due to tension or compression), not by direct moment transfer (Rotation of column and chords together). Hence my initial assumption of using pin connections: Basically a pair of forces acting at top and bottom chords that causes moment on the column and tension or compression on the truss members. Could you please tell me if this is correct, as this is the basis of my analysis.

Sadly, were I am right now I have no access to an engineer with knowledge in this subject I can rely on. That's why I decided to ask on this forums instead of just carrying on with my assumption.

 

[BAretired]

Hence my initial assumption of using pin connections: Basically a pair of forces acting at top and bottom chords that causes moment on the column and tension or compression on the truss members. Could you please tell me if this is correct, as this is the basis of my analysis.

It is perfectly correct.

BA

 

[IngDod]

Thank you very much for your answer, I was very worried that I was making a gross mistake in my assumption. If anyone has any further comment on the subject of the trussed frame model or the way I'm determining the connection stiffness I would appreciate it.

I was talking with a fellow engineer on another forum (he had no comments on the trussed frame issue) and he told me that the way he modeled connections which were semi-rigid was: Calculate the maximum moment capacity of the connection and if it was below the one being transferred by assuming fixed connection he would adjust the connection stiffness in his model (a semi-rigid connection)so that the moment at the connection was equal to its capacity. It seemed logical for me, since the moment transferred to a semi-rigid connection is entirely dependent on the connection stiffness; however I had never seen this method before and in the past I had used connections which were clearly defined (or at least assumed)as Fixed or Flexible in the literature and codes... I have seen several papers regarding the estimation of the initial stiffness of connections for I-Shaped members, but they were are all based on data from laboratory experiments.

 

[BAretired]

Trusses are often designed on the basis of pinned connections. It is not quite correct, but it is usually close enough. Continuous trusses are indeterminate structures, but pin connections can still be assumed at the nodes. When a truss is indeterminate, statics cannot be resolved by equilibrium alone. Strain compatibility must be considered.

Rigid connections are sometimes assumed, particularly when truss members become very stiff as in trussed steel bridges. In that case, connections are designed for the calculated moments, shears and axial loads found in the analysis.

Semi-rigid connections are another subject entirely. That is not what you are using. You are simply making a choice to use pinned connections, a choice with which I agree.

BA

 

[JAE]

The procedure that your engineer friend suggested - calculating a moment capacity, then in the model adjusting a connection stiffness to match that capacity - sounds to me to be very wrong.

Here's why.

The engineer appears to be confusing capacity with stiffness. The two are not always directly related. For a gross metaphor, a palm tree can have the exact same "capacity" against bending in the wind as an oak tree but they behave in very different fashion.

If a moment connection has X capacity in moment and you set the connection stiffness such that under the required load combinations you only reach X but not more, then the model is "modeled" with an explicit stiffness that may be less than what is really there - if your connection is IN ACTUALITY much stiffer than that phony stiffness in the model, then it can attract (in real life) more moment than X and fail.

What the engineer is assuming is that the connection will have some level of inherent flexibility to shed that moment prior to exceeding X - shedding it to other areas of the frame.

The problem is the connection may not have that inherent flexibility to allow that shedding.

 

[RFreund]

For some further info-

Google "Designing with Vulcraft" it is a helpful guide to practical design situations in regards to steel joists, joist girders and steel deck.
I believe they cover the situation for moment frames with truss girders. For some reason it seems like vulcraft's site is down or something is wrong but you should be able to find the document or order it (I think it may be free).

EIT

http://www.HowToEngineer.com

 

[CCox]

Our office just recently completed a project where we used Vulcraft joist girders as part of a moment frame. Generally, the bottom chord of the joist girder is not positively attached to the column, but straddles a stabilizer plate. To create a moment frame connection, the bottom chord is welded to this plate. The plate and welds must be designed to transmit the axial load. The top and bottom chord axial loads are supplied to Vulcraft for the design of the girders.

If you are not creating a load path for a force couple (as in welding the bottom chord to the column), then there is no way for the connection to be assumed as anything but pinned.

Cody

 

[IngDod]

BAretired: Thanks for your thorough explanation. It is perfectly clear.

JAE: Thanks for your answer, from what you are saying I understand that this approach is wrong because by giving the connection an arbitrary stiffness that makes the Moment at the connection to be equal to the connection capacity you condition the model to a stiffness that is not necessarily real. The moment capacity of the connection is accurate but the stiffness determined in this fashion is completely phony and entirely dependent on the load applied... not on the connection.... Furthermore it promotes a situation in which if you assume a fixed connection and the moment you are getting in your analysis is greater than the capacity of the connection you intend to use the connection would start behaving semi-rigidly instead of just failing.... basically this guy is saying that connections dont fail... they just become less and less rigid until they eventually become a pinned connection...

I have yet another question then... I have tried to find a reliable criteria for classifying connections as either pinned or fixed, from what I have read in books and papers the concept is inherently linked to the connection capacity, and also the connected member plastic moment capacity.. What I understood was that a connection regardless of its components exhibits a rigid behavior if its capacity is greater than the connected member capacity.. and connections where the member has higher moment capacity are semi-rigid, if the member capacity is much greater (how great is enough?) than the connection, then the connection would behave as a pin. My greatest doubt in this subject is how is the rotation of the connection related to this criteria, obviously connections that have no resistance to rotation are flexible but what about welded connections for example.. how can the welds, which must experience physical separation to rotate, be rigid, semi-rigid or flexible depending on the connected member. Please read my answer to CCox below regarding the hollow tube connection.

RFreund: thanks for the tip, I am trying to find it.

CCox: Thanks, this was very helpful. The connection would be direct welding of the chord and column as I am dealing with hollow tubes, this will probably require a stiffening plate. I have read in the CIDECT guides (European institute dealing with hollow tubes) that welded connections between hollow tubes shall be taken as rigid if the moment capacity of the welded connection is higher than the moment capacity of the chord.

 

[barrywilsons]

If you are connecting a truss to a column via a couple then there should also be a diagonal member. It is normal to consider this as a pinned joint because the members butting into the column transmit loads only along the axis of the member. The diagonal in the truss carries the shear force. If you are not having a diagonal member then you are creating a vierendeel truss and the horizontal members are taking the shear force.

If you are welding hollow sections to the column and you wish to use butt welds you have to place an inner tube inside the hollow section otherwise the weld will not fully penetrate. I have seen these things fail. For myself I avoid butt joints in hollow sections because the fabricators rarely do them properly. I use fillet welds only and external doubler plates (fish plates) where required.

 

[RFreund]

Here it is.

http://www.vulcraft.com/products/catalogs/designing/document.pdf

EIT

http://www.HowToEngineer.com