Truss Joints: Pinned or Rigid? 1

[Matrix]

Do you assign the joints of a truss-type structure as pinned or rigid?

I always use the joints as rigid but the static matrix is bigger and the analysis procedure is slower in order to calculate bending moments, shear and axial stresses.

Does anyone noticed any problems in modelling trusses with all joints assigned as pinned?

I know it's faster but only axial stress are given by the analysis soft, and I'm concerned about possible eccentricity of loads on truss joints in the assembly stage.

Thanks,

MP

 

[prex]

If your structure is really a truss-type and not a frame-type you should design it with pinned connections.
You can evaluate the effects of eccentricities by using the axial load calculated with pinned connections multiplied by the maximum eccentricity. Anyway a good design should avoid excessive freedom for adding eccentricities at erection.


prex
motori@xcalcs.com

http://www.xcalcs.com

Online tools for structural design

 

[JAE]

We typically model trusses with pinned joints in the web members, and continuous chords without pins. We do this because that is how most of our trusses are fabricated. Rarely do you see a chord member that is dis-continuous across a panel point, unless it is a very big truss.

 

[Ron]

We model trusses the same as JAE described. On occasion there is a need to analyze with fixed joints (welded tubular structure), but rarely.

 

[Herny]

In theory, trusses must be designed with pinned joints.
But... actually we design "rigid joints" when detailing a real trusses. We make joints details with a steel pieces a lot more rigid than the bars they join (with a lot of bolts and go on...).
Then, rigid joins are not so far than reality.
In other hand (on my own experience), if the bars have not a big sections, the diference in the analysis results are minimal.
Regards
Hernán

 

[prex]

A point that I would like to raise as a comment to Herny's posting is that structural analysis is by no means a way of simply modeling as closely as possible the behaviour of a structure by numerical means.
It is in contrast a much more complicated activity where the behaviour of the structure is compared against suitable behavioural models in order to realistically decide if it will fail or not.
This approach is required because not all the stresses are equally important to determine the failure: an example is in the local stresses generated in the connections of a truss-type structure as a consequence of the absence of a real pinned joint. Those stresses are secondary (deformation controlled quantities) and will normally cause failure only as a consequence of fatigue, not by yielding.
This is the main reason that allows for calculating truss-type structures with unpinned (e.g. welded) connections as if they were pinned: this procedure is not only simpler and faster, but also gives results closer to reality in predicting the failure of the structure.


prex
motori@xcalcs.com

http://www.xcalcs.com

Online tools for structural design

 

[JAE]

I'm not sure about the "closer to reality" aspect, but it is certainly not simpler to add numerous pins to a truss model, as opposed to leaving them out. I would agree that adding pins is more conservative in terms of estimating maximum deflections of the truss.

 

[Gourile]

Hi!
Your question had been subject of research by early founders of structural theory, as Timoshenko. If you read his book in 'Moment Distributin Method', you will find a chapter in 'Secondary Forces on Trusses' in which the behavior of pined-joint trusses is compared with the trusses with fixed joints. Results show that bending moments are negligible and the process of taking fixed joints will affect the overall structural weight only by a small percentage. As a rule, if the axis of members are coinciding at joints, taking the joint fixed, even if the members are tubular and welded connection, will not show considerable bending moment. You can prove this by once analysing a truss with fxed joints and then re-analyse it by the joints released for rotation. If you have offset in the joints i.e. brace axis do not coincide on the same point on the chord, or if you are dealing with fatigue problem as in long span bridges, you should respect the effect of fixity on the joints.

 

[Herny]

Prex is right. If the structure is important, we must do a deep "debugging" of the model. 3D truss cover of soccer stadium is not the same than roof truss of my garage. We ever must put our structural model on the safe side. But, in this point, let me "take the ball drop on the ground". If we are a structural designers, we must do an additional effort to give a better model. Model ever can be better, and ever must be the better construction of the reality "on paper" we can do (functionally to importance of trouble). In this specific case, pinned joint is the "more secure option model", but I say it is not a better construction of the reality becouse in 100% of trusses the joints designed have a important degree of stifness. Is economically too difficult build a pinned joints. Then, a good chance to generate an approach of reality is a 3d Finite element solid and all the bars concurrent with rigid support on it ==> Too complex!!!. We should eliminate a 3d solid and then make one bar with rigid end, and the others with spring end support (with variables spring values); now we have yet a good model, but is very difficult to generate the math model. To simplify, change the spring for pinned end support, and we have a model we know years and years ago (there are many other possibilities). Note I don't discusse about Bernoully-Navier Hypothesis, the base of Bar Theory. You say at this time "too much words for come to the same place" but the point is: each model represents differents ways to materialize the joints, then the model that we use must be compatible with our way to design the joints. The building details are the second half of our job, but must be consistent with the first half. At this point I believe we must do our best efforts to be better than Egyptians and Romans, and we have the materials and knowledge to do that (really, I love pyramids and archs, but I prefer to do other things).
At this moment I implore for your pardon (what a boring cascade of words!!!).
(I apologize too for my bad english)
Herny

 

[chandr]

Gourile's right. If the shape of the trusses you're using is the usual combination of triangle with no joint offset, even though you model the joint as rigid the deformation is mainly axial and the induced moment is very small, making almost no differences in the member selection. I usually use pin to check deflection and rigid to do buckling calculation, since most softwares I'm using don't allow pin joint in the buckling analysis. I bracket the results of pin and rigid to select the member and design the joints.

 

[prex]

My point was not understood.
I'm not saying that the calculation with pinned joints is safer, because indeed is the contrary: if you calculate the bending stresses due to joint fixities you'll get higher total stresses and then you'll estimate a smaller safety factor for the structure.
Nor am I saying that the error in taking pinned joints in place of fixed ones gives a good or bad approximation of real stresses: of course the difference is known to be quite small in most cases, but there may rare occasions where this difference is not so small (particularly when the trusses are not very slender).
What I am really saying is that the increase of stress due to joint fixities has no effect on structure's failure (at least up to a quite large limit), so that if the engineer's goal is to evaluate as best as possible the factor of safety with respect to failure (and indeed I believe this is often the case, especially for important structures) those stresses should not be included in the analysis.



prex
motori@xcalcs.com

http://www.xcalcs.com

Online tools for structural design