Why assuming pinned connection is conservative? 1

[rpand4]

Hi Team,

An elementary question coming from a green engineer...

I have read on the forum multiple times where people have said it is conservative to assume that the connection is pinned as opposed to assuming it to be a moment connection.

In university they taught us the difference between pinned and fully fixed connection. Primary difference being one resists rotation and the other does not.

Questions

1) Why is it that a pinned connection is a more conservative approach though?

2) By assuming a pinned connection are we in essence relying more on the member itself to have the capacity to bear the moment? So in other words the member will need to have sufficient stiffness to withstand the moment?

3) If (2) above is true then could it be said that if the member was designed as having a fully fixed support than the connection would bear some of the moment and some would be worn by the member? Which means the member size would be smaller then (2) above?

If anyone could help shed some light on the above questions, it will help nudge me in the right direction to understand how connections work better.

Thank you very much.

 

[phamENG]

A pinned connection is generally more conservative for the beam. Look at a single span beam. If you have a real pin at each end, what is the maximum moment in the beam? M[sub]1[/sub]=wl[sup]2[/sup]/8. If you have the same beam but fully fixed ends (infinite rotational stiffness), what's the maximum moment? M[sub]2[/sub]=wl[sup]2[/sup]/12. As you can see, M[sub]1[/sub]>M[sub]2[/sub]. So for a given beam span, a beam sized to have no fixity at its ends is going to have a stronger and stiffer cross section than one designed assuming full fixity.

Notice that I stressed that it's conservative for the beam. Shear connections (which we model as theoretical pins) do have some very modest moment capacity (unless it's a real pin). So if you design a super slender column, the service level moment that is incidentally transferred from the beam to that column could cause problems. It's pretty rare, IN PROJECTS I USUALLY WORK ONbut can be an issue. EDIT: Most steel projects I work on are only one or two floors, so by the time I satisfy constructibility for most connections, my column DCR is in the neighborhood of 0.2. So that moment that gets transferred into it is usually inconsequential for me. If you work on larger structures, or in certain high demand areas of small structures, you need to pay close attention to stiffness and actual load path as the other posts below have pointed out.

There's another thread going on essentially the same issue with some good insights: thread507-492155

 

[Strucyou]

Hi, as pointed out by phamENG, the pinned connection will give higher moments in beams so higher resisting stiffness will be required. To add another perspective in case of lateral loads on the whole structure with pinned connections displacements of the overall structure shall increase. To control these higher deflections you will again need stiffer sections thus giving you a conservative design.

 

[centondollar]

1) It is not necessarily "more conservative" in general. You need to evaluate the rotational stiffness that a joint provides. If you have a steel beam with flanges and web welded to the column, and possibly even a flange-to-column L-plate with bolts, the stiffness is likely to be "infinite" - you can of course calculate it and after some large stiffness is determined, standards will tell you if that joint is classified as "rigid" (infinite stiffness) or semi-rigid (finite rotaational stiffness). If the flanges are not connected and only a shear tab connects the web to the column, the connection is pinned - there is no resistance to rotation.

The importance of accurately gauging joint stiffnesses becomes increasingly important when doing frame analysis. If you assume pins and make stiff joints, or vice versa, the structure will not behave as assumed in the calculation model. The moral of the story is: calculate internal forces and frame stability using the boundary condition that will actually be built by the contractor.

2) Force follows stiffness. If you restrict rotation at the ends of a beam, that stiffness against rotation will attract force and moment, thus reducing the mid-span moment and causing some moment at the boundaries. You may observe this by browsing your favorite beam bending formula collection.

3) If you can provide that fixity in the actual construction, the member will be smaller. Do note that making moment connections causes the behavior of the frame to change from "braced frame" to "moment frame", and that moment connections are more expensive than pinned connections.

One final note. In steel construction, connections cost a lot, and moment connections cost more, and complicated moment connections cost a lot more. It may very well be cheaper to purchase slightly deeper beams than to design moment connections and use smaller beams.

 

[human909]

It is not necessarily "more conservative" in general.

This cannot be emphasised enough.

Likewise this:

The moral of the story is: calculate internal forces and frame stability using the boundary condition that will actually be built by the contractor.

It hasn't been uncommon for me to see in the built world and likewise in discussion here, connections that differ significantly from the modelled connection.

I once had a discussion with an senior engineer on site regarding this while we were staring at a column with a noticeable bend to it. I asked him what the connection had been design like to which he replied it is a pinned connection. I cheekily replied, "does that look like a pinned connection?" pointing to the large end plate on the end of a slender HSS column that bolted to the flange of a continuous rafter beam. In his model the moment no moment would have been transferred to the column from the beam. In reality significant moment was transferred just under dead loads.


In practice, I design most of my steel connections as pins. If I have any concerns then I check the behaviour of the connection if it was infinitely rigid or occasionally semi rigid. Most of my connections are shear tab connections, which are not as flexible as angle cleats but are still sufficiently flexible to satisfy the pinned assumption in most cases.

 

[Enable]

As phamENG mentioned the other thread going on has some legs and is getting great responses. So worthwhile to check it out.

I feel you have the gist of it down. In general if you look at any individual member, pinned end-restraints typically result in the greatest moment for that section for the reasons phamENG indicated (nothing to "share" those moments with). Hence it generally is a conservative assumption for the member if analyzed in isolation. However, for most of the structures we design the loads themselves will be attracted to the more rigid components (read: the things that dont want to move very much). Here is where the assumption can backfire and be rather unconservative. For example, if we analyzed a structure assuming a load path that went through nothing but pinned connections, but in reality those connections actually had significant moment capacity due geometry or or design or whatever, certain members would see more load than we designed for and may become overloaded as a result.

No matter what you do unmodeled rigidity will attract load to spots you did not intend. Such is the world we design in since nothing is truly one or the other (fixed or pinned). Actually this no doubt saves our bacon when we fuck up! Alternative load paths (intended or not) can help us. They can also hurt though. This is partially why we stay under certain utilization thresholds, as this allows some load to go through paths we didn't exactly identify without any real negative consequence (at least that we are aware of. And if no one is aware of it you cant be sued for it ).

It hasn't been uncommon for me to see in the built world and likewise in discussion here, connections that differ significantly from the modelled connection.

I once had a discussion with an senior engineer on site regarding this while we were staring at a column with a noticeable bend to it. I asked him what the connection had been design like to which he replied it is a pinned connection. I cheekily replied, "does that look like a pinned connection?" pointing to the large end plate on the end of a slender HSS column that bolted to the flange of a continuous rafter beam. In his model the moment no moment would have been transferred to the column from the beam. In reality significant moment was transferred just under dead loads.


In practice, I design most of my steel connections as pins. If I have any concerns then I check the behaviour of the connection if it was infinitely rigid or occasionally semi rigid. Most of my connections are shear tab connections, which are not as flexible as angle cleats but are still sufficiently flexible to satisfy the pinned assumption in most cases.

I have the same problem but from the contractors perspective. I'm often tasked with designing connections from drawings that provide very little in terms of connection loading, and assumed connection type. Even if I have loading (rarity), no mention of pinned vs moment vs whatever. And I always have to send an RFI because for me to design my connections I need to know the bloody load path the structural designer assumed. Sometimes fully welded connections (shop or field) are easier for me since we do small scale stuff where bolting is just an annoyance but I dont want to fuck it up by making a certain unintended path too rigid.

That is is the restoration world though where designers are not the best. New construction tends to be a bit better with loads, rigidity requirements, etc.

 

[human909]

I have the same problem but from the contractors perspective. I'm often tasked with designing connections from drawings that provide very little in terms of connection loading, and assumed connection type.

It must be tough living in a world where the contractor is responsible for connection design. Here in AUS it would be normal for the engineer to provide full connection details. On smaller projects where there is good trust between engineer and contractor you might not bother provided connection details, but this is not typical of most steel fabrication.

In my anecdote above it was early in my career and I was working for the contractor not as a structural engineer. We were responsible for the fabrication and erection, not for the connections. I neither had the experience nor the involvement in the predesign to challenge the connection before it was fabricated and erected. But it was blindly obvious once built what was going on. (At least to anybody who lives outside of a idealistic model!)