Beam offsets

[dozer]

Before I pose my question let me say that this is regarding frame analysis programs like STAAD, RISA, SAP 2000 etc. I'm using STAAD and I've always had problems with offsets. Let's say you have a beam framing into a column and you want to know the moment where the beam physically ties into the column. Say it's a 24" column so you could do one of two things. One, you could look at the moment 12" (half the column depth) away from the end of the beam. Every program I've used allows you to do this. The other option is you could put a 12" offset in the beam then the moment reported at the end of the beam is actually at the physical end of the beam.

In my mind either method should give you the same results. In STAAD it does not. I've used SAP 2000 a couple of years ago and the best I recall is it would give you the same answer. As a matter of fact, I believe you could tell it to treat the "dummy" part of the member as having the same stiffness as the rest of the member.

What is the "math" or theory or whatever going on with offsets? When used as I describe, I think (and as I've said I've used other programs that reflect this attitude) it would be to get the moment at a certain distance from the end of the beam rather than the moment at the node assuming the same stiffnes from node to node. When your working on an offshore rig with 60" columns this can be significant.

In STAAD however it usually hurts you. I think it's because they have a different philosophy about what's going on. They even refer to the "additional stiffness" of the column. Some of the increases in moment I've seen though just doesn't seem right. I can think of cases where saying that the distance from the node to the face of column are stiffer than the beam can't be right. For instance suppose the column is tubular section. I would think local bending of the wall would actually make it less stiff.

What are other programs doing in this case?

 

[transmissiontowers]

I don't use STAAD but use GTStrudl for frame analysis. It allows you to model the beams as you want with a joint size command.

_____________________________________
I have been called "A storehouse of worthless information" many times.

 

[dozer]

Trans, I'm not sure what you mean by "as you want".

 

[bkal]

I am not a Staad user, but I would imagine that using offset means that the part of the beam (in this case) between the centreline of the column and its face is perfectly rigid. If that is the case, it would explain why the results of reading the moment at that point with usual approach is different from using the offset.

By applying the offset, you are effectively changing stifnesses of the elements (i.e. columns and beams) by different amounts, and thus the load distribution is different.

Another explanation for the difference might be (probably unlkely) a treatment of distributed loads in the offset part.

Hope this makes sense.

 

[dozer]

Bkal, I think your first suggestion is correct. As a matter of fact I'm quite sure it is. Instead of using an offset I put in a very stiff member to simulate the offset. I got nearly identical results. OK, so that part of the mystery is solved.

Now for the more important question, is this the correct way to do it? I have my doubts. As I already mentioned, think about tubular sections framing into each other at right angles. There's no way you can convince me that from the center of the tube to the outside is "infinitely" stiff.

I think we have established what STAAD is doing but I would still like to know how other programs treat this. Keep those cards and letters coming.

 

[transmissiontowers]

The "as you want" meant the with the joint size command in GTS the user can have the beam stop at the column flange as you wanted it to do. I believe that GTS puts a rigid link between the column node and the end of the beam.

AFA the tube column example, the infinitely rigid idea is appropriate for a tube as it is with a WF. You are modeling the column as a line down the center. The beam attaches at the outside of the tube. You assume the tube acts as a unit and does not deform at the connection since the connection between the tube beam and column tube must be designed that way. Any decent program will analyze and design the members in a frame. The connections between the members is where the real fun begins. If you had assumed a full rigid connection between the 2, the connection must be designed that way. If you assumed a pinned connection in one axis and rigid in the other, you must design the connection that way.

Now if you have 2 round tubes welded together like in an undersea oil rig, you might consider a real FE model of plate shell elements to see what the stresses are at the junction.

_____________________________________
I have been called "A storehouse of worthless information" many times.