Torsional Stiffness of Wide Flange Column with Endplate Moment Conn

[connectegr]

We are designing endplate moment connections for given shear, strong-axis moment, and weak-axis moment. The connections are to the column flange. The structure is simply a fixed frame of girders and columns, with no horizontal bracing or diaphragm. Our concern is that the column flange connections will not provide adequate torsional stiffness to be considered a "fixed" support, with regard to the weak-axis moment. We are designing per the AISC Design Guides for endplate moment connections. The only addition is the weak-axis moment which we have incorporated in the analysis as an addition to the bolt tension. However, the design guides are intended to address the most common design situation. The method does not consider the torsional stiffness of the column to resist the weak-axis moment. If the column is not adequately stiff (or flexible) the connection will perform as a pinned support and change the design requirements of the beam, increasing the beam's weak-axis moment. A wide flange column is not the ideal torsional cross-section. The torsion is resisted as shear couple forces in the flanges of the column. However, in this case only one flange is loaded. The EOR's model views this as a node with given properties. It does not recognize the shape of the member or connection.

I know that I can simple add stiffener plates. But, the column is not made of jello, and there is some torsional stiffness in the column flange and web. Is there any procedure for determining the effective stress distribution and the stiffness of the column? Examples and references are appreciated.

Thanks in advance. (obviously there is no time for lab tests and finite element models)

http://www.FerrellEngineering.com

 

[weab]

Without running my own analysis, I'd say that what you are proposing will not provide the intended results. Without stiffeners, the column flange will flex about the web at the root. With stiffeners, the whole column will twist. The column can likely take the torque but will twist to the point of providing only a pinned connection for beam out of plane bending.

I would not consider the column connection torsionally restrained unless you provide a closed section for the column. Even then, any column if tall enough could twist some amount. I don't think there is much chance that the column will perform as you hope, but perhaps I am wrong.

If you would like to try, Blodgett (Design of Welded Structures) provides a method of determining the twist in an open section. You could determined the beam weak axis reaction moment, apply it to the column as torque, and determine if the column twist. If it closely matches the beam end rotation if it was pinned, then your assumed stiffness is not valid.

Good luck.

 

[connectegr]

wean
The model considers the column torsionally adequate. But simply identifies a fixed node.

I will check the Blogett book tomorrow. I may call Dr. Blogett for his thoughts.

http://www.FerrellEngineering.com

 

[BAretired]

I agree that the column is not torsionally "fixed". It will rotate a calculable amount when subjected to a torsional moment. The magnitude of rotation is dependent on the cross sectional dimensions and the height of the column.

If you are relying on the torsional stiffness of the column to restrain the weak axis bending in the beam, you would need to perform a strain compatibility check. Otherwise, you would be safe in assuming a pinned connection at the cg of the column.

BA

 

[paddingtongreen]

connectegr, as I read your post, it is not the torsional stiffness of the column itself that worries you but whether the load applied to only one flange is transmitted to the whole column.

Assuming it is the only horizontal member. If EOR's model assumes that it's length extends to the column centerline, I wouldn't worry too much. The beam has stiffness limited by it's Iyy, but the column can use a length, greater than the depth of the beam to replace that Iyy for half the column depth. Of course, if the EOR modeled the joint with offsets to the flanges then stiffeners are required for the infinite stiffness of the offset.

Michael.
Timing has a lot to do with the outcome of a rain dance.

 

[nutte]

You could do the following:

Calculate the stress in the flange from the weak-axis moment.
Multiply this stress by the flange area, resulting in an equivalent axial force from the weak axis moment.
Add this force to your other flange force from strong axis moment.
Check for local flange bending using this total flange force.

That won't tell you how much the flange deflects, but it would tell you if the flange is stiff enough to resist the force. It would be conservative, since the peak stress at the flange due to My is applied across the entire flange area. If your My is big compared to Mx, it may be prohibitively conservative.

 

[dhengr]

Connectegr:
Given the welding and fit-up problems associated with stiffeners, I wonder if two side plates, from flange tip to flange tip, on the col., running some distance above and below your beam might not be the answer. This gives you a better torsional shape to accept the Iyy loading from the beam and into the column, without the single flg. bending, twisting issues. Then it becomes a more typical torsional problem on the col. and compatibility must exist as BA suggests. In fact, the torsional stiffness of the col., and the Iyy of the beam, lead to the col. rotational angle, and a semi- rigid torsional moment on the col. and a like end moment in the beam.

Now, all you need is some welded nuts on the boxed section to complete your detail, because they will be a bit more difficult to get at. No?