Tension on top anchor at welded shear tab connection

[akamike]

We have an embed plate in a concrete wall. A shear plate is welded perpendicular to the embed plate and a beam connects to the shear plate. If the beam were bolted to the shear plate, the tension in the top anchors would be calculated by multiplying the gravity rxn by approx 3" (dist from face of plate to bolt line), then dividing by depth to bottom bolt. However, if we weld the shear plate to the beam rather than bolt the beam, have we increased or descreased the tension in the top anchors?

Two schools of thought:

1. The beam and plate welded have "become one" (the weld resolves the moment)and now the gravity rxn is at the end of the shear plate, basically at the face of the embed plate. This creates very little tension on the top anchor.

2. By welding you've moved away from the pinned connection to a moment connection which introduces a higher moment into the top anchors.

These are opposite schools of thought. Either look correct? Neither? why? Thanks.

 

[frv]

I think it would depend on the stiffness of the shear plate.

If you weld a relatively small shear plate that does not extend the full depth of the web member, then you allow the beam to rotate at the flanges and you will have a connection approaching a "pinned" condition.

If your shear plate is large and extends the full depth, then you have added significant rotational restraint and the connection could be closer to a moment connection (certainly not a full moment connection, as you will almost always need to weld the flanges for that)

 

[Lion06]

The first one is more accurate, in my opinion. It's all about stiffness. The welds are inherently much stiffer than bolts. Welding to the embed plate and bolting to the beam makes the shear tab act like a cantilever and the bolts as a pin.
When you weld the shear tab to the beam, the tab (as long as the plate/weld can take the moment) act as one beam (as you noted) and deliver mostly shear to the embed plate. The reason I don't believe it will increase the tension in the top anchor is that the elastic negative moment at the support can't be supported..... the plate capacity is small compared to the beam so it will act simple. That being said, there is some stiffness there, so some moment (and tension in the top anchor) will be there, but certainly less than with bolts.

 

[csd72]

Only a full analysis could tell you what the effect on the moment is.

 

[Teguci]

My strategy would be to limit the failure mechanism.

You do not want failure at the weld or in the embedded anchors.

Therefore, check your plastic bending in your shear plate and make sure your weld and embedded anchors can take that maximum moment.

If this works then you have a plastic hinge at the shear plate and the moment on the weld and embedded anchors will not exceed your plastic bending.

 

[Lion06]

I think some people are missing the actual question. The question was only about relative tension in the top anchors when the shear tab is welded to the beam as opposed to bolted to the beam.
I took this to mean that for a given shear tab with a given embed plate and a given beam with given loading.... will the top anchors in the embed plate have more or less tension in them if you weld the shear tab to the beam as opposed to bolting it.
I don't see how the answer can be anything other than less tension.

 

[frv]

StrEIT-

All things being equal, you are right. I agree that for the same shear plate, a welded condition would result in a smaller tensile force in the bolt at the top.

 

[UcfSE]

If you design the bolts or welds to the beam for the eccentricity, then you do not also have to design the plate for the eccentricity from the face of the plate to the bolt line or weld line. If the bolts or welds to the beam are not designed for the eccentricity, then your embed plate must be designed for it.

 

[Lion06]

I've always learned that welds are inherently much more stiff than bolts (unless you provide a slip critical connection) and as such, if you provide bolts to the beam, the shear tab will act as a cantilever regardless of whether the bolts can take the moment or not because they will allow more rotation than the welds before resisting the moment.
Is this an inaccurate statement?

 

[Teguci]

The end rotation of the connecting beam must occur in one of the following locations:
1. At the weld - unzipping and Failure
2. At the shear plate - Plastic hinging
3. At the Embedded plate to concrete interface - yielding of your headed anchors or potential concrete rupture (not recommended)
4. At the concrete itself - concrete is usually fairly stiff so this is probably not the case

I would not recommend allowing the rotation at the steel to concrete interface.

 

[WillisV]

Agree with UcfSE.

 

[271828]

NO time to read the other posts yet, but don't weld the tab to the beam. Shear tab rotational ductility is achieved by limiting the plate thickness to allow the bolts to deform the holes somewhat. If you weld it, you've eliminated the rotational ductility of the connection. Shear connection rotational ductility is not optional.

 

[csd72]

I agree with structural EIT,

The panel will be stiff compared to the beam so the connection to the panel should be designed as a stiff cantilever to support the reaction at the centerline of the bolts.

A single plate shear connector to a cross beam is a different matter as the cross beam has very little resistance to rotation from the connection.

As I stated above, the only way to check this if it is fully welded is to analyse it as a frame. Anyone who tells you otherwise is oversimplifying the issue.

the corner moment in a portal frame depends on the relative stiffness of the two members and will lie somewhere between close to zero and fully fixed depending on this ratio.

 

[271828]

The main issue here is detailing for ductility, not whether some bolt has more or less force (gotta get a reasonable value for that too, obviously). Check out the text starting on Page 9-13 of the 13th Ed. Manual. However you design and detail your connection, you must account for this.

As was seen in the Mexico City EQ a few years ago, this is not optional. Several bldgs collapsed because of shear connections with too much restraint and too little ductility.

I asked Tom Murray about this 3-4 years ago and he said he thought using shear tabs at weld plates like this one is an awful idea. The moment is hardly predictable and there are other connections. He actually recommended something like a tee welded as shown in Fig. 10-15(b). I think a single angle or double angle, welded correctly as shown in the Manual (i.e. NOT welded all around to the plate) would be fine also.

If you insist on using a shear tab, then I think your first idea:

"If the beam were bolted to the shear plate, the tension in the top anchors would be calculated by multiplying the gravity rxn by approx 3" (dist from face of plate to bolt line), then dividing by depth to bottom bolt."

is the way to go except that the anchor tension should be detrmined using the approach illustrated in Fig. 7-6. Be sure to adjust beff by Econcrete/Esteel because your plate is bearing on concrete instead of steel.

The shear tab and weld should conform to the dimensional rules given starting on Page 10-101 to ensure that the bolts can plow into the plate a little to allow the simply-supported beam rotation to develop without too much moment being generated.

 

[civilperson]

Fixed End Moment can be twice the magnitude of center of span moment if the connection to the concrete wall has equal stiffness as interior beam. The description of welded shear plate attaching to the web of the beam is more closely modeled as a pinned connection with the maximum moment transfered to the wall embed limited to the stiffness of the shear tab or the bolted connection.

 

[haynewp]

I haven't read through all these, but I would use horizontally slotted holes in the shear tab, and then design the embed for 3" eccentricity. Or else a flexible angle connection to the embed.

If you weld a shear plate all around to the end of the beam, then the embed will still receive some moment due to it acting as a curvature restraint. I would still design the beam itself for pin-pin in this case also.