Moments about a line weld with S=0? 2

[JCWilson]

Does anyone know of any documented statements regarding the loading of a line weld with a transverse moment about its weak axis? I need to convince a fabricator that their details are not acceptable, but I don't want to argue this alone. A statement from a credible source would greatly support my case. I have pondered Blodgett, AISC, AWS and Salmon and Johnson, but haven't come across anything convincing yet. I suspect that this condition is so unacceptable, that it isn't even worthy of attention in textbooks.

To elaborate, a flat plate is being welded flat to another steel plate, but only along one edge. The welds aren't even returned around the corners. The plate is then being pryed off the base plate by pulling from the end opposite the weld. This places a direct moment about the weakest axis of the weld.

 

[desertfox]

Hi JCWilson

I agree with you entirely, the weld position you describe is not satisfactory.I like you cannot find a case to cover your situation for bending of that welded joint, however looking in "Schaum's Outline Series Machine Design" they treat welds just as lines and if the contractor welded round the corners as you put it, to make an inverted 'u'section as shown:-



-----------------
| |
| |
| |


Zw= 2*b*d + d^2/(3)(for the top of the weld)

and Zw=d^2*(2*b + d)/(3*(b+d) (for the bottom of the vertical legs)

where Zw=section modulus
b= horizontal weld line
d= vertical weld lines
If you divide Zw by your bending moment it will give you the
force per unit length of weld,from which you can workout your stresses in the usual way.If these stresses prove to be okay you can ask the fabricator to weld round the joint to improve the design.

hope this is of some help

 

[GRoberts]

I think I rember AWS handbook, current edition chapter 5 on weldment design. Never putting the root of a fillet weld in tension. I also agree. The situation, if I understand it correctly is terribly nieve on the part of the fabricator.

 

[vonlueke]

JCWilson: Like you, I didn't find a specific citation yet, though I don't have AWS at the moment. In your described scenario, i.e., having no corner returns, the height of the rectangular cross section in bending will be the weld width. This width will either be (1) the weld effective throat t_e, or (2) if you're certain it fails in the adjacent base metal heat affected zone (haz), fillet weld size s. Always assuming t_e envelops either scenario, which would be the way to go (for fillet or groove weld). A side view of your scenario is shown below, where the plus symbol (+) indicates an end view of your line of weld in bending.

  ^ P
  |
  |<---- L ---->|<-- a -->|
  |
  --------------+         |/
         -----------------|/
                          |/

Thus, I = (b*t_e^3)/12, where b = length of weld line (same as also shown in desertfox's post), assumed equal to top and bottom plate width. Therefore sigma = Mc/I = 6*P*L/(b*t_e^2). Stress level R1 = sigma/S_{ta_w}, where S_{ta_w} = weld allowable tensile stress or base metal haz allowable tensile stress, whichever is less. If R1 > 100%, it indicates the weldment is overstressed.

Also, checking bending stress at fixed support, sigma = 6*P*(L+a)/(b*t₂^2), where t₂ = thickness of bottom plate. Stress level R2 = sigma/S_ta, where S_ta = allowable non-haz tensile stress of bottom plate material. If R2 > 100%, bottom plate is overstressed at fixed support. Note, if R2 > R1, failure mode would be bottom plate bending upward at fixed support prior to (or instead of) weldment failure.

No need to check shear strength in the above problem because bending stress greatly dominates, for both R1 and R2 failure modes.

Note that if the sense of P were reversed, you would of course have a completely different scenario and analysis, involving shear stress on the weld effective throat; you would have a fairly strong connection (though the analysis for R2 would remain the same).

 

[JCWilson]

Vonlueke,

Thanks for drawing the picture, you interpreted what I said just right. But to make matters more troubling, the plate of width L is continuous into the page. Point '+' is being assumed as a pinned connection with a stitch weld of about 1 1/2" weld at 16" o.c. The force P is pulling up at a point which is fixed against rotation.

The plate at the location of P has been designed and sized for M = P x L. The effect of only 1 1/2" of base metal per 16" acting to resist bending at the other end of the plate (point '+') is not even considered (because it is assumed pinned.) For magnitude, distance L is typically 1"-3", P is 200-500 pounds, and the plate is 1/8" to 3/16".

I am hoping to find some "smoking gun" documentation on weld design to dispute the comment that this has worked (and been stamped approved) for years, so why should it be changed now.

^ P
|
|<---- L ---->|
/- |
| --------------+
\- > -----
M /////

 

[vonlueke]

The assumption of a pinned support at the weld provides a conservative approach for design of the slider support connection at the left-hand end of your diagram. So this can be considered an acceptable approach for design of the left-hand end connection.

On the right-hand end of your diagram, at the weld, it's of course not pinned. I assume your given load P is the total load every 16 inches, o.c. According to your diagram, you have (or need) a completely fixed support at the weld; therefore the continuous problem can thus be modeled as a beam strip for the purpose of roughly approximating the maximum bending moment on the right-hand end.

Checking a beam table for a beam having one fixed support and one slider support constrained against rotation, it appears bending moment on the weld every 16 inches would be M = 0.50*P*L. Bending stress on weld throat would thus be sigma = Mc/I = 6*M/(b*t_e^2), where b = total weld length every 16 inches. In your stated problem, b = 1.5 inch, so for P = 500 lbf, L = 3 inch, and t_e = 0.125 inch, sigma = 192 ksi.

This would indicate the weldment is overstressed. I don't know why it worked in the past, unless the fixed support drawn on the right-hand end of your diagram is not truly a fixed support. E.g., if the bottom plate has significant flexibility, its deflection and rotation at the weld would reduce M. If this is the case, you would probably need to include the material and dimensions back to the true fixed support and solve the statically indeterminate beam problem to get a rough approximation of the actual M value at the weld. Are you sure an essentially fixed support exists at the weld, as shown in your diagram?

 

[JCWilson]

Vonlueke,

That is the $10 question of the day. If the weld is so underdesigned that it has minimal bending capacity ( or that it only carries the shearing load, P ) then it becomes so flexible that it is essentially a pinned connection. Hence, therein lies the oxymoron, and the short-sighted arguement that the connection works. That and the fact that there have not been any catastrophic failures as of yet to this product.

I agree with your analysis, and it reinforces the approach I would take, should this perhaps be acceptable. But I want to believe that there is some cardinal law of engineering/welding that says &quot;Thou shalt not permit rotation about the centerline of a weld, nor allow any bending stresses to be applied to the root of a weld.&quot;

 

[vonlueke]

Hopefully some weld experts who have exhaustive references handy will look for that cardinal law in the design codes. While waiting for that help, in the meantime, what is the length a, width b2 (if not continuous), and thickness t2 of your bottom plate? And what is its configuration, if not similar to my previous diagram and if not too complicated to describe? And what material(s) if not typical mild steel?

 

[JCWilson]

The &quot;base metal&quot; is actually 16 gage (.054 inch) steel studs (50ksi) at 16&quot; on center. And the continuous plate being welded to the flanges of the studs is an in-place 8 gage (3/16&quot; typical) cantilevered floor angle/pour stop. The upturned angle of the pour stop is being welded to the back of the flanges to support the dead load and wind load of a section of wall where it cannot otherwise bear on something else.

So as soon as the wall is loaded with insulation and finish material, and well before the wind begins to blow, the support plate has already begun to rotate due to these forces. One worst case scenario I just analyzed has a 1/2&quot; deflection of the wall due to sagging/rotation of this plate. The analysis assumes that the weld to stud is a &quot;pinned&quot; connection.

That should leave you with no question as to why I can't sleep at night. This method has been used simply because it is quick and dirty and cheap. We have alternates to this design, but they are relatively more expensive and time consuming. The more I ask around, the more I hear from people that don't like it, and don't understand it. So I just want all the ammo prepared to eliminate it all together.

 

[vonlueke]

I assume those studs have a C cross section. (1) What are the stud cross-sectional dimensions? (2) I assume the weld runs across the full width of the C cross section flange; and I assume this fillet weld is made at the top of the pour stop plate upturned leg and not below this leg. Correct? (3) What is the height of the pour stop upturned leg? (4) Are you certain the pour stop plate remains in the elastic range (will spring back undeformed if the load, P = 500 lbf, is released)? Or has permanent deformation already occurred in the pour stop plate for your observed 0.50 inch deflection? (5) What's the pour stop plate material? (6) Does the stud flange have a very obvious region of local deformation near the weld when the 0.50 inch deflection occurs?

 

[RG88]

I have watched this thread for a while and have found similiar results as the others, but no 'smoking gun'.

A few thoughts that might help substantiate your suspicion when presenting your case to the detailer:

AISC Specification 1.17.6 Lap Joints
'The minimum amount of lap on lap joints shall be 5 times the thickness of the thinner part joined, but not less than 1&quot;. Lap joints joining plates or bars subjected to axial stress shall be fillet welded along the end of both lapped parts, except where the deflection of the lapped parts is sufficiently restrained to prevent opening of the joint under maximum loading.'

From this I intrepret that it is not permissible to allow the root of the fillet weld to open. In your case, I agree with the others, on that the root will undergo tension due to bending in a fashion that will cause the joint to open.

AISC Specification 1.17.7 End Return of Fillet Welds
'Side or end fillet welds terminating at ends or sides, respectively, of parts or members shall, wherever practicable, be returned contintuously around the corners for a distance not less than 2 times the nominal size of the weld. This provision shall apply to side and top fillet welds connecting brackets, beam seats and similiar connections, on the plane about which bending moments are computed.'

This may not help in your case with the 'infinite' length of line weld. However, since it may not be 'practible' you could ask that the lap be puddled welded from above in addition to the detailed weld. This will keep the weld from opening. I imagine the detailer's case is founded on ease of construction. For example,when laying decking for a concrete floor, it is easier to weld from above rather than below.

Blodgett's Design of Welded Structures, chapter 7 Joint Design and Production may prove somewhat beneficial in proving your case, but it does not specifically state that you cannot place a weld into bending.

One other point, the typical 'Properties of Weld Treated as Line Table' seen in machine design text books does not account for bending stresses in welds. These tables convert bending stress into shear or transversly applied tensile stress on the weld on a per unit length basis. This table can be found in the aforementioned reference to Blodgett.

FYI the Blodgett reference can be purchased from Lincoln Arc Welding for $15.00. You can order from their web site or maybe send them an e-mail to check with the experts. Book stores are much more expensive than the Lincoln company store.

My AISC references where taken from the Manual of Steel Construction, 8 ed. I know it is old, but I left my current edition at work. I don't think these provision have changed or become unfounded in the time between versions.

 

[RG88]

A few other things that I forgot...

Blodgett's Design of Weldments 2.6-6 Load and Stress Analysis, Analysis of Bending, Rule #4 Place Joints in Low Stress Areas. It further illustrates an example of placing groove butt welds at a point of zero moment to avoid flexing problems. Again not a 'smoking gun' but another reference to not place a single weld in bending about the longitudinal axis of the weld.

The same reference, 3.2 Designing for Fatigue Loads, states that fillet welds and lap joints have a lower fatigue strength due to geometry. He also states again to 'avoid placing a weld in an area which flexes'

A common sense approach,

Watch a welder weld pipe. You cannot clamp your ground to the round pipe, so you fillet weld a piece of bar stock to the pipe to provide a clamp point. You use a one sided fillet weld to make it easier to remove by bending towards the weld.

Best of luck with this. The situation you describe disturbs me. Especially since this single weld (according to the drawing) appears to be taking the largest moment along this 'beam'.

 

[GRoberts]

Since it seems to be so hard to find, I finally took the time to find what I had referred to in the AWS handbook. This is what it says, and I hope what you are looking for:

&quot;Single fillet welds:
Single fillet welds are limited to low loads. Bending moments that result in tension stresses in the root of a fillet weld should not be permitted because of the notch condition. For this reason, single fillet welds should not be used with lap joints that can rotate under load. The welds should not be subjected to impact loads. When used with fatigue loading, the allowable stress range must be subject to stringent limitations.&quot;

AWS Handbook Volume 1, welding technology, eigth edition.

 

[JCWilson]

Thanks for all of the concerned responses. I have since found that RG88's reference to AISC Specification 1.17.6 Lap Joints is the same as AISC 9th Edition ASD, Specification J2.b, Lap Joints.

Combining this note with the one from the AWS handboook, I am satisfied enough to take this one to task.