Failure analysis: how to calculate eccentric weldment 1

[kingnero]

We've got an accident at work where someone got the top part of a pole on his leg, due to a weld that failed in action.
Obviously, the weld was too small, and placed on the worst possible side.

The calcs by the manufacturer did not account for eccentricity. I'm not sure how to do this, I haven't come accross this before.

see sketch: the weld sees a shear force "F", which is easily calculated.
besides that, there is a moment induced , magnitude F x e, which placed the root of the weld in tension.

The standard formula for combined stress accounts for stress that is uniform over the cross section of the weld (perpendicular tensile and shear stress),
this moment gives tension at the root and compression at the outside of the weld.

any tips please?

 

[weldstan]

The weld is on the wrong side. Do not place the root of the fillet weld in tension/bending! Or weld on both sides.

 

[Tmoose]

Is there other equipment with that weld config?

What is the value of "F" ?


I hope that everyone is alright. But I am reminded of something Mark Twain said as the speaker at an insurance company dinner.
(He allegedly was a director for the insurance company at the time)

http://www.homersmith.com/homersmithblog/?p=425

"There is nothing more beneficent than accident insurance. I have seen an entire family lifted out of poverty and into affluence by the simple boon of a broken leg. I have had people come to me on crutches, with tears in their eyes, to bless this beneficent institution. In all my experience of life, I have seen nothing so seraphic as the look that comes into a freshly mutilated man’s face when he feels in his vest pocket with his remaining hand and finds his accident ticket all right. And I have seen nothing so sad as the look that came into another splintered customer’s face when he found he couldn’t collect on a wooden leg."

 

[dhengr]

Kingnero:
As mentioned above, the root of that weld should not have been put in tension, by bending or shear. A weld on both sides of the left piece would have gone a long way toward solving the problem. That is a crappy weld detail to start with, you just have no way of knowing that you really get good penetration and fusion, without any imperfections or cracks at the root of a weld like that. This type of weld and load condition causes ripping (tension) across the root of the weld and almost any force can lead to a very high stress concentrations (tension) perpendicular to the line of the weld. It’s the stress at the tip of a crack problem, and at the tip the stress is infinite becuase of the small tip radius, its’s a fracture mechanics problem. It’s not something (a combined stress) you calc. every day, on every design problem, you just know it is something you should not do because of the potential weld cracking consequences.

 

[Guest102023]

Hold on folks, this was an industrial accident that led to a personal injury. In general, failure analysis should not be attempted by crowd-sourcing, and in this case the applicable authorities and insurers should be the first ones contacted. Any failure investigation must be done by a third-party accredited firm.

"If you don't have time to do the job right the first time, when are you going to find time to repair it?"

 

[weldstan]

kingnero,
I apologize for not reading fully your knowledge of the weld being improperly placed. You may not be able to fully calculate the stress imposed by the load due to the residual weld stress and the stress intensity at the root of the weld. You may need to duplicate the weld and measure the load to failure. You will also need to examine the fracture surface for lack of penetration/fusion and precrack size, etc.

 

[wannabeSE]

To get a feel for the stresses, you might calculate the section modulus at the weld (as if it were a plate with the thickness equal to the effective weld size) and then calculate the bending stress in the weld. I would never use this method for design!

 

[Duwe6]

Please keep in mind that there was a stress concentration at the root of the fillet. Loaded area would have been about the same at the area of the root face. For a definitive 'read' on those stresses, use FEA with a very small grid. I believe that you will find that the calculated, concentrated stress will exceed the tensile strength of the filler metal. In any case, unless the actual failure was centerline-of-weld, the limiting strength most likely have been the HAZ of one of the basemetals. Analysis of the fracture will tell which case it was.

weldstan got it right, weld is on the wrong side. Thus your approach of using the entire area of the weld is incorrect. The stress from the load had to pass through the stress-riser at the narrow root of this fillet.

brimstoner is correct too, get your insurance carrier to recommend an engineering firm to conduct a rigorous Failure Analysis. ASAP! These are the only folks I can recommend - my experience is limited:

http://www.wje.com/

 

[oldrunner]

Isn't this weld actually a flare weld? Also the AISC manual has some comments on this type of weld and try to discourage it. AWS D1.1 also may consider this as a prohibited weld. For investigative purposes, probably analyzing this in NASTRAN would be good. Also the piece should be sectioned to see if there full fusion. Usually welds are suppose to work in bending - only shear.

 

[oldrunner]

Whoops! Welds are NOT suppose to work in bending.

 

[oldrunner]

Whoops! Welds are NOT suppose to work in bending.

 

[kingnero]

@ everybody, thanks for the replies. I'll answer/clarify some of them here:

@ Tmoose, yes there are multiple devices hence the need for the calcs. Being as they're infrastructure related, the real-life testing of their integrity isn't an option. As the guy in the meanwhile has lost his leg, you (and several other posters) can be sure that the authorities take this seriously. Proper firms are currently investigating this particular problem. The other devices are not really their problem, they're my problem now.

@brimstoner, I'm not crowdsurfing for failure analysis, I'm crowdsurfing as for how to tackle this problem. Getting input from people who mostly have some real-life experience. Preliminary calcs (using sigma = M/S) show individual stresses three and four times the allowable (actually, there is two times an eccentricity: one as shown in the attachment in the first post, the other due to the offset of the self weight by about 1 1/2 foot to the side of the hinge).

Being a welding engineer, I have done many calcs but never one of this kind (I don't use such details in my designs). Neither Blodgett or any of the other references at my disposal show something similar. But as I need an idea about the stresses involved, I came here. Some were Captain Obvious replies, others made it come across as I were responsible for this design. I am not. Just trying to clean up someone else's mess. First thing I did was preventing the other hinges from "opening". Now I need to make the contractor that fabricated those, replace or improve them. Never an easy task.

Thanks for the input everybody.

 

[weldstan]

kingnero,
Other questions to ask:
Did the designer correctly or incorrectly specify the weld location?
Did production misread the weld location?
Did the installer improperly locate the hinge?

 

[kingnero]

- he spec'ed the weld as shown in the attachment in my first post, and it was executed as drawn
- nope, they did what was shown in the WPS
- nope, everything was executed as (badly) planned.

Fault lies (according to me) with the design guy, that drew this up and calc'ed the whole thing, then never looked back at it as the combined stress was a ridiculously low number.

 

[weldstan]

Yep!!!
One wonders at the competency of some of our Engineers and I have seen many more than one poor weld designs leading to failure.
Note that in some countries, the weld marking for weld location is the opposite of AWS.

 

[kingnero]

I'm in Europe, everything was done using european/iso codes (with which I am well familiar), WPS and welder's qualification was done according to EU codes, drawing was as usual, really no room for misinterpreting.
This really boils down to bad understanding of load paths and simplifying both eccentricities, so that the calc simply was shear force over weld area.
As said above, when doing some simplified calculation, I get three to four times the yield stress of the material. Neither will be 100% accurate, but as the weld failed, it must've been higher than yield.

 

[Duwe6]

[waives wet finger in the air] 3 or 4 times the yield strength of the material sounds about right, a small load greatly concentrated into a tiny area. And thank you for posting this update to your original question. If there is anything further, we'd like to have you post that too, please.

 

[weldstan]

kingnero,
The weld most probably failed in shear which will be less than the Y.S.

 

[Guest102023]

weldstan,
I've resorted the words in your statement:
"Note the weld marking of AWS for weld location is the opposite of that in some countries."


"If you don't have time to do the job right the first time, when are you going to find time to repair it?"