residual stresses due to welding

[clemsonC]

I'm looking for a residual stress value to use for the analysis of an existing structure - a beam-column (Fy=33ksi) with plates (A36) welded to each flange. (The member was rolled, then, after 40 years of service, these flange plates were added by continuous fillet welds on each edge of the plate.)

Testing for stresses isn't an option.

I haven't found very clear guidance on the stresses caused by welding the flange plates.

Salmon and Johnston, p. 286, shows typical residual stress distributions for welded shapes, ranging from 12-40 ksi. The discussion is a little thin. Based on their info, I would use the 40 ksi tensile stress in the flange tips for my structure, but that is higher than yield, and, clearly, would render my structure inadequte.

Stability Design Criteria for Metal Structures by Galambos, p. 35, states that the max tensile residual stress at a weld is equal to or greater than the yield stress of the plates. It seems sensible to conclude that this MAY occur, but that such high stresses would be relieved, to an acceptable level (<<Fy) before the member is put into service.

So, both texts say that residual stresses due to welding can be about equal to yield, but that doesn't make sense, given the implications.

What about the good old Fr = 16.5 ksi, from Ch F in the silver LRFD manual?

Any comments/suggestions/clarifications would sure be appreciated!

 

[metengr]

Do not underestimate the effect of local yielding in steel structures. During hydrostatic testing, when actual stresses are a percentage of tensile yield, areas of the component that are subject to local stress concentration will indeed exceed the yield strength. The material locally yields and the stresses are re-distributed. Stresses are stresses regardless of their source.

 

[Goahead]

If thermal stress relieve (even with a local flame heat source, and progressive to cover the whole structure) cannot be performed, sometimes Vibratory Stress Relieve (VSR) may be performed by specialized firms with good results.

http://www.welding-advisers.com/

 

[bklauba]

A new VSR technology website might be of help:

http://airmatic.com/content/view/50/70

BK

 

[bklauba]

Sorry, I omitted that all important slash at the end!!

Try: http//airmatic.com/content/view/50/70/

BK

 

[stanweld]

Weld residual stress can indeed be greater than the yield strength of the base material. Welding residual stress is often not taken into account in statically loaded systems due to local yielding with resultant stress redistribution when the material is ductile. Residual stresses truly become additive in cyclic service (fatigue)and brittle fracture. The Welding Research Council has finite element analysis profiles of residual stress distribution in groove and fillet welds. and you might use them as models for your situation.

 

[clemsonC]

Thanks for your replies.

 

[TNTBetter]

I would recommend investigating all potential vibration stress relief and finding out the truth. Why take a chance? If too much vibration is used is it possible the material could be fatigued? Is there some way a process can be tried instead of just going with what someone tells you will work? Seeing is believing. What else is being used in the industry? What experience have otehr people had? I would really question everything and I think that is what an engineer is supposed to do. If the process is used while welding does it interfere with the welder? Is it a voilent vibration like a jack hammer effect that will cause spatter and fatigue the work piece. There has to be a way to try something before you implement it.

 

[Tmoose]

I'm kind of stumbling over how residual stresses can ever be higher than yield in a stationary structure. I consider any welded structure a black box full of highly variable stresses just waiting for a chance to pull things out of line.

 

[gtaw]

Some rough calculations (and I do mean rough) indicate a delta T of about 220 degrees F is sufficient to exceed the yield strength of most carbon steels. So, if there is a thermal gradiant of more than 220 degrees F as a result of welding, cutting, or other sources, the residual stresses will be equal to the yield point of the metal.

Consider:

Modulus of elasticity equals stress divided by strain.

Set:
maximum stress is the yield point of the base metal.
strain is equal to coeff. of expansion times delta T.

E = Yield Point divided by delta T times coeff. of expansion, solve for delta T

Then consider the instantanious temperatures involved when you weld a piece of plate. The temperatures between the heat affected zone and slightly beyond can easily have temperatures exceeding the delta T derived from the equations above.

The secret is the delta T between the immediate area of the weld joint and the unaffected base metal surrounding the weld. The cooler unaffected base metal acts as a restraint against the metal being heated by the welding operation. As it is being heated, it expands, but is restrained, thus the maximum force it can exert is equal to the yield strength of the base metal (at temperature). At that temperature plastic flow is experienced because the yield strength of the base metal has been reached. As the metal contracts as it cools, the compressive forces become tensile in nature. Thus the residual stresses are equal to the yield strength of the base metal.

Hope this helps to explain what happens.

Best regards - Al

 

[clemsonC]

I have the specs for my structure, but no submittals from construction. The specs show the 2 welding processes that were allowed. So, per current AWS reqmts, the preheat and interpass temperature was either 50 or 150 deg, depending on which process was used, and the temperature outside would have been between 50 and 100 deg. So, based on Al's method, delta T would have been a maximum of 100 deg, resulting in stresses no greater than half of yield.

Stresses < 0.5Fy is certainly better than stresses > Fy, but it would be nice to know for sure. I just orded Welding Research Council bulletin 383 on nondestructive means for determining residual stresses, so we may use one of those methods to know for sure what's going on with this structure.

 

[gtaw]

Any welding process permitted by AWS D1.1 will involve temperature gradients much greater than 220 F. The molten weld puddle will be on the order of 3000 F, the surrounding base metal will be much lower. Thus the residual stresses will be of the same magnetude as the yield point of the base metal.

Assuming the modulus of elasticity for all carbon steel is the same, the delta T will be influenced by the actual yield strength of the alloy selected for fabrication. Still, the delta T required to develop stresses equal to the yield point are much lower than most people would expect.

This is put to use by fabricators to camber or put sweep into beams and girders. It is also used to remove excess sweep, camber, or to straighten beams that have been damaged during shipment or erection.

Best regards - Al