CTE for weld metal?

[Althalus]

I've got a project in Alaska where our steel turned out warped upon delivery. We've tried to determine the causes. The only think I can figure is that the weld metal has a different CTE than the structural steel does.

I can't find the Coefficient of Thermal Expansion (CTE) for welding metal. Is it any different than A572 or A992 steel?

 

[dvd]

What preheat/postheat temperature was applied?

What thickness of steel?

What was ambient temperature?

What electrode was used?

 

[weldstan]

They will have the same CTE if the weld filler metal is much the same chemistry (lower CE though) of the base metals joined; i.e. low carbon steels. Some distortion is inevitable because the base metal not far from the weld and HAZ is at a low temp compared to the weld so growth along the weld will be greater than the much colder base metal, which is trying to restrain growth. There are a number of ways that distortion can be minimized. Obviously your supplier did not use any of these techniques to do so.

I assume that you didn't provide any inspection during manufacture or prior to delivery since you only discovered the condition upon receipt. Too bad.

 

[MintJulep]

A lot of unfounded assumptions there weldstan.

In other threads Althalus has stated that they are fabricators. Therefore we might instead guess that "our steel" is something that they fabricated.

We might further guess that "turned out warped upon delivery" infers that it was not warped at shipment.

But it would be better is we didn't need to guess.

Regardless, welding can impart stress, and stress can impart deformation.

It is possible that there were welding stresses imparted that were just under yield. And that loads applied during shipping pushed things beyond yield.

 

[dhengr]

Althalus:
Weldstan is right on the money. You always get some distortion from welding. If you put a flange plate down and weld a vert. standing web to it, fillet welds both sides of the web to the flange, making a ‘tee’ section, the entire section will end up cambered upward, touching the table only at the ends. If you flip that new section 180° onto a new flange and gag it down, and apply two new fillet welds to the new flange, this new welding will tend to straighten things out, but likely not completely. Because with the second welding and flange, the cambered “tee” section is a much stiffer section so the same welding will not have as much distorting effect. All of this is due to residual stresses and the effects of expansion and contraction due to the welding heating. This same process can work to your benefit when you want to build camber into a built-up beam member. And/or you can take camber and distortion out of a member by heating it, heat straightening. Several of Lincoln Electric’s welding books, along with Omer Blodgett’s books on welding, “Design of Welded Structures” for one, and some AWS books cover this topic in some detail. Managing this during fabrication is quite an art and science, needing considerable experience.

 

[Althalus]

To update:

I'm the engineer for our products. I call for inspections as required for standard circumstances. But I don't actually perform them. I only visit the fab yard about once a month if I'm not busy.

The particular project was welded in the shop at ambient temperatures in Texas. At that time of year, it would have been 70s or 80s (possibly 90s). I believe we called for 70ksi electrodes. But we have no way of knowing if they use rods, wire, robotic welding... whatever.

The item in question was a WT (web pointed upward)to the bottom side of a WF beam. Continuous fillet weld on both sides of the WT web. When it was put together in the shop it was all fine and dandy. It was placed on a truck with the dunnage and strapping placed in such a manner that if any shipping issues caused the distortion, it would have bowed downward.

Instead, we released the straps and it bowed upward. I don't remember the dimensions. But I calculated the required bolt force it would take for the ends to come back down to level. It took (10) 1" dia A325 bolts (it was a rather hefty composite beam section). We went with shims instead. But I kept thinking it had to be something to do with the weld that I did not understand. And the cold weather in Alaska at time of delivery was about -30 F. So, over 100 deg difference from fabrication to offloading from the truck.

If the CTE is the same (and I don't know what CE is) then it is most likely something to do with the heating of the metal and partial contraction, with some yielding. And that is making me nervous. If there was yielding (especially around the HAZ) then we might have compromised steel.

 

[dhengr]

MintJulep:
Your last para., “It is possible that there were welding stresses imparted that were just under yield. And that loads applied during shipping pushed things beyond yield.” may be true in a few cases, but generally it is completely bass-acwards. There is yielded material around all welds, that’s what sets up the residual stresses which cause the distortion problems. I’m not suggesting that they can’t break or harm something in transit or by the way they tie it down. But, more often than not, you will tend to straighten long cambered pieces, or remove distortions from other pieces during transport. That’s called vibratory stress relieving.

 

"It happened during shipping" is the industrial equivalent of "the dog ate my homework".

I once had a supplier blame slag inclusions on shipping...

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

 

[dvd]

My suspicion is that the WF Tee was welded to the WF beam from above, i.e., upside down to orientation of erection, in weld position 2F, not overhead in position 4F. It would be like crowning the beam at the deflection due to self-weight, so maybe not surprised at the camber.

 

[Althalus]

My suspicion...

I'll have to check. It wasn't my design. And it was about 6 months ago, so my memory may be cloudy on it.

...

I looked through the files. Yup. I had it backwards. The WT was on top.

So, I'm guessing by the responses here that they probably didn't properly pre-heat? Or would this have happened anyway because of such an extreme temperature change?

Would skip welding have alleviated some of this effect?

 

[weldstan]

If somehow you could pin a load on some part or parts of the assembly such that movement is fully restrained, shrinkage of the assembly from the abient in sun temp of Texas (~110F) to the -30 F Alaska temp. could produce yielding at some point or points along welds.

 

[Althalus]

Would skip welding (instead of continuous welding) help with this?

 

A photo would really help this discussion

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

 

[weldstan]

Since we really don't know how it was welded in the shop or how it was assembled, one really cannot give any credible answer. Did they use a strongback or a comealong or other mechanism to acheive dimensional correctnss and thus establish high directional loads? Who knows?

Ditto ironic metallurgist.

 

[Althalus]

A photo would really help this discussion

Not really. The camber upward is so small compared to the overall module that you wouldn't be able to see it.

Just imagine a fairly flat smile.

 

[kingnero]

Depending on the size of the WT and WF, warping could go both ways. Without details, impossible to comment. I do agree however that transport and/or temperature is unlikely to have a significant influence on the deformation.

 

[MintJulep]

@dhengr
I accept your explanation.

At the same time, I have seen a case where a lot of clamps were used to close gaps between different sub-assemblies before they were welded together.

It all looked fine after welding and all the clamps were removed. For a while.

But in the course of moving the thing from one location to another within the factory things went "sprang!" and the stresses imparted by forcing the parts together redistributed themselves, resulting in the whole thing being distorted.

I guess those are not properly "welding stresses" because they were not caused welding (the process of fusing metal together by melting it an letting it re solidify), but rather during welding (a fabrication technique for joining metal things together).

 

[gtaw]

I would give some thought to the way the members were loaded and strapped to the bed of the trailer. There's a lot of bouncing going on during a trip of several thousand miles. If the steel is not secured properly, the bouncing could cause the members to bend. Just a thought.

Best regards - Al

 

[dvd]

My comments -

1. vibratory stress relief due to vibrations from transport
2. residual stress in the HAZ were relieved during welding, and residual stresses in non-welded flange caused deformations due to the imbalance
3. molten weld pool solidification results in a phase transformation that deforms the weldment

 

[dhengr]

MintJulep:
I thought you meant stresses caused by the welding, but now I think you might actually mean the stresses in the weld, or welds, due to the various loads. I can certainly imagine a situation like you are describing, but would need a lot more details to really comment more definitively. I would be surprised that you didn’t see any weld failures, weld cracking, parent metal cracking, etc. if you could see significant movement or deformation of the combined parts. Did you actually hear sounds ("sprang!", whatever) when things seemed to move/change/fail? If you had two stiff parts and had to really clamp them together for fit-up before welding, needing large clamping forces, then you were adding a whole new loading (set of stresses) to the welds for which they were not designed. And, those stresses along with the stresses from lifting and moving, or for which the welds were originally designed could combine for some yielding or problems. But, we do this kind of fit-up clamping, etc. every day, usually without problems when done judiciously. Many times we have to heat treat (stress relieve) a large/stiff weldment before we do any machining on it or it will move around during machining and we can’t hold tolerances. The machining relieves, removes, affects some of the locked-in stresses and this causes the weldment to keep change shape slightly.

Part of what’s happening with yield stresses caused by and around welds is that the yielded material is usually well confined by surrounding material so it can’t go anywhere and some yielding isn’t an instant failure mechanism, but it does induce a bunch of residual stresses to accomplish this. Steel can continue to work just fine after having yielded, as long as you can tolerate the strains/deformations caused by the yielding. You’ve undoubtedly seen this on a stress/strain curve for steel, or can find it in many Strength of Materials/Theory of Elasticity texts. The stress at a point (or small volume) climbs the elastic part of the curve, starts to yield, and moves out on the plastic part of the curve; then when unloaded the stress moves down (maybe back to zero) from ‘this higher stress point’ essentially at the same slope as elastic part of the curve, but at a new strain. Reloading (restressing) causes a climb up this new slope line to ‘this higher stress point’ again, and if the loading continues the stress will move further out on the plastic part of the curve, and the cycle will/can repeat.

In my earlier example, a WT made from a flange and a web, and then made into a WF by adding a second flg., in each cycle of welding the two fillets, web to flg., the welds want to shrink/shorten as they cool, and the only way that can happen is for the member to take a curved shape (circular curve) with the weld line being a shorter circumferential arc than the free edge of the web, in the first case. Assuming I’ve welded the web to the flg. downhand, the WT will camber up, lifting the center of the WT up off the table because of this curvature change. This whole process leaves residual stresses, is held in equilibrium by these residual stresses. Now, in the second case, the WT added to another flg., again welding downhand, the same thing happens, but the WT is much stiffer than the web alone was, so the curvature due to the new welds cooling and shortening is less. To make the final WF straight, I would put the second flg. on the table with several graduated shims under it, max. shim thk. at center. Then, I’d press the WT down to fit the new flg. partly straightening it out, and the second fillet welding cycle would do the rest of the straightening. My beam is not very stiff over its length, so the pressing forces are not too great, as they are ultimately reacted by the new welds. Otherwise, this is not unlike you having clamped your parts together to weld them. But, your parts must have been much stiffer, thus needing much greater clamping forces.

The OP’ers. beam is all steel and the CTE is essentially a constant (.0000065 “/”/°F) for the base matr’l. and the weld deposits, so the temp. change (Texas to Alaska) would not cause deformation or curvature change. It would just cause a fairly uniform length change of the whole part as a function of the temp. change. We just don’t know enough about his built-up beam to make a judgement about what happened. I suspect it didn’t leave Texas as straight as he is being told it was, and the faber doesn’t know what its shape was either, or is hiding what it was. Still, if the piece was put on a very flexible flatbed trailer, and blocked to the trailer bed, or put on a fairly stiff trailer and not blocked well, and then a bunch of load was put on top of it you could get some ‘vibratory stress relieving’ which could cause some shape movement.