Interesting Material Failure 7

[dik]

I have also posted this in the Welding forum.

We are doing some remedial work to a structure that is approx 40 years old. The steel used is specified as A36, but we have not confirmed this.

The structural Tees are welded to a bulkhead (all around)and cracks have developed in the flange on either side of the web. They appear to be intermittent and in some areas extend approximately 4' from the end. The span of the ST is approximately 10'. The cracks also appear to have initiated at the bulkhead. Loading is relatively high and the hydraulic load is applied to the thin plate at the bottom shown in the Detail Elevation. There is also a slight vibration that may have initiated a fatigue type of failure as well as the steel being subjected to zero degree C /32F temperatures. Welding appears to be sound and has a good appearance with no apparent 'splattering' or inclusions.

We are in the process of obtaining additional testing information to determine the full extent of the cracking as well as the depth; we don't know if the crack is the full depth of the flange, yet.

We are thinking that the cause of the cracking is from a triaxial stress condition caused by welding the ST all around to the bulkhead.

Any comments or suggestions?

 

[hokie66]

That is a puzzle. But I doubt that the welding is to blame.

Can you give a bit more information about how the tees have been loaded? I am not sure what you mean by the hydraulic load being applied to the thin plate. Are the tees stiffeners for the bottom of some sort of tank?

 

[JAE]

The cracks might be from some previous distress on the flange/web intersection - bending of the flanges relative to the web during shipping?

Also, at the flange/web intersection there is, I think, the highest amount of residual stress due to milling.

But the fact that the cracks are in the flanges (which are thicker) and not through the web (which is thinner) might be a big defining clue.

 

[dik]

hokie66

The loading on the ST is from the thin plate on the bottom and the force is in an upward direction. This thin plate acts as a 'bottom flange' for the ST.

The crack is running parallel to the direction of maximum stress.

JAE

The high stress at the fillet area of the flange-web was one of the reasons I was thinking of a triaxial stress initiated failure caused by the all around fillet weld.

Thanks, any other thoughts?

Dik

 

[hokie66]

You mentioned low temperature. Could at some stage some of the "compartments" between the tees have been filled with water and frozen? That conceivably could have caused a horizontal shear failure in the flange.

 

[dik]

Hokie66
Cracking observed is in the lower ST's is more pronounced and these are submerged for the most part. I'll have to check this out to see if it is possible or if there is a correlation to the location of the failures and thermal issues. In the spring, the lower plate is exposed to sunlight and can likely reach temperatures of about 90 degC. Thanks...

Dik

 

[msquared48]

For some reason, my guts tell me to look to Naval Architectural Design for vessels recommendations for the answer to the cracking, and I'm not talking brittle fracture. The devil is in the details here. Vibratory or cyclical loading?

Mike McCann
McCann Engineering

 

[dik]

Thanks Mike... there could be some vibration and the Naval Architecture Design is something I'm in the process of doing some 'skimming'... aka 'sorta reading'...

There could be mechanisms at play that I'm not familiar with. I'm also quizzing the testing tekkies...

Dik

Dik

 

[samdamon]

Looking at your plan detail, perhaps the steel plate welded to the web and flange of each ST end is expanding and contracting laterally at different rates than the tee flange is during cyclic temperature swings. That would tend to strain the end of the tee flange, but across across the "bf", and one could expect the stress to concentrate at the change in thickness of the flange at the fillets. In simpler terms, the end of the tee flange perhaps is being opened up like a fan. If this equipment is exposed to the sun you could have two cycles a day of thermal changes.

Also, the ST may stay at a more even temperature and thus want to move less during weather fluctuations because of the insualting effect of the thermal mass of whatever is contained below. Any, just a thought.

How exposed is this assembly, and where is it located?

 

[miecz]

Looks like a difficult weld to make (and clean) at the underside of the tee. Any inclusion at that location could initiate a fatigue crack, given the temperature swings. What is the temperature of the stiffener when the plate reaches 90[°]C? Is this ductwork with gas flowing through?

 

[dik]

miecz

The assembly was shop fabricated and welding was not a problem. The welds are actually quite well done from a visual perspective.

The ST's are essentially stiffeners for the bottom plate and there is no gas flowing through the assembly. The assembly is raised in the spring time for a few weeks/months? and for the balance of the time, it is submerged in water.

Dik

 

[msquared48]

Are you sure that the connection of the tee to the stem was welded, ie making it a built-up Tee as opposed to a cut, rolled section?

The reason I ask is that many Tees are merely W shapes that have been cut in half horizontally. Hence the connection would be integral as it would be a rolled section - no welding.

Just thinking...

Mike McCann
McCann Engineering

 

[hokie66]

From the sketch in dik's original post, the tees were shown as ST 7 x 17, which would be half of a WT 14 x 34, or whatever designation you are using in the US these days. I don't believe any of this is a welding issue, but rather due to some manner of overload. Determining how it was loaded is the issue, and until otherwise convinced, it looks to me like longitudinal shear due to lateral bending.

 

[miecz]

The elevation and section show the tee welded to the end plate. The crack begins at the weld and propogates back along the tee. Visual inspection of a weld is not all that reliable. I see no other reason for this tee to fail in this way.

 

[dik]

I think the crack initiated at the top of the flange of the ST, at the point that connects to the end plate.

I had another thought at 5AM... When the ST's were initially cut, they may have distorted and been straightened (by means of a fiddler or by heat); this could have set up internal stresses, aggravated by the welding.

Also have to chase down the possibility that the intersection of the web and flange, caused by the rolling process, gives rise to 'false' positive cracking.

Dik

 

[hokie66]

dik,
Good to see that someone else thinks about curious things like this while they sleep.

I withdraw my suggestion that this was a shear failure due to lateral loading. I did the numbers, and the shear stress would have not been very high when the flange would have yielded in bending.

I still suspect that somehow ice formation and the resulting 4% expansion of water could be the culprit, but not knowing more about the history, I don't know. Could the cracks be due to bending of the flange in the transverse direction?

 

[dik]

There was no ice formation. I learned, today, that the area had been subject to a fire a several years ago... another factor, additional thermal stresses... I'll post an elevation to outline the extent of the area affected. The 'ballooned' areas are areas that have been NDT and all but one has cracks.

Dik

 

[dik]

Forgot to add that the cracks are not through; the NDT tekkie thought that they were approximately 1/4" deep... The flanges are only partially cracked and not through. Some cracking is along the centre of the web, some along one side (at the fillet) and some are on both sides.

Dik

 

[hokie66]

So the fire may have expanded the flange in the transverse direction, but the web and the fillets restrained it on the bottom side, and the top cracked in bending?

 

[btrueblood]

Weird is right. Not a structures guy, but I play one here at my current place of work. As such, my nomenclature is probably all wrong, but in reference to your section, I call the top of the "T" the part that mimics the part of the letter as you read it. Sigh, still not clear but here goes:

I assume from what you've written that the loading puts the top of the "T" section in tension. Poisson's ratio would tell you this section wants to contract, but the weld at the end of the section is restraining if from doing so, resulting in tension across the section (perpendicular to the cracks).

Not clear to me why the welding was done all-around on that part of the section, vs. just two welds down both sides of the vertical part of the T. You want to transfer the shear stresses from the T-beams into the end plates, and most of the shear is being carried by the vertical leg of the T (web?) and not the top cap?