Recently I observed some longitudinal cracking in structural steel box columns.
The question is "what is the first step to stop growing of cracks?"
I should inform that the type of joint and weld of columns are Corner Joint Fillet weld.

My vote is solute rejection. Low melting point constituents are rejected to the center line of the weld during solidification. While they may have gone undiscovered during fabrication, the cracks may have been present, but too small to be detectable by visual examination alone.

The probability of delayed cold cracking diminishes over time as the diffusible hydrogen effuses. Generally, the weld centerline is not where I would expect hydrogen cracking to originate since the weld deposit is formulated to have a low carbon equivalency, good ductility, and slow cooling if proper preheat is employed. I would expect the delayed cold cracking to be located in the HAZ since it is likely to experience higher cooling rates (compared to the weld) and may have a microstructure more prone to delayed cracking due a higher carbon equivalence (compared to the weld deposit), faster cooling, and increased likelihood of a martensitic microstructure. 

The possibility of center line cracking due to the aspect ration of depth to width is unlikely in this case. The width in this case appears to be about twice the depth. If the depth was greater than the width of the face, I would concur, but that isn’t the case as cited in the post.

Back to the solute rejection; I would review the material test reports if they are available and look closely at the actual chemistry of the base metal. Look to verify the sulfur and phosphorus content when summed is <0.04%. Both elements are bad actors. There are other elements to consider, i.e., copper for one. Any element that has a melting point below 2000°F is suspect.

The welds may have been preheated, but box columns usually employ thick sections that should be considered highly restrained. As such, higher preheat, that is, higher than that listed in Clause 3, should be employed. I recommend calculating preheat using the alternative offered in the Annex of AWS D1.1 and go with the condition of high restraint. Usually the need for high preheat is associated as a means of mitigating the potential for delayed cold cracking, but slow cooling mitigates the formation of a hard brittle microstructure. That's usually a benefit regardless of the potential for hydrogen assisted cracking.

Magnetic particle testing using wet fluorescent magnetic particles can be used and will provide a high level of sensitivity to surface breaking cracks that may not be easily seen with the naked eye.

I see no need for stop drilling as it will only add to the difficulty of repairing the cracks. I would initiate the excavations no less than 50 mm beyond the crack tip as indicated by magnetic particle testing. Stop the excavation about mid length of the crack and then excavate beginning no less than 50 mm from the opposite crack tip. Alternate from end to end and excavate to the depth necessary to completely remove the crack. More than likely the cracks will extend to the joint root. If the cracks are intermittent and are separated by less than 200 mm, it would be best to consider the crack to be continuous and remove the affected welds including the length between the short  intermittent cracks.

Interesting case and great photographs.

Best regards - Al

I cannot say with any certainty what the root cause of the problem was. If the base metal had the chemistry required by the material specification the carbon equivalency should have been relatively low, thus an undesirable microstructure should be mitigated. If the proper preheat was used, slow cooling should have been assured and again, an undesirable microstructure should have been avoided. We have no means of ensure the condition noted were met. That is, what level of QC was instituted and provided by the contractor during fabrication?  We don’t know and have no way of assessing.

Well, that is not entirely true. The contractor or the owner can perform some metallurgical examinations to determine exactly what the product chemistry is and whether there is evidence of martensite in the HAZ and weld. However, I go back to my original assessment; the HAZ is the region most likely to experience rapid cooling and most likely to contain martensite if the chemistry was off or if insufficient preheat and interpass temperature was maintained. I suspect that the welder training at that facility was no better and no worse than what we find in many of our local contractor’s facility. Many welders associate the moisture observed on the surface of the steel during initial preheating to be the physical evidence they are successful in driving moisture out of the steel. Once the moisture is eliminated, additional preheating or the maintenance of interpass temperature is unnecessary. That is a fallacy many of us have had to contend with on more than one or two occasions.

If my supposition is correct, the weld is unlikely to be hard or to contain martensite. Thus, the weld is not the region I would expect to see cracking issues. The chemistry of the weld filler metal is such that it is unlikely to contain martensite if even marginal preheat was used.

Eliminating the probability of martensite in either the HAZ or weld deposit leaves me with two alternatives; the first being residual stress and the second is chemistry that does not meet the limitations of the base metal specification. The issue of residual stress can lead to cracking, but it is usually associated with highly restrained joints or intersecting welds. There was no mention of intersecting welds in this post, so I am not considering it to be a factor. The residual stresses in the transverse direction relative to the axis of the weld are on the order of the yield strength of the base metal or weld. However, weld filler metal is formulated to provide sufficient ductility to accommodate both shrinkage and contraction of the weld as it solidifies and cools to ambient temperature. The steel in this case is not a high strength steel with a high carbon equivalency, thus the residual stress transverse to the weld should not be cause for alarm. The aspect ration of weld face versus penetration were on the order of 2/1, so again, centerline cracking should not be an issue because of an undesirable aspect ratio.

That leaves me with base metal chemistry. If the base metal chemistry was a little out of spec., i.e., the quantity of LMPC was on the high side, their cumulative effect could result in a condition where the solute rejection phenomena could have produced the intermittent center line cracks exhibited by the photographs provided. It should be a relatively simple matter to secure a sample of the base metal for chemical analysis. The investigator would have to provide direction to the laboratory as to which elements to test for. Most laboratories will report those five or six elements that are most prominent. In this case, we are most interested in determining the quantity of LMPC present, some of which would not be reported in a typical chemical analysis.

Mind you, this is purely speculation on my part. I am trying to eliminate the causes that can be easily discounted and leave the less obvious causes as avenues for further investigation. There are still questions in my mind. One question looms large; when did these cracks form? Another question; what examinations were performed and by who? Was the visual examination thorough as dictated by the code or was it cursory or worst, random? How many times do we see visual examinations that are at best cursory or simply not performed by the contractor? Instead, it is left to the welder to “self inspect”.

Merry Christmas everyone.

Al