Hello JA, I'm going to take a stab at that one, just a stab mind you. With the fillet weld, unless there is a preset gap, you have a restriction to the shrinkage that can occur after the weld has been done, thus a greater possibility of an underbead crack, or possibily a centerline crack. You also don't generally have 100% fusion to the root of the weld, in a lot of cases this can set up a crevice or type of undercut at the root that can promote cracking. In most instances when a groove weld is being done there is not nearly as much restriction when shrinkage occurs(on an open root groove weld the root gap can close up or shrink slightly without causing excessive stresses), thus there is less likelihood that you will experience cracking, underbead or otherwise. Just a stab mind you and likely much better educated members of the forum will give you a better explanation. Regards, aevald

In general, there are two cracking mechanisms related to underbead cracking, (1) hydrogen induced cracking and (2) lamellar tearing.  Cracking associated with butt welding of components in the same orientation (such as butt welding of two plate sections or pipe welding) normally would not see the lamellar tearing phenomonem as it is normally associated with poor through thickness, material properties of steel plates.

Hydrogen induced cracking requires (1) hydrogen, (2) stress and (3) a suseptible microstructure.  Without any one of the 3, HIC cannot occur.  HIC is normally seen in the HAZ of high strength materials, but can also contribute to cracking in the base metal area related to lamellar tearing.  Lamellar tearing is normally seen in Tee weld applications, and can result from partial pen, full pen and fillet welds.  This cracking requires (1) suseptible microstructure/poor through thickness properties, (2) stress and (3) sufficient restraint.  Hydrogen can also contribute to lamellar tearing.  Lamellar tearing really can't happen in thin materials as distortion/yielding would happen before cracking.  This is normally seen in thicker section base plates, with a fairly heavy weld connecting a perpendicular member.

The attached photo shows classic lamellar tearing and was seen in the TWI article, with the following link:
http://www.twi.co.uk/j32k/protected/band_3/jk47.html

Hope this helps.

Also, fillet welds will tend to cool off quicker than groove welds since the heat can be transfered out of the weld into the base metal in three directions instead of two for butt welds.  Of course the faster cooling can contribute to hydrogen cracking.