One of the authors inspected the Exxon Valdez in the drydock in San Diego seven months after the grounding. The real eye opener was not the damage, as extensive as it was, but the welding. Just about everywhere in the damaged areas the stiffener webs had pulled cleanly out of the welds leaving the welds on the plating. The welds were tiny. They looked like continuous tack welds. When I asked the Exxon superintendent who was showing me around if the welds were legal, he angrily nodded yes. This sort of failure absorbs almost no energy and may be a reason the Valdez rode so far up on Bligh Reef.
In the bad old days, the standard filet welds in a large tanker cargo tank had a throat thickness of 6.4 mm (9 mm leg length). Now much of the filet welding in the tank area has a required throat thickness of 3.2 to 4.0 mm. And Class allows the yards a 10% negative margin. (This is explicitly written into the IACS "Quality" Standards.) So often the actual weld is less than 3 mm throat. To expect any penetration at all with such welds is crazy talk. Those are the welds we saw in San Diego. Miniscule. And stupid. The yards can easily lay down 5 mm plus throat welds with a single pass. When Hellespont asked DSME to increase all welds in the ULCC cargo tank area to 5.3 mm throat -- after signing the contract -- the price was $44,000; and it cost the yard far less.
Whatever argument -- probably nothing more than the down ratchet -- led to the current weld sizes, it should be abandoned. Welds corrode much more rapidly than the other steel. And as the Valdez showed, these tiny welds fall apart on impact.
A particularly important weld in double hulls is the lower hopper corner. First problem is that the weld is difficult to do because of the restricted space. The only reasonably sure way of laying down the root pass is manually (semi-automatic in yard parlance). This should be a requirement. Otherwise it is nearly certain that the root pass will not fill in the bottom of the notch. BP encountered this problem on a class of VLCC's built at Samsung in 1998/99. However, the biggest problem is that a great deal of stress is trying to turn a corner at this weld. We have no idea why we don't roll this corner as if it were the bilge radius. But if we are going to have a sharp corner and a natural stress concentrator, we must radius this weld very carefully.
Given a series of failures in hopper welds in double hull tankers operating in the North Sea, LR is finally starting to address this issue. LR performed a super detailed FEM of the hopper corner with a mesh size of T/12 where T is the thickness of the inner bottom. Unsurprisingly, the minimum radius of the weld is critical to the maximum stress in the corner. With a "Class standard" weld, the max stress was 660 N/mm2, almost three times yield. With a 15 mm radius weld, the max stress was 265 N/mm2. With a 30 mm minimum radius, the max stress was 210 N/mm2. It seems to us obvious that we should adopt a minimum weld radius of at least 30 mm for at least 300 mm on either side of each web. This is essentially a more careful version of the new LR weld profile 6.
Another area where the rules have fallen apart is heat treatment. The most egregious case we know of is the yard's refusing our LR surveyor's request to stress relieve the critical welds between the upper rudder casting and pintel casting to the 82 mm thick plate that connects them on our ULCC. This request was refused despite a haphazard welding sequence which thermally cycled the steel several times.
The yard was just following standard practice these days. In defiance of good engineering practice for non-critical areas, stress relief is not explicitly required under either ABS or LR rules. In this case, the LR surveyor strenuously and courageously objected. But when the yard pointed out that stress relief was no longer explicitly required in the rules, Lloyds Register failed to support its own surveyor. Not only are the days when the surveyor's word was law long since gone; the yard put Lloyds on notice that the surveyor's behavior was unacceptable and asked for his dismissal. In the yards' view, the surveyor has no right to require anything that is not explicitly in the rule. Sadly that is essentially the societies' position as well.
Hellespont had to pay extra for the heat treatment. The cost was trivial, $2,000. The yard was just making the point that a surveyor had no right to exercise his judgement. Obviously, if a yard doesn't have to heat treat this critical weld, there are almost no welds that it does. In the 1970's, Class required stress relief on all welds in excess of 40 mm thickness. The down ratchet has been working big time in this area. All welds involving castings and very thick plates should follow a carefully prescribed, conservative heat treatment.