Forging press repair weld failure

Centerpiece worked example for the Fourier-number argument. Welders preheated a 12-inch-thick press for 10–12 minutes instead of the ~2.5 hours required by the Fourier number; got residual stress, hydrogen embrittlement, cracking. Repeated, cracked again. Third attempt with full-penetration weld held until loaded, then cracked. Press scrapped — $5M loss, ~$50M to replace.

Now, remember I brought in this 12- or 13-inch-thick weld with hundreds of weld passes — it weighed about ten pounds when I passed it around. That's one where they had to preheat the metal. About 12 inches of steel — well, 12 inches isn't 30 centimeters, that's 10 inches, but it's close to 30 centimeters; I'm rounding because this is an approximate analysis. Here's the Fourier number — x squared, the distance the heat travels, is alpha T. Square 30, I get 900, divide by a tenth which is alpha, the time is 9000 seconds, which is two and a half hours. Well, that's why that weld cracked. The section I showed you wasn't cracked, but they lost about a five million dollar press because they'd been using this forging press for fifty years. Over time it wore out, developed a fatigue crack, and cracked.