Weld Preheat Temperature Calculator

Hydrogen-induced cold cracking - also called delayed cracking, or underbead cracking - requires three conditions simultaneously, and removing any one of them prevents it.

The first is a susceptible microstructure, which means martensite in the heat affected zone. Martensite forms when steel that can harden is cooled quickly through the transformation range, and how readily a steel hardens is what the carbon equivalent measures. The second is dissolved hydrogen, which every arc process supplies from moisture in the electrode coating, oil or rust or paint on the plate, and humidity in the air. The third is tensile stress, which comes from the weld shrinking against restraint.

Preheat attacks the first condition and helps with the second. Warming the plate slows the cooling rate through the critical range so the transformation goes to softer products rather than martensite, and it also holds the joint hot for longer, giving hydrogen time to diffuse out rather than becoming trapped. Low-hydrogen consumables and dry, clean steel attack the second directly. Welding sequence and releasing fixtures attack the third.

The single most dangerous property of this failure is that it is not immediate. The martensite forms as the weld cools, and the hydrogen is trapped in it, but the crack does not initiate until enough hydrogen has diffused to a stress concentration to embrittle it locally. At room temperature that diffusion takes hours - commonly a day, sometimes several.

So the weld is inspected, looks perfect, passes, and cracks over the weekend. It is why AWS D1.1 requires that inspection of higher-strength steels be delayed, typically by 48 hours, and why "I checked it and it was fine" is not evidence about a steel with a carbon equivalent above about 0.45.

It also means the failure tends to be discovered late and expensively - after the assembly has been moved, painted, or installed. Preheat is a small cost applied before the fact against a large one applied after it, and the asymmetry is the whole argument.

What preheat really controls is cooling rate, and cooling rate depends on how much cold metal surrounds the weld rather than on the nominal thickness of one member.

A butt joint gives the heat two paths out - one into each plate. A T-joint gives it three, because the stem conducts in one direction and the flange in two. A cruciform gives four. So two half-inch plates in a T-joint draw heat away like a considerably thicker section, and cool proportionally faster.

This is why fillet welds crack on plate thicknesses where the equivalent butt weld would be untroubled, and it is the most commonly skipped step in an informal preheat assessment. Every serious preheat standard - AWS, BS EN 1011, the various national codes - handles it explicitly, usually by summing the thicknesses of the members meeting at the joint. It is not a refinement; on a heavy fillet it can move the requirement by a hundred degrees or more.