Welding Heat Input

Heat input is the energy the arc puts into every unit length of joint. Volts times amperes is the electrical power at the arc; arc efficiency takes off the share that radiates away, blows off as spatter or warms the torch; and dividing by travel speed spreads what is left over the length welded. Written in the units a shop actually uses it comes out as $H = 60VI\eta/(1000S)$ with $S$ in millimetres per minute — but the 60 and the 1000 are unit conversions and nothing else. Strip them away and the relation is simply power in, divided by how fast the arc walks.

The single most important thing to understand about this number is that it does not, by itself, decide anything. What governs the microstructure of the heat-affected zone is how long the steel spends cooling through the transformation range — the $t_{8/5}$ time between 800 and 500 °C — and heat input is only one of the terms that sets it. Plate thickness matters as much. Joint geometry matters. Preheat and interpass temperature matter enormously. And whether heat is escaping into two dimensions or three changes the entire form of the relationship. The same 1.5 kJ/mm laid on 6 mm plate and on 40 mm plate gives two welds with nothing metallurgical in common: the thin one cools slowly and the thick one quenches itself. Quote a heat input to a procedure and you have said something useful. Quote it as though it settled the question of whether the weld will crack, and you have said nothing at all.

Which is why codes limit heat input to a range rather than a ceiling. Too little and the joint cools too fast, hardening the heat-affected zone and inviting cold cracking. Too much and the grain in the coarse-grained heat-affected zone grows, toughness falls away, distortion rises, and on quenched-and-tempered steels you begin to undo the heat treatment the plate was bought for. There is a window, and it is narrower on some materials than on others.

Now the unit trap, because it is the reason this page exists in the form it does. A North American procedure sheet says 30 kJ/in. A European one says 1.2 kJ/mm. Those are close to the same weld, and the factor between the two systems is 25.4. Every value from about 0.5 to about 50 looks like a plausible heat input in one system or the other, so there is no sanity check anywhere in the arithmetic that catches the mistake — an answer of 30 kJ/mm looks like a heavy submerged-arc pass rather than what it actually is, a kJ/in figure read in the wrong system. This site types heat input as a real unit with its own picker precisely so the conversion cannot get lost. Enter the number with the unit it was written in and the engine carries the 25.4.

The second trap is closer to home and costs a factor of 60. Travel speed is quoted in millimetres per minute or inches per minute, and a number written down as mm/min and later read as mm/s is out by a factor of sixty in the wrong direction. The picker on this page offers metres per minute and feet per minute, because those are the units the engine has: 300 mm/min is 0.30 m/min, and 12 in/min is 1.0 ft/min. Say the number out loud before you enter it.

Three cautions on the terms themselves. The efficiency $\eta$ is a process property and a contested one — published values for the same process differ by ten points depending on how the calorimetry was done, with broad ranges around 0.6 for gas tungsten arc, 0.8 for shielded metal arc, 0.85 for gas metal arc, and 0.9 and up for submerged arc, where the flux blanket traps nearly everything. Worse, some codes define heat input with $\eta$ omitted entirely and call the bare $VI/S$ figure arc energy to keep the two apart. If you are comparing a number against a code limit, find out which of the two the code means. The voltage belongs to the arc, not to the machine's panel meter, and on a long or warm cable the difference is real energy going into the cable rather than the plate. And with a weave, the travel speed that counts is net progress along the joint, not the speed of the hand.