What heat input controls in a weld
Heat input is energy per unit length: how many joules the arc puts into each inch or millimetre of joint. It is not a measure of how hot the arc is, and it is not the same as arc power. Two welds made at identical amps and volts can have completely different heat inputs if one is run twice as fast as the other.
What it governs is cooling rate. A high heat input means a large volume of parent metal reaches high temperature, and that mass then cools slowly, giving grains time to grow in the heat-affected zone and letting brittle or embrittling phases form in alloys that are sensitive to them. A low heat input cools fast, which in plain carbon and low-alloy steels risks hard, crack-prone martensite in the HAZ, particularly at higher carbon equivalents.
That is why a procedure caps heat input at one end and preheat at the other. Quenched and tempered steels have an upper limit because over-slow cooling destroys the properties the quench created. Duplex stainless steels have both an upper and a lower limit, because too much heat coarsens ferrite and too little prevents austenite from re-forming. Austenitic stainless is limited to control distortion and sensitisation. Each of those limits appears in the WPS as a heat input number, and the welder controls it almost entirely through travel speed.
The formula, and the two conventions that use it
Arc power is volts times amps, in watts — joules per second. Travel speed in inches per minute divided by 60 gives inches per second. Divide power by speed and the seconds cancel, leaving joules per inch:
Arc energy (J/in) = 60 × I × E / S
That is the whole derivation, and the 60 is nothing more than the minutes-to-seconds conversion. Multiplying by a thermal efficiency η accounts for the fraction of arc energy that actually enters the workpiece rather than being lost to radiation, spatter and vaporisation.
Here the two major codes part company, and confusing them is the most common error in procedure qualification. AWS D1.1 defines heat input as the quantity above with no efficiency factor — what this page calls arc energy. ISO 15614 and EN 1011 apply a thermal efficiency factor k, with values in ISO/TR 18491 of about 0.6 for GTAW, 0.8 for SMAW, GMAW and FCAW, and 1.0 for submerged arc. The same weld therefore carries two legitimate heat-input numbers that differ by 20 to 40 percent. Set the efficiency selector on this page to 1.00 and you reproduce the AWS figure exactly.
Rearranging for travel speed is what a welder actually needs at the bench. If the procedure caps heat input, the fastest permitted amps and volts imply a minimum speed, and the calculator reports it directly: Smax = 60 × I × E × η / (limit × 1000) with the limit in kJ per inch. Because heat input is inversely proportional to travel speed, small speed changes move it a long way — slowing from 12 to 10 in/min raises heat input by 20 percent.
Worked example: a GMAW fillet at 200 A and 24 V
You are running a flux-cored fillet at 200 A and 24 V, and you time the weld: 12 in of joint in one minute, so travel speed is 12 in/min. The WPS caps heat input at 40 kJ/in.
- Arc power. 200 × 24 = 4,800 W.
- Arc energy per inch. 60 × 200 × 24 ÷ 12 = 288,000 ÷ 12 = 24,000 J/in, or 24.0 kJ/in. This is the AWS D1.1 heat input.
- Apply thermal efficiency. FCAW takes η = 0.8, so 24.0 × 0.8 = 19.2 kJ/in. That is the ISO heat input.
- Convert to kJ/mm. 19.2 ÷ 25.4 = 0.756 kJ/mm.
- Check against the limit. 19.2 ÷ 40 = 48% of the cap, with plenty of margin.
- Find the speed at the cap. 60 × 200 × 24 × 0.8 ÷ (40 × 1,000) = 230,400 ÷ 40,000 = 5.76 in/min. Any speed above that is inside the limit.
Now suppose the material is a duplex stainless with a cap of 1.5 kJ/mm, which is 38.1 kJ/in. The heat input at 19.2 kJ/in is comfortably inside it. But if the welder slowed to 5 in/min to fill a wide gap, heat input rises to 60 × 200 × 24 × 0.8 ÷ 5 ÷ 1,000 = 46.1 kJ/in, or 1.81 kJ/mm — over the cap by 21 percent. Nothing on the machine changed. Travel speed is where heat input is won and lost.
What is a normal heat input, and what the limits mean
For manual arc welding on structural steel, heat inputs between roughly 20 and 60 kJ/in (0.8 to 2.4 kJ/mm) cover most of what you will see. Stick welding a root pass at 100 A and 22 V at 6 in/min gives 22 kJ/in. A heavy submerged-arc fill at 600 A and 32 V at 20 in/min gives 57.6 kJ/in at η = 1.0. Those are the two ends of ordinary practice.
Where a limit exists it comes from the material, not the process. Quenched and tempered steels carry upper caps because the properties came from a controlled quench that slow cooling undoes. Duplex and super-duplex stainless steels carry both a floor and a ceiling, usually quoted in kJ/mm, because the ferrite-austenite balance depends on cooling through a specific window. High-strength low-alloy steels with tight toughness requirements often cap heat input to protect Charpy values in the HAZ.
Read the utilisation figure with the variability of the process in mind. A machine-carried weld holds travel speed to a few percent; a manual weld does not. If you are at 95 percent of a cap with a hand-held gun, an ordinary hesitation puts you over it, and heat input is measured on the qualification coupon rather than on the production weld. Procedure qualification records typically record a range, and production monitoring is checked against that range.
Heat input also does not tell you the temperature anywhere. Cooling rate depends on plate thickness and joint geometry as well as energy per unit length, which is why codes pair heat input limits with preheat and interpass temperature requirements. A 30 kJ/in weld in 1/4 in plate cools far more slowly than the same weld in 2 in plate, because the thick section conducts heat away in three dimensions.
Heat input at typical parameters, η = 0.8
| Typical job | Current (A) | Voltage (V) | Travel (in/min) | Heat input (kJ/in) | Heat input (kJ/mm) |
|---|---|---|---|---|---|
| SMAW root, 1/8 in electrode | 100 | 22 | 6 | 17.60 | 0.693 |
| SMAW fill, 5/32 in electrode | 140 | 24 | 8 | 20.16 | 0.794 |
| GMAW short circuit | 150 | 19 | 14 | 9.77 | 0.385 |
| GMAW spray | 260 | 28 | 18 | 19.41 | 0.764 |
| FCAW fillet | 200 | 24 | 12 | 19.20 | 0.756 |
| FCAW heavy fill | 300 | 28 | 10 | 40.32 | 1.587 |
| GTAW root (use η = 0.6) | 120 | 12 | 4 | 17.28 | 0.680 |
| SAW single wire (use η = 1.0) | 500 | 30 | 25 | 28.80 | 1.134 |
The last two rows are shown at η = 0.8 for comparability of the arithmetic. Recompute them with the correct efficiency for the process — 0.6 for GTAW gives 12.96 kJ/in, and 1.0 for SAW gives 36.00 kJ/in.
Errors that make a heat input number meaningless
- Using wire feed speed as travel speed. They are different quantities in different places. Travel speed is how fast the arc moves along the joint, and the only reliable way to get it is to time a measured length of weld.
- Reading voltage at the machine. Long or undersized leads drop voltage that never reaches the arc. Meter as close to the arc as the setup allows, or use the machine's arc-voltage sensing if it has it.
- Mixing the AWS and ISO conventions. The same weld is 24.0 kJ/in by AWS D1.1 and 19.2 kJ/in by ISO at η = 0.8. Quoting one against a limit written for the other is a 25 percent error in whichever direction hurts.
- Averaging pulsed waveforms with an ordinary meter. Pulsed GMAW and waveform-controlled processes need instantaneous power averaging, not average volts times average amps. ISO/TR 18491 covers the measurement requirement.
- Treating heat input as a proxy for temperature. Cooling rate also depends on thickness, joint geometry, preheat and interpass temperature. Codes limit all of them together for that reason.
- Computing per-pass energy and comparing it to a joint total. A limit is normally per pass. Ten passes at 20 kJ/in each do not breach a 40 kJ/in per-pass cap, though they certainly change the thermal history of the joint.
Which code applies
AWS D1.1, Structural Welding Code — Steel, defines heat input without a thermal efficiency factor and requires it to be recorded for procedures qualified with heat-input limits. ISO 15614-1 and EN 1011-1 apply a thermal efficiency factor to arc energy, with values given in ISO/TR 18491. Check which one governs your work before comparing a calculated number with a specified limit, and record which convention you used on the WPS.
Heat input alongside the other welding numbers
Heat input controls metallurgy; it says nothing about whether the joint is strong enough. Weld size does that, and the fillet weld strength calculator gives the capacity of a given leg size and length against electrode strength. The two interact in practice: increasing weld size to gain capacity usually means more passes or a slower pass, and the second option raises heat input.
Preheat is the other half of the cooling-rate story. Where heat input sets how much energy goes in per inch, preheat sets the temperature the surrounding plate starts from, and thicker sections need both. AWS D1.1 tabulates minimum preheat by material group and thickness, and hardenability — usually expressed as a carbon equivalent from the mill certificate — decides how sensitive a steel is. If you only have a hardness reading rather than a certificate, the hardness conversion calculator converts between scales and estimates tensile strength, which at least identifies whether you are working with a plain carbon steel or something considerably stronger.
Distortion is the visible consequence of heat input. More energy per inch means a larger heated volume, more shrinkage on cooling and more angular distortion in a fillet joint. Balanced welding sequences and back-step techniques manage it, but the underlying driver is the energy the calculator on this page reports. On thin sheet the distortion limit often binds well before any metallurgical limit does, which is why sheet fabrication favours fast, low-energy processes.
Finally, for anything cut rather than welded, the same energy reasoning applies to thermal cutting: the kerf and the heat-affected zone at the cut edge scale with power divided by cutting speed. Plate that will be welded after cutting inherits both heat-affected zones, and specifications for demanding service sometimes require the cut edge to be machined or ground back before welding for exactly that reason. Blank sizing for those plates is covered by the steel plate weight calculator.