Preheat is the most misunderstood requirement on most welding procedures. Welders often read it as a way of drying the joint or making the arc easier to start, and both of those are side effects rather than the point. Preheat exists to control the rate at which the weld and the metal around it cool down, because cooling rate decides what microstructure forms and how long hydrogen has to escape before the metal is hard enough to crack.
Interpass temperature is the same lever pointed the other way. A minimum interpass keeps the benefit of preheat alive through a multi-pass weld; a maximum interpass stops the accumulated heat from ruining the properties the procedure was qualified to deliver.
This guide covers what preheat physically does, the three conditions that make it necessary, how the temperature is chosen, how it is measured and verified, and what the inspector is actually checking. It describes general practice. The temperatures for any real job come from the qualified procedure and the code or contract documents governing that work.
What preheat actually does
Every weld is a small casting surrounded by a very large heat sink. The colder and thicker that surrounding metal is, the faster it pulls heat out of the weld and the heat-affected zone. A fast quench through the transformation range produces harder, more brittle microstructure in the heat-affected zone, and hardness in that region is the precondition for most of the cracking an inspector worries about.
Raising the temperature of the surrounding metal before welding narrows the gap the weld has to cool across, which slows the cooling rate and softens the resulting microstructure. It also holds the joint above ambient for longer after the arc moves on, and that extra time at temperature is when dissolved hydrogen can diffuse out of the metal instead of being trapped in it.
The third effect is mechanical. A joint welded cold develops steeper thermal gradients, and steeper gradients mean higher local stresses as the weld and the base metal contract at different rates. Preheat flattens the gradient and takes some of that stress out of the equation before it can combine with a hard microstructure.
- Slows cooling through the transformation range, producing softer heat-affected zone microstructure
- Extends time at temperature, allowing dissolved hydrogen to diffuse out
- Reduces thermal gradients and the residual stress they generate
- Drives off surface moisture, which is a genuine benefit but not the reason preheat is specified
The three ingredients of hydrogen cracking
Hydrogen-assisted cracking needs three things present at once: hydrogen in the weld metal, a susceptible hardened microstructure, and tensile stress. Remove any one and the mechanism stops. Preheat is one of the few controls that acts on two of the three at the same time, which is why it appears on procedures rather than being left to the welder.
Hydrogen arrives from moisture in electrode coverings and fluxes, from damp or contaminated joint surfaces, from oil and paint, and from humidity. Low-hydrogen consumables and their storage rules attack this ingredient directly, which is why they and preheat so often appear on the same procedure as two controls aimed at one mechanism.
Susceptible microstructure comes from steel chemistry combined with cooling rate, which is what carbon equivalent formulas try to summarise in one number. Restraint supplies the third ingredient: a joint that cannot move as it contracts builds tensile stress by itself, so a heavily restrained connection needs more caution than the same joint welded free.
How a preheat temperature gets chosen
Preheat is not a single number for a material. It is set from the combination of steel chemistry, section thickness, the hydrogen level of the consumable, and the degree of restraint. Codes handle this either with prescriptive minimums arranged by material group and thickness, or with a calculated approach that takes carbon equivalent and hydrogen level as inputs.
Thickness matters twice. Thicker sections conduct heat away in three dimensions rather than two, which quenches the weld faster, and they carry more restraint because there is more metal resisting contraction. This is why a procedure can require no preheat on thin plate and a substantial preheat on the same steel in a heavy section.
The chemistry side is what carbon equivalent captures. Higher carbon and higher alloy content shift the steel toward forming hard structures at cooling rates that a plainer steel would tolerate. Two plates that meet the same strength specification can behave quite differently, which is why the material test report matters and why substituting material without re-examining preheat is a real risk.
Interpass temperature: a floor and a ceiling
On a multi-pass weld the preheat requirement does not end when the first bead is deposited. The minimum interpass temperature is the same requirement restated for every subsequent pass: the joint must still be at or above that temperature when the next arc is struck, or the benefit has been thrown away partway through.
The maximum interpass temperature exists for the opposite reason. Each pass adds heat, and on a heavily welded joint the accumulated temperature can climb well past what the procedure qualified. Too much heat means a slower cooling rate than intended, which coarsens grain structure in the weld metal and the heat-affected zone and typically costs toughness. On quenched and tempered steels it can also undo the heat treatment the base metal relies on.
So the procedure defines a band rather than a threshold, and the welder has to work inside it. In practice the minimum is a scheduling problem on cold days and the maximum is a scheduling problem on hot ones or on high-deposition work, and both are worth watching rather than assuming. Neither end of the band is advisory and neither is self-enforcing.
- Minimum interpass - carries the preheat benefit through every subsequent pass
- Maximum interpass - prevents accumulated heat from coarsening structure and costing toughness
- Both are procedure values, checked before each pass rather than once at the start
- Waiting for a joint to cool below a maximum is legitimate work, not lost time
Measuring it properly
Three tools cover almost all of it. Temperature-indicating crayons melt at a rated temperature and give a clear pass or fail against that one value. Contact pyrometers read a number directly and are the usual choice where a band has to be verified rather than a threshold. Non-contact infrared instruments are fast and convenient but are sensitive to surface condition, since a bright or scaled surface reflects rather than emits and reads low.
Where you measure matters as much as what you measure with. Reading directly on the joint immediately after heating gives the temperature of the surface the torch was pointed at, which is not the temperature of the section. The usual practice is to measure at a stated distance from the joint, and on thicker material to check the opposite face, because what the requirement is really asking about is whether the mass of metal is warm rather than whether its skin is.
Time is the other half. Heat needs a soak period to travel through the section, and a joint measured the instant the torch is removed will read high on the surface and cold in the middle. A procedure that specifies a soak time is specifying it for that reason, and it is one of the easiest requirements in welding to skip without anyone noticing until a crack turns up.
- Confirm the procedure's minimum preheat and its interpass band before heating starts
- Heat the area the procedure specifies, not just the weld line
- Allow the specified soak time for heat to reach through the section
- Measure at the distance from the joint the procedure states
- Check the opposite face on thicker sections where access allows
- Re-check before each subsequent pass against both the floor and the ceiling
Where it goes wrong on the floor
The most common failure is heating quickly with a torch, reading the bright spot, and striking an arc immediately. The surface passes the check and the section underneath is still cold, so the cooling rate is essentially unchanged and the requirement has been satisfied on paper only. The reading was honest; the method was not.
The second is losing the preheat during a break. A joint that met the minimum before lunch will not meet it afterwards, and picking up where the last pass stopped without rechecking is a silent departure from the procedure. Long joints have the same problem along their length, since the far end can drop below the minimum while the near end is still being welded.
The third is treating a maximum interpass as advisory. It reads like the kind of limit that exists for comfort, and it is not: it protects the mechanical properties the procedure was qualified to produce. Waiting for a joint to cool is real work with a real purpose, and an inspector who has never once seen a joint being allowed to cool on a busy shop is probably looking at an uncontrolled variable.
What the inspector is checking
Before welding, the checks are documentary and physical together. Does the procedure state a preheat and an interpass band, and does it match the material and thickness actually present? Is the heating method appropriate and controllable? Is the measuring equipment right for the job and in calibration where calibration is required? Has the joint area been heated, soaked, and verified at the correct location?
During welding, the checks are about persistence rather than achievement. Preheat is easy to reach once and easy to lose repeatedly, so the useful observations are made after breaks, at the far ends of long joints, on the second shift, and when the ambient temperature drops. Sampling those moments tells you far more about compliance than watching the first pass.
Recording is the last piece. Temperatures measured, where they were measured, when, and by what method belong in the record alongside the parameters, because if a hydrogen crack appears weeks later the cooling history is the first thing anybody will want and the hardest thing to reconstruct after the fact.