Why Preheat Exists and What It Is Controlling

Why this matters

Preheat is widely explained as warming the metal so it welds better, which is not a mechanism and leads shops to skip it whenever the metal already feels warm. It is two specific controls at once, and it is not free: it slows the joint's cooling so the band beside the weld transforms into something softer, and it holds the joint warm long enough for dissolved hydrogen to diffuse out before the metal is cold enough to crack, while heating a larger volume of the assembly and therefore usually increasing distortion, which the distortion card prices. A shop that understands only the first half stops preheating on a warm day. A shop that understands only the second half stops as soon as the arc is out. The number itself is not a shop decision, and this card is about producing the three inputs that let somebody else give you a defensible one.

Before applying heat to anything: never heat a sealed hollow member, a closed vessel or any assembly that can trap liquid or vapour without venting it first, because heating a sealed volume raises its pressure until something opens, and a hollow structural section with water in it is a common and violent version of this. A container that has held a flammable or its residue is not heated, cut or welded until it has been cleaned or inerted by an established procedure, which 29 CFR 1910.252(a)(3) makes a condition of the work. Preheating with fuel gas in an enclosed or poorly ventilated space produces carbon monoxide, an inhalation route with a limit in 29 CFR 1910.1000 Table Z-1, and it consumes oxygen, so it needs mechanical ventilation and, in a tank, pit or vessel, the permit space controls at 29 CFR 1910.146 with the welding ventilation requirements at 29 CFR 1910.252(c). Heating a painted, primed or galvanized surface drives off the coating as fume before the arc is ever struck: lead under 29 CFR 1910.1025 or 29 CFR 1926.62 in construction, hexavalent chromium from chromate primers under 29 CFR 1910.1026 or 29 CFR 1926.1126, zinc oxide from galvanizing, each an inhalation route needing local exhaust or a respirator issued under a written program meeting 29 CFR 1910.134. Preheated steel radiates and stays hot long after it stops glowing, so a temperature-indicating crayon or a contact pyrometer, not a hand, establishes whether a piece is safe to move.

What preheat is actually doing

It slows the cooling rate through the transformation range. A hardenable steel that cools quickly from welding temperature forms a hard, brittle structure in the band beside the weld. Start the joint warm and the same energy leaves more slowly, the band transforms into something softer and tougher, and the microstructure that hydrogen cracking needs is largely not there to crack. Why the joint's own mass and geometry dominate that cooling rate belongs to the heat-affected-zone card.

It holds the joint above ambient long enough for hydrogen to leave. Hydrogen diffuses out of steel quickly when it is warm and very slowly when it is cold. A preheated joint spends far longer in the range where hydrogen can escape, so less of it is still dissolved by the time the metal reaches the temperature where cracking runs, which is roughly below 300 degrees F and most active near ambient.

Those are two different effects with two different timescales, and the second one is why preheat is not finished when the arc starts. Interpass temperature is the same control continued through the weld, and on a heavy multi-pass joint a post-weld hold, sometimes called a hydrogen soak, extends it past the last pass.

What preheat is not

• It is not a drying step. Warming a plate above the shop dew point removes condensation and satisfies no preheat requirement, and the porosity card works through a case where a shop confused the two.
• It is not a substitute for clean metal or dry consumables. It reduces the consequence of hydrogen, it does not remove the source.
• It is not stress relief. Post-weld heat treatment at a much higher temperature is a different operation with a different purpose, and preheating does not partially accomplish it.
• It is not universally good. Some materials carry a maximum interpass temperature rather than a minimum, because too much time hot is the damaging condition. Austenitic stainless is the everyday example, and the card on what a welded joint changes about the parent metal explains why.

The three inputs, and why none of them is enough alone

The steel's chemistry, expressed as a carbon equivalent. Carbon equivalent rolls the hardening contribution of the alloying elements into one figure so a single number can stand in for the whole analysis. The formula in common use is carbon, plus manganese divided by 6, plus chromium plus molybdenum plus vanadium divided by 5, plus nickel plus copper divided by 15. It needs a mill certificate or a laboratory analysis; it cannot be read off a plate.

The combined thickness at the joint. Not the thickness of one part. Combined thickness sums the thicknesses of every element radiating from the joint, measured a short distance out from it, because that is what actually carries heat away. A fillet on a tee has three paths, not one.

The consumable's diffusible hydrogen level. Filler metals carry a hydrogen designator in their AWS A5 classification - H16, H8, H4 and H2, meaning a maximum of 16, 8, 4 and 2 millilitres of diffusible hydrogen per 100 grams of deposited metal. A lower designator earns a lower preheat requirement, and it only holds if the consumable has been stored and handled to keep it there.

Each input moves the answer in a known direction: preheat rises with carbon equivalent, rises with combined thickness and rises with hydrogen level. What it does not do is come out of any one of them. Two of the three plus an assumption is how a shop arrives at a number it cannot defend.

The determination sheet, filled in

Joint: a 1/2 inch web plate fillet-welded both sides to a 3/4 inch flange, tee configuration, shop conditions, restrained by adjacent work.

Carbon equivalent, computed:

• Carbon: 0.20
• Manganese term: 1.10 divided by 6 = 0.183
• Chromium plus molybdenum plus vanadium: (0.10 + 0.02 + 0.00) divided by 5 = 0.024
• Nickel plus copper: (0.05 + 0.20) divided by 15 = 0.017
• Total: 0.20 + 0.183 + 0.024 + 0.017 = 0.424, call it 0.42

Combined thickness, computed, and the correction printed. The obvious entry is 0.75 inch, the thickest part. That figure already contains one heat path and nothing else, so using it is not conservative, it is a different quantity. Re-basing it to the joint: heat leaves this tee through the web and through the flange in both directions, so the combined thickness is 0.50 plus 0.75 plus 0.75 = 2.00 inches. That is 2.7 times the single-plate figure, and preheat tables move upward with it. A shop that looks up 0.75 on a thickness axis reads a lower requirement than the joint has, and the error runs in the flattering direction every time.

What the sheet produces, and who owns the number. Carbon equivalent 0.42, combined thickness 2.00 inches, hydrogen designator H8, starting temperature 58 degrees F. Those four entries are what a preheat requirement is looked up against. The value itself comes from the preheat provisions of AWS D1.1 in the edition your project specification or adopted building code invokes, from the welding procedure specification qualified for this joint, or from the engineer of record where no procedure covers it. This card will not supply a temperature, because the requirement changes with the steel category, the edition and the governing document, and a number carried in from memory is exactly the failure the sheet exists to prevent.

The dew point line is on the sheet and is not part of the answer. 58 degrees F metal against a 51 degrees F dew point means the plate is not collecting condensation, which is a separate condition that has to be true and which no preheat requirement covers. It is on the sheet so it gets checked, and it is segregated so nobody mistakes satisfying it for satisfying preheat.

Check against the sibling rules, with the figures printed. The porosity card states that warming metal above the dew point is a moisture control and not preheat; this sheet records 58 degrees F metal against a 51 degrees F dew point in its own row and draws no preheat conclusion from it. The cracking card states that hydrogen-assisted cracking needs hydrogen, a hard microstructure and tensile stress together; this sheet evidences the hydrogen condition (H8, oven-current) and the hardening condition (carbon equivalent 0.42 at 2.00 inches combined thickness) and notes the restraint, so all three are addressed rather than assumed. The heat-affected-zone card states that combined mass at the joint governs cooling rate; the 2.00 inch figure here is computed on that basis and not on the single-plate 0.75.

What would change the answer. Swap the H8 electrode for an H4 and the same chemistry and thickness earn a lower requirement, which is often the cheapest lever a shop has. Move the same joint outdoors onto 20 degree F steel and the starting temperature drops the answer in the other direction, and most codes carry a floor below which welding is not permitted at all without heat. Change the web to a 1/4 inch plate and combined thickness falls to 1.75 inches, which is a smaller move than most people expect, because the flange is carrying most of the heat away and the web was never the dominant term.

How to verify the preheat you applied is the preheat you got

• Measure before you strike the arc, and measure out from the joint, not on it. A common code convention takes the reading at a distance from the weld equal to the thickness of the thickest part with a minimum of about 3 inches, on the face opposite the heating source where you can reach it. Read the exact convention out of the document that governs your work.
• On heavy sections, wait before reading. Torch heating warms the surface far faster than the interior, and a reading taken immediately after the flame comes off reports the surface only. A common rule is to remove the heat source and wait about one minute per inch of thickness before reading, so on the 2.00 inch combined thickness above that is a couple of minutes, not a couple of seconds.
• Check the whole preheat zone, not one spot. Torch preheat on a long joint is uneven by nature, and the cold end is where the crack starts.
• Re-check after any interruption. A break, a fit-up correction, a bottle change: the joint cools while the paperwork says it is preheated. Interpass temperature is the same requirement running through the weld and it is checked between passes, not once at the start.
• Log the reading with the time and the instrument. A temperature-indicating crayon gives a pass or fail at one value and a contact pyrometer gives a number; both are defensible and only one of them tells you how much margin you had.

References

• AWS D1.1 structural welding code for steel, in the edition invoked by your project specification or the adopted building code, for its preheat and interpass provisions, its preheat measurement convention and its consumable exposure requirements
• AWS A5 series filler metal specifications, which define the H16, H8, H4 and H2 diffusible hydrogen designators and the storage conditions that keep a consumable at its designator
• ASME Boiler and Pressure Vessel Code Section IX with the applicable piping or vessel code, as adopted by your jurisdiction, where the work is pressure-retaining
• 29 CFR 1910.252(a)(3) for containers that have held flammables; 29 CFR 1910.252(c) and 29 CFR 1910.146 for ventilation and confined spaces; 29 CFR 1910.1000 Table Z-1 for carbon monoxide; 29 CFR 1910.134 for respiratory protection
• See related: What the Heat-Affected Zone Is and Why It Governs; What Cracking Tells You About When It Happened; Why Porosity Appears and What It Tells You