What a Root Pass Has to Achieve

Why this matters

The root pass is the only bead in the joint deposited onto cold, unfused, fully restrained parent metal with nothing behind it. It has to fuse two root faces it can barely see, produce a back surface nobody will inspect until later if at all, and then survive the shrinkage of every pass laid on top of it. Get it wrong and you have built a crack-shaped discontinuity at the exact location where the joint is most highly stressed, and covered it with sound metal. That is the defect a shop finds two years later on a radiograph, at which point the fix is to remove all the good welding above it.

Two hazards belong to root welding specifically and neither is the arc. A back purge with argon inside a pipe or vessel displaces oxygen and argon is heavier than air, so vent the purge exhaust to open air away from anybody's breathing zone, never purge into a space someone is working in or below, and treat any pipe or vessel interior as a permit-required confined space under 29 CFR 1910.146 before anyone enters or reaches in. Back gouging with air carbon arc throws molten metal several feet, generates a heavy fume plume needing local exhaust or respiratory protection under a written program meeting 29 CFR 1910.134, and produces noise inside a hearing conservation program under 29 CFR 1910.95. Both sit on top of the ordinary arc controls: filter shade for the process and current with screens per 29 CFR 1910.133 and 29 CFR 1910.252(b)(2), and fume control matched to the base metal and coating.

The three duties, in the order they are lost

Fuse both root faces. Not one and not most of one. Fusion at the root is the joint; a root pass sitting on top of two unfused faces is a bead in a slot.

Produce an acceptable back surface. The back side is a real inspected surface with real limits: excessive penetration, suck-back (a concave dip on the back), and root concavity all have acceptance criteria owned by the code your WPS invokes, most often AWS D1.1 in the adopted edition for structural work or the governing piping or vessel code with ASME Boiler and Pressure Vessel Code Section IX for pressure work. None of those criteria are "looks fine."

Survive what comes next. Every following pass shrinks onto the root. A thin or concave root has less cross-section to carry that shrinkage, and it cracks along its centreline while the fill is going in. This is why a root bead that a welder describes as "just tacking it together" is a false economy: it is the most loaded pass in the joint during fabrication.

Why the root is the stress concentration

A lack of penetration at the root is not a void, it is a sharp planar discontinuity oriented through the thickness, which is the geometry of a crack. In a statically loaded member it may carry service life indefinitely. In anything cyclically loaded it is a fatigue initiation site sitting at the surface of highest stress, and in a pipe carrying pressure it is the notch under the hoop stress. That is the reason acceptance criteria at the root are tighter than criteria for the same size discontinuity in the cap, and it is not an inspector's preference.

The four root defects and the geometry behind each

Three of the four are dominated by fit-up, which is the joint preparation card's territory, though each carries a technique term as well. Suck-back is the one where the welder's minute-to-minute control is the dominant term.

One requirement, two ways to meet it

The requirement never changes: both root faces fused, an acceptable back surface, enough section to survive the fill. What changes is who does the work.

Open root. The welder controls the keyhole and the back bead directly, from one side, with nothing behind the joint. Gas tungsten arc with a back purge and a filler rod, or a cellulosic covered electrode run open root, are the two common routes. The fit-up tolerance is tight because the welder has no second chance, and the position test behind it is the harder one. This is the only option when the back is unreachable, which on any closed pipe is always.

Backed, or back gouged. A backing bar or backing ring supports the root so the opening can be larger and burn-through is not the constraint, or the joint is welded from one side, back gouged to sound metal from the other, and back welded. The gouge has to reach sound metal and be verified visually, and where the code requires it, by penetrant or magnetic particle before the back weld goes in. The welder's control matters less; the second operation matters a great deal.

The trade is operations against skill and access. Backing adds a fit-up operation or adds gouge, grind and back weld, so three extra operations on a joint that could have been done in one. Open root eliminates them and moves the entire risk into one pass by one person on one fit-up.

What decides it: access first (on a closed pipe there is no choice), then the code and the drawing (a permanent backing bar is a detail the designer has to accept, because it leaves a crevice and changes the fatigue category), then the qualification you actually hold. A shop whose welders are qualified with backing is not qualified without it, and that direction only runs one way. The welder qualification card owns that rule.

Worked example: a 6 in pipe spool, open root, checked at four clock positions

A shop is fitting a 6 in pipe butt weld, open root, gas tungsten arc root with a back purge, filled with covered electrode. The WPS specifies a root opening of 1/8 in plus or minus 1/32 in, and a root face of 1/16 in plus or minus 1/32 in.

Fit-up check with a taper gauge at 12, 3, 6 and 9 o'clock, before tacking.

Reading the sample honestly. Four points measured, one out of tolerance. That is 1 of 4 sampled positions, not 25 percent of the circumference: a taper gauge at four clock positions is a sample, and the joint between 6 and 9 o'clock was never measured. If one of four sampled points is out, the correct statement is that the fit-up is not under control, not that three quarters of it is fine.

What was predicted from the numbers, before welding. At 9 o'clock, a 3/16 in opening against a 1/16 in root face means the arc has 3/16 in of gap to bridge with a land only a third of that gap thick. That is the burn-through geometry in the table above. At 12 o'clock, sitting at the bottom limit with a full 1/16 in land, the risk runs the other way and the root will need the upper end of the qualified current range to reach full penetration.

What was done, and what happened when it was not. The fitter re-set 9 o'clock and re-tacked. On the next spool the same 3/16 in reading was accepted with the argument that the welder would "just carry it," and the root burned through over about 1 in of length at that position. The repair was to grind out the burn-through to sound metal, re-fit with a bridge and re-weld: one repair on one joint, against a re-fit that takes a fraction of the time and leaves no repaired root under the fill.

Correction, printed. The WPS's plus and minus is part of the specification, so the acceptance question is 3/16 in against an upper limit of 5/32 in, not 3/16 in against the 1/8 in nominal. Comparing to the nominal alone would have reported the joint as 1/16 in wide when it is in fact 1/32 in beyond the limit, and the two conversations lead to different decisions.

What would flip the whole approach. Move this joint to a plate assembly where the back is reachable and backing or back gouging becomes available, and a 3/16 in opening stops being a defect and becomes an ordinary fit for a backed joint. The requirement did not change; the way it is met did, and the tolerance came with the method rather than with the requirement.

Check it against the rules the siblings state

• Tolerance band printed with every acceptance call: 3/32 in and 5/32 in are the computed limits from 1/8 in plus or minus 1/32 in, and the 3/16 in reading is judged against 5/32 in, not against the nominal.
• Count word checked against the table: four clock positions measured, one out of tolerance, and the prose refuses the 25 percent reading because a four-point check is a sample of the circumference.
• Root face and opening treated as a pair, per the joint preparation card, rather than as independent numbers: the 9 o'clock and 12 o'clock predictions each name both figures.
• The qualification direction is stated the way the welder qualification card states it: a test with backing does not qualify work without backing, and the example does not assume otherwise.
• No number appears in the example that the general sections did not establish: 1/8 in, 1/16 in and their tolerances all come from the WPS named before the fit-up check.

References

• AWS D1.1, structural welding code for steel, in the edition invoked by your project specification or the adopted building code, for root acceptance criteria, backing details and back gouging requirements
• ASME Boiler and Pressure Vessel Code Section IX with the governing piping or vessel code, as adopted by your jurisdiction, for pressure work root criteria and purge requirements
• 29 CFR 1910.146 for permit-required confined spaces where a purged pipe or vessel is entered, 29 CFR 1910.95 for arc gouging noise, 29 CFR 1910.134 for respiratory protection, and 29 CFR 1910.133 with 1910.252(b)(2) for arc radiation
• See related: Why Joint Preparation Decides More Than Technique; What a Welder Qualification Covers and What It Does Not; How to Read a Welding Procedure Specification