Why Grinding and Gouging Are Part of the Weld

Why this matters

Removing metal is treated as the boring half of a repair, which is why repairs fail twice as often as original welds do. An excavation is a joint preparation somebody cuts freehand, in a restrained assembly, usually in a hurry, and the dimension that decides whether the repair weld will be sound is not how much metal came out. It is the sidewall angle left behind, because that angle is the maximum angle at which an electrode can be aimed into the sidewall at all. Cut a narrow slot and the geometry has already decided you will get sidewall lack of fusion, whatever the welder does.

Before removing metal from anything installed: confirm with the engineer of record or the party responsible for the structure that the member is unloaded or independently supported before any of it comes out, because an excavation reduces the section while it carries load. Air carbon arc gouging is loud enough to put the work inside a hearing conservation program under 29 CFR 1910.95, throws molten metal several feet so the floor below is cleared and the hot work permit's fire prevention and fire watch apply under 29 CFR 1910.252(a) or 29 CFR 1926.352 in construction, and produces a heavier fume load than the weld it is removing, an inhalation route needing local exhaust at the arc or a respirator issued under a written program meeting 29 CFR 1910.134. Grinding puts the base metal and any coating into the air by the same route, with capture at the tool: stainless and chromate primers give hexavalent chromium under 29 CFR 1910.1026 or 29 CFR 1926.1126, old paint can give lead under 29 CFR 1910.1025 or 29 CFR 1926.62. A cutting or grinding wheel run above its rated speed or mounted on a machine without the guard is a burst hazard, so match the wheel rating to the tool and keep the guard on, per 29 CFR 1910.243(c) and 29 CFR 1910.215.

The repair that failed twice

A 3/4 in plate splice came back from radiography with a rejectable elongated indication about 2 in long, at mid-thickness. The shop marked it, gouged it out, re-welded, re-shot. The second radiograph showed a rejectable indication in the same place. They removed more, re-welded, re-shot. Third radiograph, same result, and now the assembly was three heat cycles into a repair that started as one.

At that point the arguments in the shop were all about the welder, the consumable and the machine. All three had been changed at least once across the three attempts, and none of the changes helped, which is the tell that the variable being changed is not the variable that matters.

What the third look measured

Somebody put a rule and a profile gauge into the cavity before the fourth attempt. The excavation was 1/2 in deep. It was 3/8 in wide at the top and 3/8 in wide at the bottom, which is to say it was a slot with vertical walls, cut by running the gouging electrode straight down the marked line three times.

That is the whole finding, and it is geometric rather than metallurgical.

The access criterion, stated as a number

An electrode has to be aimed into a sidewall to fuse it, not merely held near it. The angle available for that is set by how far the sidewall leans back from vertical, minus the room the electrode's own body takes up.

For a cavity of depth d, a sidewall inclined at an angle from vertical, and an electrode of radius r held with its tip in the bottom corner, the work angle into that sidewall is limited by:

tangent of the work angle, at most, equals the tangent of the sidewall angle minus r divided by d.

Run the original slot through it. Sidewall angle zero, so its tangent is zero. Electrode radius 0.0625 in for a 1/8 in electrode, depth 0.5 in, so r divided by d is 0.125. Zero minus 0.125 is negative. There is no positive work angle into that sidewall. The electrode could not even be laid parallel to the wall without touching it, let alone pointed at it. Every one of those three repairs was welded into a cavity that made sidewall fusion geometrically impossible, and the flaw the radiograph kept finding was a new one each time, in the same place, for the same reason.

Now size a cavity that works. Ask for a modest 20 degree work angle into each sidewall. Tangent of 20 degrees is 0.364. Add r over d, and use the finished depth after cleanup grinding, 0.5625 in, so 0.0625 divided by 0.5625 is 0.111. The sum is 0.475, and the angle whose tangent is 0.475 is about 25.4 degrees. So the sidewalls must lean back at least about 25.4 degrees from vertical, and cutting them at 30 degrees leaves margin in the safe direction.

The layer the gouge leaves behind

Air carbon arc gouging melts metal and blows it clear with compressed air, using a carbon electrode. Some of that carbon transfers into the melted surface, leaving a thin carbon-enriched skin on every gouged face. Weld onto it and the deposit picks up carbon it was not designed for, which raises its hardness and its cracking risk. The carbon-enriched layer is ground off before welding, on the order of a sixteenth of an inch, with the actual figure owned by the governing document or the equipment manufacturer's instructions rather than by habit.

Two related pickups belong to the same operation. Copper from a copper-coated gouging electrode can transfer to the cut surface and cause cracking in the deposit, so a gouge that ran rough or stubbed gets inspected as well as ground. And an excavation in a completed assembly sits under more restraint than the original joint did, which is why the preheat for a repair can be higher than the preheat for the original weld; that number belongs to the repair procedure or the engineer, never to the shop floor.

Grinding is not neutral either

It has a direction. Grinding marks are small parallel grooves, and in a fatigue-loaded member they behave like the notches they are. Where a surface is being dressed on a member that sees cyclic load, the final pass runs parallel to the direction of the stress, not across it.

It generates heat. A wheel pressed hard into one spot heats a local patch fast and the cold metal behind quenches it, which is a miniature version of what a cutting torch does to an edge. Temper colours on a ground surface are the visible end of that.

It conceals as easily as it removes. Dressing the mouth of a crack makes the crack invisible without shortening it. After a defect is removed, the cavity gets re-examined by a method that can find what you were removing, before any metal goes back in.

Back-gouging is the planned version of all of this. Removing the root from the second side of a double-sided groove to reach sound metal is the same operation with the same criterion, done on purpose, and it is why a back-gouged groove is cut as a U with leaning sides rather than as a slot.

The fourth repair, cut to the geometry

The cavity was reopened to a U with a rounded bottom: 1/2 in wide at the bottom, sidewalls at 30 degrees from vertical. Then every gouged surface was ground back a sixteenth of an inch to take out the carbon-enriched skin, which is a re-basing rather than an addition, because it changes the cavity the criterion has to be checked against.

• Depth after the cleanup grind: 1/2 in gouged plus 1/16 in ground off the floor, so 0.5625 in.
• Bottom width after the grind: 1/2 in.
• Top width required for 30 degree sidewalls at that depth: 0.5 plus twice 0.5625 times the tangent of 30 degrees, which is 0.5 plus 0.6496, or 1.1496 in. Called out on the sketch as 1-3/16 in, rounding up rather than to the nearer 1-1/8 in, because rounding a required opening down is the flattering direction.
• Sidewall angle actually achieved at 1-3/16 in top width: the top half-width is 0.59375 and the bottom half-width is 0.25, so the lean is 0.34375 over a depth of 0.5625, a tangent of 0.611 and an angle of about 31.4 degrees. That clears the 25.4 degrees the criterion asked for.

Magnetic particle examination of the finished cavity confirmed the original defect was gone before any metal went back. The joint was preheated per the repair procedure, filled, and radiographed clean. The variable that fixed it was the sidewall angle, and it had been available to be measured before the first attempt.

Check the example against the rules the siblings state

• The criterion is applied to the finished geometry, not the as-gouged geometry. Depth 0.5625 in after the cleanup grind, not the 0.5 in as gouged, and the required sidewall angle of 25.4 degrees was computed on 0.5625.
• Arithmetic re-derived. tan 20 degrees is 0.364; 0.0625 over 0.5625 is 0.111; the sum 0.475 gives 25.4 degrees. Top width: 2 times 0.5625 times 0.5774 is 0.6496, plus 0.5 is 1.1496 in. Achieved lean: 0.34375 over 0.5625 is a tangent of 0.611, or 31.4 degrees.
• Neither rounding flatters the repair. The required 1.1496 in top width was called out at 1-3/16 in (1.1875) rather than 1-1/8 in (1.125), and the sidewall was cut at 30 degrees against a computed minimum of 25.4.
• The original slot is scored against the same criterion. Sidewall angle 0 degrees, tangent 0, minus 0.125 for r over d at 0.5 in depth, giving a negative result and therefore no achievable work angle. That is why it is called impossible rather than difficult.
• Direction check against the siblings. The lack-of-fusion card has narrow grooves and poor sidewall access producing sidewall lack of fusion, and has radiography weak on that flaw. Here radiography did catch it, three times, because the slot's sidewalls were vertical and the shot was vertical, so the flaw plane lay along the beam rather than across it. That is radiography's best case by the method card's own gate, and it is the same rule the sidewall flaw in a bevelled groove fails, not an exception to it. That is consistent with the method card's rule rather than an exception to it, and it is stated so a reader does not take radiography as reliable for this flaw in general.

References

• AWS D1.1, structural welding code for steel, in the edition invoked by your project specification or the building code your authority having jurisdiction has adopted, for repair procedures, removal of defects and re-examination requirements
• 29 CFR 1910.95 (hearing conservation), 29 CFR 1910.134 (respiratory protection), 29 CFR 1910.215 and 1910.243(c) (abrasive wheel guarding and speed rating), 29 CFR 1910.252(a) and 1926.352 (hot work)
• See related: What Lack of Fusion Is and Why It Hides; What Thermal Cutting Does to the Edge You Are About to Weld; Why Joint Preparation Decides More Than Technique; What Each Nondestructive Method Can and Cannot See