The fit-up check that happens before the weld goes in

The plate sits on the bench bevel side up, the shop drawing is taped to the back of a sawhorse, and the taper gauge is already out of the pouch because it is the first tool this check needs. Note S1-014, shop bench, checked before the welder pulled his hood down. Two plates, 1/2 in thick, 12 in wide, set up for a single-V groove splice in the flat position. What follows is what the instruments read on this joint, in the order they were read.

What the procedure called for before anything got measured

Fit-up has no opinion of its own. It gets checked against numbers written down before the plates were tacked, so the first thing on the bench is not a tool, it is the welding procedure specification. Ours calls for a single-V groove, 60 degrees included, 30 degrees of bevel per plate, root face 1/16 in, root opening 3/16 in, E70XX electrode at 1/8 in, no preheat required at this thickness. The detail sits under AWS D1.1. The plate arrived with a mill certificate listing a heat number, an A36 chemistry, yield at 39.8 ksi and tensile at 71.2 ksi. That certificate says what was poured. It says nothing about whether the plate is flat, square or the thickness it claims, which is why the caliper comes out later in this same check.

The root opening, read with a taper gauge

Taper gauge, leaves from 1/16 in to 1/2 in, worked into the gap at three points along the joint. Near end: 7/32 in. Mid-length: 3/16 in. Far end: 3/16 in, maybe a hair under. The procedure wants 3/16 in, so two of the three readings sat on the number and one stood 1/32 in proud. That is not a crisis by itself, but it changes how much metal goes into the groove at that end, and on a joint welded complete joint penetration the volume of the groove is the whole argument. The three numbers went onto the back of the drawing next to a small sketch of the joint, with the time beside them.

How do you check a bevel without trusting your eye?

A bevel protractor, zeroed on the plate face, then read against the cut edge. Left plate: 27 degrees. Right plate: 31 degrees. Combined included angle, 58 degrees, against a procedure that calls for 60. The eye would have passed both plates. Three degrees shy on one plate is invisible on a bench and expensive in a groove, because the bevel is what lets the electrode reach the root in the first place. The bevel on these plates was put on with a track torch and cleaned with a grinder, and the edge shows it: a light chamfer on the top corner where the cut ran out, and a drag line underneath. The 27 degree plate did not come off the track that way. It came off the grinder, where somebody took a little off the toe to clean up dross and kept going.

Root face, the land that decides where the root sits

Caliper here, not a gauge, because the land is small and wants a real reading. The procedure calls for 1/16 in of root face, the flat at the bottom of the bevel that keeps the arc from blowing through before the welder is ready for it. Left plate read 1/16 in. Right plate read 3/32 in. That extra 1/32 in is the one that costs. A heavy land pushes the root pass up out of the groove, and then the welder either turns the machine up to burn through it and risks a keyhole, or he leaves lack of penetration at the root where nobody sees it until an inspection that is not on the schedule. Plate thickness came off the same caliper on the same pass: 0.495 in and 0.503 in against a nominal 1/2 in, both inside what a mill is allowed to ship.

Hi-lo, or how far the two plates disagree about where the surface is

Straightedge across the joint, taper gauge underneath the low side. The two faces are not coplanar. One plate sits 3/32 in above the other at the joint line. That step is the mismatch, sometimes called hi-lo, and it is not a gap anybody can fill, it is a ledge. The procedure allows up to 1/8 in of mismatch at this thickness. At 3/32 in the joint was inside the limit, so it was written down and left alone, and the welder was told which side was high so he could favor the low side on the root pass. Worth saying plainly: 1/8 in is the number on this procedure, not a number handed down to all steel everywhere. Read your own.

What got re-prepped, and what did not?

Two things came off the bench. The right plate went back to the grinder to bring the 3/32 in land down to 1/16 in, and it was checked again with the caliper before it came back. The left plate had its bevel opened from 27 to 30 degrees, read with the protractor at both ends of the bevel, because a bevel that is right at one end and shy at the other is worse than one that is uniformly shy. The 7/32 in root opening at the near end was closed with a wedge and a strongback rather than by grinding, which is the cheaper fix and leaves the bevel alone. The 3/32 in mismatch stayed. Bench time for the check and the re-prep together: about forty minutes, most of it grinding.

What does this note not cover?

This is a fit-up note. It does not certify the weld. It does not say the joint will pass ultrasonic testing, it does not speak to the welder's qualification or the procedure's qualification, it does not check the consumable lot, and it does not match the mill certificate against the heat number stamped on the plate (that is a separate check done when the material is received). It does not cover what happens after the root pass. It does not measure distortion, because the joint has not been welded yet and distortion is a thing the joint does afterward. And it covers one joint. A second joint on the same drawing could read differently and would need its own note with its own number.

What a GC or an owner can ask for

Ask for the fit-up numbers rather than a statement that the joint was fitted. A tape, a taper gauge, a protractor, a caliper, three readings and a time, written on the back of the drawing, cost almost nothing and they are the only record that the groove about to be filled is the groove the procedure describes. Ask which joints were re-prepped and why, and keep the answer. The measuring in this note took under ten minutes. The re-prep took forty. Both are cheaper than a repair on a joint that is already painted and sitting behind a finished wall.