What the cut edge says about the pass that made it

The plate came off the table at ten past six, still warm, and landed on the bench: half inch A36, a base plate for a school gym job, ten nested on one 4 by 8 sheet. The fitter turned it over and looked at the bottom edge. Then he said the table was running fast on the back half of the sheet.

That is the whole check. Everything after it is writing down what the edge already said.

What was read on the bench, and with what

Protocol first. One base plate, 1/2 in A36, 18 by 18 in, cut on the 130 amp plasma table at 0.06 in standoff and the chart speed for that thickness. Four tools came with it: a 6 in scale, a dial caliper to 0.001 in, a 12 in machinist square, a 4 ft straightedge. Both went against the cut edge, not the mill edge.

Three readings came off. Kerf: 0.075 in about 2 in in from each end, 0.082 in at the middle. Top edge rounding: roughly 0.015 in, read by feel and in the caliper jaws. Bevel: about 3 degrees off square on the right side of travel, under a degree on the left. Dross: light and spotty, heaviest at the lead-in and where the plate crossed a slat.

Readings off one plate, not a table of what plasma does.

Why is there dross on the bottom and not the top?

Dross is metal the jet melted and did not blow clear. It lands on the bottom because that is where the arc exits. The cut-quality charts the torch maker publishes at Hypertherm sort it into kinds that point opposite ways.

Heavy, bubbly dross that knocks off with a scraper is low speed dross: the arc dwelled and the jet could not carry the melt out. Thin, tightly stuck dross that takes a grinder is high speed dross: the arc ran ahead of the cut and left metal behind. Dross on top, usually splatter at the pierce, comes from too much standoff or a worn nozzle.

This plate showed the second kind on the last four parts: 130 amps on 1/2 in mild steel charts around 45 ipm, and the back half of that sheet ran a nozzle that had been on since morning.

Bevel and kerf: the two numbers nobody writes down

Kerf is the width of material the jet removed: at 130 amps on 1/2 in, usually between 1/16 in and 3/16 in. What matters is which side of the line the machine put it. Cut on the line with no kerf compensation and every part comes out undersized half a kerf per edge: 0.08 in of kerf is 0.04 in a side, 0.08 in overall. Against a fit-up tolerance of plus or minus 1/8 in, read off the erection drawings and consistent with the AISC Code of Standard Practice, that passes, though it spends margin the next check wants back.

Bevel bites later. Three degrees off square sounds like nothing: on a 1/2 in plate it drops the far edge about 0.026 in, a shade over 1/64 in. On a 1 in plate it is twice that, and a bevel ground out of a groove weld face is grinding time nobody bid.

What does three degrees cost at fit-up?

Stand a stiffener cut with that bevel against a column flange and the gap is on one side only. The welder fills it, and on a fillet that is the end of it. It stops being the end of it where the drawing calls a groove detail, on a part that shows, or on anything that bears.

Bearing is the line a plasma edge does not cross here. A column bearing on a base plate, a strut on a gusset, any end the drawing shows as full contact: those get saw cut or milled. A plasma edge is not a bearing surface however clean it looks.

When the edge passes as-is

Most cut edges in a structural package pass. The read on this plate: pass, with the dross scraped off the bottom and the last four parts marked for a second look.

An edge passes when it is not a fit-up face (the outside of a clip, the profile of a fillet welded stiffener), when the bevel lands on the side the weld goes, when the dross comes off in one pass of a scraper and leaves no gouge, and when nothing bears on it and no hole sits closer than about 1/4 in to it.

Coating is part of the pass. Dross left on the steel is not steel: paint bridges it and flakes off it, and in a galvanizing kettle loose dross comes away and leaves a bare spot that shows as a rust bloom a year later. Ten seconds with a scraper settles it.

When it gets ground

It gets ground when the edge is a fit-up face on a joint that will be inspected, when the part is exposed architectural (handrail, an exposed column in a hospital wing addition lobby, anything an owner will run a hand along), when the drawing calls a groove weld the welder has to see into, and when the finish coat is the final surface.

Grinding takes off the dross, the top edge rounding, and the heat-affected layer the cut left. It does not fix a part cut 1/8 in small, and it does not square a beveled edge unless somebody grinds it square on purpose.

What this note does not cover

It does not cover the other nine plates on that sheet. One was read.

It does not cover the material. The mill certificate speaks for the chemistry and the mechanical properties of the heat: read it against the spec on the drawing, not the edge.

It does not cover the weld. The edge is not the weld, and passing this check says nothing about whether the joint that follows meets the acceptance criteria in AWS D1.1. That is a different inspection with different tools.

It does not cover hardness or the heat-affected zone: no hardness tester came out, and no section was cut and etched.

It does not certify the part dimensionally. Overall dimensions are a separate check with the tape, the hole pattern another after that.

And it is not a certification of the shop or of the work. The standards named here are named as references, the way a drawing names them.

What the next guy can do

Before calling a plasma cut good, turn the part over. Run a scraper along the bottom edge: nothing, powder, or a scab that fights you. Then lay the square on the cut face and look at the light under it.

Write the line on the back of the drawing: thickness, amperage, speed, one word for the edge. Months from now, when the same thickness comes up on another job, that line is the only record of what the table did. The edge will tell the next guy anyway. It is better when the drawing agrees with it.