Cut marine angle steel to the drawing and the members still miss fit-up. Rework and weld gaps follow. Every cause is physical, and every one can be named and controlled.

Most dimensional errors when cutting marine angle steel come from four sources: kerf width nobody programmed in, stops and feeders that drifted out of calibration, movement inside the hot-rolled section itself, and incoming bars already at the edge of EN 10056-2 tolerance. Name those four, control them, and sawn lengths hold within a millimetre.

I have sold marine angle steel into shipyards for years, and the pattern never changes. The buyer checks the mill certificate, checks the grade, checks the price per tonne. Then the steel arrives, the yard cuts it, and somebody calls me to ask why the brackets do not fit. In almost every case the steel was fine. The cut was not. And the frustrating part is that the four or five causes behind bad cuts are all visible before a single chip flies, if you know where to look. So in this article I will walk through each cause the way I walk through it when a customer calls: first the kerf, then the machine setup, then the steel itself, and finally what belongs on the purchase order so none of this becomes your problem.

Where Does the Kerf Turn a Correct Drawing into a Wrong Cut?

The saw removes steel as it cuts, and that removed strip is the kerf. If the kerf is not accounted for, every piece comes off the machine short by exactly the kerf width. On one piece it is invisible. On four hundred pieces it is a missing metre of steel.

The kerf is the strip of steel destroyed by the blade or the plasma arc. A band saw takes roughly 1.1 to 1.6 mm, a cold saw 2.0 to 2.5 mm, an abrasive wheel 2.5 to 3.0 mm, and plasma 1.5 to 4 mm. Program that width into the stop position, or every cut length lands short by the same amount.

Kerf Width by Cutting Method

The first table I show buyers is the one below, because it settles the argument about whose numbers are wrong. The saw has not lied. The program simply never told it how much metal disappears into the cut.

Notice the plasma row. The kerf is not a fixed number; it widens as the nozzle wears and the standoff drifts. A yard that calibrates kerf compensation once a month is really calibrating last month’s nozzle. On our own fabrication floor in [Zora will replace], we re-test the kerf on scrap whenever the nozzle or blade changes, and I would push that habit on any yard cutting angle steel in volume.

Bundle Cutting Multiplies the Error

Kerf becomes dangerous when it stacks. Cutting singles, one missing kerf of 2 mm per piece sits inside almost any tolerance. Cutting nested bundles, the loss compounds along the bar. I worked through this arithmetic in our earlier post on how to calculate the number of marine angle steel pieces required per project: four hundred cuts at 3 mm per kerf is 1.2 metres of steel, which decides whether your final member fits or gets cut from the next bar. Bundle cutting adds one more trap. In a strapped bundle, only the top bar is against the stop unless the bundle is clamped as a unit. The bars behind it can shift a few millimetres during the cut, and the saw operator never sees it because the top bar measures perfect. The fix is boring and non-negotiable: bundle clamps rated for the profile, and a first-piece length check on the bundle, not just the top bar.

My Personal Rule on Kerf and Stops

Here is the port-side observation I give every customer who will listen. The kerf error is the only dimensional error that is perfectly predictable, perfectly repeatable, and still the most common one I see. It requires no skill to fix, only the discipline to set the kerf value in the program and confirm it with a test cut on scrap. I tell yards to treat the kerf setting like the blade itself: it belongs to the tool, and it changes whenever the tool changes. A two-minute test cut has saved more steel than any negotiation I have ever run.

Why Do Lengths Drift Between the First Piece and the Last?

The first piece off a setup measures correct, and by the end of the shift the same job produces pieces a millimetre or two out. Nothing was changed on purpose. So what moved?

Length drift across a shift comes from feeder and stop calibration, bars slipping in the vise, blade wear changing both kerf and squareness, and burrs that make the measuring tape read short. A first-piece check catches the setup error; a last-piece check catches the drift that built up since morning.

Stops, Feeders and the First-Piece Check

On a production saw, length lives in three places: the length stop or servo feeder, the program’s kerf value, and the operator’s habit of bringing the bar hard against the stop. Every troubleshooting guide I trust lists the same failure chain, because it is the chain that actually happens. The feeder calibration drifts, or the bar does not seat against the stop, or the kerf value in the program was never set after the last blade change. The countermeasure costs two minutes: cut the first piece of every job, measure it cold, and adjust before committing the bar. I would add the less popular half: measure the last piece too. A job that starts perfect and ends short has a drifting cause, usually blade wear widening the kerf or the stop loosening under vibration, and the last-piece check is the only thing that will find it.

Clamping, Blade Wander and Squareness

Length is not the only dimension that goes wrong. Squareness of the cut face fails first, and it fails quietly. A band saw is a long flexible blade held straight by tension and guides; feed too hard, space the guides too wide, or run a dull blade, and the band wanders in the cut. The rule of thumb from sawing practice is that a band saw cut counts as straight when it leans no more than about 0.2% of the cut depth, which on a 200 mm angle is already 0.4 mm of lean across the face. That lean becomes your fit-up gap later. The vise must grip on both sides of the cut, as close to the blade as the jaw allows, and the jaws must match the profile. Angle steel rolled into an unclamped cut will strip teeth and throw the piece. And one habit matters more than any machine feature: squareness should be checked on the first cut after every blade change, because a fresh blade mounted on dirty flanges or a vise knocked out of square will ruin the whole batch before anyone measures it.

Burrs and Measurement Habits

The last cause in this group is the cheapest to fix and the most ignored. A saw cut leaves a burr at the exit corner. Measure across a burr and your tape reads short, or your caliper jaws bridge it and read long. Burrs differ by method: abrasive cuts leave heavy hot burrs, cold saws leave a light one. The discipline is simple. Deburr before measuring, measure at two points across the face, and measure cold. I watched a yard blame an entire shipment for short lengths that turned out to be burr bridges on a hot afternoon, and the supplier shipped replacement steel for nothing. Nobody enjoyed that conversation.

How Much Does the Steel Itself Move Before You Ever Switch the Saw On?

Some dimensional errors are already inside the bar before the saw starts. Incoming tolerance, residual stress from rolling, and temperature all move the steel by amounts that a correct cut cannot recover.

Hot-rolled marine angle steel arrives with leg, thickness and straightness tolerances of its own, typically ±0.5 to ±1.2 mm on thickness and up to 0.4% of length on straightness under EN 10056-2. Rolling stresses release at the first cut, and thermal expansion adds about 11.7 micrometres per metre per degree Celsius. Measure a hot bar and the number is fiction.

Incoming Tolerance You Cannot Cut Away

The saw holds its own tolerance, but it holds it relative to wherever the bar actually is. If the incoming bar sits at the far edge of the mill tolerance, a perfect cut transfers that edge straight into your fabricated member. The bands that matter for angle steel are worth keeping on one card:

The straightness row is the one that bites hardest. A bar bowed 20 mm over its length does not measure 12,000 mm along the tape held against the curve, and the stop-to-blade distance changes depending on where the bow sits relative to the vise. This is the same incoming-tolerance logic I covered in our post on what causes fit-up problems with marine angle steel sections, and it is worth repeating here because cutting does not fix it. The only real defences are buying from mills that hold the tighter band, inspecting straightness at receiving, and planning cuts away from the worst bow.

Residual Stress Release at the First Cut

Hot rolling leaves locked-in stress across the angle’s asymmetric section. The first cut releases it, and the two legs move independently, most often in a bow or a twist toward the newly cut end. Published shop-floor figures put that spontaneous movement in the range of 0.5 to 1.2 mm on ordinary hot-rolled sections, and on asymmetric profiles like angles I would plan for the upper end. It is not a defect and it is not the mill’s fault; it is physics. What separates a disciplined yard from a frustrated one is sequence and support: support the section on both sides of the cut so the offcut does not drop and pry the kerf open mid-blade, and treat the first cut on each new bar as a measurement event, not just a production step. If the first cut piece shows a sudden bow, the following cuts need re-clamping and re-measuring, not blaming.

Heat, Expansion and Measuring Too Soon

Steel expands about 11.7 micrometres per metre per degree Celsius, and this number settles more arguments in cutting shops than any other. Run the arithmetic. A 12 m bar that warms 20 degrees above the reference temperature has grown almost 2.8 mm. A 6 m bar 15 degrees hot has grown about 1.05 mm, which is your entire cut tolerance. Oxy-fuel and plasma cutting dump serious heat into the cut zone, so the piece you measure five minutes after the cut is not the piece you will weld next week. My standing advice, and it is the one personal rule I refuse to bend: never measure a flame-cut piece hot, and never set the stop from a hot measurement. Let pieces cool to shop temperature, keep the tape and the steel at the same temperature, and hold the tape straight and tensioned along the leg. On long members a sagging tape alone can eat a millimetre, and the saw gets the blame.

What Should a Buyer Put in the Order to Keep Cut Tolerances Honest?

Everything above lives in the fabrication shop, but the buyer holds one lever that decides how much of it matters: what the purchase order actually says about cutting.

A cut-to-length order is only enforceable when it names the tolerance, the standard, and the verification method: state the cutting method, the length tolerance class, the squareness requirement, first-piece verification, and who inspects. A verbal “cut to length” leaves every error above unpriced and undisputed.

Tolerance Class and Standard on the Purchase Order

The most expensive words in steel purchasing are “cut to length” with nothing after them. They carry no tolerance, no method, no standard, so every party assumes their own number. I tell buyers to specify four things in one line: the cutting method acceptable for the job, the length tolerance band (for example ±1 mm or ±2 mm, remembering that sawing practice comfortably holds ±0.75 mm with a careful setup but promising ±0.1 mm invites trouble), the squareness requirement if the cut face registers against anything, and the inspection standard the pieces will be checked against. When we quote cut services for export projects, this one line is the difference between a delivery that passes receiving inspection the day it lands and a dispute that costs both sides a fortnight. It costs nothing to write and everything to omit.

Test Pieces, Spares and Verification Bands

Even a specified tolerance needs a verification plan, and the plan needs to survive contact with production volume. Sampling every piece is unrealistic above a few hundred cuts; the practical middle ground is a first-piece check per setup, a last-piece check per shift, and a defined inspection band in between, for example one measured piece in every fifty with a written reaction plan when a sample misses. Our Gulf Metal Solutions cooperation taught me why this belongs in the order rather than in trust. Their project team asked for SGS inspection before shipment, and the dimensional check on cut members was written into that scope. Nothing needed rework because the expectations were on paper before the first bar moved. Buyers who write the verification band into the purchase order get two things at once: a supplier who prices the accuracy honestly, and a receiving team that knows exactly what “good” looks like when the container opens.

One closing observation from my side of the desk. The buyers who never call me with dimensional complaints are not the ones with the best saws. They are the ones who asked these five questions before the order was placed, and got answers they wrote down.

Conclusion

Cutting errors trace to kerf, machine drift, steel movement and vague orders. Specify the tolerance, verify the first and last piece, and marine angle steel arrives cutting-accurate.

Tags:  #marine-angle-steel   #steel-cutting-tolerance   #kerf-allowance   #shipyard-fabrication   #dimensional-accuracy