Scrap and yield: where material cost actually goes

A 2,500 x 1,250 mm sheet of 2 mm CRCA weighs 49 kg. Forty parts at 300 x 200 mm come off it in a clean nest, and those parts weigh 37.7 kg. The other 11.3 kg is skeleton, edge margin and offcut, and it leaves the plant as scrap before a single forming defect is counted.

That is a 76.8% yield, which is respectable for a part that size. The number is also the whole argument about material cost, because on a laser or a turret the steel is 40% to 60% of the piece price and every other cost is smaller. Scrap and yield are not housekeeping figures. They are the difference between a quote that holds for a year and one that is renegotiated in six months.

This guide works through the nesting arithmetic, the gap that decides how many parts fit, the defects that turn good blanks into scrap, and the weighing records that make the yield number real instead of estimated.

What yield does a sheet actually give?

Yield is the area of good parts divided by the area of sheet consumed, and both halves have to be measured, not assumed. For a single-part nest the calculation is short enough to do on the drawing:

yield = (number of parts x part area) / sheet area

Take the 300 x 200 mm part on a 2,500 x 1,250 mm sheet. Usable area is 2,480 x 1,230 mm once a 10 mm edge margin is taken, which is 3.05 m². With an 8 mm gap between parts, each part occupies a 308 x 208 mm cell, 0.064 m². Eight columns by five rows fits forty parts, 2.4 m², or 76.8% of the full sheet.

Change the gap to 15 mm and the cell becomes 315 x 215 mm. Seven columns by five rows now fits thirty-five parts, 2.1 m², which is 67.2%. Nothing about the part changed. The yield fell 9.6 points, or 12.5% of the parts, and the scrap rose from 0.73 m² to 1.03 m² per sheet.

That is the first decision in every nest: the gap is set by the process, not by comfort. A 12 kW fibre laser cutting 2 mm mild steel needs 5 to 8 mm between parts to clear the kerf, avoid heat build-up in a narrow web and give the nozzle room. A 3 mm gap is possible on thin gauge with good nesting and high speed, and it is a false economy on 6 mm plate where the cut edge can move.

Related reading: the 5S on a fabrication floor: what it changes in practice and the Calibration register: how instruments stay traceable take the same problem further. See the process and inspection scope on the capability pages, or send the drawing through the RFQ route.

Kerf is smaller than the gap, and buyers over-weight it

A fibre laser kerf on 2 mm mild steel is about 0.3 mm. Forty parts at 300 x 200 mm carry 1,000 mm of perimeter each, so 40 m of cut, and the kerf removes 0.3 mm x 40,000 mm = 12,000 mm², or 0.012 m². On a 3.125 m² sheet that is 0.4%.

The gap is twenty times the kerf. An 8 mm gap between forty parts costs about 0.4 m², which is 7% of the sheet. Buyers negotiating a cutting price often ask about kerf compensation and ignore the gap, which is where the money is.

The exception is a part with a long internal cut-out, where the kerf and the lead-in pierce both consume material and time. A 1,800 mm slot in a 2 mm panel needs a pierce at one end and a lead-out at the other, and the slug has to fall clear rather than weld itself back in.

How thickness changes the scrap number

Yield is a percentage and scrap is a mass, and the mass is what the invoice sees. Flat-rolled steel weighs 7.85 kg per square metre per millimetre of thickness, so:

The table explains why a gauge reduction is the largest single lever on material cost, and why it is also the most dangerous. Dropping a bracket from 2 mm to 1.5 mm removes 25% of the steel mass, but it also removes 58% of the bending stiffness, because stiffness goes with the cube of thickness. A part that was stiff enough at 2 mm may deflect visibly at 1.5 mm under the same load. Run the deflection check before accepting the saving.

The Ganesh column is a route note rather than a limit: all of these gauges cut on the 12 kW fibre laser, which handles up to 20 mm plate on a 40 ft x 16 ft bed, and form on the 200 T brake to 10 mm. The thickness range for custom parts runs 0.8 mm to 40 mm across the plant, with plate above 20 mm going to the iron worker.

Where the scrap actually comes from

Skeleton scrap is the biggest mass and the least discussed. It is also the most predictable: a nest either fits another part or it does not, and the skeleton is what is left when it does not. The decisions that reduce it are commercial as much as technical, because they depend on whether another job can share the sheet.

The second source is the remnant. A sheet that gives forty parts from an eight-column nest leaves a strip 30 mm wide after the margin, which is not worth keeping, while a nest of three large panels can leave a 400 mm strip that will cut a hundred brackets. A remnant policy that says "keep anything over 300 mm wide and 600 mm long, label it with grade and gauge, and use it within 60 days" turns a waste stream into inventory. Without that policy the strip is cut up for scrap because nobody can find it later.

The third source is the defect scrap, and it behaves differently:

• Setting scrap is the first parts after a tool change or a program change, before the process is stable. On a progressive die it is typically three to six strokes; on a laser it is the first part on a new nest, checked for hole position and edge quality.
• Forming defects are cracks at a bend, springback beyond tolerance, galling, and edge splits in a draw. These are process failures and appear as a rate per thousand strokes, not per sheet.
• Weld rework is not scrap but it consumes the part a second time. A frame that needs a burn-out and re-weld has already absorbed its cutting, forming and fixture time.
• Post-coat rejects are the expensive ones. A 3 kg canopy door that is scrapped after powder coating carries the steel, the cutting, the forming, the welding and a coating slot that cannot be recovered.

The mass of post-coat scrap is small. In a plant running 240 t of steel a year, coating rejects might be 1.5 t, or 0.6% of the mass, and still represent 15% of the scrap cost. That is why a single scrap percentage is useless as a management number.

What the record has to contain

A scrap record that can change a process needs five fields per event and one per month.

The event fields are date and shift, part number and operation, weight in kilograms, cause code, and the disposition. The monthly fields are input tonnage, good output tonnage, scrap by cause, and the reconciliation difference against purchases. Add the nest identifier for cutting scrap, because a skeleton-heavy month is a nesting problem and a first-part-heavy month is a setting problem, and the two look identical in a total.

The frequency matters more than the format. Weighing every shift is the only way to catch a die that started cracking flanges on Tuesday rather than at month end. A monthly total tells you that last month was bad; a shift record tells you which press, which stroke count and which operator.

Reconciliation is the discipline that keeps the record honest. If the plant bought 20 t of CRCA in a month, shipped 12.8 t of parts and recorded 1.9 t of scrap, then 5.3 t is unaccounted for. Some of it is work in progress, some is remnant stock, and some is material that became a part and was never weighed. A plant that never reconciles will always report a flattering yield.

What the numbers look like on an annual basis

Take 240 t of CRCA a year at an average 68% yield. Good output is 163 t and scrap is 76.8 t. Raise the yield to 73% through better nesting, remnant reuse and a gap policy that suits the gauge, and output becomes 175 t from the same purchase with scrap at 64.8 t. Twelve tonnes of steel comes back into the business without a price negotiation.

The same 240 t at a 2% defect scrap rate is 4.8 t, and most of it is recoverable with process control rather than with buying skill. A die that is sharpened on schedule instead of on failure stops producing burr rejects; a nest that keeps a 6 mm gap on thin gauge stops producing heat-discoloured edges that fail at coating.

There is a limit to how far this goes. A 60% yield on a small bracket is not a nesting failure; it is geometry. Small parts have a fixed gap and margin cost per part, and a 200 x 80 mm bracket on a shared sheet with four other jobs may still only reach 62%. Quoting that part at 85% yield and discovering the truth in production is how a supplier loses money for a year.

How Ganesh runs the nest and the scrap book

Cutting runs on a 12 kW fibre laser with a 40 ft x 16 ft bed, which takes a 2,500 x 1,250 mm sheet six at a time and lets a long panel and its brackets share one program. The 6 kW and 3 kW machines handle the smaller and thinner work, and the 250 T punching line takes the repeat blanks that would otherwise consume laser time. Running all three in one nest plan is what makes remnant reuse possible: the leftover strip from a panel job becomes a bracket job on a smaller bed rather than a scrap bin.

Scrap is segregated by grade at the machine, not at the yard. CRCA, galvanised, stainless and aluminium each have their own bin, weighed at the end of the shift and coded to the job. Coated rejects go to a separate stream because the powder has to be accounted for as well as the steel. The laser cutting capability covers the beds and the thickness range, the nesting article covers how a program is built, and the thickness and price piece explains why the gauge decision outweighs the nesting decision on most parts. The wider plant and quality hub collects the records side of this: what is measured, how often, and what it prevents.

What to put on the drawing and in the enquiry

1. Annual quantity and release pattern, because a nest only pays when the parts repeat.
2. Material grade and gauge, with the standard the grade is ordered to.
3. Any feature that forces a wider gap: a long slot, a sharp internal corner, a heat-sensitive edge.
4. Whether the part can share a sheet with a family of other parts, and which family.
5. Permitted grain direction or rolling direction, if the part is formed.
6. Remnant policy: whether offcuts are returned, retained or scrapped.
7. Whether the price is to include the scrap allowance or be quoted net of recovered scrap value.

The last point causes more arguments than the first six. A quote that assumes the supplier keeps the scrap is not the same as one that credits the buyer, and the difference on a 20% skeleton can be several percent of the piece price. State which one applies.

When this calculation misleads

Yield percentages flatter large parts and punish small ones, so comparing two suppliers on yield alone is meaningless unless the part mix matches. A plant that runs 80% large panels will always report a higher yield than one that runs brackets, and neither is better at what it does.

The recovered scrap value is also not the purchase price. Mixed or contaminated scrap sells at a fraction of new steel, and a bin with stainless mixed into CRCA can be downgraded entirely. Recycling revenue should be treated as a credit, never as an offset that makes a bad nest look acceptable.

The biggest error is treating yield as a purchasing problem. Buying a lighter gauge that the part cannot carry, or negotiating a cheaper grade that will not form, moves the scrap from the skeleton into the press shop and the reject bin, where it costs more. If the drawing is not settled, the yield number is not either. Send the drawing with the annual quantity and the family it belongs to, and the quotation will state the nest assumption, the yield and the scrap allowance behind the price.