How cut list optimisation works
A cut list optimiser answers one question: given a pile of parts and the sheets they have to come from, how many sheets do you buy and where does each part sit? It is a packing problem, and it is a packing problem with one constraint that changes everything.
Every cut must run edge to edge across the piece being cut. That is not a simplification — it is what a table saw, a track saw and a panel saw physically do. You push a whole panel past a blade and it comes back as two pieces. There is no way to stop halfway and turn a corner.
That constraint is called a guillotine cut, and it is why the layouts here look different from the ones a generic bin-packing tool produces. Those tools will happily drop a part into a pocket surrounded on three sides, because the algorithm only checks that rectangles do not overlap. The yield looks better. The layout cannot be built. A denser number from an unbuildable plan is not a better answer, it is a wrong one.
This tool runs the same parts through forty-eight different packing strategies — four sort orders, three ways of choosing which offcut a part goes into, and four ways of splitting what is left — and keeps whichever result wins on sheet count first, then on how much of the last sheet is used, then on the fewest cuts. That takes about eighteen milliseconds for a hundred parts, which is why it happens on every keystroke.
Why kerf matters more than it looks
A saw blade does not divide material, it destroys a strip of it. A thin kerf blade takes about 2.4 mm, a standard 10 in blade about 3.2 mm, and a panel saw more. That sounds trivial until you count the cuts.
Thirty-eight cuts at 3.2 mm is 121.6 mm of material turned into dust — more than the width of many shelves, and enough to lose a part entirely at the end of a tight layout. The classic demonstration is two 610 mm parts across a 1220 mm sheet. At zero kerf they fit exactly. At 3.2 mm they need 1223.2 mm, and the second one goes on a second sheet.
Kerf is taken at every cut except the outer edges of the sheet, because those edges already exist and no blade passes through them. A part exactly the size of the sheet still fits at any kerf setting. Getting that wrong in either direction — applying kerf at the edges, or not applying it at all — is the most common fault in cut list software.
Grain direction: when it matters and when it does not
Sheet goods with a face veneer have a direction. Plywood, veneered MDF and melamine with a wood-effect print all read differently along the grain than across it, and a cabinet side rotated ninety degrees will announce itself the moment the piece is finished, especially under a clear coat.
When grain is locked, no part rotates: every part's length runs along the sheet's length. That costs material, sometimes a great deal. Six parts of 1210 by 700 mm fit three to a sheet turned and only two upright, so locking the grain on that job takes three sheets at 57% yield instead of two at 85%.
The per-row override exists because grain is rarely all-or-nothing. Cabinet sides and doors need consistent grain; the bottom, the back and anything that will be painted or hidden do not. Tick those rows and let them turn, and you often recover the sheet that locking the grain cost.
Worked example: a cabinet carcass from one sheet
A simple base cabinet from a single 2440 by 1220 mm sheet of 18 mm ply, with a 3.2 mm kerf and grain locked:
The interesting number is the offcut. Everything fits on one sheet with room to spare, and what is left is a single strip 1657 mm long — a usable shelf for the next job rather than four scattered rectangles. Switch the objective to largest offcut and the same parts still take one sheet at the same 71.6% yield, but the remainder comes back as an 854 by 837 mm panel instead: two more cuts, and a squarer piece that a long strip cannot replace. Neither answer wastes less material. They put the waste in different shapes.
Common mistakes
- Setting kerf to zero. The commonest error and the most expensive. Zero kerf is only correct for a CNC nesting job with a compensated toolpath, or for scoring by hand.
- Entering finished sizes with no trim allowance. If a part needs edge banding, a rebate or a clean-up cut, add it to the part dimensions before packing. The optimiser plans what you tell it.
- Assuming the sheet is exactly nominal. A 2440 by 1220 sheet is often a few millimetres under, and factory edges are not always square. On a layout with no slack, measure your actual sheet.
- Locking the grain out of habit. Half the parts in a typical carcass do not care which way they run, and letting those turn can save a whole sheet.
- Reading the diagram as a cutting order. The numbers on the cut lines are the sequence. Cutting in a different order can leave you holding a piece too large to handle safely.
When these layouts do not apply
- Damaged or out-of-square sheets. The layout assumes a true rectangle. A sheet with a damaged corner or a bowed edge needs its usable area measured and entered as a smaller sheet.
- Sheets with a defect or a feature to work around. Nothing here knows about a knot, a patch or a print you want centred. Split the sheet into usable rectangles and enter those.
- Material that is not flat. A bowed or twisted sheet cannot be cut accurately whatever the plan says, and forcing it through a saw is how kickback happens.
- CNC nesting. A router with a compensated toolpath is not bound by the guillotine constraint and will beat these layouts on yield. If you are cutting on a CNC, use its own nesting software.
- Solid timber. Boards are sold by volume and carry defects that sheet goods do not. Use the board foot calculator for those.