Cutting stock and trim loss
Every offcut you throw away is material you already paid for. Cutting stock optimisation keeps more of it.
- Free 30-minute call
- No commitment
- You own the code
In short: A cutting stock optimisation plans how to cut ordered pieces from rolls, bars or sheets so that as little material as possible ends up as waste. It chooses the cutting patterns, how often to run each one and which offcuts are worth keeping.
The problem
On Monday morning, the cutting shop’s order book lists dozens of lengths and sizes, and the operator starts with whatever order is on the screen. Each bar gets cut to fit that job, then the next one. By Friday the scrap bin is full again. Nobody can say how much of it was avoidable.
Signs your cutting is wasting material
Offcuts pile up in a corner because nobody writes them down. You buy fresh stock while usable remnants are stored on the rack. Every order gets its own cutting pattern, its own layout of pieces on the material, so the saw is reset again and again. When your best operator is on holiday, the bin fills faster. And nobody knows the trim loss, the material cut off and thrown away, across the whole order book.
Why order-by-order cutting fails
In most shops, each operator arranges the pieces of their own orders on the material by eye. Then everyone hopes that little waste is left overall. That rarely works. The savings are in combinations across orders, and you can’t see those one bar at a time. Which pattern is best depends on three things: which pieces are ordered, which remnants are already on the rack and how often you have to reset the machine. Trim loss minimization software has to see all three together.
What changes
Today, the operator cuts order by order, and the offcuts go in the bin. Our model looks at the whole order book at once. It proposes a handful of cutting patterns, how often to run each and which remnants to use first. The operator checks the plan, swaps a pattern if a machine is down and starts cutting.
Cutting stock optimisation: what you get
You get cutting patterns that meet every order with the least material in the bin. Each one runs on the machine you have, whether that’s a shear, a saw or a router. Offcuts that used to be scrapped now fill orders. For the same order book, you see how much raw material you no longer need to buy. Because the model weighs waste and setup together, a few patterns do the job instead of dozens.
We build the model on your own stock sizes, order book and remnants, and test it on your own orders before it goes live. We pick the weeks when orders changed at short notice. The model has to keep waste low there, because that’s exactly when nesting by hand falls apart.
- A model written around your own cutting stock and trim loss rules.
- Results measured in the numbers you already track.
- It slots into how you already work, and you can check every answer it gives.
How it works
Your orders, your bar lengths or sheet sizes, the offcuts on the rack and what your machines can cut go into one model. It plans all orders together instead of one at a time. You get a few cutting patterns that fill every order with the least trim loss, and your team checks the plan before the first cut.
What the cutting stock problem is
Cutting stock means cutting large pieces of raw material into the smaller pieces customers order, with as little waste as possible. It’s an old problem. Leonid Kantorovich wrote it down in 1939, and Gilmore and Gomory published the standard method for it in 1961. Both think in cutting patterns. A pattern is one layout of pieces on one bar, sheet or block.
The model chooses a set of patterns and how often to run each one. Together they meet every order at the lowest total waste. With more than a few sizes, the number of possible patterns explodes. So the model doesn’t list them all. It builds new patterns as it goes and keeps one only if it lowers the cost. In optimisation, that trick is called column generation.
One, two and three dimensions
The variants differ by the shape of the stock. One-dimensional cutting cuts linear stock along one axis: steel bars, pipes, paper rolls and fabric bolts. Two-dimensional cutting turns sheets, plates and panels into rectangles. Here, the layout decides how much of each sheet ends up in parts. It matters whether pieces may be turned, how tightly they are packed and how well the edge is used. That’s the job of a 2D bin packing algorithm: it fits rectangles onto a sheet with minimal gaps.
Three-dimensional cutting cuts blocks into smaller solids: timber beams, stone, foam or metal billets. Each extra dimension multiplies the number of layouts to search. A good model handles all three, so you don’t treat them as separate problems.
Guillotine or non-guillotine cuts
The cut geometry has to match your machine. A guillotine cut splits a piece in two with one straight cut all the way across, like a paper guillotine or a plate shear. Every further cut must also go straight and all the way across its piece. Non-guillotine cuts may stop in the middle of the material or turn a corner. A shear can’t make those.
A CNC router, a computer-controlled cutter, can make non-guillotine cuts, and so can a band saw. They often raise yield, at the cost of more complex patterns. For sheet metal cutting optimisation, that means one simple rule: every pattern the model returns has to run on your shear or your router, as it’s set up today.
Setup cost per pattern
Every pattern means a changeover, whether that’s loading a sheet, positioning a bar or retooling the saw. Manual nesting ignores that cost. A model that only minimises trim does too, and it happily produces dozens of one-off patterns. The material it saves goes straight back into changeovers.
A published textbook case shows how much room there is. A paper machine makes 5,600 mm master rolls, and 13 widths must be cut from them. There are 308 possible patterns. The least waste needs 73 rolls, at 0.401% waste, and at least 10 different patterns. Nineteen different 10-pattern plans hit that same minimum, so you can pick among them on setup alone.
Which offcuts are worth keeping
The model treats every leftover above a set length as stock. It records the lengths, shapes and sizes on hand and uses them before it opens fresh material. But keeping every scrap isn’t free. Each remnant needs a rack slot, a label and someone to find it again.
So most cutting software asks for a minimum remnant length. Anything longer goes back into stock, anything shorter is scrap. Some tools let you forbid offcuts of medium length: too long to throw away, too short for most orders. The makers of these tools warn that you then generally need more bars. You get fewer useless remnants on the rack, but you buy more stock. We’d rather calculate which one costs less than guess.
Say a 6 m bar has 400 mm left after cutting. If you rarely order pieces under 500 mm, that remnant will sit on the rack for months. The model reads your order history and sets the threshold per profile. Your rack then holds pieces you’ll actually cut, and the rest goes to scrap without a second look.
Joint optimisation with MILP
We solve all of this as one problem with mixed-integer linear programming (MILP). MILP writes the problem as equations with whole-number decisions, such as how often to run a pattern. The model minimises trim loss plus setup cost. It meets every order and respects the guillotine rules and the remnants on the rack. Because it sees every order at once, it finds combinations a person working bar by bar can’t see.
The maths is the same whether you cut paper rolls, steel bars, glass panes, timber or fabric. In every one of those trades, each offcut in the bin is material you already paid for. So the model first checks whether a remnant fits an order before it opens new material.
Common questions.
Something else? Ask us directly.
Yes, and for a single job an off-the-shelf cut-list tool does that well. You enter sheet sizes, parts and the width of the saw cut (kerf), and it returns cutting layouts. It gets harder across all your orders, with remnants on the rack, several machines and the cost of each changeover. That’s where a model built on your own data is worth it.
Kerf is the width of material each cut removes. Measure the blade you use, not the catalogue figure, and set it per material if your blades differ. The optimiser then subtracts it at every cut. On a bar with ten cuts and a 4 mm blade, that’s 40 mm gone, which can decide whether one more piece fits.
There’s no fixed length. It depends on which pieces your customers order regularly. A remnant is worth keeping if a piece you regularly order fits in it, and if it gets used before the rack overflows. Your order history shows both. We set the threshold per profile or sheet type, and revisit it when the product mix changes.
Yes. Grain is the direction of the fibres in wood. Some parts may be turned by 90 degrees, others must run along the grain. The model takes that rule per part and per material. It does the same with the rule of your saw that every cut runs straight from edge to edge (guillotine cut). A layout that ignores grain can look efficient on screen. You just can’t use it.
Sometimes it can, and if that’s the case, we’ll tell you. Standard software is built for an average operation, though. Yours runs on its own rules: your contracts, your exceptions, the things you know by heart. A custom model writes those rules in, and that’s where the better plan comes from.
You don’t need either to start. We build the model from the data you’ve already got, even if it lives in three spreadsheets and one person’s head. You get a system you run yourself. We train your team until they no longer need us.
You see a first result with the prototype, before you commit to anything bigger. We start with a free 30-minute call about your operation and the model that would fit it. Then we build a working prototype and run it on your numbers.
Similar pages.
Terms on this page
Every abbreviation explained in plain words, in our article “Metrics and methods”.
Tell us what you’re planning.
In a free 30-minute call with a specialist, you find out whether optimisation can help.
- Free, no commitment
- You talk to a specialist directly
- A clear go or no-go