REEL orders are the core of paper mill and converter production. They are also the hardest planning problem to solve manually; the number of valid width combinations grows exponentially with order count. Deckle Plus solves it exactly, in under one second, for any deckle width.
A winder with a 3600mm deckle processing 12 customer reel orders of different widths has thousands of possible width combinations to evaluate for a single pass. A human planner can evaluate 10–20 combinations in an hour. Deckle Plus evaluates all of them in milliseconds and then selects the combination set that minimizes total trim across the entire order queue, not just one pass.
Consider a 5600mm winder with a maximum of 8 blade positions, processing nine customer reel orders: 1200mm (15 rolls), 900mm (22 rolls), 850mm (10 rolls), 750mm (30 rolls), 600mm (18 rolls), 500mm (25 rolls), 400mm (40 rolls), 350mm (15 rolls), 250mm (20 rolls).
A human planner might identify an obvious combination: 1200 + 900 + 900 + 850 + 750 = 4600mm, leaving 1000mm trim (17.9% waste per pass). A slightly better plan: 1200 + 850 + 750 + 600 + 500 + 400 + 250 = 4550mm, using 7 blades with 1050mm trim. Neither comes close to optimal.
Deckle Plus evaluates all valid combinations and identifies: Pass 1: 1200 + 900 + 850 + 750 + 600 + 250 = 4550mm trim 1050mm. Running it 10 times fulfills 1200mm(10), 900mm(10), 850mm(10), 750mm(10), 600mm(10), 250mm(10) simultaneously. Pass 2: 900mm(12) + 750mm(20) + 600mm(8) + 500mm(25) + 400mm(40) + 350mm(15) + 250mm(10) at 0.22% trim. The full solution achieves 0.3% total trim across all passes, a result no manual planner would find in a reasonable time.
Result
Deckle Plus consistently achieves sub-0.5% trim on order sets where experienced manual planners produce 2–4%.
The core algorithm is a two-phase solver based on the classical cutting stock problem from operations research. Phase 1 (column generation) enumerates all feasible width combinations that fit the deckle within blade count constraints. Phase 2 (knapsack solve) selects the minimum-waste subset of patterns that fulfils all order quantities.
Every blade position counts individually, including duplicates of the same width. A pattern of [900, 900, 600] uses 3 blade positions. The optimizer never generates a pattern that exceeds your machine's configured blade count, even if doing so would reduce waste further.
Optionally configure a Max Waste Target (%) for the full run or a Per-Pattern cap for individual passes. The optimizer respects these targets when finding patterns, useful when a customer contract specifies a maximum trim loss for a specific grade.
Most real order sets require multiple winder passes. Deckle Plus plans the complete set of passes needed to fulfil all order quantities, not just one pass. Each pass is a separate pattern in the run, with its own blade positions and repetition count tracked through to floor execution.
When an order is added, modified, or removed after the initial plan is set, the planner can re-optimize in seconds. The Sandbox feature lets them test the change against the original plan before committing, essential when a major customer revision arrives mid-shift.
Specific questions about how Deckle Plus handles this use case.
Deckle Plus uses a two-phase algorithm. Phase 1 (column generation) generates all mathematically valid cutting patterns for the given order widths and machine constraints: all combinations of customer widths that fit within the deckle width and blade count limit. This can produce hundreds or thousands of candidate patterns. Phase 2 (knapsack solve) selects the minimum-cost subset of those patterns that collectively fulfils all order quantities. The solution is optimal or near-optimal for all practical production order sets, and completes in under one second for typical inputs.
In Deckle Plus, the blade count limit is configured per machine. If your winder physically supports up to 12 blades for Kraft orders but only 8 blades when running heavier Duplex board due to increased blade pressure, you can configure two separate machine profiles, one per grade configuration. The optimizer enforces the blade count of whichever machine profile the run is assigned to.
There is no hard limit on the number of order lines in a single optimization. In practice, production order sets of 8–25 different order widths optimize in under one second. Order sets of 25–50 widths may take 2–5 seconds. Very large order sets (50+) are uncommon in practice; if encountered, the optimizer produces a near-optimal solution within its time budget and returns it with a note that exact optimality is not guaranteed for sets above this size.
Each distinct width is entered as a separate order line in Deckle Plus, even for the same customer. For example, if a customer needs 900mm (20 rolls) and 750mm (15 rolls), these are entered as two order lines, both attributed to the same customer, both visible in the same production run. The optimizer treats each order line as an independent width requirement and satisfies both quantities across the minimum number of winder passes.
Minimum quantity per width per pass is not currently a configurable optimization constraint in Deckle Plus. The optimizer focuses on minimizing total trim waste across all passes. If a specific width requires a minimum batch size per pass for quality reasons (e.g., a minimum 5 rolls per width to achieve consistent tension), this can be enforced by the operator on the floor execution screen; the operator marks a pattern as complete only when the minimum batch has been run.
Book a 20-minute call. We'll confirm Deckle Plus fits your specific workflow and show you a live optimization on your order data before you invest any further time.