Planning patterns, small mills and a yield improvement routine

Pattern catalogues and generation

Many sawmills choose from a fixed catalogue of cutting patterns that specify the central cant and side boards for a log of a given diameter. Mena-Reyes and colleagues (2025/2026, Mathematics, Chile) argue that fixed historical catalogues can misalign with demand and log supply. They propose generating new patterns on demand by column generation: a master model minimises logs processed subject to supply by diameter class and product demand, and a pricing model proposes one new pattern at a time, using dual prices, with the cant and boards fitting inside the circular cross section (a Pythagorean constraint with kerf allowance).

The portal could read only the abstract, introduction and part of the methods. The abstract reports log consumption reductions of up to 31 percent and near-optimal solutions with only about 20 to 25 patterns, with more patterns adding set-up time and complexity. The conditions and baseline of the 31 percent are not known, and the model ignores taper, defects, scheduling, and assumes deterministic supply and demand. Treat the results as promising but unverified.

A small bandsaw in Romania

Borz and colleagues (2021, Forests; Romania and Italy, Europe) studied a small mill with a manually driven horizontal bandsaw, mostly Norway spruce and silver fir, over seven days: 86 logs, about 26 m3 over bark (73 spruce and 13 fir), diameters 12 to 56.5 cm. Output was 695 pieces, about 18 m3 of lumber and about 8 m3 of residues. Recovery varied from 38.8 to 95 percent and averaged around 69 percent. Net productivity was about 0.5 m3 per hour, gross about 0.4. The efficiency rates were 2.045 h/m3 net and 2.539 h/m3 gross.

Where the delay-free cycle time went (approximate):

Effective sawing was only about a quarter of the cycle, so handling, not saw speed, limited output. Cycle time was explained well by log diameter (83 percent) and volume (88 percent), but net productivity only moderately (about 44 to 47 percent).

Worked checks. Recovery from the totals: 18 / 26 = 69.2 percent, consistent with the reported average of about 69. Productivity from the efficiency rate: 1 / 2.045 = 0.489 m3 per hour net, and 1 / 2.539 = 0.394 m3 per hour gross.

A yield measurement routine

1. Fix the definitions: over or under bark, how diameter is measured, lumber state.
2. Choose the sample: straight sound logs of a defined size to isolate the machine, or all logs over a period for overall performance.
3. Track logs through the line with dimensions, species and products per log.
4. Measure the saw line: thickness along each board.
5. Separate the effects: report yield by log size class, product mix and by-product share.
6. Report volume and value, with price basis and date.

Benchmark only if unit, lumber state, log distribution, product mix and measurement convention match.

Improvement order

1. Cut sawing variation and oversizing.
2. Match pattern to log and product.
3. Position and rotate sweepy logs.
4. Use thinner kerf where the line can hold it.
5. Widen the product mix and optimise edging (Steele: 9.7 percent from adding one product on one log).
6. Weigh throughput against yield.
7. Look upstream and downstream.

Limits of this course

This course is AI-drafted and based on the portal's already fact-checked articles. Some sources are old simulations, small samples, or abstract-only. Most figures are regional (Europe or North America) and do not transfer automatically between species, logs and mills. Check figures against standards, your contracts and your own measurements before acting.

Key points

• Column generation research suggests a small pattern set (about 20 to 25) can be near-optimal, but only abstract-level results were read.
• In the small Romanian mill, effective sawing was only about 27 percent of the delay-free cycle; handling dominated.
• A reliable yield figure needs a fixed definition, a defined sample, per-log tracking and size-class reporting.
• Benchmark only comparable figures; build an internal history first.
• Start with variation and oversizing; then pattern, rotation, kerf and product mix.