Radial sawing of oak and value yield

The question

Pedunculate oak (Quercus robur) is valued for radially oriented boards, which the Croatian authors describe as dimensionally stable with reduced warping during drying. But producing more radial boards means dividing the log into sections before sawing, which in the simulation lowered volume recovered. At what log size does radial sawing pay?

The Croatian simulation

Ištvanić and colleagues (2025, Forests; Croatia and Bosnia and Herzegovina, Europe) used the simulator RadSawSim. No logs were sawn physically. The design combined 16 mid diameters (25 to 100 cm), four board thicknesses (25, 32, 38, 50 mm) and five methods, giving 320 simulated logs. All logs were 4 m long with taper 1.5 cm per metre, and a band saw with fixed 3 mm kerf was assumed. Only centre-part boards of at least 8 cm nominal width were counted.

Methods: live sawing (baseline) and four radial techniques named thirds, fourths, fifths and sixths, in which the log is halved lengthwise and the halves divided into sections (fourths is quarter sawing).

For value, boards received quality weights: radial 1.0, half-radial 0.9, tangential 0.8. The conclusions depend on these assumed weights.

Results

• Volume yield: live sawing was highest at every size, rising from about 0.60 to nearly 0.75 m3 of boards per m3 of log as diameter grew. The gap to radial sawing was widest up to 45 cm and narrowed above. The text speaks of 50 cm and the abstract of 45 cm as the limit, so the paper is not consistent.
• Radial share: live sawing about 45 percent; thirds 40 to 100, fourths and fifths 60 to 100, sixths 80 to 100 percent, depending on diameter.
• Log-value yield combines volume loss and quality gain. Below about 45 cm live sawing was generally best; above it radial sawing was generally better, with exceptions depending on thickness. For example, live sawing won at 25 cm for all four thicknesses, but at 45 cm radial won for 25, 38 and 50 mm boards while live still won for 32 mm.

All radial intervals lie above live sawing, while the radial variants overlap each other. So the choice of radial versus live matters more than the choice among radial subdivisions.

Worked examples

Weighted quality of a board mix. Suppose 50 percent of sawn volume is radial, 30 percent half-radial and 20 percent tangential. Weighted quality = 0.5 x 1.0 + 0.3 x 0.9 + 0.2 x 0.8 = 0.50 + 0.27 + 0.16 = 0.93. The mix is invented; the weights are the paper's. A simple illustration of why a high radial share raises value even if volume falls (the paper's own log-value yield uses its earlier formulas).

Relative gain. Fifths against live: 0.645 / 0.544 = 1.186, about 18.6 percent higher mean value yield. Fourths: 0.634 / 0.544 = 1.165, about 16.5 percent.

Hardwood context

The Europe-focused hardwood article adds perspective, some of it general practice rather than opened sources:

• In a Romanian beech sample (Campu 2010, nine trees, one felling area), red heartwood occurred in about 34 percent of round-wood volume and the author estimated downgrading cut revenue by 40 to 50 percent. It shows the mechanism, not a European average.
• No species-specific European sawing yield for oak, beech or ash was found in the open sources. The number mills should measure themselves is top-grade yield by log class and supplier.

Checklist: applying the 45 cm rule

• Treat 45 cm as a starting rule for sorting oak, not a law; the result varies with board thickness.
• Replace the paper's 1.0/0.9/0.8 weights with your own price ratios.
• Validate on real logs; the authors list this as a next step.

Key points

• Radial sawing raises radial share and value but lowers volume yield; value is decided by log-value yield.
• Simulation result: live sawing below about 45 cm, radial (fourths, fifths) above, with exceptions by thickness.
• Mean value yield: 0.544 for live against 0.634 to 0.645 for radial variants.
• Results rest on assumed quality weights and idealised logs; no cost side.
• For European hardwoods the articles give no species-specific yield; measure top-grade yield by log class.