Oak radial sawing simulated: why 45 cm log diameter is the tipping point

Introduction

Pedunculate oak (Quercus robur L.) is a high-value hardwood in Europe, prized for furniture, flooring, construction and cooperage. Its value depends heavily on how boards are oriented relative to the growth rings. Radially sawn boards (rings roughly perpendicular to the face) shrink and swell less, resist cupping and show attractive medullary rays, so they command higher prices than tangential boards. The problem for sawmills is that producing more radial boards usually means breaking the log into sections before sawing, and that tends to reduce the overall volume recovered.

A 2025 paper in Forests, by a team from the Faculty of Forestry, University of Zagreb, and collaborators, addresses this trade-off with computer simulation. Using a purpose-built simulator called RadSawSim, the authors compare conventional live (through-and-through) sawing with four radial sawing techniques that divide each half-log into thirds, fourths, fifths or sixths. They evaluate the techniques on three yardsticks: volume yield, lumber value yield and log-value yield, over a wide range of log diameters. The headline question is practical: at what log size does radial sawing start to pay?

Research methodology

Simulation only. No logs were physically sawn. The authors explain that test sawing of oak is costly and time consuming, which is why they used a simulator. RadSawSim (version 1) was developed at the University of Zagreb; its mathematical model is described in an earlier publication by the same group (Ištvanić et al., 2010), and the details of the model and its formulas are in that work rather than reproduced in full in this paper.

Simulated logs. The experimental design combined 16 mid-diameters (25 to 100 cm in 5 cm steps), four board thicknesses and five sawing methods, giving 320 simulated logs. Every log was 4 m long with a fixed taper of 1.5 cm per metre.

Board parameters.

• Edged boards of nominal thickness 25, 32, 38 and 50 mm, with nominal dimensions defined at 12% moisture content.
• Minimum nominal board width 8 cm.
• A log band saw with a fixed kerf of 3 mm.
• The model maximises the number of boards, counting only centre-part boards that meet the minimum width and length; outer boards were not counted.

Sawing methods.

After the division, each section is sawn into an optimal even or odd number of boards.

Outcome measures. The paper reports four quantities: volume yield (volume of sawn boards per unit of log volume), the share of radial boards in the total sawn volume, lumber value yield, and log-value yield. For value calculations, boards received quality weights: radial 1.0, half-radial 0.9 and tangential 0.8. The exact formulas follow the group's earlier work. For comparing value yield across methods the authors also give Bayesian posterior means with 95% credible intervals, although the model specification is not described in the portion of the article we reviewed.

Key findings

Volume yield. Live sawing gave the highest volume yield at every size. For live sawing the yield rose with log diameter from about 0.60 to nearly 0.75 m3 of boards per m3 of log. Volume yield fell as the number of radial sections increased, in the order live, thirds, fourths, fifths, sixths. The gap between live and radial sawing was widest for logs up to 45 cm in diameter and narrowed above that. For logs over 50 cm, radial sawing recovered roughly 0.5 to 0.7 m3/m3. The authors describe radial sawing of logs up to 50 cm as questionable in volume terms.

Radial board share. Live sawing produced about 45% radial boards. Radial thirds gave 40 to 100%, fourths and fifths 60 to 100%, and sixths 80 to 100%, depending on log diameter. More sections meant a larger radial share.

Lumber value yield. Value followed the radial share: live sawing was lowest, then thirds, fourths and fifths, with sixths highest.

Log-value yield. This is the decisive measure, as it combines volume loss and quality gain. The authors conclude that for logs below about 45 cm in diameter there is no justified reason to use the radial technique, while above 45 cm radial sawing is more efficient. They stress, however, that the picture in their detailed table is mixed near the threshold. For example, live sawing was best at 25 cm for all four thicknesses (log-value yields 0.521, 0.560, 0.524 and 0.456 for 25, 32, 38 and 50 mm boards), whereas at 45 cm radial methods won for 25, 38 and 50 mm boards (0.578, 0.572 and 0.563) and live sawing won again at 50 cm for 50 mm boards (0.560). Radial fourths were the best choice for many larger diameters, for example 65, 85 and 95 cm with 32 mm boards (0.638, 0.675 and 0.676) and 75 and 100 cm with 38 mm boards (0.640 and 0.680); radial fifths won at 60 and 90 cm with 32 mm boards.

Bayesian summary of value yield.

The intervals of the radial variants overlap each other, but all lie above live sawing, so the choice between radial and live matters more than the choice among radial subdivisions.

Comparison with other studies. The authors report that their simulated yields align with industrial studies and, for 25, 38 and 50 mm boards, were higher than earlier oak figures. They attribute this to optimised sawing parameters and sequences in the simulator, which may reduce waste compared with real practice. The expected decline of yield with decreasing log diameter agrees with earlier live-sawing results.

Authors' recommendation. Among the radial methods, fourths and fifths performed best, with board thicknesses of 25, 32 or 38 mm suggested depending on log diameter. (Readers may note that sixths tie with fifths in the Bayesian table, so the preference for fourths and fifths should be read together with the practical complexity of more sections.)

Practical applications for sawmills

From the paper.

• A diameter threshold of roughly 45 cm is the central guide: below it live sawing maximises log-value yield, above it radial sawing is generally preferable.
• The threshold depends on board thickness and is not a hard rule; results near 45 to 50 cm differ by thickness.
• Radial fourths and fifths gave the best value performance among radial variants, and radial sawing needs equipment adjustments in Croatian sawmills, according to the authors.
• The authors propose combining the simulator with 3D computed tomography (CT) log scanning for real-time analysis of internal log properties.

Portal interpretation (not claims of the paper).

• Mills sorting oak by diameter could use the 45 cm figure as a starting rule for deciding which logs go to a radial line, then refine it with their own board prices. The simulation assumes quality weights of 1.0, 0.9 and 0.8; a mill with a stronger or weaker premium for radial boards would reach a different threshold.
• The cost of extra handling, turning and set-up for radial sawing is not part of the model. A mill should add its own labour and throughput costs before changing its cutting plan.
• A simulator of this kind is also a cheap way to test cutting plans on paper before committing logs, which is attractive for expensive oak.

Limitations

• Simulation, not experiment. Results were not validated by sawing real logs. The authors themselves list validation at a Croatian sawmill as a next step.
• Idealised logs. Length (4 m), taper (1.5 cm/m) and kerf (3 mm) were constant, and a band saw was assumed. Real oak logs vary in sweep, ovality and defects; the simulation setup described in the paper does not appear to model such variation.
• Counting rules. Only centre-part boards of at least 8 cm nominal width were counted, which affects volume yield, particularly for small logs.
• Assumed quality weights. The 1.0 / 0.9 / 0.8 weights drive the value results; different market prices would change the conclusions and the 45 cm threshold.
• No cost side. Machine time, handling and capital investment for radial sawing are not included in the value yields.
• Information gaps in the paper. We could not see the Bayesian model specification, and the yield formulas are given in earlier work. Some citations in the paper appear loosely matched (for instance a beech study used for oak comparison), and the authors' preference for fourths and fifths sits beside a table where sixths ties fifths.
• No dedicated limitations section. The limitations above are drawn from the methods and discussion.

Conclusions

The simulation study of 320 virtual oak logs confirms a classic trade-off: dividing the log into more radial sections raises the radial share and lumber value but lowers volume yield. When both effects are combined in a log-value measure, live sawing wins for logs below about 45 cm, and radial sawing, especially fourths and fifths, wins above that size, with exceptions depending on board thickness. The findings give mills a defensible starting point for diameter-based allocation of oak logs, but they await validation with real sawing trials and scanning-based log data.

Original research

• Authors: Josip Ištvanić, Dario Pervan, Alan Antonović, Krunoslav Piljak, Murčo Obućina, Miljenko Klarić
• Title: Simulation of the Radial Sawing Technique for Pedunculata Oak (Quercus robur L.) Logs
• Journal: Forests (MDPI), volume 16, issue 10, article 1538
• Year: 2025
• DOI: 10.3390/f16101538
• Link: https://www.mdpi.com/1999-4907/16/10/1538
• Licence: Creative Commons Attribution (CC BY); the MDPI page states CC BY 4.0 in its footer

This article is an independent summary prepared by the portal; the original paper should be consulted for full details and exact figures.