Aspen 2x4 lumber from an Alberta mill: stiffness, tensile strength, knot-size sorting and MSR and CLT grade yields

Introduction

Canadian sawmills cut mostly softwood, yet the country's hardwood forests are only partly used. The authors of this 2024 paper in the journal Materials cite a Canadian hardwood harvest of about 25.5 million cubic metres in 2020 against a sustainable supply of 56.4 million cubic metres, which they present as unused capacity. Trembling aspen (Populus tremuloides) has historically gone mainly to pulp, paper and composite boards. The study asks whether aspen sawn into 2x4 lumber has physical and mechanical properties good enough for engineered wood products (EWP) such as cross-laminated timber (CLT), glulam and I-joists.

The work comes from the Wood Science and Technology Centre at the University of New Brunswick (Fredericton) together with the Department of Civil and Environmental Engineering at the University of Alberta (Edmonton). The lumber itself was sawn at a mill in Hines Creek, Alberta, so this is a study of real Canadian mill output rather than laboratory-prepared boards. The paper was published on 17 June 2024 in the special issue "Modern Wood-Based Materials for Sustainable Building".

For a sawmill audience the interesting question is practical: if a mill produces aspen studs and 2x4s, how much of the output could be sorted into useful structural grades, and which grading levers matter most? The paper does not claim to answer sawing-pattern or drying questions; it concentrates on what the finished, kiln-dried lumber can do.

Research methodology

Material. The study tested 368 pieces of kiln-dried trembling aspen 2x4 lumber (actual size 38 x 89 mm). The logs averaged 8.5 in (215.9 mm) at the small end. According to the paper, the lumber was kiln-dried at 95 C for 96 hours to about 16 percent moisture content. Visual grading followed the NLGA Standard Grading Rules and was carried out by an inspector from the Alberta Forest Products Association. Two grades were sampled: Select Structural (SS) in lengths of 8 ft (39 pieces), 10 ft (49 pieces) and 12 ft (80 pieces), and No. 2 grade in 12 ft lengths (200 pieces). The 8 ft SS pieces were intended for finger-jointing, so only stiffness (not tensile strength) was measured on them.

Stiffness (modulus of elasticity, MOE) measured three ways.

• Machine stress rating (MSR) with a Cook-Bolinder Tecmach SG-AF unit, which bends the board flatwise at the centre point and reports mean, maximum and minimum MOE.
• A handheld longitudinal stress-wave (LSW) device, the Brookhuis/TNO MTG-820, which estimates MOE from the resonant frequency and the density.
• Edgewise third-point static bending (EWB) according to ASTM D198, using the Forintek In-Grade procedure, as the reference method.

Tensile strength (UTS). Boards were tested in axial tension following ASTM D198 on a Metriguard 401 machine, with failure location and mode recorded relative to the maximum strength-reducing defect (MSRD).

Moisture content and specific gravity. A defect-free block near each failure was used for moisture content and specific gravity following ASTM D4442; properties were adjusted to 15 percent moisture content following ASTM D1990.

Analysis. The authors compared the three MOE methods, correlated nondestructive measurements with strength, re-sorted the pieces into four tiers by maximum knot size, compared the aspen results with spruce-pine-fir (S-P-F) values, and computed grade yields for MSR settings and for CLT grades defined in PRG-320.

Key findings

Density. Oven-dried specific gravity averaged about 0.40, which the authors say is about 3.5 percent above the Wood Handbook value for aspen. Average moisture content at testing was about 7.0 percent.

Stiffness depends on the method. Mean MOE across the three methods ranged from 8,032 to 10,673 MPa. SS lumber exceeded No. 2 lumber with every method. The methods differed significantly at the 95 percent confidence level, and MSR and LSW values were consistently higher than the EWB reference. Correlation with EWB was stronger for MSR (r = 0.82, R-squared 0.67) than for LSW (r = 0.68, R-squared 0.46). The authors conclude that MSR and LSW results should be calibrated against EWB. Piece length (8, 10 or 12 ft in SS) changed mean MOE by only 0.6 to 4.5 percent within each method, which the authors regard as not significant.

Tensile strength and failure. For SS, 114 pieces were tensile tested, of which 15 were rejected because they broke in the grips; the mean UTS was 22.88 MPa (coefficient of variation 29.1 percent). For No. 2, 174 pieces were counted, 26 were rejected for grip breaks, and the mean UTS was 12.92 MPa (COV 38.9 percent) as stated in the text and Table 3. About 12 percent of each group was rejected for breaks inside the grips. The minimum MOE from MSR correlated with UTS at r = 0.75. Failure occurred at the predicted MSRD location in 43.9 percent of SS pieces and 58.6 percent of No. 2 pieces, and most failures were at knots.

Knot size is the dominant sorting variable. The lumber was divided into four tiers by maximum knot size:

According to the authors, raising the permitted maximum knot size by half an inch reduced the minimum MSR MOE by about 8.8 percent and the mean and fifth-percentile UTS by about 20.1 and 29.8 percent. The R-squared values between knot size and MOE or UTS ranged from 0.26 to 0.54, and the authors note that the T1 tier is small.

Comparison with S-P-F. Aspen was weaker than S-P-F, especially in tension and in the lower grade:

Note that the No. 2 mean UTS is 12.92 MPa in the text and Table 3 but 12.85 MPa in the comparison table; the page does not explain the difference.

Grade yield with MSR. Three combinations of MSR settings were evaluated. The third combination gave 44 percent of pieces at 1450f-1.3E, 18 percent at 1200f-1.2E and 38 percent rejects, so about 62 percent met 1200f-1.2E or better. The authors say this agrees with a published figure of 60.5 percent for 1450f-1.3E reported for hybrid poplar. Combinations that set 1650f-1.5E as the top grade (18 percent yield) left rejection at 52 to 56 percent.

CLT grades. Under PRG-320, an E1 grade was achievable only in the minor-strength direction (23 percent yield, 77 percent rejected). For grade E3, yields were 34 percent (major direction) and 37 percent (minor direction), with 29 percent rejected. The authors present E3 as showing good potential for aspen and suggest pre-sorting in the mill to cut rejection.

Practical applications for sawmills

This section is the portal's interpretation and is not a claim made by the paper.

• Knot-based pre-sorting. Because knot size correlated with both stiffness and strength, a mill that already scans or visually grades could add a knot-size limit as a cheap first sort for material destined for engineered products. The paper's tier table gives a sense of how much would fall into each tier (roughly half of the tested pieces had knots above 1 1/4 in).
• Expect a split between MSR settings and visual grade. The MSR grade-yield figures suggest that a mill grading aspen by machine would find a large share in lower stress grades; any mill considering this should run its own calibration, because the study covers one mill.
• Calibrate handheld tools. Mills using stress-wave tools for quick checks should not take the values at face value: in this study MSR and LSW read higher than edgewise bending, and the authors explicitly advise calibration against EWB.
• Product positioning. The CLT results (grade E3) may interest mills near aspen resources that wish to supply lamination stock. The 29 percent rejection rate shows that a sorting step and a use for the rejects (for example in non-structural products) would need to be planned.
• Transfer outside Canada. The paper contains no European or other non-Canadian data, so its numbers should not be transferred to other aspen or poplar resources without local testing.

Limitations

• All lumber came from a single sawmill in Hines Creek, Alberta, so the results describe one resource and one process.
• The 8 ft SS pieces were not tensile tested, and about 12 percent of UTS specimens were discarded for grip breaks, which reduces the sample for the strength results.
• The shortest knot tier (T1) contained only 19 pieces, which weakens regressions that include it; knot-size R-squared values were modest (0.26 to 0.54).
• The three MOE methods gave significantly different results, so absolute MOE values depend on the method.
• The paper has no dedicated limitations section; the points above were drawn from the text by the portal.
• The No. 2 mean UTS differs slightly between the text (12.92 MPa) and the S-P-F comparison table (12.85 MPa), and the portal could not resolve this.
• The paper does not address drying quality, sawing recovery or economics, so nothing here shows whether aspen conversion is commercially attractive.
• The portal read the article through a summarised fetch of the page; figures were not viewed and some details may be abbreviated.

Conclusions

The study shows that Alberta trembling aspen 2x4 lumber has a specific gravity of about 0.40 and mechanical properties that fall below those of S-P-F, mainly in tension and in No. 2 grade, but that with machine grading and knot-based sorting a considerable share of the output can reach structural stress grades. Knot size is the clearest practical sorting criterion, MSR is a better stiffness predictor than the handheld stress-wave method, and grade E3 CLT lamination stock appears feasible with about 29 percent rejection. The authors recommend a species-specific grading approach rather than applying softwood rules directly (a point from the discussion as summarised for the portal). Larger, multi-mill sampling would be needed before generalising.

Original research

• Authors: Dawei Wang, Mengyuan Zhang, Meng Gong, Ying-Hei Chui
• Title: Evaluation of Major Physical and Mechanical Properties of Trembling Aspen Lumber
• Journal: Materials, Volume 17, Issue 12, Article 2952
• Year: 2024 (published 17 June 2024)
• DOI: 10.3390/ma17122952
• Link: https://www.mdpi.com/1996-1944/17/12/2952
• Affiliations: Wood Science and Technology Centre, University of New Brunswick, Fredericton, Canada; Department of Civil and Environmental Engineering, University of Alberta, Edmonton, Canada
• Funding: Alberta Innovates (project 212201625)
• Licence: Open access; page footer states Creative Commons Attribution 4.0 (CC BY 4.0), while page metadata lists CC BY 3.0

This article is an independent summary prepared by the portal; the original paper should be consulted for full methods, tables and data.