Hardwood sawing carries a heavier footprint than softwood: LCA results from New Brunswick mills

Introduction

Wood is widely promoted as a low-carbon construction material, but the climate case for any timber product rests on what happens before the board leaves the sawmill gate. A team from the University of New Brunswick and the USDA Forest Products Laboratory has now published a cradle-to-gate life cycle assessment (LCA) of hardwood lumber in New Brunswick, Canada. The study appeared in the journal Buildings in February 2025 and is notable because most published lumber LCAs deal with softwood. Hardwood sawmilling differs in important ways: logs are smaller and more variable, a large share of the harvest is low-grade material, and recovery of lumber from the log is far lower.

The authors set out to do three things. First, to quantify the environmental performance of producing rough green hardwood lumber, from tree harvesting to the mill gate. Second, to build a life cycle inventory (LCI) for hardwood lumber in the region, a data gap that has limited the ability of designers and manufacturers to prepare environmental product declarations. Third, to compare the result with an existing cradle-to-gate assessment of softwood lumber from the same region.

For European readers, the paper is interesting less because of New Brunswick itself than because of what it shows about the physics and economics of hardwood conversion. Many European hardwood mills (beech, oak, birch, maple) face a similar combination of low-grade log supply, modest lumber yield and substantial electricity use in the sawline. The numbers below should not be transferred directly to a European plant, but the structure of the result is instructive.

Research methodology

Goal, functional unit and boundary. The functional unit is 1 cubic metre of rough green hardwood lumber, meaning sawn but undried and unplaned. The system boundary covers tree harvesting, transport of logs to the sawmill, debarking and sawing. Drying and planing are excluded, as are forest management activities such as planting, thinning and fertilisation. Use phase and end of life are outside the scope, so the assessment is strictly cradle-to-gate.

Primary data. Data were collected through survey questionnaires, follow-up emails and site visits for the reference year 2022. The sample is small: one logging company and two hardwood sawmills in New Brunswick. The two mills are deliberately different. One produces pulp chips alongside lumber from low-quality logs, while the other produces lumber mainly from high-quality logs. The logging company harvested 305,000 m3 of logs including bark. The species mix was dominated by yellow birch, white birch, sugar maple and red maple (together 70 percent), with aspen at 20 percent and a small softwood fraction. By log grade, roughly 10 percent were high-grade saw logs, 30 percent pallet-grade and 60 percent pulpwood. The sawmills together produced about 68,429 thousand board feet (161,493 m3) of rough green lumber, converted at 2.36 m3 per thousand board feet. Harvest residues were estimated at 0.165 m3 per m3 of roundwood and assumed to decay on site.

Background data and tools. Secondary data, for example for diesel, electricity and transport processes, came from the DATASMART LCI database, which is built on the US LCI database (version 1.60) and ecoinvent (version 2.2). Modelling was done in SimaPro 9.6.0.1. Impacts were calculated with the TRACI method for six midpoint categories (global warming, acidification, eutrophication, smog formation, ozone depletion and fossil fuel depletion) and with Cumulative Energy Demand for primary energy.

Mass balance and allocation. Moisture content and specific gravity were determined following ASTM D143; the average moisture content of logs was 67 percent and the average green specific gravity 0.532. Burdens were shared between lumber and co-products (chips, sawdust, bark) on the basis of oven-dry mass. Logs were assumed to travel an average of 118 km one way by diesel truck, equivalent to 104 tonne-kilometres per cubic metre of lumber.

Sensitivity analysis. Key parameters were varied by one standard deviation to see how strongly the global warming result depends on them. The two tested were the cubic lumber recovery (CLR) and the electricity used in sawing.

Key findings

Yield is low. Producing 1 m3 of rough green lumber requires about 2.75 m3 of logs, because CLR averages 0.36 (a volume conversion of 36.4 percent). In mass terms, 1,461 oven-dry kilograms of logs yield 532 oven-dry kilograms of lumber, a reduction of about 64 percent. The remaining material leaves the sawline as co-products: by mass roughly 47 percent green chips, 8 percent sawdust and 9 percent bark, against 36 percent lumber.

Global warming potential. The cradle-to-gate result is about 40.7 kg CO2-equivalent per m3 of lumber (reported as roughly 41 kg in the abstract), excluding biogenic carbon storage. The split by stage is:

Energy. Total primary energy demand is 736 MJ per m3. Manufacturing accounts for 54 percent of it, harvesting for 28 percent and transport for 18 percent. The paper describes the energy supply as 98 percent non-renewable and only about 2 percent renewable, with 11 MJ/m3 reported as renewable. Note that the paper also quotes 583.6 MJ/m3 for the non-renewable fraction, which does not match 98 percent of 736 MJ; the text does not explain the difference (it could reflect other categories such as nuclear), so readers should treat the 98/2 split as the authors' headline statement and the underlying breakdown as unresolved.

Hotspots by impact category. Sawmill manufacturing, including the log yard and sawing, dominates several categories: about 92 percent of ozone depletion, 58 percent of eutrophication, 42 percent of global warming and 39 percent of fossil fuel depletion. Harvesting, driven by diesel machinery and fuel for crew transport, holds the largest share of smog formation and acidification, only slightly ahead of manufacturing. Mill electricity matters because grid power is largely generated off site; the authors note that the relevant grid is roughly 30 percent fossil and close to 40 percent nuclear.

Stored carbon dwarfs process emissions. Assuming wood is 50 percent carbon by oven-dry mass, one cubic metre of lumber holds 266 kg of carbon, equal to about 975 kg CO2-equivalent. That is roughly 24 times the cradle-to-gate emissions. This is a simple storage figure and not a full carbon accounting: it ignores what happens to the carbon later, and what would have happened to the forest without harvest.

Sensitivity. Varying CLR by 58 percent (one standard deviation, baseline 0.36) raised total global warming potential by about 11 percent. Varying sawing electricity by 54 percent raised it by roughly 8 percent. Both parameters also materially influenced ozone depletion and primary energy use. In other words, yield and electricity are the levers that move the result most.

Comparison with softwood. Against the softwood cradle-to-gate study for the same region, hardwood lumber performs worse on every headline indicator:

• total global warming potential 54 percent higher;
• manufacturing global warming potential 69 percent higher;
• smog formation 68 percent, acidification 63 percent and eutrophication 40 percent higher;
• primary energy 736 versus 520 MJ/m3 (41 percent higher), mostly from non-renewable fossil sources;
• electricity in sawing 61 kWh/m3 for hardwood versus 47 kWh/m3 for softwood.

The authors attribute the gap to lower lumber recovery, more log processing per cubic metre of lumber, higher diesel and electricity use and longer haul distances.

Mill-to-mill spread. The reported hardwood figures are weighted averages of two very different operations. According to the authors, the high-grade mill used almost half as much electricity per cubic metre as the pallet-lumber mill and achieved double the CLR. So the average hides a wide range of performance.

Practical applications for sawmills

From the paper:

• Yield is the strongest lever. The sensitivity test showed that a change in CLR moves global warming potential more than any other tested parameter. The authors recommend improving recovery through advanced sawing technology, including artificial-intelligence-based optimisation.
• Use wood residues for on-site energy. The authors point to biomass energy from mill residues as a way to reduce the fossil share of manufacturing energy, which is where most of the burden sits.
• Electricity intensity in the sawline deserves attention, since hardwood sawing required 61 kWh/m3 against 47 kWh/m3 for softwood in the comparison.
• Log grade matters. The mill working mainly with high-quality logs showed far better electricity and yield performance than the pallet-lumber mill.
• Collecting site-specific data is the necessary first step for any environmental declaration; the authors call for future cradle-to-grave studies with more participating sawmills.

Portal interpretation (not claims of the paper):

• European hardwood sawmills could use the paper's structure as a checklist for an internal footprint screening: log transport distance, diesel use in the yard, electricity per cubic metre sawn and recovery rate are all data a mill typically already holds.
• Because the study stops at the green lumber stage, the kiln drying step, which is often a large energy user in practice, would need to be added for any comparison with a kiln-dried product or an EPD. The paper itself does not quantify it.
• Buyers asking for product carbon data will increasingly compare hardwood with softwood. Mills that can document higher yield and lower electricity intensity, and a high share of renewable energy for process heat, will be able to present a stronger case.
• The 24-to-1 ratio of stored carbon to process emissions is a useful communication point, but it should be presented with the caveat that it covers storage only and not the wider forest carbon balance.

Limitations

The authors name several limitations, and a few more are worth noting.

• Only two sawmills and one logging company contributed data, because other operators had incomplete data or were reluctant to share it. The results are therefore a regional snapshot for 2022, not a statistical average of the sector.
• The two mills differ greatly in what they produce, and the weighted average may not describe any real mill.
• Drying and planing are outside the boundary, as are forest management, use and end of life.
• Wood waste burned in mill boilers was not reported in the dataset.
• Biogenic carbon storage was estimated on a simple assumption of 50 percent carbon in oven-dry wood.
• The twelve data tables of the original paper contain the full inventory and impact values. They were not accessible in the text extraction used for this summary, so the inventory quantities of diesel, gasoline and electricity per cubic metre and the full impact table are not reproduced here.
• As noted above, the paper's statement of the non-renewable energy share is not fully consistent with the absolute value given for it.

Conclusions

For rough green hardwood lumber from New Brunswick, the cradle-to-gate carbon footprint is about 41 kg CO2-equivalent per cubic metre, with sawmill manufacturing the largest single contributor, followed by harvesting and then log transport. Energy demand of 736 MJ/m3 is almost entirely fossil-based according to the authors, and hardwood lumber is clearly more burdensome than softwood lumber from the same region on every headline indicator tested. The main reasons are lower recovery, higher electricity demand in sawing and longer transport. The stored carbon in the lumber is about 24 times the process emissions, which supports the general climate case for wood products, but it does not remove the need to improve the process. The recommended directions are higher yield, on-site biomass energy and larger data sets covering more mills and the full product life cycle.

Original research

• Authors: Nadia Zahabi, Meng Gong, Hongmei Gu, Janet Blackadar
• Title: The Cradle-to-Gate Life Cycle Assessment on Hardwood Lumber Production in New Brunswick, Canada
• Journal: Buildings, Volume 15, Issue 3, Article 499
• Year: 2025 (published 5 February 2025)
• DOI: 10.3390/buildings15030499
• Link: https://www.mdpi.com/2075-5309/15/3/499
• Licence: Open access, CC BY (the 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.