Life cycle assessment of sitka spruce forest products grown in Ireland

The reference functional unit of the forestry model analysed here is 1 m³ of Sitka Spruce Log Product Over bark (OB) at the forest road or delivered to the factory gate depending on the system boundary assumed. After thinning and clearfell, the harvested forest products are divided into three distinct product categories based on their size including sawlog, pulp wood and pallet wood as defined in Table 1.

In any given LCA, the functional units and their properties need to be clearly defined in order to meet ISO standards and also to ensure the LCA is comparable with other studies. For a harvested wood product, these properties include the moisture content (MC) and the density of the product.

The moisture content and both the green and dry density of the log products when harvested vary significantly between products. In order to represent a typical Sitka spruce log from an Irish forest, typical values were assumed. The density and MC of the harvested forest products have a significant influence on a number of factors in the study. For example, the fuel required in log haulage is directly proportional to the green density of the product. The MC of the logs also has a significant impact on the drying operations during any processing that may be required. Finally, the dry density of the logs also directly defines the mass of carbon stored per unit volume of product. The density of the log products tends to vary depending on many factors such as forestry conditions, forest management practices, age of forest stock and also the product type. Studies conducted across Ireland report different mean dry densities for Sitka spruce produced in Irish forests. Some found a mean density of 350 kg/m³ (Gallagher et al. 2004) while others report a calculated mean of 447 kg/m³ (Kent et al. 2011). However, this is primarily due to the considerable variability of the density and the low sample size used in these studies. For the purpose of this study, the dry density of a typical Sitka spruce log was assumed to be 370 kg/m³. This is relatively consistent with the meta-study undertaken by Harte et al. (Harte et al. 2014) and other Irish-based studies (Gil-Moreno et al. 2022; Mockler 2013; Simic et al. 2019) which found the dry density to be 359 kg/m³, 372 kg/m³, 381 kg/m³ and 386 kg/m³, respectively. It should also be noted that fast-grown logs typically have a lower density than slower-grown logs. In order to quantify the MC of the harvested log products in Ireland, the green densities of the products were calculated, which when combined with the dry density can provide the MC. The mass and volume of over 22,000 deliveries in 2011 and 2012 to sawmills and boardmills in Ireland were used to calculate the typical green density of the harvested logs. This data was provided from personal communication with sawmills and product manufacturers in Ireland (OSB Manufacturers 2015; Sawmills 2015) which allowed the typical green density and, in turn, the MC (wet basis) of the softwood logs extracted to the forest road to be calculated. The results were found to be 890 kg/m³ and 58%, respectively. This is consistent with the advice provided by a harvesting expert in Coillte Teoranta, the commercial semi-state custodian of Ireland's forests, who suggested that a harvested MC of 55% is a reasonable assumption (Coillte 2015a).

The system boundary adopted here follows a conventional ‘cradle-to-gate’ approach which assesses all major forestry operations associated with one full forest rotation from the seedling production stage to the extraction of the logs to the forest road. The ‘cradle-to-gate’ system boundary of the proposed forestry model is illustrated in Fig. 1 within the dashed black line. A second, expanded, system boundary was also considered, which includes the delivery of the forest products from the forest roadside to the factory gate for further processing. This is known as a ‘cradle-to-factory gate’ LCA within the solid black line in Fig. 1 and is common in forestry-based LCAs. The system boundary of the proposed forestry model illustrated in Fig. 1 shows the forestry production procedure in Ireland which has been subdivided into six distinct operations as follows: 1) Seedling Production, 2) Site Preparation, 3) Seedling Planting and Stand Maintenance, 4) Road Construction and Maintenance, 5) Harvesting Operations and 6) Timber Haulage. Other minor processes were ignored due to the scope of the study where such processes would be expected to have an insignificant effect on the overall impact of the forestry operation. Some of the excluded forestry processes were the collection and delivery of the seeds from the national seed centre to the seedling nurseries, forest management operations such as the use of fencing and certain tending processes such as pruning, and the utilisation of forest residue including stump collection.

The forestry industry typically has multiple product outputs both in terms of wood-based products and also non-wood products and services such as food products and recreational services. For the purpose of this model, 100% of the environmental burden is assumed to be distributed across the forest products defined in Table 1. Based on a review of over 28 studies that examined forestry products using LCA, Klein et al. (2015) found that several allocation approaches were considered, the three most popular of which were mass, volume and economic value. Klein et al. (2015) recommended the mass-based allocation approach as the most suitable approach in a forestry-based LCA as forestry product prices vary substantially over time and between countries. Despite this many authors opted to undertake an economic allocation approach as recommended generally by the ISO standards (ISO 2006a, b). Ultimately, the major difference associated with each approach in the context of forestry products is that the economic allocation approach would allocate greater environmental impacts to the higher value products such as the sawn logs whereas the mass or volume allocation approach would assume approximately the same impacts for the same volume of goods regardless of value. There is still no consensus throughout the industry as to which approach should be used. For the purpose of this study, a mass-based allocation approach was implemented.

An additional factor that must be taken into account is the fact that the three forest products considered are produced from both the thinning and clearfell operations, which are considered separately as shown in Fig. 1. Therefore, two sets of each product type (saw, pallet and pulp logs) are produced, one from the thinning operations and one from the clearfell operations, which influences how the forestry system is modelled. As the inputs associated with each of the thinning and clearfell processes differ slightly, both sets of products will ultimately have different environmental impacts. Issues to be considered and addressed here include whether some of the direct inputs associated with the thinning operations (fuel and urea usage) should be allocated towards the final clearfell operation. Furthermore, it must be determined whether the thinning operation is solely undertaken to produce the thinned harvested products, or is undertaken to improve the quality/yield of the harvested clearfelled products; therefore in the latter case, the thinning operation should attribute some of its impacts towards them.

Figure 2a depicts forestry model 1 and the issue described above. As can be seen, the provisional impacts associated with the plantation and growth of the forest site are 100% attributed to the thinning operation, as is the use of fuel and urea for the thinning. However, the thinning operation then has two major outputs, the thinned log products and the thinned forest which is clearfelled at a later stage. This modelling approach requires that a certain proportion of the impacts due to the thinning operation be allocated to the clearfell operation and products. This would imply that the clearfelled logs would also take a proportion of the urea and fuel used in the thinning operations. This model implies there is a significantly higher environmental impact associated with similar clearfelled logs than thinned logs as a proportion of the impacts associated with the thinned logs are allocated to the clearfelled logs.

Figure 2b shows a second allocation model of the same system. The major difference shown here is that the impacts associated with the managed forest are directly allocated between both the thinning and clearfell operations as opposed to being 100% allocated to the thinning. Using this model, the forest management impacts are distributed to the thinning and clearfell products processes similar to the previous model. However, this allows the direct fuel and urea usage from the thinning and clearfell processes to be attributed to the thinning and clearfell harvested logs, respectively. This model ultimately assumes that the grown forest can be subdivided into two given areas, namely for thinning and for clearfelling and allows the differences between the thinning and harvesting operations to be reasonably assessed.

Figure 2c demonstrates a third approach which combines both the thinning and clearfell operations together into one single harvesting operation. This model may be representative of the real-life situation, however, as shown, only the impacts associated with the combined harvested forest products can be analysed therefore significant detail is lost. Most forestry-based LCAs make this simplification as most upstream processes such as the manufacturing of sawn timber or OSB are only concerned with the average harvested product itself, regardless if it came from a thinning or clearfell operation. However, the difference associated with the thinning and clearfell operations in terms of fuel and urea usage cannot be assessed from this approach which would be beneficial in highlighting the environmental hot spots throughout the forestry industry in Ireland. Upon assessing the advantages and disadvantages of each allocation approach, the model depicted in Fig. 2b was assumed as the default model when analysing the total environmental impacts associated with the production of the harvested log products throughout this study. This version allows the varying effects of both the thinning and harvesting operations to be clearly assessed and also reduces the level of allocation required, which is in agreement with the ISO standards which state that allocation should be avoided when possible (ISO 2006a).

The primary data was gathered predominately from experts within Coillte Teoranta, the commercial semi-state custodian of Ireland's forests. However, further data was also sourced from Teagasc, the State Forestry Advisory Board that deals primarily with private forests (Teagasc 2015), and from local forest managers and forestry contractors. Secondary data was then derived from published studies and databases where primary data was not available as detailed in Section 3.4. This combination of data supplied from both the public and private sectors was crucial in order to ensure a complete and representative model of the Irish forestry industry throughout the country. It should also be noted that the most influential forestry processes in terms of environmental impact were identified prior to initiating the data collection phase of the study. Such processes were the fuel used throughout harvesting processes, the log haulage distances from the forest road to the manufacturing plants and the gravel usage associated with the establishment and maintenance of forest roads for harvesting. The sampling intensity while collecting data was biased towards these processes. While every effort was made to source reliable data from numerous sources for defined product systems given the pre-defined system boundaries and inherent limitations of data gathering, some unknown environmental impacts may not be captured in the analysis.

The LCI of the forestry model developed here was conducted using LCA software SimaPro (V8.0.3.14) (Pré 2015), which incorporates the Ecoinvent LCI database (V.3.3) (Ecoinvent 2016) adapted to European conditions. Using this software tool and database, the total list of emissions to air, water and land and also the resources used throughout each procedure in the forestry process was quantified.

The ReCiPe midpoint hierarchic perspective, which is incorporated within the SimaPro software, was chosen as the LCIA methodology. ReCiPe is a comprehensive methodology and includes over 18 midpoint indicators such as ozone depletion, acidification, eutrophication and climate change and is widely accepted throughout the LCA community and the wood-based product LCA community.

Individual models for each of the forestry products described here were analysed using the ReCiPe methodology. The results are then discussed further with a specific focus on the global warming potential (GWP) impacts.

The first process considered in this study is defined as the ‘seedling production stage’. Many life-cycle studies have been conducted on the seedling production process, however, the most detailed have focused on indoor/greenhouse-based nursery conditions (Aldentun 2002; Schlosser et al. 2003; Timmermann and Dibdiakova 2014) which required far more electricity but less fuel usage compared to outdoor nurseries which are the primary source for Sitka Spruce seedlings in Ireland. Coillte operates four nursery centres, which supply between 21 and 25 million plants per year to the forest market sector (Coillte 2021). The primary data gathered in relation to seedling production in Ireland was provided directly from Coillte (Coillte 2015a) and includes all significant material and energy inputs associated with the total production of 24 million seedlings from the Irish nurseries across Ireland over a twelve-month period between 2013 and 2014. The seedlings' growth was assisted at the Coillte nurseries using water and a number of fertilisers, totalling approximately 230 t, over the twelve-month period (a combination of primarily 12,6,18 NPK, a nitrogen/sulphur top dressing and a small number of micronutrients). Some herbicides and pesticides were also used in order to protect the seedlings at their early stage of production. These include cypermethrin and Flexcoat (a plant polysaccharide) used to prevent pine weevil damage and also metam sodium used for soil sterilisation. Coillte has found that the combination of cypermethrin and Flexcoat has facilitated a 25% reduction in cypermethrin usage in its nurseries (Coillte 2015b). Other major inputs for seedling production include the fuel usage required for the tractors, forklifts and other machinery used on site. Over 24 tractors were used year-round over the nursery sites, which span over 220 ha. Electricity usage was also a major input to be considered as it powered both the electric heaters required to aid in the growth of the plants and also the refrigeration in order to preserve the plants when required. The delivery of the seedlings to their forest site was also considered. It was estimated that the average distance from the nursery to the forest sites was approximately 175 km.

The inputs and outputs considered in the production of the seedlings are quantified in Table 2. Of the 24 million seeds, approximately 1.5 million were exported while the remainder were delivered to pre-prepared sites in Ireland, ready for planting.

Once the grown seedlings are ready to be planted, they are transported to a pre-established forest site to begin the forest rotation. Site establishment is essential in both the afforestation of a greenfield site and the reforestation of an existing clearfelled forest site. The aim of this process involves preparing the ground for planting to increase the probability of the planted seedling’s survival and ultimately, to maximise the timber yield for the forest site. Site preparation or site establishment refers to cleaning, draining and all other preparation of the forest site prior to planting. Presently in Ireland, approximately 2,000 ha of afforestation occurs each year, which, compared to the 808,848 ha of existing forest sites, is a relatively small change to the overall forest estate in Ireland (< 1%) (Department of Agriculture Food and the Marine 2022). For this reason, for the forestry model developed here a reforestation site is considered, which is also consistent with the steady-state assumption outlined above.

In Ireland, a combination of windrowing and mounding is considered the most popular method for site establishment. Based on data provided by both Coillte (Coillte 2015a) and Teagasc (Teagasc 2015), it was found that over 60% of re-established sites use this combined approach, whereas approximately 30% use windrows only. Finally, less than 10% of sites use other methodologies such as ripping or straight planting. Due to its widespread usage, it was decided that the windrowing and mounding approach would be assumed for the forestry model developed here. Besides being the most popular method, it is also the most energy-intensive of the methodologies used and hence is considered the most conservative approach for the model.

Data collected from experts (Coillte (Coillte 2015a) and Teagasc (Teagasc 2015)) showed that it takes approximately 8 h with an excavator to prepare 1 ha of forest site for planting using the windrowing and mounding methodology. Fuel usage data from the excavators were also provided by Coillte forest managers which assumed approximately 12 L of fuel was used per hour per excavator. Therefore, it was possible to calculate the total fuel required to re-establishing a typical forest site in Ireland as shown in Table 3.

Although stump removal and residue recovery are being tested on certain trial sites in Ireland, it is not common practice in Ireland. Ideally, after clearfell, stumps of no more than 5–10 cm are left which can be ignored during site preparation.

Once the sites are prepared, 2 to 3-year-old seedlings are planted by hand. This happens within 3 years of the site being clearfelled. In Ireland, the standard planting density associated with Sitka spruce is 2,500 seedlings per hectare. It takes several years until forest crops can be considered ‘well-established’ and during this time they are monitored carefully to ensure they reach their maximum potential in terms of timber quality and yield. Fertilisers, herbicides and pesticides are typically used throughout each forest rotation, which can have a significant environmental impact on the surroundings depending on their location, particularly in terms of water and soil quality. The quantity of pesticides considered in the forestry model was calculated based on detailed data provided by Coillte on their annual pesticide and herbicide usage data by volume and by active ingredient (Coillte 2015a). The average amount of pesticides and herbicides required to maintain one hectare of a typical Irish forest over one full rotation is shown in Table 4 in terms of the combined mass of their active ingredients, cypermethrin and glyphosate.

In Ireland, a significant amount of the land available for forestry is deficient in nutrients, therefore on such nutrient-deficient sites, several fertilisers are used in order to ensure the forest can grow successfully and ultimately maximise the potential yield from forest sites. Fertilisers are typically applied initially as the seedlings are planted and then as required throughout the remainder of the forest rotation. A number of fertilisers are used throughout Ireland, the most common of which is phosphorus-based, typically distributed in the form of rock phosphate. Other fertilisers are used when considered necessary such as potassium- and nitrogen-based fertilisers. Shown in Table 4 is the average amount of fertilisers required to maintain 1 ha of Sitka Spruce over one forest rotation based on Coillte’s total annual fertiliser usage data between 2011 and 2014.

The construction and maintenance of forest roads are often considered outside the scope of typical forestry-based LCA studies. However, when considered, they are regularly found to have a significant impact on the total environmental footprint of forest operations, particularly in relation to gravel usage. In Ireland, typical forest roads are between 3.4 m to 5.5 m wide depending on the purpose of the road and on average are constructed approximately 10–15 years after planting. An analysis of data on the percentage of road stock upgraded and financial standards by Coillte shows that they are expected to last approximately 32 years until a subsequent upgrade (Coillte 2015a). After the initial road construction, routine inspections and necessary maintenance are also essential in order to maximise the operational lifespan of the road.

Standard guidelines are in place in relation to the spacing and density of forest roads in Ireland (Ryan et al. 2004). However, each site must be considered individually, and its requirements must be carefully assessed and adhered to. Based on data provided by road engineers and forest managers involved with forest road construction (Coillte 2015a), it was calculated that forest harvest roads are typically constructed at a density of 21.23 m/ha in Ireland. The Coillte forest road network consists of approximately 8100 km of forest roads with 40 km of forest road built each year and a further 75 km upgraded. All roads are also maintained as required. Various road construction methodologies are used when constructing harvest roads, however, the vast majority of roads are constructed from crushed gravel or stone as excavated roads or floating roads. These two types of road represent approximately 60% and 30% of all road construction in Coillte forests, respectively. Excavated roads were estimated to require approximately 1–1.5 tonnes of crushed gravel per linear meter of construction. Floating roads involve constructing the road over the existing soil. This road construction methodology is estimated to take between 3—3.5 tonnes of crushed gravel per linear meter of construction. Primary material and energy usage data were collected from Coillte road engineers and forest managers in addition to direct material usage data from the ‘Coillte Forest Information Systems’ (Coillte 2015a). However, due to the huge variability associated with road construction and maintenance between forest sites and the difficulty in gathering suitable primary data, only the total gravel usage and the fuel usage of the excavator were considered in the scope of the study. Other materials such as geotextile grids, PVC pipes, and bridge construction were not considered as part of this study as they are only required on a small percentage of forest sites.

The incorporation of the ‘Steady-State’ assumption into the forestry model developed here assumes that the same quantity of gravel, fuel and other materials are used consistently each year in the construction and maintenance of the forest road network. If the steady-state assumption was assumed true in the case of Coillte’s forest road network, it would be valid to assume that the average volume of gravel reportedly used per year was distributed over Coillte’s entire road network, of approximately 8,100 km. However, when examining historic data it is also clear that the young nature of Irish forests and the rapid decrease in the Coillte afforestation programme in the mid-90 s make it difficult to apply such an assumption and risk underestimating the impacts if focusing only on recent data. To avoid this issue, the material usage data that was provided by the Coillte at a national level was combined with primary data on the required material and fuel use of both excavated and floating roads in Ireland provided by road engineers and forest managers from site-specific scenarios. This data was used to quantify the typical gravel and fuel requirements at each harvest road operation stage throughout the forest life cycle. This was then incorporated into the proposed forest model developed here. Using this hybrid approach that combines both national and site-specific data it was determined that approximately 2.66 and 1.09 tonnes of gravel are required to construct and upgrade one meter of a typical harvest road in Ireland, respectively. Roads are maintained as needed, which can range from every 2 years for busy roads to seldom, if ever, for minor, spur roads so it is difficult to define the exact volume of gravel required to maintain any given road. Therefore, it was assumed, based on the overall gravel usage by Coillte between 2010 and 2014 that 0.004 tonnes of gravel were used each year on each meter of road constructed. Excavators were reported to construct 7 m of a typical forest road on average per hour. This equates to approximately 62 MJ per meter of construction. This is assumed to reduce to approximately 31 MJ per meter of construction for upgrading forest roads.

For the purpose of the model described here, it was assumed that the life expectancy of the forest road is three forest rotations e.g. 120 years, which is in line with other forestry-based LCA studies. The final requirements associated with the construction and maintenance of a typical forest road in Ireland over one rotation was calculated to be 1.87 tonnes of delivered gravel and 45 MJ of fuel usage for excavator work as shown in Table 5. All secondary data such as the delivery of the gravel were assumed using the Ecoinvent database.

Various harvesting methodologies are used in Ireland and both thinning and final clearfell processes are discussed here together due to their operational similarities and similar data sources. In Ireland, a mechanised cut-to-length (CLT) felling system and a forwarder extraction methodology are the most commonly used for both thinning and final clearfell practices. CLT logging or harvesting is a felling system in which trees are delimbed and crosscut to their desired length directly at the stumps prior to extraction. It accounts for over 90% of harvesting operations across Ireland (Phillips 2004). Once the trees are cut, the harvested logs are extracted to the forest road using a forwarder.

The felling operations during a rotation length can be subdivided into two distinct processes; the thinning process (which can be further subdivided into the initial thinning and then subsequent thinnings) and the final clearfell at the end of the forest rotation. The total volumes of the harvested log by each product harvested for 1 ha of a typical Sitka spruce forest in Ireland are shown in Table 6. Once the trees are harvested at each thinning stage or at the final clearfell, they are sorted on site into product types based on their size (Table 1).

In order to quantify the total environmental impact associated with the harvesting operations detailed fuel usage and efficiency data for harvesters and forwarders were collected directly from private harvesting contractors (Harvesting Contractors 2015) and were combined with harvesting machinery data collected by Teagasc (2015). Combining both data sources, 20 harvesting and forwarding machines were assessed. Based on the types of machines used, their respective power, average hours of use in a given year and the average volume of logs harvested per machine, the energy requirement associated with harvesting and extracting 1 m³ of log to the forest road was calculated and is shown in Table 7.

In order to avoid butt rot and the spread of diseases, in particular, Heterobasidion (Fomes/Butt Rot), stumps are required to be treated using urea directly after felling in both the thinning and harvesting operations. The urea usage during felling operations was estimated to be approximately 0.75 l/m³ of harvested log based on details provided by Teagasc and harvesting contractors (Harvesting Contractors 2015; Teagasc 2015). As the urea usage is directly proportional to the volume harvested it is considered here rather than as part of the forest maintenance process.

The most common form of log transport from the forest road to the processing facility in Ireland is by articulated truck or truck and trailer. To accurately quantify the environmental impact associated with the transportation of the harvested forest products from the forest site using the SimaPro software, specific haulage data was required in terms of ‘Vehicle Category Class’ and ‘Gross tonne-kilometre hauled’ per delivery (Ecoinvent 2016; Pré 2015; Spielmann et al. 2007). The GWP impacts associated with 1 tkm of log product depending on Vehicle Category Class are presented in Table 8.

To quantify the timber haulage impact associated with the harvested forest products produced in Ireland, detailed log delivery weights and distances were collected from sawmills and the OSB boardmills. Detailed sawn and pallet log delivery data was provided in terms of individual deliveries from two of the five largest sawmills in Ireland (over 23,000 specific deliveries quantified using on-site weigh-bridges) and average delivery data from another sawmill, which cumulatively represents the delivery of over 1.1 million m³ of roundwood logs. The mean delivery distance from each forest to the sawmill was found to be 95.00 km, 89.11 km and 92.16 km for each of the three sawmills that provided data. The overall weighted average distance was then found to be 92.14 km per cubic meter of log based on the volume of logs each sawmill processed in that given year. A further 18,000 deliveries from 246 forests were analysed, which provided pulp logs to an OSB manufacturing plant. The average distance to each forest was found to be 183.2 km. The resultant weighted average distance per load was calculated to be 169.84 km. Using the primary weigh-bridge delivery data and mass and moisture content data provided from the aforementioned sawmills and OSB manufacturing plant (OSB Manufacturers 2015; Sawmills 2015), the average weight of the load and, in turn, the ‘Vehicle Category’, was quantified for the log haulage considered here. It was found that the average volume of the load per delivery is approximately 28.03 m³. This equates to an approximate mass of just less than 24 metric tonnes depending on the density and moisture content of the load. As this weight excludes the weight of the truck itself, it is safe to assume that the log haulage considered in the model was undertaken using a truck of weight class > 32 tonnes (GWP = 0.110 kgCO2eq.).