Introduction

With climate change recognised as a global threat, reductions in emissions of anthropogenic greenhouse gases (GHGs) are crucial [1]. Replacing fossil fuels with biomass is considered a viable approach, and this has increased the demand for biomass for energy conversion [2]. Transforming biomass to pellets provides improved characteristics, such as higher energy density, lower moisture content and a more homogeneous shape [3], making the material easier to transport, store and use. The global wood pellet market has grown sharply in recent decades and further growth is expected. Wood pellets are traded internationally, with large trade flows from North America and Russia to Europe, which is currently the main market for wood pellets. This is partly a consequence of the European Union’s target to reduce GHG emissions and increase the share of renewable energy sources [3]. Using wood pellets for electricity production in new dedicated bioenergy plants or for co-firing in existing fossil fuel-fired plants has been shown to be a relatively economically and technically straightforward solution to mitigate GHG emissions [2, 4, 5].

In addition to non-torrefied wood pellets, interest in pellets made from torrefied biomass (torrefied wood pellets) is increasing [6, 7]. In the production of torrefied wood pellets, a torrefaction step, in which biomass is exposed to temperatures between 220 and 300 °C in a low-oxygen atmosphere, is added before densification of the biomass [8]. The torrefied biomass typically contains about 90% of the initial energy content, but only 70% of the initial mass [9,10,11]. Consequently, torrefied wood pellets are more similar to coal in terms of handling, transport and milling [6, 7]. Compared with non-torrefied wood pellets, they have higher energy density, are less moisture sensitive and require less energy for grinding [7, 8]. In existing coal-fired power plants, co-firing rates of up to 10–15% for non-torrefied wood pellets are possible without major modifications. However, with torrefied wood pellets, the co-firing rate can be up to 50% and they can thereby replace more coal [6, 7, 11,12,13]. Besides woody biomass, other biomass feedstock is suitable for torrefaction (e.g. herbaceous biomass) [10, 14, 15]. Great improvements have been made in the torrefaction technology during the past decade and the main challenge today is to move from demonstration to industrial scale [15]. However, with the inclusion of an extra process and increased raw material demand, the production costs will most likely increase.

Increasing the use of biomass to replace fossil fuels is a central component in many climate change mitigation strategies [16]. It is important to have knowledge about the climate effects when new energy systems are being developed. As a major contributor to GHG emissions, coal generates about 40% of the world’s electricity [17]. In existing coal-fired power plants, both non-torrefied and torrefied wood pellets are seen as a potential replacement for coal.

Life cycle assessment (LCA) is a standardised method (ISO 14040/44) that assesses the potential environmental impact throughout the whole life cycle of a product or process. The method is often used to evaluate the climate effects of bioenergy systems [18,19,20], with global warming potential (GWP) being the most commonly used metric to describe the climate impact. GWP is defined as the integrated radiative forcing (RF) due to a pulse emission relative to the integrated RF for a reference gas (most commonly carbon dioxide (CO2)) over a defined time horizon. Radiative forcing (W m−2) describes the energy balance on Earth due to altered GHG concentrations, with a positive RF reflecting a warming climate response and a negative RF a cooling climate response [21].

Climate benefits of using non-torrefied or torrefied wood pellets instead of fossil alternatives have been shown in several studies [4, 22,23,24,25,26]. Most LCAs of bioenergy systems include only GHG emissions released during the production chain, including harvest, upgrading and transport of the biomass. However, these systems may also be connected to land use changes causing altered biogenic carbon stocks in both soil and living biomass. Furthermore, the general assumption that bioenergy is carbon-neutral (i.e. assuming that the same amount of CO2 as released during combustion is sequestered during regrowth of new plants) has been questioned for disregarding the time lag between emission and uptake of biogenic CO2 [27, 28]. The importance of including these effects in climate impact assessments of bioenergy systems has been repeatedly emphasised [29,30,31].

In LCA, all emissions from the system are usually summed up into a single pulse, irrespective of when they occur. With this approach, net changes in biogenic carbon stocks during the study period can be captured in the climate impact assessment, but not the temporary fluxes. In order to include these temporary fluxes, both the timing and magnitude of the GHG fluxes need to be considered in the climate impact assessment. Several approaches have been suggested to handle these temporal effects of GHG fluxes connected to bioenergy systems, e.g. expressing climate impact as radiative forcing or as temperature change [19].

Traditionally, sawdust and shavings have been the main raw materials used for wood pellet production [32]. However, in many Western and Central European countries, these residues from the sawmilling industry are already utilised to a large extent. Their availability is also dependent on the shifting demand for timber products. This has increased the interest in alternative raw materials, such as bark, wood from thinnings, forest residues and even prime log wood [32]. Sweden has large forest resources [33] and potential to increase the use of forest residues which would otherwise be left in the forest after final felling or thinning [34]. Forest residues are already used for heat and power production, but pelleting the biomass produces a more energy-dense and homogeneous product, which could open up new market possibilities.

The aim of this study was to analyse and compare the energy efficiency and climate impact of production and use of non-torrefied and torrefied wood pellets. The wood pellets, produced from logging residues extracted from a boreal coniferous forest stand (Norway spruce (Picea abies (L.) H. Karst)) in northern Sweden, were assumed to be exported and finally used in a power plant. In order to include annual GHG fluxes of both fossil and biogenic origin, a time-dependent LCA, expressing the climate impact as global temperature change over time, was performed.

Method

System Description

The production system included supply of raw material, upgrading, transport to the end-user and final use of the non-torrefied and torrefied wood pellets in a power plant (Fig. 1). In a base scenario, a transport distance of 600 km by train, from the pellet production plant to a harbour in Central Sweden, followed by a transport distance of 3000 km by ship, was assumed. These distances represent export to the UK or the Benelux countries. The wood pellet systems were also compared with a coal-based reference scenario.

Fig. 1
figure 1

Overview of the production system for non-torrefied and torrefied wood pellet systems based on logging residues, compared with a coal-based reference scenario

Full size image

For raw material supply, a boreal coniferous forest stand (Norway spruce) in Västerbotten in northern Sweden (64° N) was assumed for extraction of logging residues, with a rotation interval of 120 years. The analysis included extraction of logging residues, allocating emissions occurring prior to final felling to timber production. Ash was assumed to be spread in the stand according to Swedish recommendations to avoid soil depletion [35].

A wood pellet production chain comprising drying, fine milling, pelleting and finally cooling was assumed. For production of torrefied wood pellets, a torrefaction step was added before milling of the biomass. During torrefaction, about 30% of the initial biomass is converted to torrefaction gases (mainly water, organics and lipids) and non-condensable gases (mainly carbon monoxide (CO) and CO2) [7]. These gases can be recirculated and combusted to partly cover the heat requirement within the process, with current technology covering at least 60% of the heat requirement according to Batidzirai et al. [23]. Even for relatively dry raw materials, the main components of the torrefaction gas are incombustible water and CO2 [11]. To increase the combustion properties, the raw material is often pre-dried to a moisture content of about 20% before torrefaction [23, 32]. Compared with untreated biomass, torrefaction results in a 50–85% reduction in the electricity requirement for milling, depending on the degree of torrefaction applied [6, 10, 11]. Moreover, depending on the degree of torrefaction required, binders may be needed in the pelleting process [7]. Under optimal conditions, the electricity requirement for densification may be similar to that for non-torrefied wood pelleting, but values up to threefold higher have also been reported [6].

Mass and Energy Efficiency

To assess the overall energy efficiency of a system, the energy ratio (E r), describing the primary energy input per unit output energy, is a frequently used indicator [36]. Here, E r was calculated by dividing the energy in the non-torrefied or torrefied wood pellets produced (E out) by the primary energy input to the system (E in), based on lower heating value (LHV) of dry biomass adjusted for the specific moisture content:

$$ {E}_{\mathrm{r}}=\frac{E_{\mathrm{out}}}{E_{\mathrm{in}}} $$
(1)

Part of the incoming raw material was assumed to be used within the non-torrefied and torrefied wood pellet production process, to cover the heat requirement for drying and torrefaction. To assess the mass and thermal energy efficiency of non-torrefied and torrefied wood pellet production, the mass ratio (M r) and thermal energy ratio (E THr) of the processes [7] were calculated as follows:

$$ {M}_{\mathrm{r}}=\frac{M_{\mathrm{out}}}{M_{\mathrm{in}}} $$
(2)
$$ {E}_{{\mathrm{TH}}_{\mathrm{r}}}=\frac{M_{\mathrm{out}}\times {\mathrm{LHV}}_{\mathrm{out}}}{M_{\mathrm{in}}\times {\mathrm{LHV}}_{\mathrm{in}}} $$
(3)

where M out is the biomass output in the form of non-torrefied or torrefied wood pellets and M in is the biomass input including both the densified biomass and the biomass used for drying and in the torrefaction process, based on dry weight. The energy in the non-torrefied or torrefied wood pellets produced, LHVout, is equivalent to E out, while LHVin is equivalent to the energy in the unrefined biomass.

Climate Impact

First, GWP was calculated for the different systems using a time horizon of 100 years. The functional unit in the GWP calculations was 1 GJ fuel (produced in year 1). The GWP was expressed in CO2 equivalents (CO2-eq.) per functional unit, using characterisation factors of 28 and 265 kg CO2-eq. kg−1 for CH4 and N2O, respectively [21]. Biogenic CO2 fluxes were not included in the GWP calculations.

The climate impact was also expressed as the global mean surface temperature change, referred to as ∆T S in this study, at a specific point in time, following the time-dependent LCA methodology described in the study by Ericsson et al. [37]. Besides the RF, this method also considers the inertia of Earth’s processes, which delays climate effects. To calculate ∆T S, a time-dependent life cycle inventory is required, which makes it possible to include annual net biogenic CO2 fluxes between atmosphere and biogenic carbon stocks in biomass and soil, as well as GHG emissions from the production system. Annual fluxes of the three major GHG gases (CO2, methane (CH4) and nitrous oxide (N2O)) were quantified in a time-dynamic life cycle inventory including GHG emissions from harvest, upgrading and transport and biogenic CO2 emissions. Biogenic carbon fluxes were defined as the yearly difference between harvesting forest residues (combustion) and leaving the residues in the forest (decomposition).

A time step of 1 year was used for calculation of ∆T S and a temperature response function was applied. This function, referred to as the absolute global temperature change potential (AGTP) [38], represents the global temperature change due to RF. The atmospheric GHG concentrations were estimated based on the atmospheric perturbation lifetime of the gases, using simple exponential decay for CH4 and N2O [38]. To model the more complicated decomposition of CO2, the Bern carbon cycle model was used [39, 40]. The indirect effect of CH4 oxidation to CO2 was included by adding the oxidised fraction in the following year [38]. Baseline atmospheric GHG concentrations (CO2 390 ppm, CH4 1803 ppb, N2O 324 ppb) reported by IPCC and representing mean values for the year 2011 were used [41].

The forest management system was simulated using the Heureka Forestry Decision Support System (Heureka), and the forest carbon balance was modelled with the Q model [42]. The Heureka system is a software developed for forest planning analysis [43] and can be used for projection of forest growth when modelling live biomass stocks. The Q model can be used to simulate carbon stock changes in Heureka [44]. The Q model describes the decomposition over time for different litter fractions of a certain litter quality and requires data on the annual input of litter and annual weather.

The temperature change, ∆T S, for non-torrefied and torrefied wood pellets was expressed per gigajoule fuel (including emissions from large-scale combustion) and per gigajoule electricity (including the end-use efficiency). A stand view perspective was applied, meaning that the impact of a single harvest was studied.

Data Collection and Assumptions

Supply of Raw Material

Performance data for the raw material supply operations, including extraction, chipping and transport of logging residues to the pelleting plant, were obtained from the study by Hammar et al. [29]. An extraction level of 70% of the available biomass was assumed, resulting in a harvest level of 33.5 Mg dry matter (DM) per hectare. After final felling, the logging residues were assumed to be forwarded to the roadside for storage for a period of 8 months before chipping by a truck-mounted chipper. The chipped raw material was then assumed to be transported by truck to the pelleting plant. Based on the average transport distance of forest fuels in northern Sweden, the length of the round trip between the forest stand and pelleting plant was set to 145 km [29]. The truck was assumed to have a load weight of 34 Mg.

Dry matter losses during storage were set to 1% per month and chipping losses to 3.6% [45]. A moisture content of 45% (on a wet weight basis) at chipping was assumed and a LHV on a dry weight basis of 19.2 MJ per kg DM for forest residues [45, 46]. Fuel consumption for all activities associated with raw material supply (forwarding, roadside chipping and transport of logging residues) was assumed to be the same as reported by Hammar et al. [29]. Emission factors for diesel fuel were 77.8 g CO2 MJ−1, 0.034 g CH4 MJ−1 and 0.003 g N2O MJ−1 [47] and the primary energy factor was 1.09 [48].

Upgrading

For non-torrefied wood pellet production, part of the incoming raw material was assumed to cover the heat requirement for drying the raw material from 45 to 10% moisture content (wet basis). For torrefied wood pellet production, the heat requirement in the torrefaction process and for pre-drying the raw material to a moisture content of 20% was assumed to be covered by combustion of the torrefaction gases released and by direct combustion of part of the incoming raw material. Assumed electricity and biomass use for non-torrefied and torrefied wood pellet production in this study are shown in Table 1. In addition, 6.3 MJ diesel per GJ non-torrefied or torrefied wood pellets was assumed to be used for raw material handling at the pelleting plant according to Uasuf and Becker [49]. Emission factors for electricity, assuming Nordic electricity mix, were 19.3 g CO2 MJ−1, 0.162 g CH4 MJ−1 and 0.0023 g N2O MJ−1 [50], and the primary energy factor was 1.5 [51]. The carbon stored in biomass used for drying the raw material was assumed to be released as CO2 to the atmosphere on combustion. For CH4 and N2O, emission factors were 0.011 g MJ−1 and 0.006 g MJ−1, respectively [52]. The DM losses at the pelleting plant (for both non-torrefied and torrefied wood pellets) were set to 3% for handling and storage of the raw material and 1% in the pelleting process, according to the work by Sikkema et al. [26].

Table 1 Electricity and biomass use for non-torrefied and torrefied wood pellet production, based on a lower heating value (LHV) of 19 MJ kg−1 dry matter for non-torrefied wood pellets [29] and of 21 MJ kg−1 dry matter for torrefied wood pellets [7, 15]
Full size table

Transport

A transport distance of 600 km by rail and 3000 km by ship was assumed for the wood pellets, which represents export to the UK or the Benelux countries. For handling and transport by rail, DM losses for non-torrefied and torrefied wood pellets were set to 1%, while for ocean transport, additional 2% DM losses were added, based on the study by Sikkema et al. [26]. Cargo capacity and energy use of the transport options are described in Table 2. Emission factors and primary energy factor for heavy fuel oil were the same as reported by Gode et al. [54]. No return trips for rail and ship were included in the calculations.

Table 2 Cargo capacity and energy use for transport by rail and ship; maximum cargo capacity was assumed based on weight [55]
Full size table

Energy Conversion

Combustion of the fuels was assumed to occur within a few months of delivery to the energy conversion plant, releasing the carbon stored in the biomass as CO2 to the atmosphere (including also CO2 emissions from raw material used within the wood pellet production process). Emission factors for non-CO2 emissions in large-scale combustion of non-torrefied wood pellets were 0.01 g CH4 MJ−1 and 0.006 g N2O MJ−1 [50]. The same emission factors were used for combustion of torrefied wood pellets. Emission factors for production and distribution [56] and combustion [52] of coal result in total emissions of 100 g CO2, 0.04 g CH4 and 0.01 g N2O per MJ fuel.

The electrical efficiency of using non-torrefied wood pellets in a dedicated biomass power plant was assumed to be 35% for non-torrefied wood pellets and was set to 45% for coal, based on the study by Giuntoli et al. [56]. With the properties of torrefied wood pellets being more similar to those of coal, the efficiency for torrefied wood pellets was assumed to lie between that of coal and non-torrefied wood pellets (40%).

Forest Carbon Balance

Data on forest soil carbon balance and biomass harvest were obtained from a previous study [29] in which biogenic carbon stock changes were estimated when using logging residues in different Swedish climate zones for energy conversion. In that study, biomass stock changes were simulated using the Heureka forestry decision support system where the Q model is used for simulating decomposition [28]. The Heureka system is based on an empirical relationship of forest growth. Information on the forest management regime was retrieved from the forest planning tool INGVAR, and average values for site productivity and understory cover were calculated based on the Swedish Forest Soil Inventory and the Swedish National Forest Inventory [57, 58]. In this study, a forest stand located in northern Sweden (64° N) with a rotation interval of 120 years and with one thinning at year 65 was studied. The productivity (maximum tree height at age 100 year) was 20 m and the understory was blueberry and mosses.

Analysis of Alternative Scenarios and Changed Calculation Assumptions

With the higher energy density of torrefied wood pellets, more efficient transport is possible than with that of non-torrefied wood pellets. Different transport alternatives and distances were analysed to assess the effect on the total climate impact of the different pellet systems (Table 3) compared with the base scenario (S1). Export to destinations farther away was investigated (S2) assuming an increased transport distance by ship (25,000 km), which represents export to Asia where the consumption of pellets more than doubled in 2014 [59]. In scenario S3, use in southern Sweden or nearby countries was assumed (1200 km by rail).

Table 3 Transport distance (km) by different modes for transport scenarios S1–S3
Full size table

Torrefied wood pellet production is a relatively new industrial process compared with production of non-torrefied wood pellets. Therefore, the sensitivity of some of the assumptions made for the production process for torrefied wood pellets was studied:

  • A change in emission factors of ± 20% for non-CO2 emissions for large-scale combustion of torrefied wood pellets (these values were set to 0.01 g CH4 MJ−1 pellets and 0.006 g N2O MJ−1 pellets in the base scenario).

  • A change in thermal energy efficiency (E THr) of ± 5%-units (this value was set to 91% in the base scenario).

Furthermore, a sensitivity analysis was performed in which the fuel use in field operations and electricity use for pellet production was varied (± 20%). For torrefied pellets, systems in which the use of electricity for pellet production was 7 and 25 MJ per GJ pellets, respectively, were also investigated (this value was set to 16 MJ per GJ pellets in the base scenario).

In addition, the total temperature response per gigajoule electricity for co-firing wood pellets with coal was assessed, including emissions originating from both wood pellets and coal. Different co-firing rates were assumed: 5, 10 and 15% for non-torrefied wood pellets and 40, 50 and 60% for torrefied wood pellets. Based on findings by Zhang et al. [5], co-firing was assumed to result in 0.5% lower efficiency for every 10% non-torrefied wood pellets compared with using 100% coal. For torrefied wood pellets, a smaller reduction in efficiency can be expected [23], and therefore, a reduction of half that used for pellets was assumed. Electricity consumption for pulverisation of the fuel mix (coal and wood pellets), which is common in direct co-firing with coal, was not included in this study.

Results

Life Cycle Inventory of the Base Scenario

Mass and Energy Efficiency

The primary energy input and energy ratio were approximately the same for non-torrefied and torrefied wood pellets (Table 4). Higher primary energy requirement for upgrading and slightly higher energy requirement for supply of raw material for torrefied wood pellets compared with non-torrefied pellets were found. However, this was partly compensated for by more energy-efficient transport.

Table 4 Primary energy input per gigajoule fuel and the energy ratio for non-torrefied and torrefied wood pellets, not including end-use efficiency
Full size table

When upgrading the biomass into non-torrefied or torrefied wood pellets, part of the raw material was assumed to be used to cover the heat requirement within the process. For non-torrefied wood pellet production, the thermal efficiency was calculated to be 96% and for torrefied wood pellet production, it was 91%. The mass efficiency for non-torrefied and torrefied wood pellet production was calculated to be 87 and 74%, respectively.

Greenhouse Gases

The production system (including non-CO2 emissions from large-scale combustion but not biogenic CO2 emissions) for non-torrefied and torrefied wood pellets contributed similar levels of CO2 emissions per gigajoule fuel (Table 5). However, the CH4 and N2O emissions were somewhat higher (in a percentage perspective) for torrefied wood pellets compared with non-torrefied wood pellets.

Table 5 Total greenhouse gas emissions (carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O)) from production and use (including non-CO2 emissions from large-scale combustion, but not biogenic CO2 emissions) of 1 GJ non-torrefied and torrefied wood pellets
Full size table

All emissions associated with the production systems were assumed to be emitted in year 1, while biogenic net CO2 emissions (∆CO2) were taken as the difference in CO2 emissions between combustion of the logging residues in year 1 and leaving the residues in the forest to decompose over time, as shown in Fig. 2. The positive ∆CO2 in year 1 mainly represents CO2 emissions from combustion, while the negative ∆CO2 during the rest of the time frame represents the decomposition of the logging residues over time.

Fig. 2
figure 2

Differences in biogenic carbon dioxide (CO2) emissions (Mg CO2 ha−1) over time, referred to as ∆CO2, between direct combustion of logging residues in year 1 and leaving the residues in the forest to decompose over time

Full size image

Life Cycle Impact Assessment of the Base Scenario

Global Warming Potential

The GWP for the non-torrefied and torrefied wood pellet systems studied, not including biogenic CO2 emissions, was calculated to be 8.5 and 8.8 kg CO2-eq. GJ−1, respectively. In comparison, coal resulted in a much higher GWP, 114 kg CO2-eq. GJ−1 fuel.

Temperature Change

The climate impact assessment showed that, per gigajoule fuel, both non-torrefied and torrefied wood pellets contributed to lower global mean surface temperature change (∆T S) during the whole study period compared with coal (Fig. 3). Non-torrefied wood pellets had a slightly lower ∆T S value than torrefied wood pellets. The difference in ∆T S between the non-torrefied and torrefied wood pellets peaked after 10 years and then decreased over time. The highest global temperature effect for all fuels was obtained about 10–15 years after the emission impulse due to combustion in year 1. This delay is due to the inertia of the Earth’s climate processes. For both non-torrefied and torrefied wood pellets, the ∆T S curves declined faster over time compared with coal, partly because of the net negative CO2 emissions included for logging residues, as shown in Fig. 2. The forest stand studied had a rotation period of 120 years. However, the greatest global temperature changes took place during the first 50 years of the study (Fig. 3), and therefore, the remainder of this results section focuses on that period.

Fig. 3
figure 3

Global mean surface temperature change (∆T S) over time for 1 GJ non-torrefied and torrefied wood pellets produced from logging residues combusted at a power plant compared with 1 GJ coal (produced and used in a power plant in year 1) over one rotation period (120 years)

Full size image

Net emissions of biogenic CO2 accounted for by far the largest part of the global temperature effect for both non-torrefied and torrefied wood pellets (Fig. 4). They were also found to be the main cause of the higher ∆T S for torrefied wood pellets compared with non-torrefied wood pellets. The higher raw material demand for production of torrefied wood pellets resulted in greater net emissions of biogenic CO2, as these are fixed per hectare. Comparing only the differences in climate impact of the production systems, torrefied wood pellets had a higher global temperature impact in raw material supply and upgrading, while the global temperature impact for transport to the end-user was lower from torrefied wood pellets than from non-torrefied wood pellets (Fig. 5).

Fig. 4
figure 4

Global mean surface temperature change (∆T S) over time for 1 GJ non-torrefied and torrefied wood pellets produced from logging residues and used in a power plant, and temperature change for only biogenic carbon stock changes (∆Bio)

Full size image
Fig. 5
figure 5

Global mean surface temperature change (∆T S) over time for the production system for 1 GJ non-torrefied wood pellets (WP) or 1 GJ torrefied wood pellets (TOP), divided into supply of raw material, upgrading and transport to end-user

Full size image

Higher electrical efficiency was assumed for torrefied wood pellets compared with non-torrefied wood pellets, which can be expected since it has characteristics more similar to coal. On including the energy conversion efficiency, this resulted in a lower ∆T S being obtained per gigajoule electricity produced for the torrefied wood pellets (Fig. 6).

Fig. 6
figure 6

Global mean surface temperature change (∆T S) over time for 1 GJ electricity produced in year 1 from non-torrefied and torrefied wood pellets produced from logging residues used in a power plant

Full size image

Analysis of Alternative Scenarios and Changed Calculation Assumptions

The results showed a small difference in GWP between non-torrefied and torrefied wood pellets for all transport scenarios investigated. In contrast to the shorter transport scenarios (S1 and S3), a slightly lower GWP value was obtained for torrefied wood pellets in the longest transport scenario studied (S2) compared with non-torrefied wood pellets (Table 6). On the other hand, for all transport scenarios, the total ∆T S was found to be lower for non-torrefied wood pellets than for torrefied wood pellets. This is explained by the dominant effect of biogenic CO2 emissions on the results, which is not included in the GWP values. Nevertheless, the long transport distance in scenario S2 resulted in higher ∆T S and significantly higher GWP values for both non-torrefied and torrefied wood pellets compared with scenarios S1 and S3 (Table 6 and Fig. 7).

Table 6 Global warming potential (GWP) for production and use (including non-CO2 emissions from large-scale combustion, but not biogenic CO2 emissions) of 1 GJ non-torrefied and torrefied wood pellets delivered to a power plant for no transport and for transport scenarios S1–S3
Full size table
Fig. 7
figure 7

Global mean surface temperature change (∆T S) over time for 1 GJ non-torrefied wood pellets (WP) and torrefied wood pellets (TOP) produced from logging residues and transported and used in a power plant for the base transport scenario S1 (600 km rail, 3000 km ship) and the transport scenarios S2 (600 km rail, 25,000 km ship) and S3 (1200 km rail)

Full size image

The same emission factors for combustion at a large-scale power plant were assumed for both non-torrefied and torrefied wood pellets in this study. In the sensitivity analyses, changing the emission factors for CH4 and N2O by 20% for torrefied wood pellets was shown to have little impact on the total ∆T S. In contrast, a change in thermal energy efficiency of 5%-units for the torrefaction process had a larger impact on the total ∆T S (Fig. 8). The thermal energy efficiency affected the degree of torrefaction and thereby also the raw material demand in the process, which had a large impact on ∆T S from biogenic CO2 emissions per gigajoule torrefied wood pellets.

Fig. 8
figure 8

Global mean surface temperature change (∆T S) for 1 GJ fuel produced in year 1 of non-torrefied wood pellets (WP) and torrefied wood pellets (TOP) produced from logging residues and used in a power plant for the base scenario and with a change in assumed thermal energy efficiency of 5%-units (from 91% to 86 or 96%)

Full size image

A sensitivity analysis of the use of fuels for field operations and of the use of electricity for pellet production showed that these factors had a small impact on the total global mean surface temperature (Table 7). It was also noted that the difference between non-torrefied and torrefied pellets was negligible. Little impact on ∆T S was also found for a larger change in electricity demand (7 and 25 MJ GJ−1 pellets, instead of 16 MJ GJ−1) within the production of torrefied wood pellets.

Table 7 Change (%) in global mean surface temperature (∆T S per GJ pellets produced) in a time perspective of 50, 100 and 120 years, as a result of changes in the use of fuels and electricity for non-torrefied and torrefied wood pellets
Full size table

The ∆T S from co-firing wood pellets with coal (including GHG fluxes from both pellets and coal) was substantially lower for torrefied wood pellets than for non-torrefied wood pellets. This was due to the expected higher co-firing rates for torrefied wood pellets (studied rates 40, 50 and 60%) than for non-torrefied wood pellets (studied rates 5, 10 and 15%) and thus more coal being replaced in the former alternative (Fig. 9).

Fig. 9
figure 9

Global mean surface temperature change (∆T S) over time for 1 GJ electricity generated in year 1 for different scenarios for co-firing non-torrefied wood pellets (WP) (5, 10 or 15%) or torrefied wood pellets (TOP) (40, 50 or 60%) produced from logging residues with coal (including emissions from both the wood pellets and coal) compared with using 100% coal

Full size image

Discussion

In this assessment of the energy efficiency and climate impact of production and use of non-torrefied and torrefied wood pellets made from logging resides, by far, the largest impact on global temperature was found to be caused by biogenic CO2 emissions (Fig. 4). These were calculated as the net biogenic CO2 release between harvest and use of logging residues (in year 1) compared with on-site decomposition over time of the biomass. This indicates that choice and origin of the raw material and efficient use of the biomass are important factors when assessing the climate impact of wood pellet systems. It also confirms the importance of biogenic carbon fluxes in climate impact assessments of bioenergy systems, as discussed in several other studies (see [31, 37, 60]). Harvesting forest residues for energy releases the CO2 earlier in time (compared with the slower process of decomposition), which has a warming climate impact. In contrast, biomass grown directly for wood pellet production, such as willow and poplar established on former agricultural land, can have a cooling effect on mean global temperature, as shown by Porso and Hansson [30]. This is due to biogenic carbon sequestration in soil and biomass. In such cases, previous land use and its initial carbon stock are crucial factors, as they determine whether the system is going to be a net carbon sink or emitter [30, 31].

In order to produce a fuel more suitable for storage and transport, a torrefaction step could be added in the pellet production chain. However, a larger share of the incoming raw material was shown to be required in the production process for torrefied wood pellets compared with non-torrefied pellet production. One of the main benefits of producing torrefied wood pellets is energy-efficient transport, but with net biogenic CO2 emissions accounting for most of the global mean surface temperature change (∆TS), efficient use of the raw material is more important in a climate perspective. This explains why torrefied wood pellets had higher ∆T S values than non-torrefied wood pellets per gigajoule fuel in this study (Figs. 4 and 7).

On the other hand, when benefits in the end-use phase of the torrefied wood pellets were included, such as potentially higher electrical conversion efficiency and higher co-firing rates, a lower ∆T S value for torrefied wood pellets per gigajoule electricity was obtained compared with non-torrefied wood pellets (Fig. 6). However, in the long term, building new dedicated power plants for biomass combustion or enabling high co-firing rates may reduce the benefits of using torrefied wood pellets compared with non-torrefied wood pellets, as discussed by Koppejan et al. [6].

Factors such as thermal efficiency (Fig. 8) and the degree of torrefaction affect the performance of the torrefied wood pellet production process. Thermal efficiency is an important indicator of the technical performance of a process and is determined by thermal losses, moisture content and heating value of the raw material used. In the long-term perspective, Batidzirai et al. [7] point out that the thermal efficiency is likely to increase due to expected technical improvements in the torrefaction process, as well as more optimised use of torrefaction gas. Improved torrefaction technology may also result in wider use of torrefaction gas, replacing petroleum-based products. This could potentially result in both economic and climate benefits.

As mentioned, the largest part of the global temperature change was found to be caused by biogenic CO2 emissions. By assessing the climate impact using GWP with a fixed time horizon, these emissions would not be considered. Expressing the climate impact over time as global temperature change, as done in this study, or as instantaneous or cumulative radiative forcing over time [37, 61] enables inclusion of these temporal biogenic fluxes. All these studies report increased climate benefits over time for use of forest residues compared with a fossil alternative, as the difference in carbon stocks between extraction of the residues and leaving them to decompose in forest decreases over time. By including the temperature response, the inertia of the Earth is also considered, resulting in a delayed temperature response after radiative forcing. While additional uncertainties are introduced when presenting the results as temperature change rather than radiative forcing, temperature change values may be easier for policymakers to interpret.

However, using GWP and assuming biomass to be carbon-neutral (not including biogenic CO2 emissions) is the most common approach used to assess the climate impact of different non-torrefied and torrefied wood pellet production chains in earlier studies [23,24,25, 30]. In the present study, the GWP value was approximated to 7–26 kg CO2-eq. per GJ non-torrefied wood pellets and 7–25 kg CO2-eq. per GJ torrefied wood pellets for the different transport scenarios (Table 6). The GWP value for wood pellets used in Sweden is reported to range between 2 and 25 kg CO2-eq. per GJ non-torrefied wood pellets, including both national and imported pellets [24]. Roder et al. [25] reported a GWP value for non-torrefied wood pellets from forest residues produced in the south-east USA and exported for use in the UK, of approximately 15 kg CO2-eq. per GJ pellets. Agar et al. [22] compared the climate impact of non-torrefied and torrefied wood pellets from logging residues in Finland used for co-firing in Spain and found small differences between these two types of pellets, which is in line with findings in the present study (recalculated to 12–13 CO2-eq. per GJ fuel for non-torrefied and torrefied wood pellets). However, as also pointed out by Ehrig and Behrendt [4, 24] and Hansson et al. [4, 24], the design of the supply chain determines the climate impact and energy efficiency of a system. Different assumptions regarding e.g. raw material used, transport alternatives and electricity origin mean that the results of different studies of pellet production chains are not directly comparable. Factors with a large influence on the results include GHG emissions due to electricity mix, transport and emissions from drying the raw material depending on fuel used and moisture content [24, 25]. There are also great uncertainties in CH4 emissions during storage of raw materials, as discussed by Roder et al. [25].

Concerns regarding potential negative effects when removing additional forest biomass have been raised, especially regarding the risk of nutrient removal and its implications for future forest productivity and the risk of biodiversity loss [62,63,64]. A recent study by de Jong et al. [65], which examined the environmental sustainability of energy for forest residues, found that the highest risk of biodiversity loss is when deciduous forest residues are harvested. Furthermore, it is difficult to estimate the level to which the harvest should be limited, although the risk of species extinction increases when more than 50% of tops and branches are harvested at landscape level (at final felling).

A review by Egnell [66] found that harvesting tops and branches under Nordic conditions may have a moderate negative impact on growth in Norway spruce stands. This risk was increased when residues were harvested at thinnings. To avoid nutrient removal and potential negative effects on site productivity and acidification, in the present study, harvest was limited to the final felling; all needles were fallen to the ground before the material was chipped and ash was assumed to be recycled to the forest stand. Ash recycling could also be complemented by N fertiliser [65]. No long-term effects on carbon stocks were considered in this study. A review by Lippke et al. [67] concluded that carbon accumulation in forest soil depends on moisture content in the soil, carbon-nitrogen dynamics and climate, and not necessarily on the amount of dead biomass in the form of forest residues left at the forest site. However, it is important to increase knowledge of these long-term effects on soil carbon stocks and the effects on future forest productivity.

In conclusion, the present analysis revealed that:

  • Replacing coal with non-torrefied or torrefied wood pellets made from logging residues could mitigate climate change.

  • Torrefied wood pellets are better from a climate perspective (per GJ electricity), due to an assumed higher electrical efficiency and a higher co-firing rate compared with non-torrefied wood pellets.

  • Biogenic CO2 emissions are the greatest contributor to the global mean surface temperature change in non-torrefied and torrefied wood pellet systems.

  • For both short and long transport distances, total ∆T S is lower for non-torrefied wood pellets than for torrefied wood pellets (per GJ pellet fuel). This is explained by the dominant role of biogenic CO2 emissions for the outcome.

  • The energy output of these systems is about sevenfold the primary energy input.