Analysis of the Global Warming Potential of Biogenic CO₂ Emission in Life Cycle Assessments

Abstract

Biomass is generally believed to be carbon neutral. However, recent studies have challenged the carbon neutrality hypothesis by introducing metric indicators to assess the global warming potential of biogenic CO₂ (GWP_bio). In this study we calculated the GWP_bio factors using a forest growth model and radiative forcing effects with a time horizon of 100 years and applied the factors to five life cycle assessment (LCA) case studies of bioproducts. The forest carbon change was also accounted for in the LCA studies. GWP_bio factors ranged from 0.13–0.32, indicating that biomass could be an attractive energy resource when compared with fossil fuels. As expected, short rotation and fast-growing biomass plantations produced low GWP_bio. Long-lived wood products also allowed more regrowth of biomass to be accounted as absorption of the CO₂ emission from biomass combustion. The LCA case studies showed that the total life cycle GHG emissions were closely related to GWP_bio and energy conversion efficiency. By considering the GWP_bio factors and the forest carbon change, the production of ethanol and bio-power appeared to have higher GHG emissions than petroleum-derived diesel at the highest GWP_bio.

Introduction

Biomass is generally considered as a carbon neutral energy resource because the emissions from biomass combustion will be absorbed by plants through photosynthesis^1,2. The first comprehensive guideline for estimating greenhouse gas (GHG) emissions and sinks alleged that “CO₂ emission resulting from bioenergy consumption should not be included in a country’s official emission inventory”³. According to the guidelines compiled by the Intergovernmental Panel on Climate Change (IPCC), CO₂ emission from bioenergy sources should not be counted in national greenhouse gas inventories because the emission from bioenergy sources is already fully included in the Agriculture, Forestry and Other Land-Use (AFOLU) sector⁴. Therefore, bioenergy is always referred as a carbon neutral source of energy and promoted by government policies as a substitute for fossil fuels. For the same reason, no carbon tax is applied for burning biomass in any country around the world.

Due to the assumption of zero climate change potential of biomass combustion, the methods for assessing its global warming impact in environmental analysis tools (e.g. SimaPro, GaBi) usually do not include the biogenic CO₂ emission and even treat biogenic CO₂ emission as a negative impact^5,6,7. In addition, most of the life cycle assessment (LCA) studies conducted on bioenergy systems also claim that CO₂ emission from biomass has no global warming potential (GWP). Cherubini and Strømman reviewed 94 LCA studies of bioenergy systems and found that only one single case study included an account of the climate change impact of biogenic CO₂ emission⁸. Shonnard et al. reviewed 74 LCA case studies in the Pan American region and found that most of the articles assumed carbon neutrality of biomass combustion⁹. This carbon neutrality assumption, which could reduce the GWP of a bioenergy system, might be suitable if the rotation length of biomass were as short as in the case of perennial grass. However, the assumption may not hold when the rotation lengths are long, especially such as those of boreal forests which can be 100 years¹⁰. The exclusion of biogenic CO₂ emission in LCA also brings an unfair comparison of the GHG emissions from bioenergy and fossil fuel systems, although biogenic CO₂ emission is fully accounted for in the AFOLU sector.

More recently, researchers have become aware that CO₂ emission from biomass combustion may have, to some extent, a climate change impact because the CO₂ emitted by biomass combustion is a one-time pulse and remains in the atmosphere for several years. Johnson strongly asserted that biomass fuel might not always be carbon neutral and could in fact have a far greater impact than fossil fuels¹¹. Searchinger et al. also questioned the carbon neutrality assumption and suggested that shorter rotation lengths would be able to lower the GWP by absorbing the emissions after biomass combustion¹². Several recent studies have proposed a method to calculate the GWP of CO₂ emission derived from biomass, providing estimates of GWP_bio factors for different scenarios^{8,10,13,14,15}. The GWP_bio was calculated by the relative radiative forcing effect during its stay in the atmosphere. With a 100-year time horizon, their estimates of GWP_bio fell in the range of 0.34–0.62 when the rotation length was 100 years. Cherubini et al. estimated GWP_bio for rotation lengths from 1to 100 years and revealed a negligible GWP_bio for short rotation lengths¹⁰. If the forest stands were harvested for long-lived wood products, the GWP_bio could be as low as -1if the wood products were collected for bioenergy after decommissioning¹³.

The application of GWP_bio factors is straight forward. These factors can be considerably lower than 1, implying an advantage of bioenergy compared to fossil fuels in a climate change perspective. However, Holtsmark articulated the weakness of the previous methods which neglected the effect of harvesting on the dynamics of the major carbon pools in forest stands, such as logging residues and soil carbon¹⁶. He defined a no residue harvest baseline scenario and determined the GWP_bio factors of other scenarios by comparing them to this baseline. The derived GWP_bio in a 100-year time horizon was 1.54 for no residue harvested and 1.25 for 25% of residues harvested. In another study, Holtsmark included the albedo effect by collecting residues, which slightly reduced GWP_bio¹⁷.

These earlier studies provided fundamental results to bring a fair comparison of CO₂ emissions from bioenergy and fossil fuel. However, two considerations need to be clarified for a more accurate accounting: 1) the residue, if not collected, will decompose slowly and emit CO₂ into the atmosphere over the following years, and 2) the harvest of biomass or timber will result in further emissions from dead organic matter. Thus, a method to estimate GWP_bio accurately should address these considerations. In our study, we analyzed the GWP_bio by applying an accounting model considering long-lived wood products and the decomposition of wood residues. The GWP_bio factor also applied to five LCA studies of bioproducts to estimate the potential of biomass utilization in reducing carbon emissions.

Materials and Methods

Forest Stand Modeling

A forest stand storing 80 tC/ha live biomass was harvested at the start (t = 0). The biomass accumulation after harvesting was estimated by a Chapman-Richards function: , where B(a) is the mass of biomass measured in tC/ha (tC: metric ton carbon equivalent) and a the stand age; the b_i are empirical parameters assumed to be known. The harvest activity resets all the living biomass to zero. This function provides a reasonable growth of the stand as a function of its age. Three sets of parameter configurations were used to allow the stand to grow back to its previous carbon level after harvesting in 30, 50 and 100 years (Table 1).

Table 1 Parameter settings for the Chapman-Richards function.

The average merchantable timber is assumed to be a proportion of live biomass. We are using the proportion θ = 0.48 proposed by Løken et al.¹⁹. The merchantable timber is thus calculated as: T(a) = θB(a), where T(a) is the merchantable timber and B(a) is live biomass. The unit of T(a) and B(a) is metric ton in carbon equivalent (tC). In this study, we assumed a proportion ω = 0.25 of residues is collected for bioenergy. The biomass not harvested is considered to be dead organic matter (DOM). The decomposition process is simulated by Yasso07 with average amounts of compartments in the residue (See Table S1 in the supporting information). Yasso07 is a widely-used model for simulation of biomass decomposition²⁰. The decomposition rate is shown in Fig. S1.

Biogenic CO₂ in the Atmosphere

Once fossil fuel derived CO₂ is emitted to the atmosphere, the decay rate is not following a simple trajectory. The fraction of the initial pulse of CO₂ at time t is labeled as y(t) and calculated as follows:

where y_i and τ_i are estimated parameters. This model is based on the Bern2.5CC carbon cycle model using a CO₂ concentration of 378 ppm in the atmosphere²¹. The parameters are average values of a set of climate models and set as y₀ = 0.217, y₁ = 0.224, y₂ = 0.282, y₃ = 0.276, τ₁ = 394.4, τ₂ = 36.54, τ₃ = 4.304²¹. Equation (1) takes into account the CO₂ uptake from the oceans and the terrestrial biosphere. Because y₀ > 0, there is always a portion of CO₂ remaining in the atmosphere at any time.

When residue is collected for bioenergy, the initial pulse of CO₂ is E(ω) = ω(B(a_h) − T(a_h)), where a_h is the stand age at harvest. After harvesting, stand age a equals t. Let E_h(t) as CO₂ remains in the atmosphere after the pulse of E(ω) at time t, and E_d(t) as CO₂ emission from decomposition at time t if the residue is not collected. Hence, we have the following equations:

where σ_b(t) and σ_d(t) are mass allocation coefficients for the residue if the residue is used for bioenergy or is not removed respectively. B′(t) is the increment of live biomass after harvesting. Because the stand is harvested for timber and these timber products usually do not emit CO₂ to the atmosphere immediately, we consider two scenarios. In the first case, the timber products have a pulse of CO₂ right after harvesting (η_l = 0%); in the other case the timber products do not emit any CO₂ before the end of the rotation (η_l = 100%). η_l is percentage of long-lived woody products in the total merchantable timber. The detailed calculation of the mass allocation coefficients are presented in the supporting information. E_h(t) − σ_b(t)B′(t) and E_d(t) − σ_d(t)B′(t) will be set to zero once these value are lower than zero because CO₂ absorbed by biomass regrowth can fully compensate for the CO₂ remaining in the atmosphere. If the residues are not collected, they decompose and emit CO₂ into the atmosphere. Thus, the emission from decomposition is subtracted and the CO₂remaining in the atmosphere at time t is as follows:

Global Warming Potential

The Global Warming Potential is used in this study as a standard estimation to compare the climate change impact of GHG emissions. The GWP is introduced as a relative measure by comparing the amount of heat that a greenhouse gas traps in the atmosphere to the amount of heat trapped by a similar mass of CO₂. The GWP relies heavily on radiative forcing where the radiative forcing of CO₂ () is 1.37 × 10⁻⁵ W m⁻² ppb⁻¹ by Myhre et al.²². The absolute global warming potential (AGWP) of a CO₂ pulse of E(ω) is calculated as follows:

where T is the time horizon for integration in the GWP. The IPCC introduced three time horizons for this purpose, i.e., 20, 100 and 500 years. In this study, we considered T = 100 as the time horizon which is generally used to estimate the overall GHG emissions in most LCA studies.

Similarly, the AGWP_bio is calculated based on the remaining fraction of the initial pulse of CO₂ by subtracting the decomposition of DOM. The AGWP_bio is:

and the GWP_bio is then defined as:

LCA Case Studies

In order to analyze the effect of this adjusted GWP_bio factor, five biomass to bioproducts pathways were tested: biomass to ethanol (BTE), biomass to liquid fuel via fast pyrolysis (BLFP), coal and biomass to liquids (CBTL), biomass to bio-power and biomass to pellet fuel. The cradle-to-grave assessments included residue collection, transportation, storage, preprocessing, bioproduct conversion, distribution, final usage and waste disposal. These studies focused on GHG emissions derived from fossil fuels and biomass combustion. The functional unit (FU) of the system was 1,000 MJ of energy equivalent bioproduct produced. The GHG emissions of petroleum-derived diesel with 98.8 kg CO₂ eq/FU were used as a baseline to determine the advantage of biomass on climate change impact²³. The impact of GHGs was calculated using 100-year global warming potentials¹⁹. All the GHG emissions were calculated in CO₂ equivalent (kgCO₂ eq). The system boundaries are defined in the Supplemental Information and the detailed life cycle inventory data are also available in Supplemental Information²⁴. Traditional LCA quantifies the emissions related to the production of bioproduct and the harvest of biomass. Several researchers suggest that emissions related to land use change and forest carbon change should be included^25,26. Although the forest stand in this study is allowed to regrow and the land use does not change, forest carbon change induced by biomass collection was accounted for in the LCA studies. See supporting information for details regarding the method of calculating forest carbon change.

Results

GWP_bio factors

As explained in the methods section, we examined two scenarios: 1) η_l = 0% where the timber products emit CO₂ at the same time as biomass combustion, and 2) η_l = 1000% indicated all timber products do not emit CO₂ before the end of the rotation. Compared to fossil fuel derived CO₂ which decays with the interaction of ocean-atmosphere systems, these scenarios allow an increasing decay rate of biogenic CO₂ in the atmosphere if the harvested stand is re-established and allowed to regrow (Fig. 1).

Figure 1: Carbon decay patterns for different scenarios.

(a) Rotation length 30-year, η_l = 0%; (b) Rotation length 30-year, η_l = 100%; (c) Rotation length 50-year, η_l = 0%; (d) Rotation length 50-year, η_l = 100%; (e) Rotation length 100-year, η_l = 0%; (f) Rotation length 30-year, η_l = 100%.

In Fig. 1, E_h(t), E_d(t) and y(t)E(w) are the amount of CO₂remaining in the atmosphere from biomass combustion, decomposition and fossil fuel, respectively. The curve representing E_h(t)-E_d(t) is not a decay curve but the curve that we used to calculate GWP_bio. Higher values of σ (representing a longer lifespan of timber products) allow higher decay rate of biogenic CO₂ (Fig. 1b,d,f). Higher biomass growth rate could also effectively increase the decay rate of biomass derived CO₂ (Fig. 1). It took 21, 32 and 49 years, assuming η_l = 0%, to eliminate all the biomass derived CO₂ when the rotation length was 30, 50 and 100 years. If η_l = 100%, the corresponding periods were 14, 20 and 30 years.

The GWP_bio factors in this study were significantly lower than 1 and related to the rotation length and the mass allocation coefficient (Table 2). All our assumptions led to lower GWP_bio factors when compared to earlier studies, especially when wood products have a long lifespan (Table 2). Sensitivity analyses were conducted on rotation length and η₁. The GWP_bio varied between 0.13 and 0.32 as the rotation length changed from 30 years to 100 years (Fig. S3). The GWP_bio was linearly related to η₁ as it changed from 0% to 100%.

Table 2 GWP_bio factors for different scenarios with a 100-year time horizon.

The model allowed the modification that the CO₂ absorbed in previous years was excluded from the atmosphere over the following years. The model also subtracted the emission from the decomposition of residues if they were not harvested. Decomposition was considered to be slow but could have a positive global warming impact in a young stand where the biomass accumulation is slow (Table S3). These modifications allowed a logical explanation of the benefits from regrowth of a forest stand.

GHG emissions of Case Studies

Without considering biogenic CO₂ emission, all five bioproducts had lower GHG emissions in comparison with petroleum-derived diesel (Fig. 2a). The production of pellet fuel had the lowest GHG emissions, at only 4.8% of petroleum-derived diesel. The production of ethanol and bio-power also produced low GHG emissions. Liquid fuels produced by fast pyrolysis had higher GHG emissions compared to the production of ethanol, bio-power and pellet fuel. The highest emissions were estimated from CBTL, amounting to 90% of the emissions from petroleum-derived diesel.

Figure 2: GHG emissions of biomass to bioproduct pathways under different scenarios.

(a) no biogenic CO₂ emission; (b) RL = 30; (c) RL = 50 and (d) RL = 100. Note: Fossil – GHG emissions from fossil fuel; Biogenic – accountable biogenic GHG emissions from biomass; FC – accountable forest carbon change; RL – Rotation length.

When considering the biogenic CO₂ emission in LCA, we found that most of the biogenic CO₂ emission was produced in “conversion” and “final usage” (Table 3). All the energy usage in “feedstock collection”, “transportation” and bioproduct “distribution” was assumed to be provided from fossil energy. In the production of bio-power in particular, 99% of the total biogenic CO₂ was accounted for by emission in the “conversion” process. In the production of bio-power, all biomass was combusted at the facility site and no further emissions were assumed in the end usage.

Table 3 Biogenic CO₂ emission from all processes.

The GHG emissions from ethanol and bio-power were very low compared to other bioproducts if the GWP_bio factor was zero, but both appeared to have high GHG emissions when the rotation length was 100 years and η_l = 0%. When forest carbon change was accounted for in the LCA study, both had higher GHG emissions than petroleum-derived diesel. The GHG emissions of CBTL did not exceed those of petroleum-derived diesel in all cases due to the low percentage of biomass required to produce liquid fuels.

Generally, shorter rotation lengths and higher mass allocations benefit the environment more in terms of GHG emissions (Fig. 2b,c,d). To produce the same bioproduct, the GHG emissions were the lowest when the rotation length was 30 years. It was 1.4–41.4% and 2.5–39.2% less than the GHG emissions when the rotation length was 50 years and 100 years, respectively. To produce the same bioproduct, the GHG emissions when η_l = 0% were 1–25% higher than the GHG emissions in the η_l = 100% scenario in all cases.

Discussion

GWP_bio factors

Our model assumed that the primary objective of harvesting is for timber products which usually have a longer lifespan than biomass. It was also assumed that the dynamics of soil carbon would not be affected by a small percentage (≤25%) of residue removal. The portion of biomass regrowth accounted for CO₂ absorption depended on the percentage of CO₂ remaining in the atmosphere from biomass combustion. We examined two mass allocation scenarios. If the wood products emitted CO₂ before the end of the rotation, the mass allocation coefficient would fall between these two scenarios. A long lifespan of timber products allows a high mass allocation coefficient, and short rotations imply rapid regrowth of a forest stand. Thus, the decay rate of biogenic CO₂ increases with decreasing rotations and increasing lifespan of timber products. The forest carbon change due to harvesting was attributed to timber and biomass collection and allocated based on their mass.

The GWP_bio factors in all scenarios are significantly lower than 1. This indicates the advantage of biomass when compared to fossil fuels. An even lower GWP_bio factor would be possible if growth would be more vigorous at a younger age of the forest stand. The GWP_bio factors correlates with rotation length, but this relationship should not be over exploited. It merely implies that higher growth rates may accelerate the decay of CO₂ in the atmosphere. In reality, the rotation length of a forest stand is relatively long. For fast growing tree species such as eucalypts, the rotation length is at least 15–18 years^27,28. The pulse of CO₂ from biomass combustion can be neglected in the case of perennial grasses which rotation length is 1 or 2 years^10,29.

The high GWP_bio factor in Holtsmark’s results could be reasonable if, as assumed in his study, all the trunks and harvested residues were used for biomass combustion. This suggests that all the changes of major carbon pool in the stand should be accounted for by biomass combustion. Nonetheless, in most cases stands are harvested for timber. In Cherubini et al.’s study, the forest model was modified to allow more vigorous growth in younger stands leading to slightly lower GWP_bio factors^10,17. Recent estimates of GWP_bio factors may be too optimistic because no negative effects on the growth of biomass were included, like natural disturbances and climate change. Natural disturbance, like wild fire, drought, insect and pathogen outbreaks, will reduce biomass growth and even induce more emission to the atmosphere³⁰. These negative effects will increase the possibility of high GWP_bio factors, but relevant scenarios may be more realistic for research and policy decisions. The impact of climate change on the growth of biomass is complex. Both negative effects^31,32 and positive effects³³ were found in earlier investigations. Moreover, harvest level may also be increased for biomass because of high energy demands in the future, which in turn will lead to a higher GWP_bio factor.

The GWP_bio factors were not calculated for longer time horizon (>100 years), but previous studies have shown that these values may be even lower^10,16. Thus, biomass will be a good substitute of fossil fuels in the long run^10,17,23. Growth of biomass can even bring negative GWP if longer time horizons are considered, because biomass will continue to grow and absorb CO₂. But in this study, the CO₂ remaining in the atmosphere was set to zero once the increment of biomass growth fully compensates the CO₂ emission. Part of the reason for that strategy is the assumption that the forest stand absorbs CO₂ if it is not harvested, which implies that the extra absorption of CO₂ is not due to residue collection. Additionally, the extra absorption of CO₂ is external to the studied system (LCA boundary). Considering these CO₂ reductions may involve some kind of double accounting.

GHG emissions of Case Studies

Without considering biogenic CO₂ emission, the life cycle GHG emissions of the five cases closely resembled previous studies^{34,35,36,37,38,39}. The production of pellet fuel had the lowest GHG emissions because of the high energy conversion efficiency and low energy consumption. The reason that the production of ethanol and bio-power had low GHG emissions was the low energy demand in the production of ethanol and the self-sufficiency of power plants on energy. Liquid fuels produced by fast pyrolysis required high electricity input, which led to higher GHG emissions compared to the production of ethanol, bio-power and pellet fuel. CBTL emitted the highest GHG because the large proportion of GHG emissions from the combustion of coal could not be ignored even in conventional LCA studies.

Although the CO₂ emission from burning biomass is compensated by the growth of biomass, the CO₂ emission will remain in the atmosphere for a certain time. Thus, it is necessary to consider the GWP of biogenic CO₂ when conducting LCA. Otherwise, the total GHG emissions will be systematically underestimated. By calculating biogenic CO₂ in the LCA studies, we found CBTL had a small amount of biogenic CO₂ emission compared to the other four bioproducts. This is because a small portion of biomass (8%) was needed to produce 1,000 MJ of energy equivalent liquid fuels. Given that biomass was the only feedstock for the other four bioproducts, their biogenic CO₂ emission were closely related to their energy conversion efficiencies which were 25.6% for BTE³⁵, 73.6% for BLFP³⁶ and 24% for bio-power⁴⁰. Usually no waste was assumed in the production of wood pellets⁴¹. By increasing the energy conversion efficiency, biogenic CO₂ emission could be effectively reduced.

When the biogenic CO₂ emission was multiplied by a nonzero GWP_bio factor, higher GHG emissions were expected in the LCA case studies. This effect was more pronounced when the biogenic CO₂ emission was high in the production of a bioproduct, such as ethanol and bio-power. The consideration of forest carbon change added more emissions to the total GHG emissions²⁵. When a cradle to grave LCA was modeled, the LCI should quantify all the emissions associated with the activities and their effects to produce the product. By adding biogenic CO₂ emission and forest carbon change, an unbiased comparison could be made between biofuel and fossil fuel, although a significant increase of GHG emissions from biofuel should be expected. However, in this study, the forest carbon change was analyzed at forest stand scale because the timber harvest and residue collection are all occurred in stand scale. In addition, the focus on stand scale fits the objective of this study as well. Here, the study is an attributional LCA model which will not consider effects from outside of the system boundary. Once the study is expanded to landscape level, the forest carbon change could be different.

Policy Implications

Renewable fuel standard (RSF2) indicates that forest residue makes up less than 1% of the total feedstock under current scenario⁴². Most of the biofuels are produced from corn, soybean, agricultural residue, and grasses. These feedstocks have short rotation period and their GWP of biogenic CO₂ can be ignored¹⁰. However, as the renewable fuel volume increases year by year, more forest residue need to be collected for biofuels. To be qualified as renewable energy for cellulosic fuels, the life cycle GHG emissions should be 60% less than the life cycle GHG emissions of the 2005 baseline average gasoline or diesel fuel (92 kg/1,000 MJ)⁴². Thus, the direct influence of GWP_bio is that some cellulosic fuel or biopower turn to be non-renewable energy, such as BTE, BLFP and biopower. The other influence is that more short rotation biomass will be used in the future scenarios. Therefore, the GWP_bio provides a solid scientific reference to policy makers when making decisions for the utilization of forest residues. However, it is not a simple work to decide the best pathway for forest residue and bioproducts. Many factor should be considered, such as residue availability, cost, market demand, convenience for distribution and usage, GHG emissions and existing facilities. If only GHG emissions are emphasized, pellet fuel is the best choice because it has low emissions in all scenarios.