Not carbon neutral: Assessing the net emissions impact of residues burned for bioenergy

Meeting the Paris Agreement goal of limiting global temperature increase will require fast deployment of zero-emissions energy and greatly increased carbon sequestration. In developing pathways to limit atmospheric CO₂, many climate mitigation models include a doubling or more of bioenergy to at least 100 EJ in the coming decades [1–3], with much of the fuel assumed to come from forestry and agricultural residues [3]. Though oxidizing 100 EJ of biomass would emit about 9 Gt of CO₂ each year, most mitigation models assign bioenergy zero net emissions.

The assumption of bioenergy carbon neutrality underpins many renewable energy investments, including in the EU, UK and Asia where dried wood pellets are imported as a replacement for coal. Such policies, and the lucrative subsidies they provide, have driven rapid growth in the wood pellet sector in North America, with US exports growing from less than 0.1 Mt in 2008 [4] to 4.9 Mt in 2016 [5]. Canadian pellet exports increased 46% from 2015–2016 [6], and US pellet exports are projected to double or triple from 2016 levels by 2025 [5, 7].

Biomass power plants tend to emit more CO₂ than fossil fueled plants per MWh, and as shown by a number of studies, net emissions from bioenergy can exceed emissions from fossil fuels for decades [8–12]. Nevertheless, some studies conclude rapid carbon benefits from burning wood pellets by employing various assumptions: that forest planting will increase in response to demand for wood [13]; that replanting occurs immediately after harvest [14]; or that forest growth elsewhere compensates for emissions from harvesting and combusting trees [15, 16] (for a review, see Ter-Mikaelian et al 2015 [17]). Some discussions of bioenergy in mitigation modeling include similar assumptions that burning 'sustainable,' 'optimal' [1, 3] or 'surplus' [18] forest wood can reduce net CO₂ emissions as long as forest carbon stocks are increasing. Such assumptions often disregard the role of the forest carbon sink, thus the controversy around bioenergy carbon accounting continues.

However, on one aspect of bioenergy carbon accounting there is wide agreement: that when biomass is sourced from residues from forestry, wood products manufacturing, or agriculture, net carbon emissions are properly assessed as the difference between emissions from their use as fuel (which can include emissions from fuel manufacturing and transport), and emissions from an alternative fate, such as leaving material on-site to decompose or burning it without energy recovery [8–10, 12 19–23].

Studies using this approach generally conclude net bioenergy emissions are not zero over varying periods of time. Nonetheless, many policies still treat bioenergy as having zero or negligible emissions. European Commission guidance for the EU carbon trading program explains bioenergy emissions should be 'taken to be zero,' and that wood pellets consist of 'processing residues from forest based industries' [24]. The IPCC acknowledges harvesting trees for fuel can increase cumulative emissions for years to centuries, but concludes that 'agricultural and forestry residues can provide low-carbon and low-cost feedstock for bioenergy' [3]. The IPCC renewable energy report identifies potential for 100 EJ of bioenergy specifically from residues [18] and does not discuss potential emissions.

For the assumption that residues have negligible net emissions to be reasonable, at least two conditions must be met. First, biomass classified as residues must actually be residues—that is, materials generated by some other process, where the alternative fate is decomposition or burning without energy recovery. Second, net emissions from bioenergy, that is, the cumulative additional CO₂ emitted from processing and burning biomass versus from an alternative fate, must be low, if not negligible, within a timeframe meaningful for climate mitigation.

What should 'low net emissions in a meaningful timeframe' mean? Most scenarios for climate change mitigation that constrain temperature rise consistent with Paris Agreement goals require emissions to peak between 2020 and 2030 and decline to less than half 2010 levels by 2050 [25], with negative emissions shortly thereafter. Actions that reduce or end emissions in the next ten years are thus essential, given that elevated CO₂ is already driving essentially irreversible polar ice loss, permafrost melting, and ocean acidification, along with thermal sea-level rise, which has been shown to respond to temperature changes from short-lived climate pollutants in a ten-year timeframe [26].

Here, 'low net emissions' from bioenergy implies a comparison to gross or 'direct' emissions from manufacturing and burning biomass. This study uses a simple model to calculate a new metric, the 'net emissions impact' (NEI), which is the ratio of cumulative net emissions to direct emissions from burning residues for energy. The NEI expresses the proportion of direct CO₂ emissions that contributes an additional warming effect over a fifty-year period. Fuel and feedstock use, net emissions and the NEI are calculated for three main case studies: the existing US bioenergy sector, new wood-burning plants using chipped wood, and wood pellets that are exported to the EU to be burned as a replacement for coal.

Built in Excel, the model calculates cumulative net emissions as cumulative direct emissions (CO₂ from combustion for energy plus CO₂ from harvesting, producing, and transporting biomass, or 'HPT emissions'), minus cumulative counterfactual emissions (what emissions would be if the biomass were left in the field to decompose or were burned without energy recovery). The net emissions impact (NEI) is the ratio of cumulative net emissions to cumulative direct emissions.

HPT emissions are calculated as explained below. Direct combustion emissions are calculated as joules of heat input for each fuel multiplied by fuel-specific CO₂ emission factors [27] (non-CO₂ greenhouse gas emissions are not included in this version of the model). The spreadsheet sums cumulative counterfactual emissions from biomass collected in each year in columns, then sums across columns to calculate cumulative emissions by each year from all biomass collections up to and including that year.

Counterfactual carbon emissions (with conversion to CO₂ at the last step) are calculated as:

$$\begin{equation} {\rm PE}^{\prime}(t) = 1 - (e^{-k^{\prime} t}) \end{equation} \tag{ 1 }$$

$$\begin{equation} {\rm cE}^{{\prime}} (t) = {\rm BC}^{{\prime}} * {\rm PE}^{{\prime}} (t) \end{equation} \tag{ 2 }$$

$$\begin{equation} {\rm CE}(t) = \sum_1^t {\rm cE}^{\prime} (t) \end{equation} \tag{ 3 }$$

where

PE^'(t) = proportion of carbon from biomass collected in a given year that has been emitted by year t

k^' = rate-constant for decomposition of biomass collected in a given year

cE^'(t) = carbon from biomass collected in a given year that has been emitted by year t

BC^' = carbon content of biomass collected in a given year

CE(t) = carbon emitted by year t from biomass collected in all years

To evaluate emissions from the US bioenergy industry, the model uses bioenergy data from the Energy Information Administration (EIA) for 2001–2016 [28] and CO₂ combustion emission factors used by the US Environmental Protection Agency (EPA) for power sector modeling (original units short tons mmbtu⁻¹; converted here to metric tonnes GJ⁻¹) [27]. Alternative fate emissions are calculated using k-constants particular to each fuel (table 1).

The model includes HPT emissions for forestry residues and other wood as equivalent to 4% of the carbon content of green chips, based on Domke et al (2012) and reviews of other studies [9, 29]. The model assumes the alternative fate for agricultural residues is decomposition, as crop burning occurs on less than 1% of agricultural acres in the US [30]. Selecting an HPT factor for agricultural residues is not straightforward, as emissions from harvest, transport, shredding, baling, and sometimes pelletizing can be significant. Depending on how system boundaries are drawn, emissions from crop cultivation, including N₂O from fertilizers, can be ascribed to residues [31]. Storage also imposes lifecycle emissions because agricultural materials can only be collected at fixed intervals. Most importantly, removing agricultural residues can deplete soil carbon [32]; some estimates of total HPT emissions including soil C loss sum to more than 100% of fuel carbon content [31]. This model used an HPT factor of 7.5% for agricultural residues based solely on harvest and transport estimates for corn stover in Whitman et al (2011) and did not include soil carbon impacts because this factor was not included for forestry residues. As agricultural residues provided a small percentage of total fuels, the choice of HPT factor had only a trace effect, but any study of large-scale use of agricultural residues should include soil carbon effects.

Data on wood use by the US pellet manufacturing sector was obtained from the forest-industry tracking company Forisk [33]. Five pellet scenarios were modeled to examine how the k-constant and changing use through time affect net emissions (details in table 2). Scenarios 1–4 estimated HPT emissions (which include harvesting, transport to plant, debarking, chipping, pulverization, pellet extrusion, drying, and oversea transport) as 322 kg CO₂ per tonne of pellets, following Jonker et al [15], similar to an estimate by Dwivedi et al [14]. Also following Jonker et al the model assumed that pellet drying consumes 0.51 green tonnes of residues per tonne of pellets, an estimate confirmed by checking the dryer fuel to pellet production ratio in permits for two industrial-scale plants in the US [34, 35]. The model assumed residues burned to dry pellets would decompose with a k-constant of 0.15 if not burned for energy. Facilities that use fossil fuels to dry pellets have much higher HPT emissions [36], but this effect was not included.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Sources of biomass burned for heat and power in the US

Sources of wood utilized by the US pellet manufacturing industry

As of late 2016, annual production capacity at operating and under-construction wood pellet manufacturing facilities in the US was 13.2 Mt, requiring about 28.6 Mt of green wood as feedstock [33], though not all plants produced at capacity. Pellet companies emphasize use of residues, downplaying the use of roundwood as feedstock [40]. However, exported wood pellets must meet specifications including restrictions on bark content [41], thus there is a limit on the amount of low-diameter branches and tops that can be used. Accordingly, industry data indicate about 56% of pellet feedstock is supplied from pulpwood (41% from softwood, 14% from hardwood); 42% from mill residues, 1% from urban wood; and just 1% from logging residues [33]. Investigations of pellet feedstock at some large US mills have confirmed that a significant portion of feedstock is bolewood (figure 2). The feedstock supply from small pellet producers in the Northeastern US also appears to avoid residues; a study of nine pellet mills in Maine found only 2% came from tops and limbs, with the remainder classified as pulpwood or small diameter trees [42].

Similarly, company-supplied data on sources of wood pellets burned in the UK indicate that the majority of US pellets burned by the coal- and wood-fired Drax power station in 2015 was sourced from logs that 'formed part of the trunk of a tree which grew for at least ten years' from harvesting that was a 'mix of clearfell and thinning' [43]. While mill residues currently constitute a proportion of pellet feedstock in the US, they will not supply a meaningful amount for future capacity, because supplies are limited [44]. The dominance of pulpwood and the documented use of bolewood thus indicate that pellets often fail to meet the first condition for low carbon residues, that feedstocks genuinely be residues that if not collected would otherwise decompose or be burned without energy recovery.

Zoom In Zoom Out Reset image size

Emissions from the US biomass industry

Emissions modeling examined the industrial and non-industrial bioenergy sectors separately. The model estimated cumulative direct CO₂ emissions from industrial facilities at 1135 Mt at year 10. However, because black liquor and other wastes provide a large proportion of industrial sector biomass, and the assumed alternate fate for these materials is combustion without energy recovery, cumulative net emissions are 224 Mt, for an NEI of 20% (figure 3(a)). Cumulative direct emissions for the smaller non-industrial sector are 208 Mt at year 10, but because the majority of fuel for this sector is wood and the weighted k-constant is lower, cumulative net emissions are 120 Mt and the NEI is 58% at year 10 (figure 3(b)). Both sectors show cumulative net emissions still increasing in the 40–50 year period, though less steeply than in the initial decades.

This analysis calculates emissions at the sector level as if all units initiated operation at the same time, while ideally, sector-level accounting would consider how long each facility has been operating. While inadequate data render this impractical, in general, the industrial sector has shrunk since the 1980s [45] and the present day NEI should be shifted toward the right (lower), as facilities are on average older. In contrast, new construction of wood burning power plants and coal-to-wood conversions in the non-industrial sector since the early 2000s (figure 1) [28] means the average age of the sector is younger, shifting the NEI to the left (higher).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Refined emissions estimates for new wood-burning power plants

Emissions from wood pellets manufactured from residues

For bioenergy to offer genuine climate mitigation, it is essential to move beyond the assumption of instantaneous carbon neutrality. The NEI approach provides a simple means to estimate net bioenergy emissions over time, albeit one that tends to underestimate actual impacts. The model finds that for plants burning locally sourced wood residues, from 41% (extremely rapid decomposition) to 95% (very slow decomposition) of cumulative direct emissions should be counted as contributing to atmospheric carbon loading by year 10. Even by year 50 and beyond, the model shows that net emissions are a significant proportion of direct emissions for many fuels. Similarly, the model concludes that for wood pellets manufactured from residues in in the US and shipped overseas, even a rapid decomposition counterfactual produces an NEI of 55% at year 10, while a slow decomposition counterfactual produces an NEI of 79%. By year 40, net emissions still represent 25% to more than 50% of direct emissions. Scenarios that increase the amount of biomass burned each year, as is currently occurring in the EU, have even larger net emissions impacts.

Models like this have their critics. The IPCC warns that using a 'simple sum of the net CO₂ fluxes over time' to highlight the 'skewed time distribution between sources and sinks,' is probably insufficient to understand the climate implications of bioenergy, which instead requires models that include temperature effects and climate consequences [3]. Bioenergy advocates have seized on models that emphasize the importance of cumulative emissions for warming, pointing out that bioenergy can reduce carbon impacts over time compared to fossil fuels [23], though they say little about carbon impacts when bioenergy displaces zero-emissions technologies.

However, since the IPCC's call for more complex bioenergy modeling was published in 2014, the intensity of the climate crisis has deepened; in the US, legislation has been enacted that compels the EPA to treat bioenergy as carbon neutral [64]; and combustion of forest wood by the EU, UK and Asia is increasing each year, unmitigated by the carbon capture and storage that some climate models say is required [65]. Also while the climate modeling community ponders, governments are making practical decisions about renewable energy funding, as in the UK, where the government provided £809 m (about $1.2 b) [66 67] in subsidies to biomass electricity in 2015, the same year it announced it was terminating subsidies for offshore wind earlier than planned [68]. Since residues would eventually release carbon to the atmosphere whether through burning or decomposition, any putative reduction in CO₂ emissions actually depends on residues-fueled bioenergy displacing fossil fuels, but in the UK, it appears bioenergy may instead be displacing zero emissions technologies, while prolonging the life of coal plants that partially switch to subsidized wood burning.

There is no time like the present to reduce emissions. Given the anticipated role for bioenergy in climate mitigation, climate-related policies should be reformed immediately to account for bioenergy impacts. Using the NEI to weight biogenic CO₂ for inclusion in US and EU carbon trading programs and to qualify bioenergy for renewable energy subsidies would reduce emissions more effectively than continuing with the current assumption of zero emissions, though for wood pellets sourced from bolewood, counting direct emissions is a more protective and accurate approach.

This work was supported by funding from the Packard Foundation and the Tortuga Foundation.