• Open Access

Energy and greenhouse gas balance of bioenergy production from poplar and willow: a review



    1. Department of Biology, Research Group of Plant and Vegetation Ecology, University of Antwerp, Universiteitsplein 1, B-2610 Wilrijk, Belgium
    Search for more papers by this author

    1. Department of Biology, Research Group of Plant and Vegetation Ecology, University of Antwerp, Universiteitsplein 1, B-2610 Wilrijk, Belgium
    Search for more papers by this author

    1. Department of Biology, Research Group of Plant and Vegetation Ecology, University of Antwerp, Universiteitsplein 1, B-2610 Wilrijk, Belgium
    Search for more papers by this author

Sylvestre Njakou Djomo, tel. +32 3 2652827, fax +32 3 2652271, e-mail: sylvestre.njakoudjomo@ua.ac.be


Short-rotation woody crops (SRWC) such as poplar and willow are an important source of renewable energy. They can be converted into electricity and/or heat using conventional or modern biomass technologies. In recent years many studies have examined the energy and greenhouse gas (GHG) balance of bioenergy production from poplar and willow using various approaches. The outcomes of these studies have, however, generated controversy among scientists, policy makers, and the society. This paper reviews 26 studies on energy and GHG balance of bioenergy production from poplar and willow published between 1990 and 2009. The data published in the reviewed literature gave energy ratios (ER) between 13 and 79 for the cradle-to-farm gate and between 3 and 16 for cradle-to-plant assessments, whereas the intensity of GHG emissions ranged from 0.6 to 10.6 g CO2 Eq MJbiomass−1 and 39 to 132 g CO2 Eq kWh−1. These values vary substantially among the reviewed studies depending on the system boundaries and methodological assumptions. The lack of transparency hampers meaningful comparisons among studies. Although specific numerical results differ, our review revealed a general consensus on two points: SRWC yielded 14.1–85.9 times more energy than coal (ERcoal∼0.9) per unit of fossil energy input, and GHG emissions were 9–161 times lower than those of coal (GHGcoal∼96.8). To help to reduce the substantial variability in results, this review suggests a standardization of the assumptions about methodological issues. Likewise, the development of a widely accepted framework toward a reliable analysis of energy in bioenergy production systems is most needed.


The progressive depletion of fossil energy sources and the growing concerns about global climate change and air quality have increased the interest in renewable energy sources that are potentially carbon dioxide (CO2)-neutral and less polluting (Rubin et al., 1992). The use of renewable energy is a way to reduce reliance on fossil fuels, to mitigate greenhouse gas (GHG) emissions, to increase energy resource diversification, and to avoid depletion risks (De Vries et al., 2006). Among renewable energies, bioenergy is considered to be relatively inexpensive and a highly promising strategy as a substitute for fossil fuels (IPCC, 2007). Biomass has received a renewed interest during the last 20 years and is attracting growing attention around the world as an abundant and available energy source (Hall & Scrase, 1998; Righelato & Spracklen, 2007). The diversity of organic materials used as renewable bioenergy sources has expanded and includes agricultural and forestry residues, municipal solid and liquid wastes, agro-industrial by-products, and cultivated biomass sources. Among the cultivated biomass sources, dedicated crops and especially short-rotation woody crops (SRWC) are the most promising (Rowe et al., 2009). SRWC such as poplar and willow are fast-growing and high-yielding woody species which can be managed in a coppice system. This biomass can be burnt or gasified to generate electricity and/or heat in combustion or gasification plants (Hughes et al., 2007). One of the advantages of SRWC is that they can be grown on abandoned and/or contaminated land. Thus, production does not necessarily have to compete with food crops for the most fertile soils and their management is usually less energy intensive than the one needed on food crops (Tillman et al., 2006; Schmer et al., 2008). However, to be ecologically and energetically viable, the energy gain from SRWC must outweigh the energy used for the production, transport, and conversion to bio-electricity as well as significantly reduce some impacts on the environment (e.g. GHG emissions).

A considerable number of studies has examined and compared bioenergy production systems from an energetic and environmental point of view using diverse approaches. For example, Turhollow & Perlack (1991) reported on CO2 emissions from bioenergy crops using an energy analysis (EA) approach. Mann & Spath (1997) published a comprehensive life cycle assessment (LCA) study of a biomass gasification combined-cycle power system. Styles & Jones (2008) used a combined LCA and economic approach to assess the environmental and economic impacts of bioenergy chains. These and other studies have advanced the understanding of the potential environmental impacts and of the energy balance of bioenergy systems. However, their sometimes significantly different outcomes and conclusions have generated controversial views among scientists, policy makers, and the public forum (Whitaker et al., 2010).

This paper reviews and synthesizes published studies on environmental impacts and the energy balance of SRWC (for the production of heat and/or electricity) where LCA, EA, or a combination of LCA and economic approaches was applied. The objectives were (i) to summarize the available information in the scientific literature about the energy and GHG balance of bioenergy production from SRWC; (ii) to identify and investigate the mechanisms that frequently lead to conflicting results while attempting to draw coherent conclusions from the published studies, and (iii) to highlight the shortcomings in the analysis of environmental impacts.

Construction of literature source database

The ISI Web of Knowledge, Web of Science, and Science Direct databases were queried for original studies published in the literature between 1990 and 2009 that reported on the environmental impacts, energy balance, and/or sustainability assessment of SRWC for the production of electricity and/or heat. The search was further extended to include grey literature such as one academic thesis, one report found by searching the archives of Wageningen University in the Netherlands, and the database of the US National Renewable Energy Laboratory. The titles and abstracts of all papers were first screened to determine their suitability; then, certain inclusion/exclusion criteria were applied to the complete articles. The bibliographies of the selected articles or reports were also examined for additional references. We attempted to contact key authors of papers that did not include the essential information needed for this review. Only published studies that reported on environmental impacts (mainly CO2 and GHG emissions) and/or energy balance, and that presented the assessment methodology were selected. Articles reporting only on economic data, secondary review papers, papers on nonwoody crops, and papers not written in English were excluded. The exclusion criteria were applied hierarchically and articles were excluded on the basis of the first exclusion criterion met. A flow chart of the selection process is provided in Fig. 1. Key data from all included studies were then extracted and converted into same units before they were entered into the tables. The full spectrum of data categories and studies used to construct the source database of this review are presented in Table 1.

Figure 1.

 Flow chart of the construction of the literature source database. The boxes represent the selection processes (i.e., identification of study, screening, and selection). n represents the number of studies. The horizontal arrows represent the studies that were excluded after each stage whereas the vertical arrows represent the link between selection processes.

Table 1.   Overview of the methodology, energy indicators, environmental impacts, system boundaries and functional unit, reference system, types of life cycle studies and species of short-rotation woody crops (SRWC) used in the reviewed studies
SB and FUConversion
Types of life
cycle study
  • *

    Only CO2 and N2O pollutant gases were included.

  • CO2, CH4 and N2O pollutant gases were included.

  • A, acidification; BD, biodiversity; CCS, carbon capture and storage; E, eutrophication; EA, energy analysis, ECA, economic analysis, EE, energy efficiency; ER, energy ratio; ERE, energy requirement; EY, energy yield; FU, functional unit; FWAE, freshwater aquatic ecotoxicity; GHG, greenhouse gas; HT, human toxicity; LCA, life cycle assessment; LU, land use; MAE, marine aquatic ecotoxicity; NEB, net energy budget; NEG, net energy gain; NEP, net energy production; NER, net energy ratio; NEY, net energy yield; ND, not defined; ODP, ozone depletion potential; OEE, overall energy efficiency; PNEY, primary net energy yield; PO, photochemical oxidation; R, resource use; SB, system boundary SW, solid waste; TE, terrestrial ecotoxicity; W, water use.

EAEECO2Cradle- to-plant; FU=NDCo-combustion, combustionCoal powerComparativePoplarBelgiumVande Walle et al. (2007)
EAEE, NEYCradle-to-farm gate; FU=NDStand alonePoplarNetherlandsNonhebel (2002)
EAERCO2Cradle-to-plant; FU=NDCo-combustionCoal powerComparativeWillow, PoplarSwedenBoman & Turnbull (1997)
EAERCO2Cradle-to-farm gate; FU=NDComparativePoplarTennessee (USA)Turhollow & Perlack (1991)
EAER, ERECO2Cradle-to-farm gate; FU=NDFossil fuel: natural gas, oil, dieselStand aloneWillow, PoplarEnglandMatthews (2001)
EAER, NEYCradle-to-farm gate; FU=NDComparativeWillowSwedenBorjesson (1996a)
EAER, NEYCO2Cradle-to-plant; FU=1 GJCo-combustion, gasificationCoal powerComparativePoplarBelgiumDubuisson & Sintzoff (1998)
EAER, NEYCO2Cradle-to-farm gate; FU=NDComparativeWillowSwedenBorjesson (1996b)
EANEGA, ECradle-to-farm gate; FU=NDComparativePoplarGermanyScholz & Ellerbrock (2002)
EANERCradle-to-farm gate; FU=NDStand alonePoplarPennsylvania (USA)Strauss & Grado (1992)
EANEYCO2Cradle-to-plant; FU=NDGasificationCoal/natural gas powerComparativeWillowSwedenGustavsson et al. (1995)
EAPNEYCradle-to-farm gate; FU=NDComparativeWillowGermanyBoehmel et al. (2008)
EA and ECAERCradle-to-farm gate; FU=NDStand alonePoplarItalyManzone et al. (2009)
LCAEEGHG*, ODP, E, A, HT, R, SWCradle-to-plant; FU=1 MJGasification (with CCS)Coal powerStand alonePoplarItalyCarpentieri et al. (2005)
LCAEE, OEEGHG*, ODP, A, E, PO, SW, RCradle-to-plant; FU=NDGasificationElectricity mix (50% coal and 50% oil)ComparativePoplarItalyRafaschieri et al. (1999)
LCAERGHG*Cradle-to-plant; FU=1 haGasificationNatural gas powerComparativeWillowBelgiumLettens et al. (2003)
LCAERGHG*Cradle-to-farm gate; FU=1 GJFossil fuel: CoalStand aloneWillowthe NetherlandsVan Bussel (2006)
LCAERGHGCradle-to-plant; FU=NDGasificationGrid electricityComparativePoplarPennsylvania (USA)Adler et al. (2007)
LCAERGHG, LUCradle-to-plant; FU=NDCo-combustionPeat, coal power and conventional croplandComparativeWillowIrelandStyles & Jones (2007)
LCANEPCO2Cradle-to-plant; FU=1 haGasificationGrid electricityStand aloneWillowIrelandGoglio & Owende (2009)
LCANERGHGCradle-to-plant; FU=1 MW hGasificationGrid electricityStand aloneWillowNew York (USA)Heller et al. (2004)
LCANERGHG, A, ECradle-to-plant; FU=1 MW hGasificationGrid electricityStand aloneWillowNew York (USA)Keoleian & Volk (2005)
LCANERGHG, A, E,Cradle-to-plant; FU=1 MW hGasificationStand aloneWillowNew York (USA)Heller et al. (2003)
LCANERGHG, R, ODP, HT, FWAE, MAE, TE, PO, A, E, WCradle-to-farm gate; FU=3.93 TJ and 1 haNatural gas, BrassicaComparativePoplarSpainGasol et al. (2009)
LCANER, EEGHG, E, R, SWCradle-to-plant; FU=1 kW hGasificationStand alonePoplarIowa (USA)Mann & Spath (1997)
LCA and ECAERGHG, LUCradle-to-plant; FU=NDCo-combustionPeat and coal powerComparativeWillowIrelandStyles & Jones (2008)

Types of life cycle studies

Two types of life cycle studies emerged from the reviewed literature. The first type of assessment – the so-called stand-alone assessment – describes a bioenergy production system, often in an explanatory way, in order to characterize some important environmental impacts of that bioenergy production system. In contrast, comparative life cycle studies compare the environmental impacts of bioenergy systems to other alternative energy systems.

Techniques and approaches used

A wide range of techniques and approaches have been used in the reviewed studies to assess the environmental effects and energy balance of SRWC (Table 1). These approaches are summarized below.


EA can be defined as a study that quantifies the energy consumed and CO2 emitted in the process of making a product or providing a service (IFIAS, 1974). It includes all processes needed to enable the manufacturing of a product, starting with the procurement of raw materials, and ending with the processing of waste. Each process of the production chain is analyzed separately. Energy and mass flow normalized per unit of product, and finally mass and energy balances are calculated for the chain as a whole. EA was one of the first techniques used in the early and mid-1990s (Turhollow & Perlack, 1991) to provide more information on the total energy used and the CO2 emissions of SRWC systems.


Another widely used method is LCA. The LCA methodology provides a consistent framework for the assessment of environmental aspects and potential impacts associated with a product or service (ISO 14040, 2006). It quantifies the environmental impacts resulting from the provision of a particular product or service (Guinée et al., 2002), and it expresses them relative to a ‘functional unit’ (i.e., a unit that measures the usefulness of this system). Its principle may be summarized by the ‘cradle-to-grave’ (ISO 14040, 2006) approach, according to which all flows of matter and energy into and out of the production system are inventoried. The specificity of LCA is that it avoids shifting the impacts from one area of protection to another. LCA is a compilation of several interrelated components: goal definition and scope, inventory analysis, impact assessment, and interpretation (ISO 14044, 2006). Unlike EA, LCA studies include a wider range of environmental impacts (e.g., acidification, eutrophication, ozone depletion, human toxicity, ecotoxicity) in addition to energy used and CO2 or GHG emissions.

Combined or integrated approaches

The combined energetic-economic analysis (Manzone et al., 2009) and combined LCA-economic analysis (Styles & Jones, 2008) are other approaches used to assess or to compare the environmental, energetic, and economic sustainability of bioenergy production systems or chains. These approaches integrate costs and LCA information into a consistent framework model. They differ from the two previously mentioned methods as they include – in addition to energy and environmental impacts – producer and consumer profitability, the financial valuation of externalities (typically CO2 avoidance benefits) associated with bioenergy crop production, transport, and conversion, as well as impacts so far insufficiently addressed.

System boundaries (SBs) and functional unit (FU)

The SB is the interface between the product (e.g., bioenergy system) and the environment (i.e., other product systems). It delineates which unit processes are included within the LCA. SBs vary among studies in the reviewed literature and one of the most striking features among studies is the number of stages in the life cycle of bioenergy systems that are assessed and compared against the lifetime energy output of the system. Most of the cradle-to-farm gate assessments include the acquisition of raw materials, cultivation and harvesting, and sometimes transport and storage at the farm gate or intermediary storage place (Table 1). The cradle-to-plant studies include the transport of biomass to the power plant, biomass fuel preparation, conversion to electricity, and treatment of waste in addition to the stages listed in the cradle-to-farm gate studies. The spatial and temporal boundaries also differ among the reviewed studies.

The FU describes the primary function fulfilled by a product system, and indicates how much of this function is to be considered in the LCA study (Guinée et al., 2002). The FU is the reference unit that forms the basis for comparisons between different systems. The FU in the reviewed studies, depending of the goal and scope of the studies, is expressed in terms of per unit land area (1 ha), per unit energy content of biomass (1 GJ), or in terms of per unit usable energy output (1 GJ or 1 kW h−1 electricity).

Conversion technologies

A number of biomass conversion technologies have been reported in the literature for converting SRWC to usable energy (i.e., electricity, heat, or both electricity and heat). These conversion technologies can be grouped into two types: (i) direct combustion technologies such as conventional combustion and co-combustion and (ii) indirect combustion technology such as gasification (Table 1). In the direct combustion system, biomass from SRWC is directly burnt to produce high-pressure steam to generate electricity, whereas in the co-combustion system, the biomass is co-combusted with coal as a small proportion of input fuel for the generation of electricity or heat. Gasification processes convert biomass from SRWC into combustible gases that ideally contain the energy originally present in the biomass. These gases are then burnt to produce electricity and/or heat.

Reference systems

System analysis is possible by comparing the bioenergy system with a targeted reference system (Schlamadinger et al., 1997), which in most reviewed studies is limited to a fossil fuel system. Five types of reference systems – fossil fuel, biofeedstock (Brassica carinata), fossil power plant, grid electricity, and previous land use – have been used in the reviewed studies (see Table 1). In the cradle-to-farm gate assessment, harvested biomass from SRWC is compared (on the energy content of the fuels) to fossil fuels such as coal and natural gas. The land area (1 ha) is also used in the study of Gasol et al. (2009) to compare SRWC with other bioenergy systems such as the B. carinata cropping system in addition to the energy content of the biofeedstock. This comparison is expressed in terms of MJ ha−1. In one study (Styles & Jones, 2007) the reference system also included the previous land use expressed in ha yr−1 in order to determine the carbon emissions from the change of land use.

In the cradle-to-plant assessment, the bio-power system is compared with conventional power systems such as a coal power plant, a natural gas power plant, a coal or natural gas combined heat and power (CHP) plant, or to regional grid mix electricity.

Environmental impacts

One of the primary incentives for producing bioenergy is its capacity to reduce GHG emissions as compared with fossil energy. However, as conventional energy production systems, bioenergy production systems cause environmental impacts. Environmental impacts are the consequences of the physical interactions between the studied system and the environment. In practice, all environmental impacts can be classified in several categories of environmental problems. These impact categories range from global impacts such as climate change (GHG balance), regional impacts such as acidification, to local impacts such as eutrophication, or ecotoxicity impacts. With regard to bioenergy from SRWC, the most common environmental impacts reported in the reviewed studies are GHG emissions, and to a lesser extent acidification, eutrophication, solid wastes, and resource use (Table 1). These impacts depend on various factors such as the SRWC cultivation practice, land management, location, and downstream processing and distribution routes.

Energy performance indicators

In the reviewed studies over the period from 1990 to 2009, 10 energy metrics were used to quantify the net renewable energy yield over the life cycle of SRWC (Table 1). Often, these energy indicators are defined differently but have the same meaning. These energy indicators are summarized below.

Energy efficiency (EE)

The EE (Mann & Spath, 1997) or overall EE (Rafaschieri et al., 1999) is defined as the ratio of the usable energy (e.g., electricity) produced to the energy contained in the biomass feedstock. Usually expressed as a percentage, the EE gives the fraction of energy in the biofeedstock that is converted to the final energy product (i.e., electricity). A higher EE indicates a more efficient conversion process.

Life cycle efficiency (LCE)

The EE as defined above does not include the energy consumed by the upstream processes. With reference to LCA, an appropriate energy metric found in the reviewed studies for system efficiency is the LCE. The LCE (Mann & Spath, 1997) or overall system efficiency (Rafaschieri et al., 1999) is defined as the ratio of the difference between the usable energy produced and the energy consumed by the upstream processes to the energy contained in the biomass feedstock. The LCE can be negative, and a negative LCE indicates the overall system energy deficit. The LCE and EE were found mostly in studies using the cradle-to-plant approach.

Energy ratio (ER)

Studies that used the cradle-to-farm gate approach (Turhollow & Perlack, 1991; Dubuisson & Sintzoff, 1998; Matthews, 2001) defined the ER as the ratio of the energy contained in biomass to the energy inputs to produce the biomass feedstock. In the cradle-to-plant studies, the net ER (Mann & Spath, 1997; Vande Walle et al., 2007) was defined as the total usable energy (i.e., electricity, heat, or both electricity and heat) produced by the system divided by the total energy input to drive the system. Typically, only fossil energy inputs are included in this ratio, whereas the renewable inputs, including biomass feedstock itself, are not included. This energy metric reveals the influence of the inputs expressed in energy units to obtain either the biofeedstock (i.e., in the cradle-to-farm gate case) or the usable energy product (i.e., in the cradle-to-plant case). The ER is dimensionless and it illustrates how much energy is produced for each unit of fossil fuel energy consumed. An ER <1 implies that the energy input is higher than the produced energy output.

Energy requirement (ERE)

The ERE (Matthews, 2001) is the ratio between the energy inputs to produce the biomass feedstock vs. the energy contained in the biomass. It is thus the inverse of the ER. The ERE of a bioenergy production system is <1 if the system produces more energy than it consumes (Matthews, 2001).

Net energy yield (NEY)

The NEY (Borjesson, 1996a, b) or net energy budget (Hanegraaf et al., 1998), also referred as net energy gain (Scholz & Ellerbrock, 2002) or primary NEY (Boehmel et al., 2008) or net energy production (Goglio & Owende, 2009) is the difference between the gross energy output produced (i.e., the energy content of the biomass at the farm gate) by the bioenergy system and the total energy required to obtain it (i.e., the fossil energy input). In bioenergy processes, this energy metric is normally related to the unit of production (e.g., 1 ha). The NEY combines productivity and EE into one value. A smaller NEY means that the bioenergy system requires more land to produce the same amount net of energy, when the surface area is used as the unit of production.

Energy use efficiency (EUE)

Finally, another energy indicator used in the cradle-to-farm gate approach to assess the direct and indirect energy required to produce a unit of energy is the EUE. The EUE (Boehmel et al., 2008) is defined as the ratio of the primary NEY (the difference between the primary energy yield and the energy consumption) to the energy consumption. As in the case of ER, an EUE greater than unity indicates that the system produces more unit energy than is consumed by the biomass production processes.

General characterization of the reviewed studies

The majority (19 of 26) of the reviewed studies were undertaken in Europe, and the remainder in the USA. Besides two studies that examined both poplar (Populus) and willow (Salix), a similar amount of studies examined either poplar or willow. Fifteen of the 26 studies quantified and compared the energetic and ecological performance of SRWC with fossil fuels or other bioenergy systems, whereas 11 of the 26 evaluated the performance of SRWC alone without comparisons. Of the reviewed studies the LCA and EA approaches were equally used (46% each), whereas the combined approach was used less frequently (8%). Sixteen studies made the cradle-to-plant assessment and the rest were cradle-to-farm gate assessments. Some of the cradle-to-plant assessments (10 studies) also presented the results of the cradle-to-farm gate stages. Thus, data for 20 cradle-to-farm gate studies could be extracted and analysed from the reviewed studies (Table 2). Of the cradle-to-plant assessments, gasification appeared to be the most applied conversion technology among the main conversion technologies reported in the reviewed studies to convert biomass to electricity and/or heat.

Table 2.   Energy ratios, CO2 and GHG emissions, biomass yield and species of short-rotation woody crops (SRWC) reported in the reviewed studies
Energy ratioCO2 and GHG emissionsBiomassSRWC
Cradle-to-farm gateCradle-to-plantYield
(t ha−1 yr−1)
Life span
Total harvestable
biomass (t ha−1)
  • *

    Values obtained after allocation of impacts to electricity production only.

  • GHG, greenhouse gases; na, not assessed.

13nanana10876PoplarManzone et al. (2009)
15nanana1615naPoplarStrauss & Grado (1992)
16na1.3 kg C GJbiomass−1na11.318252PoplarTurhollow & Perlack (1991)
16410.6 g CO2 Eq MJbiomass−1132 g CO2 Eq kW h−18.823202WillowStyles & Jones (2007)
193nana4.22074PoplarVande Walle et al. (2007)
20na3.8 kg CO2eq GJbiomass−1na15.615212WillowVan Bussel (2006)
21na0.7 kg C GJbiomass−1na924216WillowBorjesson (1996b)
22na1.1 kg C GJbiomass−1na16.8nanaWillowBoman & Turnbull (1997)
22–26na1.7–1.9 kg C GJbiomass−12.9 kg C GJ−110–1523235–345PoplarDubuisson & Sintzoff (1998)
23nanana520100PoplarNonhebel (2002)
26nanana9nanaWillowGustavsson et al. (1995)
29na1.3 g C MJbiomass−1na8–1216128–168WillowMatthews (2001)
32na9.8 g CO2 Eq MJbiomass−1na1025250WillowLettens et al. (2003)
388*2.1 g CO2 MJbiomass−158 kg CO2 GJ1*10nanaWillowGoglio & Owende (2009)
48na0.5 kg C GJbiomass−1na730210PoplarAdler et al. (2007)
50na1.9–2.0 g CO2 Eq MJbiomass−1na13.517216PoplarGasol et al. (2009)
50nanana6.920138PoplarScholz & Ellerbrock (2002)
55110.7 g CO2 Eq MJbiomass−1na13.623214.4WillowHeller et al. (2003)
55130.7 g CO2 Eq MJbiomass−139 g CO2 Eq kW h−113.623214.4WillowHeller et al. (2004)
55160.6 g CO2 Eq MJbiomass−146 g CO2 Eq kW h−113.435469PoplarMann & Spath (1997)
79nanana15.216235WillowBoehmel et al. (2008)

More than half (16) of the reviewed studies did not explicitly refer to the FU, but instead normalized the mass and energy flows per unit of product energy output. Nevertheless, the resulting unit reflects the concept correctly. Among the studies that clearly defined the FU, the land area (1 ha) or energy unit (1 GJ, 1 kW h) were chosen as the FU. All studies quantified the energetic performance of SRWC, although there were differences in the energy indicators used in the assessments. More than three-quarters of all studies provided information on the CO2 or GHG emissions of SRWC. However, in many cases only one or a few pollutant gases contributing to this impact category were included in the assessment. About a quarter of the studies did not assess any environmental impacts. Other important environmental impacts (non-GHG impacts) were less studied. For example, six studies included acidification, eutrophication, and/or resource use impacts. Only three of the reviewed studies included ozone depletion, photochemical oxidation and solid waste impacts. Land use and water use were reported the least (i.e., in only two studies).

Energy balance vs. environmental impacts

This section analyses and compares the range of results presented in the reviewed studies. Owing to the limited data extracted from the studies focusing on the cradle-to-plant assessment, the focus of the analysis and comparison is restricted mainly to the cradle-to-farm gate assessment. Given the small number of studies presenting results on impact category indicators other than GHG emissions, they were not analyzed in detail. Table 2 provides the detailed technical results on the energy indicators and on the GHG emissions. The main data on SRWC included yield, the life span, total biomass production, ER and CO2 or GHG emissions. Yields ranged from 4.2 to 16.8 ton ha−1 yr−1 and the life span varied from 8 to 35 years (Table 2). The variation in yield can be explained by the agronomic practices which vary with intensity of production, the edaphic and climatic conditions. The mean harvestable yield was 11.5 ton ha−1 yr−1 and the median 11.7 ton ha−1 yr−1. With regard to SRWC, the mean and median yields of poplar and willow were comparable (Fig. 2).

Figure 2.

 Comparison of the yield of the two tree species of short-rotation woody crops (SRWC) analyzed in this study. The boxes represent the interquartile range (IQR, i.e., the 25th to the 75th percentile), the horizontal lines within the boxes represent the medians, the small squares within the boxes represent the means, the vertical lines drawn from the edges of the IQR boxes represent the whiskers (i.e., the largest and smallest values within 1.5 IQR), the horizontal lines on the whiskers represent the outliers (i.e., values which are within 1.5 and 3 IQR lengths from the upper and lower boundaries). The number n in this figure represents the number of studies included in the analysis.

The ER values ranged from 13 to 79 for the cradle-to-farm gate and from 3 to 16 for the cradle-to-plant assessments, respectively. The ER value was lower if the final output was quantified in terms of electricity generated rather than as the energy content of the produced biomass from SRWC. There was no exception to this finding. This result is indeed consistent with the fact that expanding the boundary beyond the farm gate to include conversion to electricity should always result in a lower ER. Assumptions about energy use in biomass production and the efficiency of biomass conversion to electricity had large effects on the cradle-to-plant ER. The highest cradle-to-plant ER value (i.e., 16) was for the gasification plant that had an electrical conversion efficiency of 37.2%. The direct biomass combustion technology had a much lower efficiency (η=27.7%) as well as ER value (9.9) than the gasification technology. Despite its high electrical efficiency (η=37.5%), biomass co-combustion technology had a low ER value (i.e., 4). This was mainly due to the relatively high EREs for biomass production that were (coincidentally) assumed in the studies that used co-combustion as a conversion technology.

The mean and the median ER values of the reviewed studies (cradle-to-farm gate) were 32.5 and 24.5, respectively (Fig. 3). The variation in the ER values can be attributed to differences in yield, to the types of fertilizer used and their application rates, and to major differences in the method of harvesting.

Figure 3.

 Cradle-to-farm gate energy ratios (ER) of the reviewed bioenergy systems classified into types of short-rotation woody crops (SRWC), assessment techniques, and overall studies. Twenty studies which presented data on ER were analyzed in this graph. The whiskers boxes of this figure are explained in Fig. 2.

Table 3 presents the processes that contributed to energy input in the investigated bioenergy system of each of the reviewed studies. The components (i.e., processes) within the investigated bioenergy systems in the reviewed studies vary considerably. This variability illustrates the diversity of the systems in which SRWC can be and are grown. The total energy input ranged from 46.3 to 247.7 GJ ha−1, whereas the energy output ranged from 1418 to 6930 GJ ha−1 depending on the life span. The energy input was higher in fertilized bioenergy systems (i.e., intensive) than in unfertilized (i.e., extensive) bioenergy systems. The comparison of different energy consuming processes revealed that harvesting and fertilization (i.e., fertilizer production plus their application) accounted for the majority of the energy input to the bioenergy system. Harvesting accounted for 8–76% of the energy input in the bioenergy production across the reviewed studies followed by fertilization, which accounted for between 10% and 64% of the energy input, depending on the growing conditions. Fertilizer production constituted the major part (∼90%) of energy consumed in the fertilization step. Herbicide treatment and weeding contributed between 1% and 8% of the total energy input of the bioenergy systems in the reviewed studies. Other mechanical operations, such as tillage and planting or the removal of stumps (grubbing up), required less energy than harvesting and fertilization and mainly concerned the planting of SRWC. They involved energy inputs ranging from 2% to 19% for tillage and planting and from 1% to 9% for the removal of stumps. The contribution from the production of cuttings ranged from 2% to 9% across the reviewed studies. Transport is also an important component in the energy consumption of bioenergy systems as its contribution ranged from 2% to 15%. In general, harvesting and fertilization processes were the major contributor to energy input in the reviewed studies. However, in some studies processes such as active drying and fencing had far-reaching impacts on the energy input as well as the ER. For example, in the study of Matthews (2001), the contribution of active drying and fencing totaled 53% (Table 3). When these processes (i.e., active drying and fencing) were excluded from the SB of the analysis, the resulting ER was 60 (Matthews, 2001). Similarly, the ER reported by Goglio & Owende (2009) and that reported by Styles & Jones (2007), respectively, increase from 38 to 45 and from 16 to19 if the contribution of fencing was excluded from their analyses. It is worth mentioning that active drying of SRWC depends on the end use. Drying may not be required if the produced biomass is dried on farm; it could be performed at the conversion site (using waste heat) or not be performed if the conversion system can use wet chips.

Table 3.   Cradle-to-farm gate energy input and output, contribution of energy consuming processes (included in or excluded from the system boundaries), and species of short-rotation woody crops (SRWC) reported in the reviewed studies
gate energy
(GJ ha−1)
Process contribution in (%)SRWC
Total inputTotal outputCapital equipmentCuttings productionTransportTillage/plantingHerbicide/weedingFertilizationIrrigationFencingHarvest/chippingStorage/dryingGrubbing up
  • The sum of all contributions does not always give 100%.

  • *

    Irrigation is included in the system boundary but no value for the breakdown is available.

  • †This value includes the contribution of all farming processes, except fertilization.

  • –, the process is not included in the system boundary; na=not assessed.

46.31759. & Owende (2009)
49.53933.27.55.336.142.68.5WillowBoehmel et al. (2008)
52.42622.1na         PoplarScholz & Ellerbrock (2002)
75.21418.0nanananananananananaPoplarVande Walle et al. (2007)
79.01800. (2002)
84.24104.25.22.519.34.535.6*32.9PoplarGasol et al. (2009)
84.44053.2nananananananananananaPoplarAdler et al. (2007)
98.35434.93923.14.33938.41.2WillowHeller et al. (2003)
105.03006.23841330402WillowMatthews (2001)
113.61504.08.37.663.718.91.4PoplarManzone et al. (2009)
115.03024.0nanananaWillowGustavsson et al. (1995)
123.71860.57.42.514.275.8PoplarStrauss & Grado (1992)
126.26930.31.815.482PoplarMann & Spath (1997)
140.94509.11.911.92.21.848.326.67.3WillowLettens et al. (2003)
155.03225. Bussel (2006)
184.94198. & Sintzoff (1998)
202.04320.03.515.34.2151.124.9WillowBorjesson (1996b)
211.74761. & Turnbull (1997)
234.43663. & Jones (2007)
247.74027.31.210.48324.253.1PoplarTurhollow & Perlack (1991)

With regard to the techniques used, the cradle-to-farm gate ER values ranged from 16 to 55 for LCA and from 13 to 79 for EA, respectively. The EA technique determined a lower mean (28) and median (22.5) ER compared with LCA. The ER interquartile range (IQR) is lower for the EA technique than for LCA, but overlaps with it (Fig. 3). Results from the two techniques varied because of the difference in the types and sources of data, assumptions about farm inputs, and the computation methods. Many LCA studies combine primary data and sometimes secondary data available in the life cycle inventory databases, whereas EA uses data from producers. EA uses simple computational tools (e.g., Microsoft excel spreadsheets), whereas simple as well as complex dedicated tools (e.g., simapro, gabi) are used in LCA to model the bioenergy system.

With regard to the type of species of the SRWC, the ER values ranged from 16 to 79 for willow and from 13 to 55 for poplar, respectively. The mean and median ER values for willow and poplar were found to be nearly identical [i.e., 33.8 and 27.5, respectively, for willow vs. 31.2 and 23, respectively, for poplar (Fig. 3)]. Their ER IQR and whisker also overlap. Thus, one can conclude that, on average, willow and poplar have very similar ER values.

In general and regardless of the techniques used, the ER values reported in the reviewed studies for both willow and poplar indicate a high ER (i.e., there is a high energy return). On the basis of fossil energy inputs, SRWC improve the effective use of this finite energy source. Therefore, the cultivation of SRWC for bioenergy production can be considered beneficial from an energy perspective.

The intensities of GHG emissions ranged from 0.6 to 10.6 g CO2 Eq MJbiomass−1 for the cradle-to-farm gate and from 39 to 132 g CO2 Eq kW h−1 electricity for the cradle-to-plant assessment. The intensity of GHG emissions was larger when the final output was given as electricity generated rather than as the energy content of the biomass from SRWC. This difference is simply due to the efficiency of biomass conversion to electricity. The gasification technology had the lowest intensities of GHG emissions (39 g CO2 Eq kW h−1) due to its high efficiency (η=37.2%), followed by the direct combustion technology (52.3 g CO2 Eq kW h−1). Co-combustion technology (η=37.5%) had the largest GHG emission intensities. This high value of GHG emission intensities for the co-combustion technology was due to the relatively high GHG emissions in biomass production that were (coincidentally) assumed in the co-firing studies, and to the up- and downstream GHG emissions from coal.

The wide range of cradle-to-farm gate CO2 and GHG emissions observed among the reviewed studies can be attributed to the agrochemical input (mainly fertilizer), assumptions about N2O linked to fertilizer input, the carbon sequestration process (soil carbon and carbon pools below ground), and the N2O and CH4 associated with the decomposition of leaves and litter (Table 4). The types of fertilizer used differed among the reviewed studies. Ammonium-based fertilizer (e.g., ammonium sulfate), nitrate-based fertilizer (e.g., ammonia), and urea are some types of fertilizer used in the reviewed studies. Nitrogen fertilizer requirements varied from 40 to 138 kg N ha−1 whereas the emission factors associated to fertilizer production varied substantially depending on the production process.

Table 4.   Cradle-to-farm gate CO2 and greenhouse gas (GHG) emissions, contribution of sources and sink of GHG emissions (included in or excluded from the system boundaries), coppice cycle, and species of short-rotation woody crops (SRWC) GHG reported in the reviewed studies
Cradle-to-farm gate CO2 and GHG emissionsSources and sink of GHG emissions (%)BiomassSRWCSpeciesReferences
Net totalTotal without sequestrationManagementAgricultural inputFertilization (N2O)DecompositionCarbon sequestrationCoppice cycle
  1. The values between parentheses represent the contribution to GHG emissions, when carbon sequestration is considered.

  2. na, not assessed.

0.6 g CO2 Eq MJbiomass−184.615.47PoplarMann & Spath (1997)
0.7 g CO2 Eq MJbiomass−13.2 g CO2 Eq MJbiomas−117.8 (86)18.9 (91)22.3 (107)40.9 (197)(−381)3WillowHeller et al. (2003)
1.7 g CO2 Eq MJbiomass−1nananananana10PoplarAdler et al. (2007)
1.9 g CO2 Eq MJbiomass−14939.411.65PoplarGasol et al. (2009)
2.1 g CO2 MJbiomass−167.632.43WillowGoglio & Owende (2009)
3.1 g CO2 MJbiomass−15050naWillowBorjesson (1996b)
3.8 g CO2 Eq MJbiomass−18.4 g CO2 Eq MJbiomass−124.9 (55)19.2 (42)42.3 (95)13.6 (30)(−123)2WillowVan Bussel (2006)
3.9 g CO2 MJbiomass−144.855.36WillowBoman & Turnbull (1997)
4.8 g CO2 MJbiomass−1nana3WillowMatthews (2001)
4.8 g CO2 MJbiomass−172.827.2naPoplarTurhollow & Perlack (1991)
6.2–6.9 g CO2 MJbiomass−16733naPoplarDubuisson & Sintzoff (1998)
9.8 g CO2 Eq MJbiomass−19.713.676.83WillowLettens et al. (2003)
10.6 g CO2 Eq MJbiomass−17.747.823.13WillowStyles & Jones (2007)

Many reviewed studies overlooked N2O emissions from fertilizer application; those that included N2O used the IPCC methodology for direct and indirect N2O emissions estimation (IPCC, 1996). Two studies included the decomposition of leaves and litter in their assessments and reported GHG emissions values ranging from 1.1 to 1.3 g CO2 Eq MJbiomass−1 (Heller et al., 2003; Van Bussel, 2006).

Few reviewed studies included the carbon sequestration process (soil carbon and carbon pools belowground) in their analyses. In the small number of reviewed studies in which values are incorporated, data ranged from −2.7 to −4.7 g CO2 Eq MJbiomass−1 (Table 4). However, it is important to note that the sequestration of carbon in soil is site-specific and depends on factors such as existing soil carbon levels, climate, soil characteristics, and management practices (Keoleian & Volk, 2005). Generally, SRWC would be expected to significantly increase soil carbon in arable soils, but not in grassland soils. It can therefore be argued that accounting for carbon sequestration is not always relevant, and depends on SBs and displacement assumptions (even when planted on tillage land SRWC may ultimately displace grassland if arable production shifts onto grassland).

The intensities of CO2 emissions ranged from 2.1 to 6.2 g CO2 MJbiomass−1 for EA, the mean CO2 emission intensities was 4.7 g CO2 MJbiomass−1. EA studies solely focused on CO2 emissions from fuel combustion and CO2 emissions from farm material production and overlooked the carbon sequestration process as well as non-CO2 GHG emissions such as N2O from fertilization (Fig. 4a). The mean and median GHG emissions intensities were 4.1 and 1.9 g CO2 Eq MJbiomass−1 for the LCA technique, respectively (Fig. 4b).

Figure 4.

 Cradle-to-farm gate carbon dioxide (CO2) emissions (a), greenhouse gas (GHG) emissions (b) of the reviewed bioenergy systems classified into types of short-rotation woody crops, assessment techniques, and overall studies. Thirteen studies which presented data on CO2 and GHG emissions were analyzed in this graph. The whiskers boxes of this figure are explained in the legend of Fig. 2.

With regard to the tree species in SRWC, the intensities of CO2 for willow ranged from 2.1 to 4.8 g CO2 MJbiomass−1, whereas for poplar the range was 4.8–6.2 g CO2 MJbiomass−1. The mean and median CO2 emissions intensities for willow were 3.2 and 3.5 g CO2 Eq MJbiomass−1, respectively. For poplar, the mean and median CO2 emission intensities were identical: 5.4 g CO2 MJbiomass−1 (Fig. 4a). The intensities of GHG emissions ranged from 0.7 to 10 g CO2 Eq MJbiomass−1 for willow, whereas for poplar the range was 0.6–1.9 g CO2 Eq MJbiomass−1. The mean and median GHG emissions were higher for willow than for poplar (Fig. 5b). Based on these data values and given the fact there was not enough data for a meaningful comparison, it is difficult to determine if the GHG as well as the CO2 emission intensities of willow and poplar were similar. However, there was some evidence to suggest that these SRWC species might be comparable (Fig. 4).

Figure 5.

 Cradle-to-farm gate greenhouse gas (GHG) emissions for short-rotation woody crops (SRWC) as compared with coal. The comparison is based on GHG emissions per MJ energy content of biomass and coal from seven studies. The bars represent the values of GHG emissions of SRWC. The horizontal line above indicates the value of the reference system (i.e., coal).

Irrespective of the differences among the reviewed studies and assuming that the intensity of GHG emissions from coal to be 96.8 g CO2 Eq MJcoal−1 (Frischknecht et al., 2007), Fig. 5 shows that SRWC reduce GHG emissions as compared to coal. The achievable GHG emission reductions ranged between 90% and 99%.


This demonstrates that SRWC reduce emissions and should therefore be part of an overall strategy for achieving the minimum target for GHG emissions reduction (i.e., 50%) in the year 2017 as required by the EU Renewable Energy Directive (EC, 2008).

The intensities of CO2 or GHG emissions were related to the ER for the reviewed studies as presented in Fig. 6. The CO2 or GHG emission intensity declined exponentially as the ER increased. This finding confirms the common knowledge that a reduction of GHG emissions can be achieved via reduced energy input into the system.

Figure 6.

 Carbon dioxide (triangle) and greenhouse gas (GHG) (bullets) emissions as a function of energy ratio (ER). Each symbol (triangles and bullets) represents one specific study. The dashed and solid lines indicate the best fits through the data. R2, correlation coefficient; P, level of significance.

With regard to other environmental impacts – especially those that are characteristic of the agricultural phases of SRWC cultivation such as acidification and eutrophication – no average results can be provided because of the small number of cradle-to-farm gate LCA or EA studies that investigated these impacts. Nevertheless, one general observation can be made. For SRWC, environmental impacts such as acidification and eutrophication seem to be low. The cradle-to-farm gate acidification impacts ranged from 15.7 to 23.5 mg SO2 Eq MJbiomass−1. These values were 20–30 times lower than those of coal (476 mg SO2 Eq MJcoal−1). The eutrophication impact values ranged from 2.4 to 3.3 mg PO4 Eq MJbiomass−1. SRWC performed slightly better in terms of eutrophication impacts as compared to coal (5.2 mg PO4 Eq MJcoal−1).

Lessons to be learned

Our review revealed that the estimation of the energetic performance of bioenergy systems is complex. Not only the methodologies were different, but also various indicators were used for the evaluation of the energetic performance of bioenergy systems. These indicators prevented far-reaching conclusions from being drawn, discouraged a more transparent view of bioenergy systems, and did not facilitate immediate comparison of studies. As the results of LCA studies are increasingly being used to assist decision making at national and international levels, it is of the utmost importance to refine the ISO standards and to expand the LCA methodology with guidelines on indicators and methodologies to be used to estimate the energetic performance of bioenergy systems.

In the reviewed studies, fossil fuels (e.g., coal, natural gas) as well as biofeedstock (B. carinata) were used as reference systems. This picture however, is incomplete. To make sure that bioenergy systems do not deplete the soil carbon stock, we recommend that the SB also includes a reference land use. With this SB, it will be possible to compare the land on which the SRWC are grown to previous land use.

With regard to energy balance, three variables were identified as the main sources of diverging results among reviewed studies: the amount and types of fertilizer used, harvesting method, and assumptions about the yield per hectare. With respect to GHG balance the divergent results were due to assumptions about N2O emissions, the type of fertilizer used and its application rate, differences in the treatment of gases that contribute to GHG, and the SBs. Harmonized rules based on reasonable guidelines and assumptions on methodological issues, and how to deal with the associated uncertainty of key parameters would help to reduce the variability of LCA results.

Although the two studies that included the contribution of N2O emissions from decomposition of leaves and litter in their assessments indicated a high contribution from decomposition of leaf-litter to GHG emissions (Table 4), it is; however, important to mention that all vegetation systems result in N2O loss from leaf fall. Also, given that leaves and litter accumulate on the soil surface, their decomposition in most cases will be aerobic, and the emissions of N2O due to denitrification (an anaerobic process) will be minimized (Heller et al., 2003). Consequently, it is not always relevant to include leaf-litter N2O emissions – certainly not relevant to include all of it – in the LCA of bioenergy systems. For example, emissions from leaf-litter should not be accounted for when SRWC systems result in less litter and associated N2O emissions compared with the reference land use. In contrast, emissions from leaf-litter should be accounted for when SRWC systems result in more litter and associated N2O emissions compared with the reference land use.

Insights from this review indicated that carbon sequestration contributed to improve the GHG balance. However, there are situations when this factor (i.e., carbon sequestration) should not be accounted for in the analysis. This is the case when for example, SRWC displaces land with high carbon stock such as grassland. In contrast, carbon sequestration should be accounted for when SRWC displaces cropland, and if the latter is not shifted to grassland. Carbon sequestration should also be accounted for when SRWC are grown on abandoned land that exhibit low soil carbon stocks.

The cradle-to-farm gate results from statistical analysis showed that poplar and willow appeared to have similar mean yield and ER values while the results for the mean CO2 and GHG emissions varied substantially. This indicates different assumptions about fertilizer emission rates, transport distance, and carbon sequestration between willow and poplar. The yield values demonstrated the smallest difference in the relative variability (IQR) between the two SRWC species. The ER also showed a much lower variation. One can therefore have confidence in the results that compared the energetic performance of willow and poplar because their ER was less wide-ranging.

Difficulties arose in the course of this review. Inventory data presented in some studies were incomplete and the sources of data were not specified. Also, very few studies presented a breakdown of the processes contributing to the energy input or to GHG impacts. We therefore recommend that future studies present complete inventory data, specify their sources, and when possible, make a breakdown of processes contributing to energy use as well as environmental impacts.


Despite the wide variation in specific numerical results among the reviewed studies, it is possible to draw the following conclusions: on average, SRWC yielded 36 times more energy than coal (ERcoal∼0.9) per unit of fossil energy input, and GHG emissions were 24 times lower than those of coal (GHGcoal∼96.8). Consequently SRWC provide an opportunity to reduce dependency on fossil fuels and to mitigate GHG emissions. Harvesting and fertilization were the largest contributors to energy use across the reviewed studies, and it was found that harvesting consumed 1.2–1.3% more energy than fertilization.

Despite the fact that SRWC can play an important role in mitigating GHG emissions, some uncertainties linked to evaluating the GHG emissions from individual bioenergy systems remain. N2O emissions from fertilizer application, carbon sequestration, and the reference land use (baseline) pose the major challenges to providing a high degree of confidence in the calculated emissions.

To reduce the high variability and create some more consistency in the future studies, harmonized rules based on reasonable guidelines and assumptions on methodological issues are needed. This could be achieved by limiting the freedom of choices for dealing with carbon sequestration. It should for example not be allowed to account for carbon sequestration in LCA when SRWC displace land with high carbon stock such as grassland. Likewise, when SRWC displaces croplands, carbon sequestration should not be accounted for should the latter shift to grasslands. Conversely, carbon sequestration should be accounted for in LCA when SRWC are grown on abandoned lands that exhibit low soil carbon stocks.

Efforts should also be made to develop a widely accepted framework toward a reliable analysis of EE of bioenergy production systems. Finally, more research is needed to address insufficient knowledge of the net GHG emission fluxes from bioenergy systems.


The research leading to these results has received funding from the European Research Council under the European Community's Seventh Framework Programme (FP7/2007-2013), ERC grant agreement no. 233366 (POPFULL). O. El Kasmioui is a research assistant of the Flemish Science Foundation (FWO, Brussels). We acknowledge various authors (in particular Dr David Styles and Dr Pietro Goglio) who have helped us by providing more detailed information on their published results. We also thank Dr Rhonda Fisher for checking English grammar and language throughout this manuscript. Finally, we thank the three anonymous reviewers for their constructive comments and valuable suggestions on an earlier version of the manuscript.