Woody Biomass Energy Potential In 2050 - David Klein

2P. Lauri et al. / Energy Policy ( ) – Offermann et al., 2011). If all these resources were used for energyproduction, they could cover 10–40% of the world's primaryenergy consumption in 2050.4Woody biomass energy potential depends not only on theavailable woody biomass resources but also on the competitionbetween alternative uses of those resources and competitionbetween alternative sources of energy (Radetzki, 1997; Sedjo,1997; Berndes et al., 2003). These effects can be separated by usingthe concept of supply and demand curves. The energy wood supplycurve defines the amount of woody biomass available for large-scaleenergy production at various hypothetical energy wood prices, thatis, it summarizes all the relevant information from the biomasssector needed to model large-scale energy wood use. The energywood demand curve defines the desired amount of woody biomassfor large-scale energy production at various hypothetical energywood prices; in other words, it summarizes all the relevantinformation from the energy sector needed to model large-scaleenergy wood use. Note that the term “energy wood” refers to largescale woody biomass use for energy. Hence, energy wood does notinclude small-scale woody biomass use for energy (householdfuelwood), which is modeled separately. Household fuelwood isnot directly connected to large-scale energy wood markets becauseit often comes from different sources than large-scale energy woodand because its utilization is based on technologies that areincompatible with other forms of energy (May-Tobin, 2011).The advantage of studying energy wood supply separately fromdemand is that it provides a consistent way of analyzing woodybiomass energy potential without the need for explicitly modelingwhat happens in the energy sector. The energy wood supply curvedefines woody biomass energy potential for an arbitrary range ofenergy wood prices rather than for some scenario-specific pricesand/or quantities. As energy wood demand includes large uncertainties associated with different technology alternatives andmitigation policy options (e.g., IEA, 2010; IPCC, 2011; GEA, 2012),it makes sense to study a large range of possible outcomes. Theenergy wood supply curve can be used directly for energy policyanalysis by connecting it with different energy wood demandscenarios or it can be used as an input to the energy sector modelinstead of linking the energy and biomass supply models bycomplicated iterative procedures (Tavoni et al., 2007).Large unused woody biomass resources and an increasing needfor climate change mitigation has awoken policymakers' interestin woody biomass energy potential and has given rise to a largenumber of studies on this topic. The majority of these studies focuson regional potentials (e.g., Asikainen et al., 2008; de Wit and Faaij,2010; Moiseyev et al., 2011; Verkerk et al., 2011; Daigneault et al.,2012; Ince et al., 2012; Lauri et al., 2012). The global studies are notbased on explicit economic analysis (Parikka, 2003; Smeets andFaaij, 2007a, 2007b; Anttila et al., 2009); they lack a detaileddescription of woody biomass supply from forests (Berndes et al.,2003; Reilly and Paltsev, 2008; Hoogwijk et al., 2009; van Vuurenet al., 2009, 2010; Haberl et al., 2010; Beringer et al., 2011; Poppet al., 2011); or they lack a detailed description of woody biomasssupply from energy crop plantations (Raunikar et al., 2010; Faveroand Mendelsohn, 2013). Hence, the existing literature misses aneconomic analysis of global woody biomass energy potential,which would include a detailed description of woody biomasssupply from forests as well as from energy crop plantations in aconsistent framework with the agricultural sector.The objective of this paper is to estimate the global woodybiomass energy potential in 2050 using the Global BiosphereManagement Model (GLOBIOM) (Havlík et al., 2011; Schneider4According to energy sector assessments, world primary energy consumptionis expected to increase to about 1000 EJ/year in 2050 (GEA, 2012).et al., 2011). Our analysis differs from previous studies in four ways.First, it explicitly models the competition between alternative usesand sources of woody biomass through the market mechanism.Second, it includes a detailed spatially explicit description of woodybiomass supply from forest and plantations. Third, it separatesenergy wood supply and demand from each other. Fourth, theland-use competition between food, wood and energy production ismodeled explicitly. The focus of the paper is on global level results.However, some regional level implications will also be highlighted.This paper is organized as follows: Section 2 describes the modeland data. Section 3 presents the results of the model (i.e., estimateson energy wood supply curves). In Section 4 we compare theestimates to some other studies. Finally, Section 5 summarizes theresults and discusses some of the policy implications of energywood supply curves.2. Methods and data2.1. ModelThe Global Biosphere Management Model (GLOBIOM) is a globalpartial equilibrium model of the forest and agricultural sectors, whereeconomic optimization is based on the spatial equilibrium modelingapproach (Takayama and Judge, 1971). The supply of biomass ismodeled at 200 km 200 km resolution,5 while the demand andtrade of biomass operates at regional level. The world is divided into30 regions that can produce, consume, and trade agriculture andforest sector final products in perfectly competitive markets. Besidesfinal products, the model has several primary and by-products, whichare utilized as inputs in final products production activities. Themodel includes six land cover types: cropland, grassland, othernatural vegetation land, managed forests, unmanaged forests, andplantations.6 The term “managed forests” refers to forest area that isharvested, while “unmanaged forests” refers to forest area that is notharvested. Depending on the relative profitability of primary, by-, andfinal products production activities, the model can switch from oneland cover type to another. Biomass use for large-scale energyproduction is usually based on the POLES or MESSAGE energy sectormodels (Havlík et al., 2011; Reisinger et al., 2013), but this option wasnot used in the present study. Mean annual increments and growingstocks for GLOBIOM are obtained from the Global Forest Model(G4M), which is a spatially explicit process-based forest managementmodel (Kindermann et al., 2006, 2008). Plantations yields are basedon own calculations, as described in Havlík et al. (2011). The initialperiod of the model is the year 2000, and operation is in 10-year timesteps. More information about the model and related publications canbe found in web page 〈www.globiom.org〉.GLOBIOM is based on recursive optimization so that the solvingtimes of the model can be kept within reasonable limits. In therecursive optimization model decisions are based only on the costsand benefits of the current period, which sets some limitations onthe forest dynamics modeling compared to intertemporal optimization models (Sjolie et al., 2011). To handle forest dynamics in therecursive framework, it is assumed that all managed forests arenormal forests, which allows the model to ignore the forest ageclass dynamics and other intertemporal aspects of forest owners’decision making. Normal forests have a uniform distribution of5The supply-side resolution is based on the concept of Simulation Units, whichare aggregates of 5–30′ pixels belonging to the same country, altitude, slope, andsoil class (Skalsky et al., 2008). For the present study, the supply-side resolutionwas aggregated to 120′ ( E200 km 200 km).6There are other three land cover types represented in the model to cover thetotal land area: other agricultural land, wetlands, and not relevant (bare areas,water bodies, snow and ice, and artificial surfaces). These three categories arecurrently kept constant at their initial level.Please cite this article as: Lauri, P., et al., Woody biomass energy potential in 2050. Energy Policy (2013), http://dx.doi.org/10.1016/j.enpol.2013.11.033i

P. Lauri et al. / Energy Policy ( ) – 3material productssawlogsenergy woodmaterial productsroundwood( commercial species)pulplogsenergy woodother industrialroundwoodhousehold fuelwoodtotal tree biomass growthstemwoodroundwoodenergy wood(non-commercialspecies)household fuelwoodenergy woodharvest losseshousehold fuelwoodenergy woodbranches andstumpshousehold fuelwoodfoliage and rootsFig. 1. Available woody biomass resources from forests in the model.age-classes, which implies that mean annual increments andgrowing stocks remain constant over time and that forest ownerscan maintain constant sustainable harvested volumes over time(Reed, 1985).Deforestation is driven in GLOBIOM by agricultural land useexpansion.7 Afforestation is not modeled explicitly, but it isincluded implicitly in the deforestation. Hence, deforestationshould be interpreted as net deforestation rather than grossdeforestation.8 Deforested biomass is excluded from the availablewoody biomass resources; this implies that woody biomassdemand influences deforestation only indirectly through landuse competition between agriculture and forestry. There areseveral reasons why deforested biomass often remains unused(Smeets and Faaij, 2007a). First, deforestation happens mainly intropical areas where the deforested biomass is largely from noncommercial species. Second, burning the deforested biomass onthe spot facilitates the land-use change from forest to agriculture.Third, there is often no infrastructure for harvesting and transporting the deforested biomass away from remote areas.2.2. Available woody biomass resources from forestsThe available woody biomass resources in forests depend on forestareas, increments, and biomass expansion factors. As incrementsmeasure only stemwood growth, biomass expansion factors areneeded to estimate the amount of branches and stumps. Fig. 17There are various reasons for deforestation. However, in 80% of cases themain reason is agricultural land expansion (Hosonuma et al., 2012; Houghton,2012).8One way to complement the GLOBIOM results for forest area changes is toimplement deforestation and afforestation in tandem with the G4M model,whichwould allow the model to take the intertemporal aspects of forest areachanges into account. However,this option was not used in the present study.Table 1Initial forest areas in the model 92314136611460483381243177116251258260230756aThe EU27 includes European Union, Russia includes Russia and rest of Europe,Africa includes Africa and Middle-East, Asia includes Asia and Oceania, North-Americaincludes Canada and USA, South-America includes Central and South-America.illustrates the available woody biomass resources from forests inthe model.The location of forest areas is based on global land-cover data(GLC, 2000). To be consistent with forest inventory data, totalforest area is calibrated to match FAO forest area data (FAOSTAT,2013). Total forest area is divided into disturbed and undisturbed(¼primary) forests. The share of primary forests in total forests isdownscaled by G4M based on human activity impact on the forestareas (Kindermann et al., 2008). This approach gives 1771 Mha ofprimary forests, which is 43% of total forest area (Table 1). Thedownscaled primary forest area is somewhat higher than the 36%of total forest area given in FAO statistics (FRA, 2010).The available stemwood for each forest area unit is determined bymean annual increments, which are based on net primary productivity(NPP) maps from Cramer et al. (1999) and from different downscalingtechniques described in Kindermann et al. (2008). Mean annualincrements for forests are divided into commercial roundwood, noncommercial roundwood and harvest losses (Table 2). CommercialPlease cite this article as: Lauri, P., et al., Woody biomass energy potential in 2050. Energy Policy (2013), http://dx.doi.org/10.1016/j.enpol.2013.11.033i

P. Lauri et al. / Energy Policy ( ) – 4Table 2Average regional mean annual increments and wood biomass yields for forests aSouth-AmericaMean annual incrementHarvest l)Branches and .13.3roundwood is stemwood that is suitable for industrial roundwood(sawlogs, pulplogs and other industrial roundwood). Harvestlosses and non-commercial roundwood are stemwood that isunsuitable for industrial roundwood, but can be used for energywood or household fuelwood. The difference between harvestlosses and non-commercial roundwood is that the former hasunwanted stemwood sizes, while the latter has unwanted treespecies. The amount of harvest losses is based on G4M modelestimates and is about 20% of the increment. The share of noncommercial species is based on FRA (2010) data on commercialand non-commercial growing stocks. In tropical zones thediversity of tree species is high, and only 20–50% of the growingstock can be utilized for industrial roundwood. Most tropicaltree species have no commercial use because their propertiesare unknown and/or because the production process requirehomogenous raw material (e.g., chemical pulp production). Inboreal and temperate zones, forests are more homogenous andthe share of non-commercial species is practically zero. At theglobal level the share of non-commercial species is about 40% ofthe growing stock.In addition to stemwood, available woody biomass resourcesalso include branches and stumps, which can be used for energywood or household fuelwood. The biomass expansion factor forbranches and stumps is estimated by assuming that total treebiomass consists of 60% stemwood, 25% branches and stumps, and15% foliage and roots. There are large variations in the biomassfractions of different tree species and tree ages (IPCC, 2006;Petersson et al., 2012). However, GLOBIOM currently does notdifferentiate explicitly the different tree species and therefore weuse average fractions. The average fractions are based on coniferous and non-coniferous species with a stand age of 60 years fromboreal and temperate forests (Lehtonen et al., 2004; IPCC, 2006;Petersson et al., 2012). The fraction of stemwood is usually lowerthan 60% and the fraction of branches and stumps is higherthan 25% for tropical non-coniferous trees (IPCC, 2006; FAO,2007). However, if tropical forests are managed more intensivelyor even transformed into plantations, the fraction of stemwoodwill increase. Hence, we also apply the above values for tropicalforests, as suggested by Anttila et al. (2009).Logging residues consist of harvest losses, branches and stumps.Logging residues are by-products of roundwood harvesting, that is,they supply is connected to harvested volumes of roundwood. Thefraction of logging residues that can be removed from forestsdepends on technical and environmental constraints such asindustrial roundwood harvesting methods, tree species, soil type,etc. (Hakkila, 2004; EEA, 2007; Mantau et al., 2010). As this type ofdata is not available at the global level, the share of logging residuesthat can be recovered is commonly approximated by using recoveryratios. Recovery ratios define the fraction of logging residues thatcan be realistically harvested. The estimates of recovery ratios varyin the range of 0.25–0.75 (Gan and Smith, 2006; Smeets and Faaij,2007a; Titus et al., 2009; Offermann et al., 2011). Most studies use arecovery ratio of 0.5, which is also used in this study.Table 3Suitable areas for plantations in the model (Mha).WorldEU27RussiaAfricaAsiaNorth AmericaSouth AmericaTotalCroplandGrasslandOther natural 202621169984511112347124113332.3. Available woody biomass resources from plantationsAvailable woody biomass resources from plantations dependon plantation areas and plantations yields. Plantations are dedicated energy crop plantations, that is, they produce only energywood. Industrial roundwood and household fuelwood plantationsare included in forests because they are often closer to naturalforests than energy crop plantations in terms of their rotationtimes and increments.Plantation area expansion depends on the land-use changeconstraints and economic trade-offs between alternative land-useoptions. Land-use change constraints define which land areas areallowed to be changed to plantations and how much of these areascan be changed within each period and region (so-called inertiaconditions). Land-use inertia conditions limit the maximum feasibleplantation expansion to 5% of available areas for each period.Permitted land-cover types for plantations expansion includecropland, grassland, and other natural vegetation areas, and theyexclude forest areas (Table 3). However, plantation expansion indirectly affects forest areas through relocation of cropland and grassland. Excluding forest areas from plantation expansion can beinterpreted as a sustainability criterion, as in van Vuuren et al.(2010) and Beringer et al. (2011). Within each land-cover type theplantation expansion is additionally limited by land suitability criteriabased on aridity, temperature, elevation, population, and land-coverdata, as described in Havlík et al. (2011).Plantation expansion to cropland and grassland depends on theeconomic trade-off between food and wood production. Hence,the competition between alternative uses of land is modeledexplicitly instead of using the “food/fiber first principle”, whichgives priority to food and fiber production and allows plantation tobe expanded only to abandoned agricultural land and wasteland(Smeets and Faaij, 2007b; Hoogwijk et al., 2009; van Vuuren et al.,2009, 2010; Beringer et al., 2011; IPCC, 2011).Plantations yields are based on NPP maps and own calculations, asdescribed in Havlík et al. (2011) (Table 4). Woody biomass supply fromplantations is not separated to stemwood and other tree parts becauseenergy crop plantations usually utilize total tree biomass and do notproduce any logging residues. Moreover, the fraction of stemwoodfrom the total tree biomass in plantations is 80–90% (FAO, 2007),which makes the share of other tree parts relatively small importance.Please cite this article as: Lauri, P., et al., Woody biomass energy potential in 2050. Energy Policy (2013), http://dx.doi.org/10.1016/j.enpol.2013.11.033i

P. Lauri et al. / Energy Policy ( ) – Table 4Average yields for plantations (m3/ha/year).WorldEU27RussiaAfricaAsiaNorth AmericaSouth America20.316.46.624.424.516.130.55household fuelwood is often used for cooking and cannot bereplaced by fossil fuels, which would require investments in newstoves (Arnold et al., 2010; IARC, 2010; May-Tobin, 2011). Hence,fossil fuels are not substitutes for household fuelwood in developing countries where most household fuelwood consumptionoccurs. Moreover, household fuelwood is often woody biomassthat is not used for energy wood for technical or economicreasons, that is, unrecovered logging residues. Hence, householdfuelwood does not have a direct connection to fossil fuels or largescale energy wood markets.2.4. Woody biomass production costs2.6. Forest industry technologies and recycled woodWoody biomass production costs consist of harvest and transport costs. Harvest costs are modeled by using spatially explicitconstant unit costs. Transport costs are not spatially explicit butare modeled by using regional level constant elasticity transportcost functions, which approximate the short run availability ofwoody biomass in each region.9 Transport costs functions areshifted over time in response to the changes in the harvestedvolumes such that in the long run only average transport costsmatters. There are several institutional factors like building forestroads and transport capacity limitations that make woody biomasssupply less elastic in the short run than in the long run (Binkleyand Dykstra, 1987).Harvest costs include planting, logging, and chipping in thecase of logging residues. Harvest costs for forests are based on theG4M model and they vary in the range of 10–40 /m3 dependingon the region and the steepness of terrain. Harvest costs forplantations are based on own calculations, as described in Havlíket al. (2011), and they vary over a range of 5–30 /m3 depending onthe region and the steepness of terrain.Transport costs include costs of accessing and moving woodybiomass from forests or plantations to domestic production unitslike sawmills, pulp mills and energy plants. Transport costs arebased on recursive regional level constant elasticity cost functions,which are parameterized by previous period harvested volumesand average transport costs from forest to mills. Average transportcosts are assumed to be 5–15 , depending on the region, based onHakkila (2004), Hamelinck et al. (2005) and regional adjustment.The elasticity of the transport cost function is assumed to be 0.1–0.5based on the country/regional level estimates of the price elasticityof industrial roundwood supply (Buongiorno et al., 2003). In thecase of plantations, the elasticity is assumed to be 1 becauseplantations are usually located close to production units andtherefore have lower transport costs than forests.2.5. Woody biomass demand for material products andhousehold fuelwoodThe forest sector has seven final products (chemical pulp,mechanical pulp, sawnwood, plywood, fiberboard, other industrialroundwood, and household fuelwood). Final product demands aremodeled by using regional level constant elasticity demand functions, which are parameterized by consumption quantities fromFAOSTAT (2013), price data from Buongiorno et al. (2003), priceand income elasticities from Buongiorno et al. (2003), and population and GDP growth data from the POLES energy sector model(Havlík et al., 2011).The demand function for household fuelwood is assumed to beindependent of energy wood and primary energy prices. Thereason for this simplification is that in low income countries9The model has recently been extended to include spatially explicit transportcosts by combining different infrastructure data sets (Mosnier et al., 2012).However, this extension concerns only the Congo Basin area. Extending the analysisat the global level remains a subject of future study.Forest industry final products (chemical pulp, mechanical pulp,sawnwood, plywood and fiberboard) are produced by Leontiefproduction technologies. Input-output coefficients for Leontieftechnologies are based on the engineering literature (e.g., FAO,2010). By-products of these technologies (bark, black liquor,sawdust, sawchips) can be used for energy production or as rawmaterial for pulp and fiberboard (Fig. 2).Initial production capacities for forest industry final productsare based on production quantities from FAOSTAT (2013). After thebase year the capacities evolve according to investment dynamics,which depend on depreciation rate and investment costs. Theinvestment pay-back time is assumed to be 10 years, which is thetime step of the model. The annual depreciation rate is assumed tobe 0.03, which approximates to 30 average capital operating times.Capital operating times and investment costs are based on theengineering literature (e.g., Diesen, 1998).Some forest industry final products (sawnwood, plywood, andfiberboard) can be recovered as recycled wood after their finalconsumption. Recycled wood can be used for energy production orraw material for fiberboard. Recycled wood recovery variesbetween 20% and 50% of the final product consumption dependingon the country (Mantau et al., 2010). Hence, it is assumed that themaximum available amount of recycled wood is 50% of sawnwood,plywood, and fiberboard consumption.2.7. ScenariosWe examine two different scenarios: baseline and environmentscenarios. In the baseline scenario all forest areas are allowed to beused for harvesting (i.e., we do not exclude protected or primaryforests from the available forest areas). The environment scenarioassumes that primary forests are set aside for protection (i.e.,primary forests cannot be harvested or deforested). For eachscenario we solve energy wood supply curves using GLOBIOMand analyze their implications for the forest and energy sectors.Energy wood supply curves display the amount of woody biomassavailable for large-scale energy production at various hypotheticalenergy wood prices.Energy wood supply curves are generated as follows. In the baseyear (2000) energy wood demand is based on IEA data regardingsolid biomass use for energy. After the base year, energy wooddemand is replaced by hypothetical energy

Woody biomass energy potential in 2050 Pekka Lauri, Petr Havlík, Georg Kindermann, Nicklas Forselln, Hannes Böttcher, . We examine woody biomass energy potential by partial equilibrium model of forest and agriculture sectors. It is possible to satisfy 18% (or 14% if primary forests are excluded) of the world's primary energy consumption in .