Availability assessment of bioenergy and power plant location optimization: A case study for Pakistan

Renewable and Sustainable Energy Reviews 42 (2015) 700–711

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser

Availability assessment of bioenergy and power plant location optimization: A case study for Pakistan Markus Biberacher a,n, Markus Tum b, Kurt P. Günther b, Sabine Gadocha a, Peter Zeil c, Rehmatullah Jilani d, Muhammad Mansha d a

Research Studios Austria (RSA), Studio iSPACE, Schillerstrasse 25, A-5020 Salzburg, Austria German Aerospace Center (DLR), German Remote Sensing Data Center (DFD), Oberpfaffenhofen, D-82234 Wessling, Germany c University of Salzburg, Department of Geoinformatics-Z_GIS, Schillerstrasse 30, A-5020 Salzburg, Austria d Pakistan Upper Atmosphere Research Commission (SUPARCO), Earth Science Directorate, P.O. Box 8402, Suparco Road, Karachi PK-75270, Pakistan b

art ic l e i nf o

a b s t r a c t

Article history: Received 12 August 2013 Received in revised form 21 August 2014 Accepted 18 October 2014

In terms of land use competition, bioenergy is often subject to controversial discussions. This paper presents a study, which addresses the scope of geographic discrete biomass growth, optimal bio-energy plant locations, and related biomass supply areas. In a first step, annual biomass growth is calculated with the BETHY/DLR model on a spatial resolution of 1 km2. In a second step, the ASECO model is utilized to identify optimal plant locations with related biomass supply areas, determined by biomass growth rates and available road infrastructure. The case study is carried out for Pakistan. Scenarios have been investigated on district level, with a special focus on total supply areas for single power plants at identified locations, as well as supply area deviations over the years due to varying biomass growth rates. Achieved results are relevant for the political debate on an optimal bioenergy strategy for Pakistan. & 2014 Elsevier Ltd. All rights reserved.

Keywords: Spatial modeling Linear optimization Bioenergy Power plant assessment

Contents 1. 2.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Method and models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. BETHY/DLR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. ASECO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Results and discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction Access to energy and a steady, stable energy supply are important requirements for socio-economic development. During the last decades, global energy resources—mainly fossil fuels—were massively exploited, to meet the growing energy demands of the growing world population, and the increase of economic welfare [33]. When

n

Corresponding author. Tel.: þ 43 662 90858 5221; fax: þ 43 662 908585 299. E-mail addresses: [email protected] (M. Biberacher), [email protected] (M. Tum), [email protected] (K.P. Günther), [email protected] (S. Gadocha), [email protected] (P. Zeil), [email protected] (R. Jilani). http://dx.doi.org/10.1016/j.rser.2014.10.036 1364-0321/& 2014 Elsevier Ltd. All rights reserved.

700 701 701 703 706 708 710 710

discussing future perspectives of energy availability, two concerns usually come in focus. One is carbon output and its influence on climate change. The other is the question of whether existing resources will meet the projected increase in energy demand, especially in non-OECD countries. When dealing with bioenergy, a third concern needs to be discussed: the dilemma of “food versus fuel”. However, although widely discussed, the finite answer to this question still remains open [14]. This paper presents an analysis for the energyfood nexus, with a special focus on energy availability and sustainable resource exploitation. Recent studies showed that areas available for bioenergy production until 2050 are highly dependent on the scenario considered. The scenarios presented in Thrän et al. [36] mainly found substantial

M. Biberacher et al. / Renewable and Sustainable Energy Reviews 42 (2015) 700–711

decreases of potential areas for bioenergy production. In contrast, Zeddies et al. [42] report increasing areas for bioenergy of up to 440.5 million ha until 2050. Both studies investigated scenarios based on the continuation of current business, diets, etc., and a variety of other scenarios comprising substantial changes in the behavior. A similar study was also conducted by Poudel et al. [30], who found comparable patterns. Further studies focused on the assessment of globally available and projected bioenergy potentials from agricultural residuals, and report a range of 27–77 EJ yr  1 [6,12,10]. When looking at regional scales, a realistic assessment of available biomass potentials becomes even more relevant. This is especially true for countries with low average incomes, since aside from access to energy, nutritional security might be limited. One example is Pakistan, with a per capita income 2011: US$1201 ([16]), on which our study is focused. Being based on an agricultural economy, the share of agriculture and fishery related employees in Pakistan was about 45.1% in 2011, contributing to the gross domestic production ([29]) with 21.1%. Pakistan is one of the ten largest wheat and the five biggest sugar cane producers. Thus huge amounts of straw and bagasse are available each year. Especially in rural areas the population (67%) is historically dependent on biomass use. A study by Mirza et al. [26] estimated an average household to consume of 2325 kg of firewood, 1480 kg of dung, or 1160 kg of crop residues per year, mainly for heating and boiling. The consistently growing urban population, however, predominately use fossil energy carries, resulting in costs which already consume 20% of the total foreign exchange [32]. With the current energy share and consumption, Pakistan is at the point where energy demand exceeds supply, which inevitably will lead to a decrease of national growth [2]. Projections estimated a tripling of the current energy consumption by 2030, which will require a massive increase of energy imports ([28]). Thus the use of renewable energy sources, especially biomass, is seen as vital to meet increasing energy demands [2]. Pakistan's government decision to increase the share of renewables until 2030 from 0 to 2.5% will save the country up to 400 million US$ [32]. The Energy Security Plan [20] suggests increasing biogas production to 4000 MW by 2030. Presently more than 5000 small sized biogas units are installed, which cannot even exploit 1% of the estimated biogas potential of 12–16 million m³ d  1 [3]. In 2010, the Alternative Energy Development Board (AEDB) was established by the Government of Pakistan to develop a national strategy, policies, and plans for the utilization of alternative and renewable energy resources. At present, six biomass power plants with a capacity of 9 to 12 MW are planned to be established ([1]). Three of them are designed to be exclusively driven by agricultural waste (i.e. straw, bagasse). The availability of agricultural side products depends on its use competitions (e.g. animal housing, soil fertilization). Furthermore, the efficiency of such biomass power plants is highly dependent on transport costs. Thus it is necessary to estimate the technical availability of biomass, and assess economically reasonable locations for a power plant. Recent studies proposed various methods to quantify biomass/bioenergy availability. Jiang et al. [17] proposed a GIS based approach and calculated bioenergy potentials from crop residues for China. Scarlat et al. [34] and Herr and Dunlop [15] used statistical methods and performed studies for Romania and Australia. Tum et al. [39] used a vegetation model to assess annual Net Primary Productivity (NPP) for Germany, and calculated energy potentials for straw using conversion factors on the yield-to-straw ratio. A further approach to assess this topic is provided by earth observation. Earth observation and in-situ data can be integrated with ground surveys, historical records and market information, and assimilated with geographic information systems (GIS) analytical tools to illuminate short- and long-term trends on bioenergy demand and supply, at regional and local scales. Optimization approaches are currently developed to find suitable locations for e.g. bioethanol refineries or biomass power plants.

701

Leduc et al. [23] used a mixed integer linear programming model and focused on optimal locations of lignocellulosic ethanol refineries in Sweden, which was also used for Finland by Natarajan et al. [27]. The model is based on a minimum cost approach, covering all levels of biofuel production and the production chain, like biomass and methanol production, transport, and investments for the production plants and gas stations [24]. A similar approach was developed by Celli et al. [7], who did a study for Sardinia, Italy.

2. Method and models The current study was subdivided into two tasks. First we assessed time-series of bioenergy potentials (2001–2010) from agricultural side products for Pakistan, on a spatial resolution of 1 km². For this we used the Biosphere Energy Transfer Hydrology (BETHY/DLR) model, operated at the German Aerospace Center (DLR), to calculate time series of Net Primary Productivity (NPP). NPP was validated using empirical data on acreage and yields, as already proposed by Tum and Günther [37]. NPP was then transferred to energy potentials, using country specific conversion factors on e.g. the above-to-below ground biomass, yield-to-straw ratios and lower heating values. In a second step we introduced the new biomass power plant location optimization tool ASECO, to identify potential power plants and their supply range. Following the suggestions of [1], we will thus contribute to the energy planning directive of Pakistan. 2.1. BETHY/DLR BETHY/DLRis a Soil–Vegetation–Atmosphere–Transfer (SVAT) model. It can be used to track the transformation of atmospheric carbon dioxide into energy storing sugars, a process known as photosynthesis. BETHY/DLR has recently been used to assess Net Primary Productivity (NPP) for parts of Europe and Asia [9,37], and has been validated and cross-compared with other process based models for agriculture [38] and eddy covariance data [40]. The scheme of modeling photosynthesis with models like BETHY/DLR is widely accepted and serves as an input to global dynamic vegetation models, such as the Jena Scheme of Atmosphere Biosphere Coupling in Hamburg (JSBACH) by Knorr and Heimann [21], and the Lund–Potsdam–Jena (LPJ) by Prentice et al. [31] and Bondeau et al. [5]. Photosynthesis is modeled using the integrated approach of Farquhar et al. [11] and Collatz et al. [8]. Following this approach, the enzyme kinetics are parameterized on a leaf level, taking into account the different behavior of C3 and C4 plants in their way of carbon fixation. Since C4 plants are able to fix more atmospheric carbon dioxide at higher temperatures than C3 plants, and Pakistan yields on average (2000–2010)  50.4 million tons sugar cane (C4 plant), and  20.8 million tons of wheat (C3 plant), this distinction needs to be taken into account. The rate of photosynthesis is then extrapolated from leaf to canopy level, respecting both the canopy structure and the interactions of vegetation with soil and atmosphere. Sellers' [35] two-flux approach is used to consider the absorption of radiation in the canopy. A detailed parameterization of the processes of evapotranspiration, stomatal conductance, soil water balance, and autotrophic respiration is also included in the model formulation. Details on the model approach can be found in Knorr and Heimann [21]. BETHY/DLR is driven by time series of the Leaf Area Index (LAI), derived from remote sensing and meteorology. LAI data is taken from geoland2, and is globally available from 1999 onwards at a 1 km² resolution. The data is provided free of charge as 10-day composites. We pre-processed the data to eliminate outliers and gaps, using time series analysis. For this we used harmonic analysis,

702

M. Biberacher et al. / Renewable and Sustainable Energy Reviews 42 (2015) 700–711

which is based on the least-squares technique [25,4]. The meteorological data are provided by the European Center for MediumRange Weather Forecast (ECMWF). These data comprise six-hourly data of temperature at 2 m above the ground, wind-speed at 10 m above the ground, cloud cover (high, medium, low fraction), and twelve-hourly data of precipitation. The cloud cover fractions are used to calculate the fraction of photosynthetic active radiation (fAPAR). In addition, static maps of elevation, land cover / land use, and soil types are needed. We used the SRTM (elevation), the Harmonized World Soil Database (soil types), and the GLC2000 (land cover), which are all freely available at a 1 km² resolution. In addition to the GLC2000, we used a land cover / land use dataset, which was provided by SUPARCO [22], to increase the accuracy of available agricultural areas. SUPARCO used Landsat ETM data to assess Pakistan's land use for 2006. The data was provided as a shape file, which needed to be rasterized to the resolution of GLC2000 (1 km²) using a geographic information system (GIS). Since the two datasets are valid for two different periods, and report agricultural areas differently (SUPARCO: 22,922 km²; GLC2000: 26,328 km²), we followed a two-step decision regarding which areas to integrate in our study. We assumed areas that are classified as agricultural areas in both datasets to be permanently reserved for agricultural management. Areas with disagreement were not taken into account for the further study, because they mainly occur in arid and mountainous areas (additional pixels in GLC2000), which might result from misclassifications. By contrast, the SUPARCO land cover map mainly reports additional agriculture for a triangular area of 75 km  150 km  150 km between the Chashma Reservoir (Northwest), Khushab (Northeast) and Multan (South). This area was also not regarded in this study, because it belongs to the Thal desert and thus misclassification in the SUPARCO land cover map is assumed. The main outputs of BETHY/DLR are time series of Gross Primary Productivity, NPP, evapotranspiration, and soil water content in daily steps with the spatial resolution of the respective land cover classification. A more detailed model description of BETHY/DLR can be found in Wißkirchen et al. [40]. Since agriculture in Pakistan is dominated by irrigation, but no further pixel-wise information was available, we assumed ideal irrigation to be applied on all agricultural areas. However, the amount of irrigation was calculated individually for each grid cell, assuming ideal irrigation (IR) to be the difference between the soil water available for plants (W) and the vegetative evapotranspiration (EP) demand. IR can thus

be expressed as: IR ¼ EP–W

ð1Þ

This assumption allows vegetation to grow under ideal conditions. To transfer the modeled annual sums of NPP to energy potentials of crop residues (i.e. straw), further processing of the data had to be performed. According to the approach of Tum et al. [39] the straw content of NPP can be determined using statistical information on acreage and mean yields, and combining them with relationships of the above- and belowground biomass and corn to straw ratios. To consider use competitions of straw (e.g. traditional use for animal housing and feeding, soil re-fertilization, etc.), a conservative share of 20% was assumed to be available for energetic use. This value was taken from Kaltschmitt et al. [18] and is assumed to represent a sustainable share. The sustainable share used in the study was

Fig. 2. Transport costs scheme from grid cell to grid cell.

Fig. 1. Balance flow examples between supply grid cells (S) and demand grid cells (D) via transport grid cells (T) for two possible plant locations with varying installation sizes (red shaded).

M. Biberacher et al. / Renewable and Sustainable Energy Reviews 42 (2015) 700–711

703

Fig. 3. Pre-selected biomass power plant sites (green asterisks) in the regions Faisalabad and Jhang. In the background relevant GLC2000 land use classes and VMAP streets are presented (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

therefore calculated by multiplying the total straw energy content with a factor of 0.2. Statistical data of the agricultural land use was taken from the Pakistan Bureau of Statistics, of the Government of Pakistan (www. pbs.gov.pk). Statistical yearbooks in Pakistan are published on an annual basis since 1952, and provide comprehensive overviews of different socio-economic aspects of the country. The most recent publication from 2011 includes data for the last ten years. Besides agriculture related topics it also covers transport and communications, money and credit, public finance, and further datasets. The agricultural surveys contain information about arable land with associated yields and are available on the province level. Once the straw content is available, lower heating values can be applied to calculate the energy potential given in TJ km² and year. Lower heating values for various cereals are presented in Kaltschmitt and Hartmann [19]. Since the range of lower heating values of cereals is very small (17.0–17.7 MJ kg  1), and wheat can be seen as the main cereal in Pakistan, we used the value of wheat (17.2 MJ kg  1) as a generalized lower heating value. These results serve in a further step as input for the ASECO model.

2.2. ASECO To identify optimal biomass power plant locations, under the prerequisite of ideal biomass supply areas, we applied and extended the Autarkic Spatial Energy Cluster Optimization (ASECO) approach (see Zaliwski et al. [41]). ASECO was initially developed to identify self-sufficient supply areas for renewables. The smallest spatial units in the model are assumed to be individual grid cells in a regular raster. The model is implemented in GAMS (general algebraic modeling system) with CPLEX as linear solver. With the extended ASECO model, the identification of optimized transport flows from single grid cells to neighboring grid cells among an entire region to meet possible geo-located demands is enabled (see Fig. 1). In a mixed integer linear optimization process, ASECO determines optimal locations for biomass power plant installations, related optimal individualized power plant size, and required biomass supply areas, in order to reach a predefined overall biomass power plant capacity in the region of interest. As ASECO is a grid based model, it pays attention to spatially discrete input data of available biomass potential (pij) for energetic

704

M. Biberacher et al. / Renewable and Sustainable Energy Reviews 42 (2015) 700–711

purposes, the individualized transportation cost (cij) from one grid cell (ij) to neighboring grid cells, and possible biomass power plant locations (lij) from a pre-selection process. As model variables, the transport (tnij) to neighboring grid cells (n ¼ 1–8) and the biomass power plant installation size in annual energy units (sij) are considered. Two alternative objective functions can be used in the ASECO model for the identification of power plant locations (2.1 and 2.2), which are formulated as follows: minimize obj ¼ ∑ij ∑8n ¼ 1 t outnij cij

ðto minimize transport effortÞ ð2:1Þ

maximize obj ¼ ∑ij sij

ðto maximise power plant capacityÞ ð2:2Þ

The constraints to be satisfied are formulated as follows: pij þ ∑8n ¼ 1 t innij Zsij þ ∑8n ¼ 1 t out nij ∑ij sij Z Sreg

only for objective function 2:1

TCost Z ∑ij ∑8n ¼ 1 t out nij cij sij ¼ sij lij

for all grid cells ij

only for objective function 2:2

for all grid cells ij

ð3Þ ð4Þ ð5Þ ð6Þ

Fig. 4. Calculated sustainable energy potential of straw for Pakistan for the period 2001–2010. Red color represents high values, yellow color moderate and blue color low energy values. White areas represent non-agricultural areas (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

M. Biberacher et al. / Renewable and Sustainable Energy Reviews 42 (2015) 700–711

705

Fig. 4. (continued)

sij Z Smin

for all ij where sij 4 0

ð7Þ

where obj TCost pij cij lij t in t out

objective of the model; the overall biomass transport cost; the usable biomass potential in a grid cell ij; the specific transport cost for one biomass unit from a grid cell ij to a neighboring grid cell; the suitability (0/1) of grid cell ij as a location for a power plant installation the biomass transport streams into a grid cell; the biomass transport streams outwards from a grid cell.

Sreg Smin

the overall biomass power plant capacity size, and the minimum capacity size to be established at a single location.

The objective function either minimizes occurring biomass transport costs (2.1) or maximizes occurring power plant capacities (2.2). The model calculates either a solution where the transport costs are minimal among the entire model region for a given overall power plant capacity, or a solution where the installed power plant capacity to be satisfied with biomass among the entire region is maximal for a given transport cost threshold. Eq. (3) guarantees the biomass balance within a grid cell ij, considering the local biomass potential, all incoming and outgoing biomass streams, and possible

706

M. Biberacher et al. / Renewable and Sustainable Energy Reviews 42 (2015) 700–711

“biomass demand” in form of a biomass power plant, to always be positive. Eq. (4) controls the decision of plant installations at individual locations ij, and is satisfied if the predefined minimum biomass power plant capacity is guaranteed. Eqs. (5) and (6) limit the power plant installation locations to the pre-selected locations and guarantee that selected installations are beyond a predefined capacity threshold which can be defined individually. Corresponding to the grid cell size of the BETHY/DLR results, the raster cell size for ASECO was set to a 1 km2 resolution. The ASECO model has been set up to identify optimal biomass plant locations out of a list of pre-selected possible plant locations. The optimization is carried out from the perspective of surrounding biomass availability and corresponding transport costs. The transport effort is assumed to scale linearly with the transported amount of biomass over a transported distance along a chosen transport route, which is triggered by a correlated specific transport effort. Although this is a fairly rough simplification, it describes the reality quite well in a first approximation—and allows the problem to be formulated as linear optimization problem. Specific location-dependent transport costs were estimated as an indicator, using the national road grid in Pakistan, which is based on the VMAP dataset (http://gis-lab.info/qa/vmap0-eng.html). It can be assumed that both the transport costs and efforts on wellestablished streets, such as major roads and highways, is less than on secondary streets. Thus we assumed transportation on secondary roads to have 1.5 times the cost of that on primary roads. Grid cells without direct connection to streets were labeled with a significant higher cost factor of 4 times the cost of primary roads (see Fig. 2). To identify possible biomass power plant locations, politically motivated guidelines ([1]) were used, which have been combined with a land cover land use map (GLC2000). Since AEDB reports five possible regions for potential power plants, each of which should have 9–12 MW capacity, we focussed our assessment on these regions (Badin, Faisalabad, Jhang, Mardan and Mirpurkhas,).

Due to data processing limitations with each single grid cell as possible plant location, a list of plant locations was pre-selected. The list consists of a number of individual grid cells with higher suitability for biomass power plants, by assuming places with good transport infrastructure, high energy demand (close to densely populated areas), or high biomass potentials to be most suitable. Fig. 3 shows the pre-selected possible locations for the two regions Faisalabad and Jhang.

3. Results and discussion The results of the study show differences in the spatio-temporal distribution of bioenergy potentials. In Fig. 4, the average (2001– 2010) of sustainable bioenergy, and the annual deviations (year minus average) for the period 2001 to 2010 is outlined on a km2 resolution. Statistical analysis revealed an average energy potential of 0.72 TJ km  2 a  1. This comes with a maximum of 10.8 TJ km  2 a  1. 2002 was the year with lowest overall energy potential (60.1 PJ) and 2008 the year with the highest energy yields (71.4 PJ). For all years the highest values can be found in the North-Eastern parts of Pakistan in an area close to the Indian border with the margins of the river Jhelam in the North, and the river Satluj in the South. Further areas with high potentials are the banks of river Indus. These areas are characterized by intensive agricultural use. Further but lower energy potentials exist in the area between the river Indus and Chenab / Jhelam, the northern parts of Punjab, and the area East of Karachi. A map showing Pakistan and its main rivers and minimum and maximum area consumptions of 5 biomass power plants is presented in Fig. 5. A reason for this bioenergy distribution can be found in agricultural practices, which are highly dependent on water availability. Since we assumed irrigation to be applied in all regions of Pakistan, a closer look to the vegetative development is required. Fig. 6 shows the

Fig. 5. Pakistan with main rivers and minimum and maximum area consumptions of 5 biomass power plants identified with the ASECO model. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

M. Biberacher et al. / Renewable and Sustainable Energy Reviews 42 (2015) 700–711

707

Fig. 6. Averaged cumulative annual LAI for the period 2001–2010 for Pakistan. Low values are represented in red to orange; medium in yellow and green; and high values in blue. White areas represent non-agricultural areas (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 7. Area consumption of the biomass supply for the five biomass power plants in Feisalabad, Jhang, Mardan, Mirphurkas and Badin in km². Diamonds represent the median of area consumption. Blue diamonds represent 12 MW and purple diamonds 9 MW power plants. In addition maximum and minimum areas (bars) are given (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

708

M. Biberacher et al. / Renewable and Sustainable Energy Reviews 42 (2015) 700–711

average annual cumulative LAI (Leaf Area Index) for Pakistan. It can be deduced that areas with low cumulative LAI are areas with low energy potentials, and vice versa. This finding leads to the assumption that regions with low cumulative LAIs either have limited access to water (irrigation), or are managed differently in terms of fertilization, planting scheme, etc. The zoom windows in Fig. 5 outline the plant locations identified as optimal with the ASECO model and related biomass supply areas for the five districts. The shape and dimension of supply areas among districts are triggered by the underlying road net and distribution of biomass availability. Power plant locations which are sparsely surrounded with biomass potentials (see district 1 or 5) cause a highly scattered supply area, which represents high collection and transportation efforts for biomass supply. Districts with higher and more equally distributed biomass growth potential around a power plant location result in nearby and compact supply areas (see district 3). Fig. 5 also shows maximum and minimum supply areas which are based on varying annual biomass growth within the considered 10 year time horizon. While the additional transport effort for the collection of biomass in years of low growth is negligible for districts with high and nearby biomass potentials (see district 2 or 3), new collection routes have to be established for districts with spatially scattered biomass potentials during these years (see district 1 or 5). Looking at the area consumption of the five regions and its variability (Fig. 7) it becomes apparent that the efficiency in terms of supply area minimization strongly depends on the location of the power plant. The two power plants in Feisalabad and Jhang consume the lowest area (median 236 km² a  1 and 224 km² a  1) with only little variation during the ten years (see also Table 1). Power plants in slightly more unfavorable areas, like Mirphurkhas, show both a higher median and variation. The Badin station, which was optimized for 9 MW, requires the largest supply area (median: 312 km² a  1]; maximum 411 km² a  1), because of the low biomass output in this area (see Fig. 4). Aside from the location and supply area optimization for biomass power plants on a district level in the five outlined districts, three scenarios were also calculated for the complete area of Pakistan. Because of the computational effort, these scenarios are based on the biomass growth potential of 2010 only. 174 possible power plant locations have been pre-selected from 57 out of 115 districts where significant biomass potential is available in Pakistan. This results in at least 2 competing possible plant locations for each of these districts. With the ASECO model for each of these districts, the most optimal site for the installation of a 12 MW biomass power plant has been chosen in Scenario 1 (see Fig. 8), summing up to 684 MW installed biomass power plant capacity for the entire country. It is based on the objective to find the minimal transport effort in arbitrarily defined units (arb.units), as outlined in Eq. (2.1) and accounting to 83,483 arb.units. Supply areas in the northern and southern districts are larger than in the central districts due to lower biomass potentials around identified power plant sites. The area required to satisfy the correlated biomass demand would correspond to 18,136 km². On average 275 km² (7192 km²) are required to supply a 12 MW power plant. The high standard deviation is due to the fact that significantly larger (maximum supply area: 1414 km²; Attok) or smaller (minimum supply area: 130 km²; Rahimyar Khan) areas are required for individual power plants (see Fig. 8). In total, 11.7 Mt (average: 206 kt) of straw biomass are required to feed the identified biomass power plants, which is about 30% more than the estimate by [1], and can thus be seen as in good concordance. Although the result indicates optimized biomass power plant locations for 57 districts, some of them can be seen as not preferable because of their area consumption, and thus transportation effort. This is mainly due to the goal of identifying an optimal power plant for each district, which has a connection to biomass potentials.

Table 1 Median, maximum and minimum area consumption to supply bioenergy power plant stations. Station

Median [km² a  1]

Minimum [km² a  1]

Maximum [km² a  1]

Feisalabad Jhang Mardan Mirphurkhas Badin

236 224 282 316 312

189 203 245 249 265

255 252 327 375 411

In Scenario 2 the restriction of a power plant in each district with significant biomass potentials has been eliminated, and the model endogenously chose the most optimal locations among the entire country to reach the condition of 684 MW installed biomass power plant capacity. In such a scenario, locations which are in or near areas with high biomass potentials are identified as optimal power plant sites (see Fig. 8—Scenario 2). Instead of 57 single power plants with a size of 12 MW each (Scenario 1), only 56 power plants are chosen by the model in Scenario 2. For the compensation of the capacity gap, six power plants with a greater capacity between 12 MW and 15.4 MW are assigned at locations with higher biomass potentials by the ASECO model. The required overall area to satisfy the biomass demand would decrease to 11,620 km², and the resulting optimized transport effort amounts to 64,564 arb.units. In a third scenario (see Fig. 8—Scenario 3), the objective function was changed from minimizing the transport effort for the given overall plant capacity, to maximizing the overall biomass power plant capacity for a given transport effort threshold amounting to 73,200 arb.units. In scenario 3 the required biomass harvesting area would sum up to 13,369 km², assigned to 63 individual power plant locations. The total maximized installed power plant capacity would be 761 MW instead of 684 MW, realized mainly by 12 MW power plants and only in the high biomass growth rate areas with plant sizes beyond 12 MW. However, since 2010 was a year with average sustainable bioenergy potentials, our scenario results can be seen to be at the lower edge of possible bioenergy potentials.

4. Conclusion Earth observation and in-situ data are integrated with ground surveys, historical records, and market information, and are assimilated with geographic information systems (GIS) analytical tools to illuminate short- and long-term trends in bioenergy demand and supply, at regional and local scales. The methodology presented for Pakistan demonstrates that the integration of different thematic data layers can yield essential information for energy planning—if these data are available and accessible. For the presented study we modeled Net Primary Productivity for Pakistan for the period 2001–2010 on a 1 km², using the vegetation model BETHY/DLR. We transferred the NPP for agricultural areas to energy potentials of straw using conversion factors and lower heating values, also taking into account use competitions of the straw. In a second step we used the ASECO model to identify preferable bioenergy power plant locations and their supply areas out of a list of pre-selected possible locations. We first focused our research on five districts for which results have been produced for the whole observation period (2001–2010). We demonstrated that it is not feasible for all five districts to build a bioenergy power plant. In a second step we expanded our approach to all districts of Pakistan (Scenario 1–3), which have significant access to biomass potentials. In these scenarios we based our calculation on the biomass potentials as calculated with BETHY/DLR for the year 2010. For 57 out of 115 districts the ASECO model, identified in Scenario 1, optimized power

M. Biberacher et al. / Renewable and Sustainable Energy Reviews 42 (2015) 700–711

709

Scenario 1: Each district with significant biomass potential is assumed to participate with one biomass power plant to the overall Bioenergy target.

28

Scenario 2: Only the most suitable biomass plant locations among the whole country contribute to the overall Bioenergy target.

Fig. 8. Scenarios on optimal biomass power plant locations (yellow dots for 12 MW power plants and turquoise dots for power plant capacities beyond) with assigned biomass supply areas (red areas) over all districts in Pakistan with relevant biomass potentials (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

710

M. Biberacher et al. / Renewable and Sustainable Energy Reviews 42 (2015) 700–711

Scenario 3: Objective has been switched from minimal transport effort to maximal power plant capacity at fixed transport effort threshold.

Fig. 8. (continued)

plant locations and supply areas, consuming a total supply area of 18,136 km² and biomass of 11.7 Mt. It is interesting to notice that in Scenarios 2 and 3—decoupled from district constraints and mainly triggered by biomass availabilityonly the district Jhang, out of the 5 initially investigated districts, would contribute to the achievement of a national bioenergy target due to relevant biomass potential. The study illustrates a novel method to determine optimal biomass power plant locations and their corresponding supply areas. The results will be useful in estimating the cost and area consumption of such installations, to assist regional and national planning efforts for a secure and stable energy provision. However large-scale energy crops may induce landuse changes that put food security at risk, or lead to the destruction of natural ecosystems [13]. Analytical tools, such as the ones presented here, allow planners to visualize existing and potential bioenergy resources, and to determine viable solutions at regional or national scales. In this study we used a coarse resolution of 1 km², however, this method could also be used for an assessment with significantly higher resolution, if all required datasets are available. Therefore it can be applied to any region around the world. In addition, better information on transportation costs could increase the accuracy of our estimates. The methodology exists, but can only produce effective results if data are maintained and accessible in a national spatial data infrastructure (SDI). National governments and donor funded development initiatives—especially in renewable energy use or energy efficiency—need to invest in SDIs to support the assessment of potential & impact of bioenergy resource use, and to provide the basis for effective evaluation of the progress made in supplying energy for a secure and stable energy supply. The Group on Earth Observations (GEO), as a voluntary global partnership of governments and international organizations, has the mandate and the

appropriate governance structure for supporting renewable energy planning worldwide. GEO offers the institutional framework for coordination and cooperation between multiple stakeholders, and facilitates the access to technical skills and methodologies required to assess potential and impact of bioenergy resources. An example would be the Bioenergy Atlas for Africa (BAfA) initiative which is ongoing since 2009 and supported by GEO and AfriGEOSS. The Global Earth Observation System of Systems (GEOSS) has among its goals to provide Earth observations to a wide variety of users together with the infrastructure and interoperability standards for free and open data sharing. The case study for Pakistan demonstrates the potential of globally available data and the requirements for data sharing principles. Acknowledgments This study was conducted under the FP7 project EnerGEO (Grant agreement no.: 226364). We thank ECMWF, geoland2 and VITO for providing data. We further acknowledge the active support by Mr. Imran Iqbal, SUPARCO and member of Pakistan's delegation to COPUOS. References [1] Alternative Energy Development Board (AEDB). 〈http://www.aedb.org〉 (Last visit: 31.1.13. [2] Amjid SS, Bilal MQ, Nazir MS, Hussain A. Biogas, renewable energy resource for Pakistan. Renew Sustain Energy Rev 2011;15:2833–7. [3] Bhutto AW, Bazmi AA, Zahedi G. Greener energy: issues and challenges for Pakistan-biomass energy prospective. Renew Sustain Energy Rev 2011;15:3207–19. [4] Bittner M, Offermann D, Bugaeva IV, Kokin GA, Koshelkov JP, Krivolutsky A, et al. Long period/large scale oscillations of temperature during the DYANA campaign. J Atmos Terr Phys 1994;56:1675–700.

M. Biberacher et al. / Renewable and Sustainable Energy Reviews 42 (2015) 700–711

[5] Bondeau A, Smith PC, Zaehle S, Schaphoff S, Lucht W, Cramer W, et al. Modeling the role of agriculture for the 20th century global terrestrial carbon balance. Glob Chang Biol 2007;13:679–706. [6] Campbell JE, Lobell DB, Genova RC, Field CB. The global potential of bioenergy on abandoned agricultural lands. Envion Sci Technol 2008;42:5791–5. [7] Celli G, Ghiani E, Loddo M, Pilo F, Pani S. Optimal location of biogas and biomass generation plants. In: Proceedings of the 43rd international universities power engineering conference; 2008. p. 1–6. [8] Collatz GJ, Ribas-Carbo M, Berry JA. Coupled photosynthesis—stomatal conductance model for leaves of C4 plants. Aust J Plant Physiol 1992;19:519–38. [9] Eisfelder C, Klein I, Niklaus M, Kuenzer C. Net primary productivity in Kazakhstan, its spatio-temporal patterns and relation to meteorological variables. J Arid Environ 2014;103:17–30. [10] Erb K-H, Haberl H, Plutzar C. Dependency of global primary bioenergy crop potentials in 2050 on food systems, yields, biodiversity conservation and political stability. Energy Policy 2012;47:260–9. [11] Farquhar GD, von Caemmerer S, Berry JA. A biochemical model of photosynthesis in leaves of C3 species. Planta 1980;149:58–90. [12] Field CB, Campbell JE, Lobell DB. Biomass energy: the scale of potential resource. Trends Ecol& Evol 2008;23:65–72. [13] GEO. How GEO can contribute to the development and management of bioenergy resources in Africa; 2010 〈http://www.earthobservations.org/docu ments/general/201006_bafa_brochure.pdf〉. [14] Graham-Rowe D. Beyond food versus fuel. Nature 2011;474:6–8. [15] Herr A, Dunlop M. Bioenergy in Australia: an improved approach for estimating spatial availability of biomass resources in the agricultural production zones. Biomass Bioenergy 2011;35:2298–305. [16] International Monetary Fund (IMF). Gross domestic product per capita; 2012. [17] Jiang D, Zhuang D, Fu J, Huang Y, Wen K. Bioenergy potential from crop residues in China: availability and distribution. Renew Sustain Energy Rev 2012;16:1337–82. [18] Kaltschmitt M, Merten D, Fröhlich N, Nill M. Energiegewinnung aus Biomasse. Externe Expertise für das WBGU-Hauptgutachten 2003: „Welt im Wandel: Energiewende zur Nachhaltigkeit“. Berlin: WBGU; 2003. [19] Kaltschmitt M, Hartmann H. Energie aus Biomasse–Grundlagen, Techniken und Verfahren. Berlin, Heidelberg: Springer-Verlag; 2001. [20] Khan M.A.A., Amir P., Ramy S.A., Munawar Z., Ahmad V. National economics and environmental development study (NEEDS). UNFCCC; 2011. [21] Knorr W, Heimann M. Uncertainties in global terrestrial biosphere modeling, Part I: A comprehensive sensitivity analysis with a new photosynthesis and energy balance scheme. Glob Biogeochem Cycles 2001;15:207–25. [22] Land use atlas of Pakistan; 2009 〈http://www.mocc.gov.pk/gop/index.php? q=aHR0cDovLzE5Mi4xNjguNzAuMTM2L21vY2xjL3VzZXJmaWxlczEvZmlsZS9 MYW5kX1VzZV9BdGxhc19vZl9QYWtpc3Rhbi5wZGY%3D〉 (Last visit: May 2014). [23] Leduc S, Starfelt F, Dotzauer E, Kindermann G, McCallum I, Obersteiner M, et al. Optimal location of lignocellulosic ethanol refineries with polygeneration in Sweden. Energy 2010;35:2709–16. [24] Leduc S, Schwab D, Dotzauer E, Schmid E, Obersteiner M. Optimal location of wood gasification plants for methanol production with heat recovery. Int J Energy Res 2008;32:1080–91.

711

[25] Lomb NR. Least-squares frequency analysis of unequally spaced data. Astrophys Space Sci 1976;39:447–62. [26] Mirza UK, Ahmad N, Majeed T. An overview of biomass energy utilization in Pakistan. Renew Sustain Energy Rev 2008;12:1988–96. [27] Natarajan K, Leduc S, Pelkonen P, Tomppo E, Dotzauer E. Optimal locations for methanol and CHP production in Eastern Finland. Bioenergy Res 2012;5:412–23. [28] Pakistan Integrated Energy Model (Pak-IEM). Final report volume III—User's Guide. ADB TA-4982 PAK; 2010. [29] Pakistan Bureau of Statistics (PBS). Islamabad; 2012. [30] Poudel BN, Paudel KP, Timilsina G, Zilberman D. Providing numbers for a food versus fuel debate: an analysis of a future biofuel production scenario. Appl Econ Perspect Policy 2012;34:637–68. [31] Prentice IC, Cramer W, Harrison SP, Leemans R, Monserud RA, Solomon AM. A global biome model based in plant physiology and dominance, soil properties and climate. J Biogeogr 1992;19:117–34. [32] Rehman MS, Rashid N, Saif A, Mahmood T, Han J-I. Potential of bioenergy production from industrial hemp (Cannabis sativa): Pakistan perspective. Renew Sustain Energy Rev 2013;18:154–64. [33] Rühl C, Appleby P, Fennema J, Naumov A, Schaffer M. Economic development and the demand for energy: a historical perspective on the next 20 years. Energy Pol 2012;50:109–16. [34] Scarlat N, Blujdea V, Dallemand J-F. Assessment of the availability of agricultural and forest residues for bioenergy production in Romania. Biomass Bioenergy 2011;35:1995–2005. [35] Sellers PJ. Canopy reflectance, photosynthesis and transpiration. Int J of Remote Sens 1985;6:1335–72. [36] Thrän D, Bunzel K, Seyfert U, Zeller V, Buchhorn M, Müller K, et al. Global and regional spatial distribution of biomass potentials—status quo and options for specification. DBFZ report no. 7. 2009 (124 pp.). [37] Tum M, Günther KP. Validating modeled NPP using statistical yield. Biomass Bioenergy 2011;35:4665–74. [38] Tum M, Strauss F, McCallum I, Günther KP, Schmid E. How sensitive are estimates of carbon fixation in agricultural models to input data? Carbon Balance Manag 2012;7:1–13. [39] Tum M, Günther KP, Kappas M. A process-based vegetation model for estimating agricultural bioenergy potentials. In: Ruppert H, Kappas M, Ibendorf J, editors. Sustainable bioenergy production—an integrated approach. Springer; 2013978-94-007-6641-9. p. 97–110. [40] Wißkirchen K, Tum M, Günther KP, Niklaus M, Eisfelder C, Knorr W. Quantifying the carbon uptake by vegetation for Europe on a 1 km² resolution using a remote sensing driven vegetation model. Geosci Model Dev 2013;6:1623–40. [41] Zaliwski A, Faber A, Pudełko R, Biberacher M, Gadocha S, Borzęcka-Walker M. Biomass supply for co-firing in main-network power stations in Poland. J Food, Agric Environ 2013;vol.11(3 and 4):2031–5. [42] Zeddies J, Bahrs E, Schönleber N, Gamer W. Globale Analyse und Abschätzung des Biomasse-Flächenpotentials. Universität Hohenheim, Institut für Landwirtschaftliche Betriebslehre 2012(410 B):141.