Life cycle water use of low-carbon transport fuels

ARTICLE IN PRESS Energy Policy 38 (2010) 4933–4944

Contents lists available at ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

Life cycle water use of low-carbon transport fuels Christopher Harto a,n, Robert Meyers a, Eric Williams a,b a b

School of Sustainability, Arizona State University, Tempe, AZ, United States Department of Civil, Environmental and Sustainable Engineering, Arizona State University, Tempe, AZ, United States

a r t i c l e in fo

abstract

Article history: Received 17 July 2009 Accepted 29 March 2010 Available online 11 May 2010

In society’s quest to mitigate climate change it is important to consider potential trade-offs in climate solutions impacting other environmental issues. This analysis explores the life cycle water consumption of alternative low-carbon energy sources for transportation. Energy sources analyzed include both biofuels used in internal combustion engines and low-carbon electricity generation methods used in conjunction with electric vehicles. Biofuels considered are corn-based ethanol, soybean biodiesel, cellulosic ethanol from switchgrass, and microbial biodiesel. Electricity sources analyzed are coal with carbon sequestration, photovoltaic cells, and solar concentrators. The assessment method used is hybrid life cycle assessment (LCA), which combines materials-based process method and the economic input– output (EIO) method. To compare these technologies on an even footing the life cycle water use to propel a passenger vehicle one mile is estimated. All technologies evaluated showed an increase in water consumption compared to unleaded gasoline when water use from vehicle manufacturing was included. Scale-up calculations showed that mass adoption of electric vehicles and some configurations of algae and switchgrass systems could potentially contribute to the decarbonization of transportation with tolerable increases in overall water consumption. Irrigated crop based biofuels however were found to have significant potential impact on water resources when scaled up to macroscopic production levels. & 2010 Elsevier Ltd. All rights reserved.

Keywords: Water Biofuels Transportation

1. Background Climate change has emerged as a key global challenge for the 21st century. Fossil fuels, the primary source of greenhouse gas emissions, also face risks related to the reliability of supply. In the US, for example, coal and natural gas are produced domestically and used primarily for electricity and home heating, but over 60% of the oil used is imported almost exclusively for transportation. Much of the world’s remaining oil reserves lie in geologically (deep under the ocean) or politically (Middle East) challenging locations. In addition, there are potential future constraints on oil supply—many scientists and geologists are predicting that global oil production will soon reach a peak, followed by a plateau and/or continuous decline. There is significant disagreement on the exact timing of the peak as it depends on a number of uncertain factors; however, nearly all estimates put the timing of ‘‘peak oil’’ before 2040, with a majority predicting peak before 2020 (Hirsch et al., 2005; Fournier and Westervelt, 2005; GAO, 2007). Oil is thus at the nexus of climate, energy security, and supply concerns. Finding low-carbon alternatives is priority, in particular replacements for gasoline and diesel fuels used in transportation.

n

Corresponding author. E-mail address: [email protected] (C. Harto).

0301-4215/$ - see front matter & 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.enpol.2010.03.074

While low-carbon domestically supplied alternatives such as biofuels or solar power can address important oil-related concerns, it is important to consider other potential environmental, social, and economic impacts of alternatives. Biofuels, for example, have been promoted on the basis that they are a ‘‘green’’ and carbon-neutral energy source, but we are increasingly learning that corn-based ethanol in particular has unforeseen significant consequences such as perturbing nitrogen (Miller et al., 2006) and other material cycles as well as land use around the globe (Searchinger, 2008). Indeed, there are calls for traditional engineering to move towards sustainable engineering in which new technologies are developed alongside a sophisticated understanding of potential systemic effects of adoption (Allenby, 2007). Life cycle assessment (LCA) is a key tool to help understand the systemic effects of technology. LCA aims to quantitatively characterize environmental impacts of a technology over the entire supply chain including extraction, manufacturing, operation and end-of-life (Baumann and Tillman, 2004; Hendrickson et al., 2006). LCA grapples with two systems issues: full supply chain and multiplicity of environmental impacts. By considering the full supply chain, LCA can reveal that a proposed alterative may be much less effective than expected at mitigating the impacts that they were intended to address (e.g. carbon emissions from corn-based ethanol) (Farrell et al., 2006; Hill et al., 2006). By analyzing a multiplicity of environmental issues along a supply

ARTICLE IN PRESS 4934

C. Harto et al. / Energy Policy 38 (2010) 4933–4944

chain, LCA can characterize potential impacts not foreseen in the original scope of the alternative, such as potential lead flows associated with mass adoption of lead–acid based electric vehicles (Lave et al., 1995). This study uses LCA as a tool to estimate one commonly overlooked impact: the water consumption associated with alternative energy sources. Growth in water demand around the world has lead to increasing concern over supply constraints. Traditional forms of energy production are known to consume large volumes of water and account for approximately 20% of non-agricultural water consumption in the US (USDOE, 2006). This percentage is likely to increase in the future as the EIA has estimated (in their reference case) that world demand for liquid fuels for transportation will grow by 50% between 2005 and 2030. An updated projection, taking into account today’s high energy prices and expecting continued increases, predicts a somewhat lower, but still significant increase in transportation fuel demand over the same time period (EIA, 2008). If the US and other countries in the world are going to meet future energy demand, they must do so while both decreasing harmful CO2 emissions and without putting significant strain on potentially limited water resources. This study attempts to provide decision makers with data to help make informed choices on which technologies are best suited to meet this challenge.

2. Literature review and case study: life cycle water consumption of alternative transport energy sources There are as yet few studies examining life cycle water use of alternative energies. A 2008 Report from the National Academies Water Science and Technology Board explored the broad water implications of increased biofuel production, but did not attempt to quantify life cycle water consumption (WSTB, 2008). King and Webber studied life cycle water use of a variety of transport fuels: gasoline, diesel, compressed natural gas, corn-based ethanol, soy-based diesel, and cellulosic ethanol from corn stover (King and Webber, 2008). We have two objectives in the context of this previous work. The first objective is to evaluate water use of a portfolio of alternative fuels that appear promising for mitigating carbon emissions. The carbon benefits of corn and soybean based fuels are increasingly under question, especially given potential induced changes in land use (Searchinger, 2008). In addition, biofuels based on traditional crops induce substantial non-energy impacts, such as effects on nutrient cycles (Miller et al., 2006) and grain prices (BBC News, 2007). Despite the mounting evidence against sustainability of corn ethanol and soy biodiesel, they are included in this study for comparison purposes. The additional candidates we consider are coal-based electricity with carbon sequestration, solar-based electricity, cellulosic ethanol from switch grass, and microbial-based biodiesel. The second objective is to draw on hybrid life cycle assessment to provide a more complete description of water supply chains. Hybrid LCA draws on both bottom–up process models of water use and top–down economic input–output (EIO) models (Suh et al., 2004; Williams, 2004). Bottom–up process models rely on facility level data while EIO models describe supply chain material use for aggregated economic sectors. The power of the hybrid method is a more complete description of processes contributing to water use: when facility data is unavailable, process water use can be estimated by the economic input–output model. LCA also involves impact assessment methods, which aim to quantify total environmental impacts and trade-offs between environmental issues (Baumann and Tillman, 2004). Here we only characterize water consumption, leaving the multi-criteria assessment of alternative fuels as future work.

3. Method Hybrid LCA combines a bottom–up process model with a top– down economic input–output model (Engelenburg et al., 1994; Suh et al., 2004; Williams, 2004). Bottom–up process models based on facility/site level data describe elements in a supply chain more precisely but lack of data leads to error due to excluded processes. Economic input–output LCA (EIOLCA) models (Hendrickson et al., 2006), based on national sectoral data, are holistic but suffer from aggregation error due to coarse graining of processes. The term hybrid generically refers to any method combining process and economic input–output analysis, there are a number of approaches to achieve this. The simplest is additive hybrid in which economic data are identified covering processes for which materials data are unavailable and associated with sectors in an EIO model (Bullard et al., 1978). An economicbalance hybrid calculates the value-added covered in a materials process model, subtracts this from the total price and estimates impacts associated with the remaining value using EIOLCA (Williams, 2004). A mixed unit hybrid model constructs a matrix with both physical and economic quantities (Hawkins et al., 2007). In this analysis the wide range of technologies being covered implies that the additive hybrid method best suits the available data. To elaborate on the additive method, let j be a set of processes for which water consumption data at the process/site level are available, Wj and k are sets of processes for which economic data are available (e.g. $ per liter of fuel spent on plant equipment), Expk. The total water use is given by water use ¼ SWj þ SExpk WkSC , where WkSC is the supply chain use of water in sector k, given by WkSC ¼ WD ð1AÞ1 , where WD is the direct use of water in a sector and A is the requirements matrix in the IO model. We use WkSC values from the Carnegie Mellon University EIOLCA.net tool (CMUGDI, 2008). Only the 1992 model includes water use thus expenditures are adjusted to the year 1992 using an appropriate price index. Details on the specific EIO data used are described in Appendix A. Note that if neither material nor economic data are available that processes may be excluded from the analysis. The system boundaries are discussed later in this section. A range of potential technologies was explored using this methodology, each falling into one of the two main categories. The first technology category was biofuels, including corn ethanol, soy biodiesel, cellulosic ethanol, and microbial biodiesel. While biofuels are currently controversial, they are the only alternative fuels as yet making any significant contribution to transportation energy. The main advantage of biofuels is that they are largely compatible with current transportation infrastructure. In addition, some more advanced biofuels do not require food crops, eliminating one of their biggest drawbacks. The second technology category is electric vehicles (EV) and plug-in hybrids (PHEV). While there are few EVs on the road today, they are expected to enter the market in larger numbers over the next few years. Highly anticipated plug-in vehicles include the Tesla Roadster, the Chevy Volt, and a plug-in version of Toyota’s popular Prius hybrid. The largest roadblock to mass adoption of electric vehicles is in the limited range due to expensive batteries with relatively low energy density. Steady progress is being made in this area, but battery cost and capacity will ultimately determine the long term market penetration of EVs. Also, while electric vehicles use no gasoline, they are only as clean as their source of electricity. Three low carbon electricity sources are evaluated: coal power with carbon sequestration,

ARTICLE IN PRESS C. Harto et al. / Energy Policy 38 (2010) 4933–4944

solar photovoltaics (PV), and concentrated solar power (CSP). We evaluate water use of each electricity generation technology individually. In practice an electric vehicle would run from a grid based on a mix of different sources. It is not our purpose here to evaluate the future of electrical grids powering electrical vehicles, rather to understand the water implications of individual technologies. For all technologies at least a high and a low estimate of the water consumption was made. The high and low estimates were calculated using the extremes of the range of data available. Neither estimate is expected to represent the true water consumption for the technology. This was done in an attempt to scope out the range within which the true water consumption is likely to fall as well as capture some of the regional variability in water consumption. Given the limited data and significant uncertainties involved, this was viewed as the best way to make the analysis as useful as possible. For technologies where the difference between the high and low estimate was significant, a third estimate was made for the ‘‘average’’ or ‘‘typical’’ value. This value can be considered our best guess at what the overall average water consumption would be from a given technology. For the purposes of this study only water consumption was considered. Water consumption reflects water evaporated, incorporated in products or waste, or dumped into the sea after use (Pfister et al., 2009). Consumption focuses on resource aspects of water. Water withdrawals measure the amount of water redirected from ‘‘natural’’ flows and tend to reflect environmental impacts on eco-systems. We choose to focus on resource aspects and thus measure water consumption. The functional unit for the analysis was gallons of water consumed per vehicle mile traveled (VMT) (1 gallon/VMT¼2.36 l/ vehicle km traveled). This was chosen for comparison purposes to compensate for different energy densities of different fuels and different efficiencies of the vehicles that use them. However, data are also presented in units appropriate to each energy source including per gallon of fuel or kiloWatt-hour (kWh) to provide numbers that are free of any bias introduced due to assumptions regarding the specific vehicle used. To account for the use phase the water consumption for the production of vehicles was also considered. For biofuels a Toyota Prius equivalent was assumed. For electricity sources a PHEV with similar characteristics to the Chevy Volt was assumed. The technologies analyzed along with the processes included are listed in Table 1. The processes included are broken down into columns based upon if they were evaluated using process or EIOLCA methodology. For some processes neither material nor economic data were identified, leading to a degree cut-off error. Excluded processes include transportation of feedstock and intermediate products and production of machinery and equipment required for the production of intermediates (for example, farm or mining equipment). However in order to preserve the value of the relative comparison between technologies, care was taken to maintain a consistent system boundary. Even given the data limitations we have significantly expanded the system boundary compared to the previous analyses (e.g. King and Webber, 2008).

4935

Table 1 Summary of included processes. Technology

Process

EIOLCA

Unleaded

Refining

Coal w/CS

Mining, washing, slurry pipe transport, transmission quartz mining, cell manufacture, transmission

Crude production, distribution and marketing, vehicle manufacture Plant construction, use

Solar PV

Solar CSP

Cooling water, transmission

Switchgrass

Irrigation, farm inputs, ethanol production

Soy

Irrigation, farm inputs, biodiesel production

Corn

Irrigation, farm inputs, biodiesel production

Algae—open

Evaporation, process water, biodiesel production Algae—enclosed Glass tube manufacture, process water, biodiesel production

Inverter, Installation, module finishing, balance of system, other costs, O&M, vehicle manufacture collector field, thermal storage, power block, other costs, O&M, vehicle manufacture Plant construction, distribution and marketing, vehicle manufacture Plant construction, distribution and marketing, vehicle manufacture Plant construction, distribution and marketing, vehicle manufacture Plant construction, distribution and marketing, vehicle manufacture Plant construction, distribution and marketing, vehicle manufacture

refining phase, it is estimated that 1–2.5 gallons of water are consumed for each gallon of fuel produced (USDOE, 2006). The oil production phase is not nearly as straightforward, since in many locations water is a byproduct of oil production while in other cases, such as enhanced oil recovery (EOR), it is consumed in varying quantities. Due to the significant uncertainty in the direct water use of petroleum production the EIOLCA methodology was used to estimate the indirect water consumption for petroleum production. The water consumption for petroleum production was calculated to be between 0.32 and 0.75 gallons of water per gallon of gasoline produced. EIOLCA was also used to estimate water consumption for distribution and marketing. Water consumption was calculated to be between 0.65 and 2.67 gallons of water per gallon of gasoline. The unexpectedly high value for distribution and marketing can be attributed to water consumption related to the large quantities of electricity used to run fuel pipelines. This same value was used for the distribution and marketing costs of all liquid fuels. While it is understood that there will be transitional costs associated with the switch to biofuels due to potential incompatibilities with the current infrastructure. When added together the total water consumption for gasoline production comes to between 2 and 6 gallons per gallon of gasoline. When converted to VMT the water consumption is 0.04–0.13 gallons per mile. It is important to mention that it is likely that net water consumption from petroleum production will increase in the future as more mature oil fields deplete and there is more dependence on water intensive EOR methods to maintain oil production.

4. Petroleum based fuels 5. Biofuels Prior to exploring the impacts of advanced transportation fuels, it is important for comparison purposes to consider the technological system they will be replacing. Water comes into the gasoline production process in three main areas: refining, petroleum production, and distribution and marketing. During the

A primary advantage of biofuels is that they can be grown and produced regionally within the US, enhancing energy security and allowing the country to produce more of its own energy. They also fit into the existing national transportation infrastructure, which

ARTICLE IN PRESS 4936

C. Harto et al. / Energy Policy 38 (2010) 4933–4944

is currently designed to service automobiles that run on gasoline or diesel—both of which, like biofuels, are combustible liquids. However, biofuels will not be easily incorporated into the existing petroleum infrastructure. New production plants must be constructed near large sources of biomass to keep transportation costs down. Biofuels would require their own pipeline systems (or else be shipped by inefficient truck or rail), though the possibility exists for them to take over existing gas pipelines. But biofuels would need to become the dominant fuel in the US for that possibility to be realized and that will be unlikely if they are shipped by more costly means (Morris et al., 2006). There are other problems associated with biofuels. Increasing demand for food crops for ethanol production can inflate food prices. Another concern is that most plants have relatively poor solar efficiencies and therefore may not represent the best use of their land from an energy standpoint. Table 2 shows the average biomass productivity in barrels of oil equivalents (boe) per hectare per year for a number of crops currently considered for biofuel production. Solar efficiencies are calculated based on an estimate of 10,000 boe average solar energy per hectare per year and are all very low. Sugarcane is the only traditional crop that has a solar efficiency above 1%, but its utilization is limited due to climate and moisture requirements. One promising alternative to traditional crops is the production of biofuels from microorganisms. There are a number of algal and bacterial strains that are capable of functioning at much higher solar efficiencies. The advantages of using microorganisms are numerous: they can grow in environments not suited for traditional crops, they can be grown in highly controlled and optimized systems, they grow quickly under the correct conditions, and they also generally contain a high percentage of lipid content that can be more easily converted into useful fuels through a relatively simple transesterification process. The main drawback, however, is higher upfront capital investment. The water consumption results for the biofuels are shown in Tables 3 and 4 for ethanol and biodiesel, respectively. The calculations are described briefly in the sections below and in more detail in the online supplement. 5.1. Corn ethanol Ethanol derived from corn is produced in relatively large quantities within the US—3.9 billion gallons in 2005 (EIA, 2007a). Corn ethanol production seems to be a logical first step in US biofuel production, since the country is the world’s largest producer of corn with 72.7 million acres harvested in 2000 (USEPA, 2007). The infrastructure and knowledge base for making and handling ethanol are already in place since corn has been distilled into alcohol for centuries and the crop is currently grown in abundance throughout the country. Unfortunately, as the scale of ethanol production from corn has increased—mainly attributed Table 2 Average biomass productivity. Biomass source

Fuel type produced

Productivity (boe/ha-yr)a

Solar efficiency (%)b

Corn Switchgrass Sugarcane Soybean Sunflower Microalgae

Ethanol Ethanol Ethanol Biodiesel Biodiesel Biodiesel

20 23–50 210–250 13–22 8.7–16 390–700

0.2 0.2–0.5 2–3 0.1–0.2 0.1–0.2 4–7

a

Huber et al. (2006), Calculated based on 200 w/m2 global average solar energy (Pinker and Laszlo (1992). b

Table 3 Ethanol life cycle water use scenarios (gallons of water consumed per gallon of fuel). Process

Corn ethanol Allocation factor Crop irrigation Farm inputs Ethanol plant construction Ethanol production Distribution and marketingi Total

Low water use

Avg. water use

High water use

0.63a 25b,c 2.1d 0.03g

0.81a 128b 4.1e 0.1e

0.93a 403b,c 6.5f 0.18g

1h 0.65

4.7h 1.3

11h 2.7

28

138

423

1.0 0 1.8l 0.07l

1.0 0 3.4m 0.18

3.4l 1.3

6g 2.7

6.5

9.6

1.0 352 0.92k 0.03

1.0 380 1.8l 0.07l

1.0 411 3.4m 0.18

1.9o 0.65

3.4l 1.3

6p 2.7

356

387

423

Cellulosic ethanol (no irrigation) Allocation factor 1.0 Crop irrigationj 0 Farm inputs 0.92k Ethanol plant 0.03 constructionn Ethanol production 1.9e Distribution and 0.65 marketingi Total 2.9 Cellulosic ethanol (drought) Allocation factor Crop irrigationa Farm inputs Ethanol plant constructionn Ethanol production Distribution and marketingi Total a

Pradhan et al. (2008), Derived from USDA (2003), USDOE (2006); d Oliveira et al. (2005), Graboski (2002), Shapouri et al. (2004), and CMUGDI (2008); e Average of low and high values before allocation. f Pimentel (2005), Graboski (2002) and CMUGDI (2008), g USDA (2002) and CMUGDI (2008), h USDA (2005), i Assumed the same as gasoline, j Assumed to be zero, k McLaughlin and Kszos (2005), Wang (2001), and CMUGDI (2008), l Geometric mean of high and low case, m Wang (2001), Pimentel (2005), and CMUGDI (2008), n No specific data, assumed same as corn ethanol, o Ayden (2007), p Ayden et al. (2002). b c

to increased subsidies—problems have arisen. In the February of 2007, Mexican citizens took to the streets to protest a price increase of 400% in the price of tortillas due to increased demand for corn used for ethanol production (BBC News, 2007). In addition, recent research into the life cycle impacts of corn ethanol for energy has determined that the energy yield is only about 25% due to significant fossil fuel inputs for planting, harvesting, transporting, and distilling. It has also been concluded that even if 100% of the US corn crop is devoted to ethanol production, only the net energy equivalent of 2.4% of US transportation fuel demand could be met (Hill et al., 2006). Life cycle water inputs for corn ethanol are calculated for several stages including crop production, farm inputs, ethanol plant construction, ethanol production, and distribution and marketing. Water credits due to co-product manufacture are taken into account through the use of energy allocation

ARTICLE IN PRESS C. Harto et al. / Energy Policy 38 (2010) 4933–4944

Table 4 Biodiesel water use scenarios (gallons of water consumed per gallon of fuel). Process

Low water use

Avg. water use

High water use

Soy biodiesel Allocation Crop irrigation Farm inputs Biodiesel plant construction Biodiesel production Distribution and marketingg Total

0.18a 11b,c 1.2d 0.03f 1b 0.65 14

0.5a 120c 11 0.05f 1b 1.3 133

0.8a 286b,c 30e 0.06f 1b 2.7 321

Algae biodiesel (enclosed) Allocation Glass tubes Process water Biodiesel plant construction Biodiesel production Distribution and marketingg Total

0.69h 1.7 28 0.03f 1b 0.65 30

– 2.8i 40i 0.04i 1 1.3 44

1.0 4.6j 57 0.06f 1b 2.7 63

Algae biodiesel (open) Allocation Surface evaporation Process water Biodiesel plant construction Biodiesel production Distribution and marketingg Total

0.54h 0 31 0.03f 1b 0.65 32

– 165k 50i 0.04i 1 1.3 216

1.0 575k 80 0.06f 1b 2.7 656

a

Hammerschlag (2006), USDOE (2006), Derived from USDA (2003) and USDA (2002), d Graboski (2002), e Pimentel (2005) and CMUGDI (2008), f EIA (2007b) and CMUGDI (2008), g Assumed the same as gasoline, h Energy allocation based on lipid vs. non-lipid biomass energy content, i Geometric mean of high and low case, j Derived from European Commission (2005), k Derived from ADWR (2006). b c

factors—water consumption allocated to co-products is assumed to be the same fractional amount as energy consumption allocated. Since only 15% of all US corn crops are irrigated, water usage for the crop production stage is calculated by accounting for the proportion of irrigated and non-irrigated corn crops in the nation (Ayden, 2007). Farm inputs are limited to pesticides, herbicides, lime, and fertilizer. Ethanol plant construction utilizes EIOLCA data from Carnegie Mellon (CMUGDI, 2008), assuming a 30-year average plant lifetime. The ethanol production stage is assumed to use the common dry mill process.

5.2. Cellulosic ethanol The cellulosic crop considered here is switchgrass, a perennial grass found primarily in the prairie lands of North America. Cellulosic ethanol, while still in its early developmental stages, is set to emerge as a major contender in the biofuels market. Rather than handling and fermenting the starches from a plant (for example, corn kernels), the more abundant lignocellulosic stalks and leaves are used for this process. There are several benefits gained from doing so: lignocellulosic product is generally considered waste or animal feed, so this process captures an existing low-grade output stream and turns it towards more productive uses. A wide variety of crops can be processed into ethanol in this manner, many of which are hardier than

4937

traditional food crops. And finally, most cellulosic crops have a better net energy balance than corn ethanol (Schmer et al., 2008). There are also drawbacks to cellulosic ethanol. Conversion facilities that transform cellulosic biomass into fuel are currently much more expensive than those dealing with corn or soy. Capital costs will likely come down with more research and experience, but for now they pose a formidable barrier to economic production. Lignocellulosic ethanol also requires more water and inputs at the fuel plant, though this increased demand is offset by the very low irrigation needs of switchgrass. Life cycle water inputs for cellulosic ethanol are calculated for crop production, farm inputs, ethanol plant construction, ethanol production, and distribution and marketing. No specific data were available for cellulosic ethanol plant construction so it was assumed to be the same as for corn ethanol. While costs are currently higher for cellulosic ethanol plants, the impacts of manufacturing plant construction on life cycle water use are low and are unlikely to be significantly different. Switchgrass does not require much irrigation in most climates and there are no national statistics indicating the proportion of switchgrass crop that would need to be irrigated in the US. Therefore, the amount of water needed in the crop irrigation stage is provided in two scenarios: no irrigation and drought conditions. The former assumes absolutely no irrigation water is needed to grow the crop and should be taken as a lower bound on water input. The latter scenario is derived from a study by Sauerbeck et al. (2008), which examined the growth of switchgrass stands in Italy during several consecutive summers of severe drought conditions. This should represent a very high water use for the crop. Actual water inputs will fall somewhere in between these two scenarios, likely much closer to the no irrigation scenario. Farm inputs included fertilizers (nitrogen and phosphorous), pesticides, and herbicides, however, lime was not included. 5.3. Soy biodiesel The US is the world’s largest soy exporter, and planted as many acres of soy as corn in 2007 (USEPA, 2007). As such, soy is a well established crop and is very attractive for large-scale fuel manufacturing. It has many of the same benefits as corn ethanol in that its production is well defined and it fits into the current national energy infrastructure. Life cycle water inputs for soy biodiesel are calculated for crop production, farm inputs, biodiesel plant construction, biodiesel production, and distribution and marketing. Co-product allocations are included here as well, utilizing the same methodology as with corn ethanol. Since only about 4% of soybean acreage is irrigated in the US, water usage for the crop production stage is calculated by accounting for the proportion of irrigated and non-irrigated soy crops in the nation. Farm inputs included the usual pesticides, herbicides, lime, and fertilizers (nitrogen and phosphorous). Biodiesel plant construction used EIOLCA data from Carnegie Mellon and assumed a 30-year plant lifetime. Finally, the biodiesel conversion process assumed here is standard transesterification. 5.4. Algae biodiesel Biodiesel derived from microorganisms is seen as a major potential source of sustainable transportation fuel. The main advantages of algae for fuel production are their fast growth rates, high photosynthetic efficiency, and high lipid content. High lipid content is preferable for biodiesel production as that is the portion of the biomass that eventually becomes fuel. Photosynthetic efficiencies are orders of magnitude higher than food crops, meaning less land is required to grow them as well. It has been estimated that as little as 1–3% of the existing US

ARTICLE IN PRESS 4938

C. Harto et al. / Energy Policy 38 (2010) 4933–4944

cropping area would be required to meet 50% of all US fuel needs compared to many times the areas of existing crop land for corn and soybean based fuels (Christi, 2007). In addition, no actual crop land would have to be used, since algae can be grown on any relatively flat land, including the desert. The biggest roadblock in the development of algae based fuels is cost. Unlike traditional agriculture, growing algae is capital intensive. Current costs are still approximately an order of magnitude higher than current market prices for fuel. Even when expected economies of scale are factored in, prices are still too high by a factor of between two and five (Christi, 2007). This cost challenge applies to both types of reactor systems currently being considered for large scale production. They are enclosed, tubular photobioreactors (PBR) and open raceway ponds. Enclosed systems have the advantage of allowing better control and potentially higher yields, at the cost of higher upfront investment. They are necessary if pure cultures are desired or if genetically altered organisms are used. Open ponds are cheaper up front, but they can easily be contaminated by undesirable organisms and generally achieve lower yields. Water use was estimated for both systems. Due to this concern, lower system productivities and lipid content were assumed in the analysis for the open system compared to the enclosed system. Processes considered for the enclosed microbial system were water used to produce the glass reactor, process water extracted with the concentrated biomass, and biodiesel production and distribution and marketing. Processes considered for the open microbial system were evaporation from the pond surface, process water, and biodiesel production. High and low estimates for each system are shown in Tables 8 and 9. Surface evaporation calculations for the high case were based upon the range of conditions in the southwest US with high solar potential and low land costs. It is expected that water use would be lower for this system located in milder and more humid climates, although there may be some tradeoff in terms of yield. To account for this, the low case included no net evaporative loss assuming any evaporation would be replaced by rainfall. Another important consideration is that all calculations assumed only fresh water would be used; however, some algae strains are known to grow well in brackish water, which could significantly decrease fresh water demand. To account for the potential co-products, the low cases used energy allocation based upon the energy content of the lipid vs. non-lipid biomass. In the high case it was assumed that any coproduct energy produced was utilized within the process itself.

6. Electric vehicles Electric and plug-in hybrid vehicles (PHEV) provide a promising alternative to petroleum or biofuel powered vehicles. One advantage is the high efficiency of electric motors, which can approach 95% (McCoy et al., 1993). Also, when running off electricity, vehicles produce no tailpipe emissions, significantly

reducing local air pollution in crowded cities. This is not to say electric vehicles are completely clean, they just have a ‘‘long tailpipe’’, as the emissions are associated with producing the original source of electricity. Generally speaking, however, it is more economical to treat pollution at a single source than at thousands or millions of point sources. In the case of coal power, power plants can be paired with technologies to capture and store carbon dioxide, significantly lowering overall emissions. Electric vehicles can also be charged using clean, renewable energy sources such as solar and wind as well, which reduces the impacts even further. The southwest US has a high potential for solar energy production with average incident solar radiation above 6 kWh/ m2/day (NREL, 2006). The two main types of solar power are photovoltaics (PV) and concentrated solar power (CSP). While PV is probably the most well known form of solar power, CSP is gaining lot of attention for utility-scale solar power. CSP operates by using mirrors to direct sunlight onto a fluid, which is heated and used to generate steam to drive a turbine instead of converting sunlight directly into electricity like PV. Both of these technologies, along with coal power with carbon sequestration, were evaluated for their life cycle water consumption. Table 5 shows a comparison of general characteristics for the three electrical generation technologies. PV panels are typically more efficient than CSP plants, but they also tend to be more expensive. CSPs can incorporate thermal storage into their design, which opens up the potential to significantly increase their capacity factor (the average percentage of their rated capacity generated over a course of a year). This also allows for shifting of loads towards peak hours, which provides a useful benefit for utilities. The combined effects of lower capital costs and higher capacity factor show up in the form of significantly lower levelized cost of electricity production from CSP compared to PV. The added cost of sequestration brings the overall cost of coal up to comparable levels with CSP, but it is unlikely that CSP will completely displace coal for base load capacity. Based upon this, CSP, at least initially, appears to be an attractive way for utilities to increase their renewable generation mix. The water consumption results for the electricity sources are shown in Table 6. The calculations are described briefly in the sections below and in more detail in the online supplement. 6.1. Solar photovoltaics The most common types of photovoltaics are based upon silicon, which can be in the form of multiple crystals, a single crystal, or amorphous, with each form having its own advantages and disadvantages in terms of cost and efficiency. In addition, there is a wide range of newer technologies utilizing different elements including copper indium selenide, cadmium telluride, and a number of gallium and indium-based designs. However, as of 2004, 94% of commercial PV shipments were still silicon-based

Table 5 Low-carbon electricity data.

Efficiency (system) Capital cost ($/kW) O&M cost ($/kWh) Capacity factor Levelized cost ($/kWh)d a

Silicon PVa

Parabolic trough CSPa

Coal +carbon sequestrationb

10–17% $5000–7000 $0.01–0.02 23% 0.20–0.25

8–16% $2500–4000 $0.01–0.05 22–56% 0.06–0.11

24–32% $2400–2900 $0.01–0.02c 80–85% 0.10–0.12

Data aggregated from Tester et al. (2005), Aabakken (2006), and Price and Kearney (2003), NETL (2007), c O&M costs do not include fuel, d Present value of life cycle cost divided by life cycle energy production. b

ARTICLE IN PRESS C. Harto et al. / Energy Policy 38 (2010) 4933–4944

Table 6 Electricity life cycle water use scenarios (gallons of water consumed per kWh). Process

Low water use

High water use

Solar photovoltaic Quartz mining Cell manufacturing Other manufacturing O&M Transmission losses Total

0.0002a 0.0080b 0.05c 0.006c 0.006d 0.07

0.0008a 0.012b 0.14c 0.02c 0.015d 0.19

Concentrated solar Cooling water Plant construction O&M Transmission losses Total

0.77e 0.02c 0.003c 0.08d 0.87

0.92e 0.08c 0.02c 0.10d 1.12

0.01e 0e 0e 0.5f 0.013c 0.05 0.57

0.075e 0.025e 0.07e 1.2f 0.025c 0.13 1.53

Coal + carbon sequestration Coal mining Coal washing Coal transportation Plant operation/cooling Plant construction Transmission losses Total a

Derived from Williams (2000) and Williams et al. (2002), BP (2004), c From cost data in Table 3 and CMUGDI (2008), d NREL (2003), e USDOE (2006), f NETL (2007). b

(Kazmerski, 2006). This study focuses only on silicon-based photovoltaics for this reason. Water consumption data for solar PV systems is somewhat limited. The only two processes in the life cycle for which water data were available were solar cell manufacture and quartz mining. The water consumption for the remaining production processes and operations and maintenance was estimated using the EIOLCA methodology. The total water use was also corrected to account for transmission losses from the generation point to the plug. The limited amount of data results in a significant level of uncertainty that is not truly reflected in the range of the high and low estimates. However, based upon the data available it appears that water consumption for PV-based electricity is relatively low. This conclusion is in line with the conventional wisdom that photovoltaics do not require much water (USDOE, 2006).

6.2. Concentrated solar power Concentrated solar power comes in three main forms. The first and the most common form is the parabolic trough. In this configuration, long parabolic mirrors are focused on a tube of liquid running down the center, heating the liquid, which is then pumped to a heat exchanger to generate steam. Another configuration is the power tower or central receiver system, where an array of mirrors focuses light onto a central tower where the heat is collected and used to generate steam. The third configuration is known as a disk engine or Sterling engine. In this configuration, a large disk-like mirror focuses light onto a small receiver that both collects the heat and generates electricity in one step, making this configuration better suited for small-scale or distributed applications (Tester et al., 2005). The parabolic trough is the most developed of the three configurations, with nine plants totaling over 300 MW of installed capacity operating

4939

since the 1980s (Aabakken, 2006). The parabolic trough configuration was selected for analysis for this reason. A parabolic trough CSP plant is made up of three main components: the collector field, the power block, and the optional storage system. For CSP, the largest component of water consumption is evaporation of cooling water during operation. Water consumption for plant construction was estimated using the EIOLCA methodology. As expected, the water use for CSP is dominated by cooling water requirements. Looking at cooling water requirements for the different configurations of CSP plants, power tower systems have been estimated to fall near the bottom of the range for parabolic trough plants. Sterling engine systems use air cooling and require no cooling water, but are also significantly more expensive than the other two configurations (USDOE, 2006). 6.3. Coal with carbon sequestration Coal power is one of the oldest forms of electricity generation. It has the advantages of an abundant resource base and the ability to generate power continuously rather than just when the sun is shining or the wind is blowing. Also, a large percentage of our electricity already comes from coal. Because of these advantages it is unlikely it will be replaced completely, or at least not for a long time. However, the main disadvantage of coal is that it produces large quantities of CO2 and other pollutants (Tester et al., 2005). One solution that is being proposed to address this problem is carbon sequestration. With carbon sequestration, the CO2 is captured at the stack and transported to a location where it is stored permanently. Storage locations can vary from geological formations, to depleted oil or gas wells, to deep under the ocean, or it can be used in the enhanced oil recovery process. The main drawback is that sequestration can add up to 80% to the cost of electricity production from coal and use between 20% and 30% of the gross electricity produced (Anderson and Newell, 2004). Water consumption in the life cycle for coal plants occurs in coal mining, coal washing, coal transportation, and in plant operation, which is generally dominated by cooling water. Two main scenarios were considered for coal plants with carbon sequestration. The first, or low case, was based upon an integrated coal gasification combined cycle (IGCC) power plant and low end estimates for coal production. The second, or high case, was based upon a pulverized coal (PC) plant and high-end estimates for coal production. Water consumption at coal plants with carbon sequestration was recently estimated by the National Energy Technology Laboratory by doing detailed modeling of a range of plant designs both with and without sequestration. Average total water consumption for IGCC power plants with carbon sequestration represented a 10–15% increase in water use compared to the baseline without sequestration. Average total water consumption for PC power plants with carbon sequestration was about double the water consumption of the baseline power plant. These numbers also incorporated energy penalties of approximately 30% in the case of IGCC and 20% in the case of PC (NETL, 2007).

7. Vehicle manufacturing water consumption Previous analyses have shown that vehicle production can be a significant contributor to the overall environmental impacts of transportation (Samaras and Meisterling, 2008). In order to account for this potential impact, the water consumption was estimated for the manufacturing of a Toyota Prius and a nominal PHEV using the EIOLCA methodology. The calculations for the Prius are based upon the difference in invoice cost for a

ARTICLE IN PRESS 4940

C. Harto et al. / Energy Policy 38 (2010) 4933–4944

Toyota Corolla and Toyota Prius, which have similar size, weight, and characteristics (Edmunds.com, 2009). The difference in cost was equally allocated between additional electronic components and batteries. To calculate the water consumption from a PHEV the battery cost was increased based upon the relative capacity between the Prius battery and the specified battery size for the upcoming Chevy Volt (EERE, 2009; GM, 2007). Water consumption for vehicle manufacturing for each vehicle was then divided by a vehicle lifetime of 152,000 miles. The resulting water consumption was 0.19–0.33 gallons per VMT for the Prius and 0.29–0.79 gallons per VMT for the PHEV. For reference, the water consumption for the base Toyota Corolla was estimated to be 0.16–0.27 gallons per VMT using the same methodology. While there is significant uncertainty in these values based upon the age of the data used for the EIOLCA, especially for the battery production sector, it does indicate that vehicle construction may contribute significantly to the total water consumption from transportation.

impacts from hypothetical low carbon energy futures, it was assumed that the vehicle fleet efficiency would improve significantly as well. Thus, these calculations were based upon the efficiency of the most efficient vehicles available rather than the current fleet average efficiency. For biofuel VMT calculations the Toyota Prius, which gets an EPA estimated 46 mpg combined fuel economy running unleaded gasoline, was chosen. To account for differences in energy densities of gasoline, ethanol, and biodiesel, ratios of fuel efficiencies for a number of 2008 models available in both gasoline and flex-fuel or diesel were averaged. The ratios used were 1.31 for diesel to gasoline and 0.73 for ethanol to gasoline fuel efficiency (fueleconomy.gov 2008). Data for electric vehicles were limited. Listed energy efficiencies for the Chevy Volt and the Tesla roadster were 0.20 and 0.18 kWh/mile, respectively (GM, 2007; Tesla Motors, 2008). Taking into account that producers tend to be optimistic with their estimates for new technologies, the higher value was used for all EV calculations. Results for total consumptive use of water for fuel production and delivery are shown numerically in Table 7. Fig. 1 combines the water consumption from fuel production with that from vehicle production. It is important to note that the error bars in this graph do not represent true uncertainty in any statistically significant way but only represent the difference between the high and low water consumption estimates. In general, biofuel technologies tend to use more water than fossil and solar technologies, although the gap is much smaller when the vehicle contribution is included.

8. Results—life cycle water use per vehicle mile for alternative vehicle fuels In order to compare all the technologies, the data in Tables 3, 4, and 6 were converted to common units of gallons per vehicle mile traveled. Since the goal of this study was to look at the water

Table 7 Summary of water use for fuel production (gallons per vehicle mile traveled (VMT)). Technology

Low

Average

High

Unleaded Coal + CS Solar PV Solar CSP Corn ethanol Switchgrass—no irrigation Switchgrass—irrigation Soy biodiesel Algae biodiesel—closed Algae biodiesel—open

0.04 0.13 0.016 0.2 0.84 0.10 11 0.23 0.53 0.50

0.07 0.21 0.027 0.23 4.1 0.19 12 2.2 3.6 0.72

0.13 0.35 0.044 0.26 13 0.37 13 5.3 10.9 1.04

9. Comparing results A comparison was performed between our results and the results from King and Webber for unleaded gasoline, corn ethanol, and soy biodiesel. Since King and Webber only considered either irrigated or non-irrigated crops, new values for corn and soy were calculated assuming 100% irrigation instead of only 15% and 4% assumed above for corn and soy, respectively. Also, King and Webber use vehicle efficiencies reflecting current average stock rather than ‘‘best practice’’ hybrid and electric vehicles. To focus the comparison on water use in making fuels, we adjust our results using their vehicle efficiencies. The values were 20.5, 15.1,

Fuel Vehicle

10

1

lo se

d

pe n l-

C

O ga e

Bi

od i

od ie Al

ga e

Bi

So Al

es e

Bi y

se l-

od ie

at io ig lrr

hg r

ra

itc Sw

hg

n

n at ig lrr

rn

Co

io

an

h Et

s-

CS

as

So

r la

ss

C

ol

P

PV

itc

U

ar

l So

o

S

+C

l oa

-N

ed

ad

e nl

se l

0.1

Sw

Gallons of Water Consumed per VMT

Life Cycle Water Consumption per Vehicle Mile 100

Fig. 1. Summary of results for consumptive life cycle water use of transportation.

ARTICLE IN PRESS C. Harto et al. / Energy Policy 38 (2010) 4933–4944

and 25.7 mpg for gasoline, ethanol, and biodiesel, respectively. The results of this comparison are shown in Table 8. The two studies show a very significant difference in both the high and low estimates of water consumption, especially for soy biodiesel. A more detailed look at the supporting information of King and Webber’s study reveals a few potential explanations. The most glaring difference is the assumption of 10.7 gallons of fuel per bushel of soy. This is an order of magnitude greater than the value of 1 gallon/ bushel that was used in the current study. A closer look at the original source of this number indicates that there may have been a unit issue as the original number is 10.7 pounds/bushel rather than gallons/ bushel. Another potential source of difference was that attempts to re-calculate the corn ethanol estimates from the supporting information resulted in higher values than presented. It is unclear if there were miscalculations or if the supporting information is simply incomplete. The remainder of the difference appears to be the result of different values used for irrigation, which is not surprising due to high spatial and temporal variability. The differences for unleaded gasoline can be traced to the inclusion of water consumption from the oil production and distribution processes, which were not included in the other study. A final observation is that while the values for water consumption for non-irrigation agricultural inputs were at least an order of magnitude higher in this study due to the inclusion of water consumption from fertilizer production, this has little impact on the overall water consumption that was dominated by irrigation requirements for both soy and corn.

4941

technology was scaled up to see the potential impact of meeting 10% and 50% of the total personal transportation demand. The impact was calculated both in terms of total acre feet of water required and as a percentage of current overall water demand. These calculations used the average combined fuel and vehicle estimates from Fig. 1. The results of this analysis are shown in Table 9. The results show that all irrigated biofuels and open pond algae systems begin to have a noticeable impact (above 1% total water demand) on overall water consumption even at a low 10% market penetration. At 50% penetration these fuels all exceed 10% of total water consumption with the exception of soy biodiesel, which benefits from high allocation to animal feed. Corn and Soy biofuels are not expected to ever reach this level of adoption due to land use issues, however switchgrass and open pond algae likely at least theoretically could. If these fuels are to scale-up significance, care should be taken when selecting production locations with lower water requirements or abundant water resources. While electrical energy sources, non-irrigated switchgrass, and enclosed algae systems have much less of an impact on water resources, they still increase water consumption compared to unleaded gasoline by 40–200%. In the case of electric vehicles the majority of the water consumption results from the vehicle manufacturing process. Given the globalization of the automobile manufacturing industry it is likely that many of these impacts may occur outside of the US.

10. Potential macro-level impacts on water use 11. Discussion The life cycle water use results were used along with data on typical transportation and water demand to determine the impact on overall water resources from scaling up the proposed technologies. Values used for personal transportation demand were 2.3 trillion miles per year for the US (EIA, 2005). Total consumptive water demand was estimated to be 110 million acre feet per year for the US (USGS, 1998). Each

Table 8 Comparison of adjusted life cycle water results with King and Webber (2008). This study (adjusted to current average vehicle) (gallons water/VMT)

King and Webber (gallon water/ VMT)

Low

High

Low

High

0.29 139 281

0.07 1.3 0.6

0.14 62 24

Unleaded 0.09 Corn ethanol—all irrigated 8 Soy biodiesel—all irrigated 12

This study showed that EIOLCA can be a useful tool in combination with process LCA for estimating water impacts. Unfortunately water is not included in most EIOLCA data sets and the most recent is from 1992 (wherein water impacts were adapted from the data from 1983). In general, water consumption data are difficult to obtain if it exists at all. As water is becoming an increasingly important issue, especially in the southwest US, better and more detailed tracking of water consumption for industries will be important in improving the accuracy of future analyses. Our results indicate significant water consumption for some alternative transport fuels, in particular, agricultural crop based biofuels and open pond algae systems located in arid regions. If adopted on a mass scale, these technologies can have macroscopic effects on water demand. Electricity based power systems showed that they would still likely increase water consumption in comparison to petroleum based fuels. However, the scale of the impacts may not put a significant strain on water systems. That being said, adoption of these systems is highly dependent

Table 9 Water demand for technology scale-up to provide 10% and 50% of current VMT. Technology

Unleaded Coal + CS Solar PV Solar CSP Corn ethanol Switchgrass—no irrigation Switchgrass—irrigation Soy biodiesel Algae biodiesel—open Algae biodiesel—closed

10%

50%

Total acre ft.

% demand

Total acre ft.

% demand

230,000 490,000 360,000 500,000 3,080,000 310,000 8,310,000 1,730,000 2,700,000 690,000

0.21 0.44 0.32 0.45 2.75 0.28 7.4 1.55 2.41 0.62

1,140,000 2,450,000 1,790,000 2,500,000 15,410,000 1,570,000 41,550,000 8,660,000 13,510,000 3,430,000

1.02 2.19 1.60 2.23 13.76 1.40 37.11 7.73 12.07 3.06

ARTICLE IN PRESS 4942

C. Harto et al. / Energy Policy 38 (2010) 4933–4944

upon the development and adoption of affordable electric or plugin hybrid vehicles. Second, generation biofuels show the potential to have similar, low levels of water consumption if they are designed and located to limit the impacts on water resources. It is important to note however that these technologies are still in the developmental stage and their economic viability is still highly uncertain. That being said, this can also be seen as a benefit as there is still time to influence the development of these technologies in ways that minimize potential impacts before technological lock in sets in. Uncertainties and open questions remain. For biofuels based on agricultural crops where new cultivation takes place makes a huge impact on water consumption. Future modeling efforts could attempt to characterize the regional impacts of future land and associated water use. Argonne National Lab has begun to look at these impacts with respect to corn ethanol production in the US and they have shown significant variability in water consumption by region (Wu et al., 2009). In addition, global supply chains require consideration of impacts beyond national borders, which adds another significant layer of difficulty to this type of analysis. New forms of low-carbon transportation energy will be important if we, as a society, are to successfully combat global climate change. However, when considering any new technology it is important to critically and systematically assess it to ensure that it will not lead to other problems or unintended consequences. In the end, water consumption is only one data point in a complex, multidimensional decision-making process. As with all decisions, there are clearly trade-offs. When considering alternatives to fossil fuels we must better understand not only the cost and carbon impact but also potential impacts on land use, natural resource, and other environmental impacts such as air pollution and eutrophication to name only a few.

Table A1 Direct water consumption of traditional energy sources. Energy source

Unit

High

Low

Electricitya Coalb Oilb Natural gasb

gal/kWh gal/MJ gal/MJ gal/MJ

2 0.0076 0.0210 0.0028

0.47 0.0019 0.0083 0.0028

a b

NREL (2003), for current US electricity mix, USDOE (2006).

Table A2 Total EIOLCA water consumption data by sector (gallon/2007$). Sector

Low

High

N and P fertilizers Pesticides and agricultural chemicals Lime Industrial inorganic and organic chemicals Crude petroleum and natural gas Pipelines, except natural gas Glass containers Farm machinery and equipment Motor vehicles (passenger cars and trucks) General industrial machinery and equip. Turbine and generator sets Electrical industrial equipment Semiconductors and related devices Storage batteries New construction Other repair and maintenance construction Engineering, architectural, and surveying services Retail trade, excluding eating, and drinking

37.19 4.01 5.03 6.96 0.19 3.60 1.75 5.75 1.47 1.06 0.89 1.42 0.54 1.15 0.63 0.57 0.11 0.20

40.26 5.01 7.24 8.19 0.44 15.06 3.69 6.61 2.50 1.89 1.67 2.57 1.14 2.36 1.10 1.00 0.25 0.59

Acknowledgements This work was supported by the Arizona Water Institute, Science Foundation, Arizona, and by the grant ‘‘Sustainable infrastructures for energy and water supply’’ (#0836046) from the National Science Foundation, Division of Emerging Frontiers in Research and Innovation (EFRI), Resilient and Sustainable Infrastructures (RESIN) program.

Appendix A—EIOLCA methodology To estimate the water use for unknown capital equipment and processes, economic input–output (EIO) data were used. This methodology aggregates all impacts of a given economic sector and allocates them based upon the economic value of the process of interest. The main advantage of EIO data is that it includes both direct and indirect impacts. This means that it considers impacts directly by a specific sector and water used in producing all inputs to that sector as well. In general, EIO analysis tends to give more complete accounting of environmental impacts compared to process-based analysis, which always suffers from cut-off error due to processes and inputs that are not accounted for. This advantage comes at a cost of aggregation, which makes it difficult to evaluate specific processes or compare similar technologies. The specific data set used in this analysis was from the US 1992 benchmark index implemented in the Carnegie Mellon online tool, EIOLCA.net (CMUGDI, 2008). This was the only recent EIO data set for the US that contains water consumption data. In

order to correct for the age of the data, all values were converted to 2007 dollars using PPI data (Economagic, 2008). The output from this model included water intake, treated discharge, and untreated discharge. In order to determine water consumption, both discharge quantities were subtracted from the water intake value. It is quite possible that a significant portion of the water that is discharged untreated may be of diminished quality, but determining what fraction for each sector was beyond the scope of this effort and thus was not accounted for. The data set also only contains process water consumption from manufacturing and does not consider water from energy inputs or agricultural inputs (two of the largest consumers of water). It does provide total energy inputs by type though, so total energy consumption for each sector was corrected by calculating the water consumption for the energy inputs. High and low estimates of water consumption from energy inputs were estimated and added to the process water requirement. The water consumption values used are shown in Table A1. For most sectors the water requirement for energy accounted for between 10% and 50% of the total water consumption (Table A1). A summary of the total water consumption data for important sectors is shown in Table A2 in units of gallons per 2007$.

Appendix B. Supplementary material Supplementary data associated with this article can be found in the online version at doi:10.1016/j.enpol.2010.03.074.

ARTICLE IN PRESS C. Harto et al. / Energy Policy 38 (2010) 4933–4944

References Aabakken, B., 2006. Power Technologies Energy Data Book Fourth Edition US Department of Energy Office of Energy Efficiency and Renewable Energy. Allenby, B., 2007. Earth systems engineering and management: a manifesto. Environmental Science & Technology 41 (23), 7960–7965. Anderson, S., Newell, R., 2004. Prospects for carbon capture and storage technologies. Annual Review of Environmental Resources 29, 109–142. Arizona Department of Water Resources (ADWR) (2006). Arizona water atlas: volume 1 introduction. /http://www.azwater.gov/dwr/Content/Find_by_ Program/Rural_Programs/content/water_atlas/ArizonaWaterAtlas_Vol1_Intro duction_Draft_June2006.pdfS (accessed 11.26.08). Ayden, A.; Ruth, M.; Ibsen, K.; Jechura, J.; Neeves, K.; Sheehan, J.; Wallace, B.; Montague, L.; Slayton, A.; Lukas, J.. (2002). Lignocellulosic biomass to ethanol process design and economics utilizing co-current dilute acid prehydrolysis and enzymatic hydrolysis for corn stover, Report no. NREL/TP-510-32438, National Renewable Energy Laboratory. Ayden, A., 2007. Water usage for current and future ethanol production. Southwest Hydrology 6 (5), 22–23. Baumann, H., Tillman, A., 2004. The Hitch Hiker’s Guide to LCA. Studentlitteratur AB. BBC News (2007). Mexicans stage tortilla protest. /http://news.bbc.co.uk/2/hi/ americas/6319093.stmS (12.3.2007). BP Solar, 2004. Verified Environmental Statement: Calender Year 2004. Sydney Olympic Park, Australia. Bullard, C., Pennter, P., Pilati, D., 1978. Net energy analysis: handbook for combining process and input–output analysis. Resources and Energy 1, 267–313. Carnegie Mellon University Green Design Institute. (2008) Economic input–output life cycle assessment (EIO-LCA), US 1992 US Department of Commerce Model [Internet]. Available from: /http://www.eiolca.netS (accessed 10.5.2008). Christi, Y., 2007. Biodiesel from microalgae. Biotechnology Advances 25, 294–306. Economagic (2008). Browse US producer price index data. /http://www. economagic.com/blsppi.htmS (accessed 5.3.08). Edmunds.com (2009). 2009 Toyota Corolla car pricing guide. /http://www. edmunds.com/new/2009/toyota/corolla/100980405/prices.htmlS (accessed 12.14.2009). Energy Efficiency and Renewable Energy (EERE) (2009) 2010 Toyota Prius–6063 Hybrid BOT battery test results. /http://avt.inel.gov/pdf/hev/batterygenIII prius6063.pdfS (accessed 12.14.09). Energy Information Administration (EIA) (2005). Household vehicle energy use: latest data and trends. /http://www.eia.doe.gov/emeu/rtecs/nhts_survey/ 2001/S (accessed 8.5.08). Energy Information Administration (EIA) (2007a) Biofuels in the U.S. transportation sector. /http://www.eia.doe.gov/oiaf/analysispaper/biomass.htmlS (accessed 7.28.2008). Energy Information Administration (EIA) (2007b). Biodiesel performance, costs, and use. Energy Information Administration (EIA) (2008). Annual energy outlook 2008: with projections to 2030, Washington, DC. Engelenburg, B.C.W., van Rossum, T.F.M., Blok, K., Vringer, K., 1994. Calculating the energy requirements of household purchases. Energy Policy 22 (8), 648. European Commission (2005). MEEuP methodology report. Farrell, A.E., Plevin, R.J., Turner, B.T., Jones, A.D., O’Hare, M., Kammen, D.M., 2006. Ethanol can contribute to energy and environmental goals. Science 311, 506–508. Fournier, D.F. and Westervelt, E.T. (2005). Energy trends and their implications for US army installations. Report no. ERDC/CERL TR-05-21, Army Corps of Engineers. General Motors (GM) (2007). Latest Chevy Volt battery pack and generator details and clarifications, /http://gm-volt.com/2007/08/29/latest-chevy-volt-batterypack-and-generator-details-and-clarifications/ (accessed 6.14.08). Government Accountability Office (GAO) (2007). Uncertainty about future oil supply makes it important to develop a strategy for addressing a peak and decline in oil production, Report no. GAO-07-283. Graboski, M.S., 2002. Fossil energy use in the manufacture of corn ethanol. Colorado School of Mines prepared for National Corn Growers Association. Hammerschlag, R., 2006. Ethanol’s energy return on investment: a survey of the literature 1990–present,. Environmental Science & Technology 40, 1744–1750. Hawkins, T., Hendrickson, C., Higgins, C., Matthews, H.S., Suh., S., 2007. A mixedunit input–output model for environmental life-cycle assessment and material flow analysis. Environmental Science & Technology 41 (3), 1024–1031. Hendrickson, C., Lave, L., Matthews, S., 2006. Environmental Life Cycle Assessment of Goods and Services: An Input–Output Approach. RFF Press, Washington DC. Hill, J., Nelson, E., Tillman, D., Polasky, S., Tiffany, D., 2006. Environmental, economic, and energetic costs and benefits of biodiesel and ethanol biofuels. Proceedings of the National Academy of Sciences 103 (30), 11206–11210. Hirsch, R.L., Bezdek, R. Wendling, R. (2005). Peaking of world oil production: impacts, mitigation, & risk management. Huber, G.W., Iborra, S., Corma, A., 2006. Synthesis of transportation fuels from biomass: chemistry, catalysts, and engineering. Chemical Reviews 106, 4044–4098. Kazmerski, L.L., 2006. Solar photovoltaic R&D at the tipping point: a 2005 technology overview. Journal of Electron Spectroscopy and Related Phenomena 150, 105–135.

4943

King, C.W., Webber, M.E., 2008. Water intensity of transportation. Environmental Science & Technology 42 (21), 7866–7872. Lave, L., Hendrickson, C., McMichael, F., 1995. Environmental implications of electric cars. Science 268, 993–995. McCoy, G.A., Litman, T., and Douglass, J.G. (1993). Energy-efficient electric motor selection handbook, Report no. DOE/CE-0384. McLaughlin, S.B., Kszos, L.A., 2005. Development of switchgrass (Panicum virgatum) as a bioenergy feedstock in the United States. Biomass and Bioenergy 28, 515–535. Miller, S., Landis, A., Theis, T., 2006. Use of Monte Carlo analysis to characterize nitrogen fluxes in agroecosystems. Environmental Science & Technology 40, 2324–2332. Morris, W.R., Griffin, W.M., Matthews, H.S., 2006. Modeling switchgrass derived cellulosic ethanol distribution in the United States. Environmental Science & Technology 40, 2866–2877. National Energy Technology Laboratory (NETL) (2007). Cost and performance baseline for fossil fuel plants—volume 1: bituminous coal and natural gas to electricity, Report no. DOE/NETL-2007/1281. National Renewable Energy Laboratory (NREL) (2003). Consumptive water use for U.S. power production, Report no. TP-550-33905. National Renewable Energy Lab (NREL), (2006). Solar resources, transmission, and load centers. /http://www.nrel.gov/analysis/power_databook/docs/pdf/ db_chapter13_5.pdfS (accessed 4.29.2008). Oliveira, D., Marcelo, E., Vaughan, B.E., Rykiel Jr., J., Edward, 2005. Ethanol as fuel: energy, carbon dioxide balances, and ecological footprint. Bioscience 55 (7), 593–602. Pfister, S., Koehler, A., Hellweg, S., 2009. Assessing the environmental impacts of freshwater consumption in LCA. Environmental Science & Technology 43, 4098–4104. Pimentel, D., Patzek, T., 2005. Ethanol production using corn, switchgrass, and wood; biodiesel production using soybean and sunflower. Natural Resources Research 14 (1), 65–76. Pinker, R.T., Laszlo, I., 1992. Global distribution of photosynthetically active radiation as observed from satellites. Journal of Climate 5, 56–65. Pradhan, A., Shrestha, D.S., Van Gerpen, J., Duffield, J., 2008. The energy balance of soybean oil biodiesel production: a review of past studies. Transactions of the ASABE 51 (1), 185–194. Price, H. and Kearney, D. (2003). Reducing the cost of solar power from parabolic trough solar power plants, Presented at the ISES 2003 International Solar Energy Conference, Hawaii Island, Hawaii, March 16–18, 2003. Samaras, C., Meisterling, K., 2008. Life cycle assessment of greenhouse gas emissions from plug-in hybrid vehicles: implications for policy. Environmental Science & Technology 42 (9), 3170–3176. Sauerbeck, G. Batcher, W., El Bassam, N., Elbersen, W., Pignatelli, V., Piscioneri, I., Sharma, N., 2008. Effects of different seeding rates, drilling rates and weed control on establishment of switchgrass (Parcium virgatum L.) varieties in northern Germany and southern Italy. /http://www.switchgrass.nl/pdf/Sw-Fi nalRep-chapter6.pdfS (accessed 7.30.2008). Schmer, M.R., Vogel, K.P., Mitchell, R.B., Perrin, R.K., 2008. Net energy of cellulosic ethanol from switchgrass,. Proceedings of the National Academy of Sciences of the USA 105 (2), 464–469. Searchinger, T., Heimlich, R., Houghton, R.A., Dong, F., Elobeid, A., Fabiosa, J., Tokgoz, S, Hayes, D, Yu, T., 2008. Use of U.S. croplands for biofuels increases greenhouse gases through emissions from land-use change. Science 319 (5867), 1238–1240. Shapouri, H., Duffield, J., Mcaloon, A.J. (2004). The 2001 net energy balance of cornethanol.In: Proceedings of the Conference on Agriculture as a Producer and Consumer of Energy, Arlington, VA, June 24–25. Suh, S., Lenzen, M., Treloar, G.J., Hondo, H., Horvath, A., Huppes, G., Jolliet, O., Klann, U., Krewitt, W., Moriguchi, Y., Munksgaard, J., Norris, G., 2004. System boundary selection in life-cycle inventories using hybrid approaches. Environmental Science & Technology 38 (3), 657–664. Tesla Motors (2008). Tesla Motors—well-to-wheel. /http://www.teslamotors. com/efficiency/well_to_wheel.phpS (accessed 6.14.08). Tester, J.W., Drake, E.M., Driscoll, M.J., Golay, M.W., Peters, W.A., 2005. Sustainable Energy: Choosing Among Options. The MIT Press, Cambridge, MA. United States Department of Agriculture (USDA) (2002) Characteristics and production costs of U.S. soybean farms, Statistical Bulletin No. 974-4. United States Department of Agriculture (USDA) (2003) Farm and ranch irrigation survey (2003), Vol. 3, Special Studies, Part 1. United States Department of Agriculture (USDA) (2005) USDA’s 2002 ethanol costof-production survey. United States Department of Energy (USDOE) (2006). Energy demands on water resources. United State Geological Survey (USGS) (1998). Estimated water use in the United States in 1995 /http://water.usgs.gov/watuse/pdf1995/html/S (accessed 11.26.2008). US Environmental Protection Agency (2007). Major crops grown in the United States. /http://www.epa.gov/oecaagct/ag101/cropmajor.htmlS (12.2.2007). Water Science and Technology Board (WSTB) (2008). Water implications of biofuels production in the United States. The National Academies Press: Washington, DC. Wang, M. (2001). Development and use of GREET 1.6 fuel-cycle model for transportation fuels and vehicle technologies, Argonne National Laboratory, Argonne, IL, ANL/ESD/TM-163.

ARTICLE IN PRESS 4944

C. Harto et al. / Energy Policy 38 (2010) 4933–4944

Williams, E. (2000). Global production chains: the case of high-purity silicon and its applications in IT and renewable energy. United Nations University Institute of Advanced Studies. Williams, E.D., Ayres, R.U., Heller, M., 2002. The 1.7 kg microchip: energy and material use in semiconductor devices. Environmental Science & Technology 36, 5504–5510.

Williams, E., 2004. Energy intensity of computer manufacturing: hybrid assessment combining process and economic input–output methods. Environmental Science & Technology 38, 6166–6174. Wu, M., Mintz, M., Wang, M., and Arora, S. (2009) Consumptive Water use in the production of ethanol and petroleum gasoline. ANL/ESD/09-1, Argonne National Laboratory, Argonne, IL.