Thermodynamic Metrics for Aggregation of Natural Resources in Life

A hierarchy of metrics is proposed that spans the range from detailed data to ... This insight is illustrated via application to the life cycle of gas...
0 downloads 4 Views 2MB Size
Download PDF
Environ. Sci. Technol. 2010, 44, 800–807

Thermodynamic Metrics for Aggregation of Natural Resources in Life Cycle Analysis: Insight via Application to Some Transportation Fuels ANIL BARAL AND BHAVIK R. BAKSHI* William G. Lowrie Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, Ohio 43210

Received August 24, 2009. Revised manuscript received November 18, 2009. Accepted November 19, 2009.

While methods for aggregating emissions are widely used and standardized in life cycle assessment (LCA), there is little agreement about methods for aggregating natural resources for obtaining interpretable metrics. Thermodynamic methods have been suggested including energy, exergy, and emergy analyses. This work provides insight into the nature of thermodynamic aggregation, including assumptions about substitutability between resources and loss of detailed information about the data being combined. Methods considered include calorific value or energy, industrial cumulative exergy consumption (ICEC) and its variations, and ecological cumulative exergy consumption (ECEC) or emergy. A hierarchy of metrics is proposed that spans the range from detailed data to aggregate metrics. At the fine scale, detailed data can help identify resources to whose depletion the selected product is most vulnerable. At the coarse scale, new insight is provided about thermodynamic aggregation methods. Among these, energy analysis is appropriate only for products that rely primarily on fossil fuels, and it cannot provide a useful indication of renewability. Exergybased methods can provide results similar to energy analysis by including only nonrenewable fuels but can also account for materials use and provide a renewability index. However, ICEC and its variations do not address substitutability between resources, causing its results to be dominated by dilute and lowquality resources such as sunlight. The use of monetary values to account for substitutability does not consider many ecological resources and may not be appropriate for the analysis of emerging products. ECEC or emergy explicitly considers substitutability and resource quality and provides more intuitive results but is plagued by data gaps and uncertainties. This insight is illustrated via application to the life cycles of gasoline, diesel, corn ethanol, and soybean biodiesel. Here, aggregate metrics reveal the dilemma facing the choice of fuels: high return on investment versus high renewability.

1. Introduction Due to the highly multivariate nature of the results of life cycle studies, several methods have been devised for comparing and aggregating the results. Aggregate metrics are appealing for communicating findings, for convenient * Corresponding author. Tel: +1-614-292-4904; fax: +1-614-2923769; e-mail: [email protected]. 800

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 2, 2010

comparison between alternatives, and for decision making. Life cycle impact assessment (LCIA) involves aggregation of various emissions into midpoint and end-point indicators by use of characterization factors, and several standardized LCIA methods are available (1, 2). Comparable approaches to aggregate diverse natural resources for the purposes of estimating their consumption and its impacts exist but are not as mature or standardized. Methodologies such as abiotic depletion potential (ADP) (3) and surplus energy (2) have been used for quantifying resource use in LCIA, but they deal mainly with nonrenewable resources such as minerals and fossil fuels. ADP aggregates nonrenewable resources by use of characterization factors that are derived from the rates of extraction of minerals and ores compared to their reserves. Surplus energy measures damage to resources in terms of additional energy required for extracting low-quality resources as abundant and highquality resources become depleted. Other methods such as ecological footprint analysis (EFA) deal mainly with renewable resources and aggregate them in terms of the global land area required to regenerate them (4). EFA does not directly measure the depletion or stocks of nonrenewable resources but indirectly quantifies fossil fuel use on the basis of the forest area required to sequester CO2 emitted from fuel combustion. In addition, thermodynamic concepts such as calorific value and exergy have been suggested for aggregation of resources in life cycle assessment (LCA) and have been particularly popular, even among lay people (5), for comparing transportation fuels (6-10). However, the reduction in dimensionality due to aggregation usually causes loss of information about the variables being combined. An implicit assumption is that the resources being aggregated are substitutable. Thus, popular metrics such as energy return on investment and net energy add the calorific values of crude oil, natural gas, and coal, implying that a joule of natural gas energy may be substituted by a joule of coal energy. Depending on the materials and their application, such substitutability may not be possible since, for example, natural gas may be converted to electricity more efficiently than coal. Thus, for generating electricity, natural gas is a higher quality resource than coal. If the resources being aggregated include both renewable and nonrenewable resources, the disparity between their qualities can be even more significant. Clearly, a joule of sunlight or biomass energy is not able to do the same extent and type of work as a joule of coal energy. The presence of such quality differences and ignoring them in energy analysis has led some researchers to even question the relevance and usefulness of such metrics for decision making (5). Similar challenges also arise in the aggregation of emissions done in impact analysis. However, impact assessment relies on characterization factors to represent the variables in terms of a common unit such as equivalents of CO2 for aggregating all greenhouse gas emissions. The issue of energy quality and aggregation of different resources has received some attention (11) and various quantities such as exergy, emergy, and monetary value have been suggested for capturing differences in resource quality. While exergy is the useful energy that can be theoretically extracted from a given system and hence deals with the energy quality issue to some extent, emergy adjusts for the quality of resources by a transformation index or transformity. This quantity is analogous to the characterization factors used in LCIA. Most resources have exergy as a fundamental property, making exergy-based aggregation more inclusive than EFA, 10.1021/es902571b

 2010 American Chemical Society

Published on Web 12/18/2009

ADP, and surplus energy. However, there is no consensus on the most appropriate approach. Studies that apply and compare multiple resource aggregation schemes to the same data are uncommon since most papers focus on mainly one scheme (7, 9, 10). The primary contributions of this work are that it highlights the assumptions implicit in various aggregation schemes and provides insight into the features, pros, and cons of thermodynamic aggregation schemes. The schemes considered in this work include cumulative energy; industrial cumulative exergy consumption (ICEC), often referred to as exergy, and its variations; and ecological cumulative exergy consumption (ECEC), which is closely related to emergy. A monetary weighting scheme is also considered. After a unified introduction to these methods in the next section, these schemes are applied to four transportation fuelsscorn ethanol, soybean biodiesel, gasoline, and dieselsto provide a comparative assessment in the context of life cycle resource use. Life cycle inventory data were obtained via hybrid models constructed from ecologically-based life cycle assessment (Eco-LCA) (12). Such applications are essential for understanding the features of various aggregation approaches and determining their appropriate use. Since aggregation leads to loss of information, disaggregated and normalized thermodynamic data are also presented along with aggregate metrics. This results in a hierarchy of metrics from detailed to aggregate that can provide unique insight about the vulnerability of selected fuels to specific ecosystem goods and services, and a holistic view for decision making.

2. Concepts and Methods 2.1. Resource Aggregation. All the common aggregation schemes, including those considered in this work, may be represented as xj )

∑λx

i i

(1)

Here, xi represents the physical property of the ith resource, usually calorific value, exergy, or mass. λi is the quality correction factor for the corresponding resource, and xj is the aggregate value for all resources. Equation 1 implies that λixi units of the ith resource may be replaced by λi+1xi+1 units of the (i+1)th resource. Results of various methods described here for the selected fuels are presented in section 3.4. 2.1.1. Energy Analysis. This approach considers xi to be the calorific value of resources such as coal, oil, and natural gas. It treats all λi to be unity, thus ignoring any differences in energy quality. One commonly used metric in energy analysis is energy return on investment, rE, defined as the ratio of the calorific value of the produced fuel to the total calorific value of the resources used for converting the raw materials into the fuel, not including the feedstock. Cleveland et al. (11) have considered various approaches for quality correction and have claimed that monetary value is most appropriate. Their argument is that market values capture many quality aspects of resources such as their abundance, human preference, cleanliness, and convenience of use. In this approach, the quality correction factor may be calculated as λi )

pi pb

(2)

where pi and pb are the prices per joule (J) of the ith resource and the fuel selected as the basis, respectively. A shortcoming of this approach is that it may not capture externalities, and because market values are dictated by subjective human factors, they may have temporal and spatial variations even though the qualities of the resources under consideration remain unchanged.

2.1.2. Industrial Cumulative Exergy Consumption Analysis. Exergy is defined as the maximum useful energy that can be obtained as the system achieves equilibrium with the reference environment. Exergy analysis has been widely used in engineering for evaluating systems and identifying opportunities for efficiency improvement (9) and is often proposed as a way of accounting for the quality of resources (13, 14) because exergy considers only that part of a resource that can be converted to useful work. Exergy is a better indicator of quality than calorific value, since it can account for differences such as those due to temperatures of heat or different chemical potentials of materials. In terms of eq 1, the quality indicator, λi, for exergy analysis may be written as λi )

Exi xi

(3)

where xi may be the calorific value, enthalpy, or mass, and Exi is the exergy content. Since both fuels and nonfuel resources have exergy, this quantity can be used to compare and aggregate many different types of resources: a distinct advantage over energy analysis. Studies based on exergy analysis of the life cycle of transportation fuels (9, 10, 15) present aggregate metrics such as renewability factor (15) and exergetic breeding factor (9). These aggregate metrics were obtained by cumulative exergy consumption or demand (CEC) analysis, also called industrial cumulative exergy consumption (ICEC) analysis to indicate the emphasis on industrial and economic activities (16). While aggregation based on exergy is appealing, the results can be misleading since a joule of exergy of different resources such as coal, natural gas, biomass, and sunlight is not fully substitutable. This problem is particularly severe in the comparison and aggregation of renewable and nonrenewable resources, as illustrated in sections 3.3 and 3.4. In response, Dewulf et al. (17) have proposed to consider only the exergy that is actually used in industrial processes. This approach calculates the cumulative exergy extraction from the natural environment (CEENE) and ignores resources that do not directly enter the selected life cycle. For example, only 2% of the incident sunlight on land is included in CEENE since that is the fraction metabolized by plants. CEENE also excludes the overburden in mining since it is not used by industry. However, the part that is not used still must be produced to get the useful fraction. That is, producing the 2% metabolized sunlight cannot be done without the other 98%. This situation is analogous to allocation in LCA. The correction factor used in CEENE is a modification of eq 3 with the numerator being the extracted or used exergy instead of the total exergy of the resource. Thus, the numerator may be written as γiExi where γi is the fraction extracted. For solar exergy, γi ) 0.02. This approach still does not account for quality differences between various sources and considers the metabolized solar exergy to be equivalent to oil exergy. While this is an improvement over traditional ICEC analysis, other resources in addition to the metabolized sunlight are needed to convert biomass into the concentrated and higher quality of fossil fuels. CEENE ignores these along with differences in the quality of resources. 2.1.3. Ecological Cumulative Exergy Consumption or Emergy Analysis. A conceptually appealing approach for addressing the issue of energy quality was developed by Odum (18) via emergy analysis. Emergy is equivalent to cumulative exergy consumption when ecosystems are also included in the calculation and may be referred to as ecological cumulative exergy consumption (ECEC) (19). This approach aims to represent all resources in terms of a common numeraire, or unit of account, usually as solar equivalents. As a simple illustration, consider a hypothetical supply chain for a biofuel where 1000 J of sunlight is needed VOL. 44, NO. 2, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

801

to produce 10 J of biomass, which is used to produce 1 J of fuel. This implies that a joule of fuel is equivalent to 10 J of biomass, which is equivalent to 1000 J of sunlight. Thus, 1 J of the biofuel is equal to 1000 solar equivalent joules (sej), and adding different resources in terms of their solar equivalents satisfies the assumption of substitutability. This approach retains information about resource quality and therefore diminishes criticism about the loss of information due to aggregation (20). The quality correction factor in emergy analysis is given as λi )

Emi xi

(4)

where xi is usually the exergy, energy, or mass of the ith resource and Emi is its ECEC or emergy. The quality indicator λi is referred to as the transformity and converts all resources into solar equivalent joules. It has been proposed that resources with higher transformities are of higher quality and may be scarcer (18). Unlike all the approaches discussed in this section, ECEC is able to quantify the contribution of many ecosystem goods and services, which may make it more appropriate for evaluation of ecological aspects. Although Odum (18) and others have calculated the solar equivalents of many resources, the emergy approach relies on knowledge about complex ecosystems, which is likely to be inaccurate and incomplete. Despite being conceptually appealing, emergy analysis has been controversial and has not been used widely as an aggregation method for environmental accounting (21). Comparative studies, as in this work, are essential to understand and select appropriate aggregation methods. 2.2. Ecologically-Based Life Cycle Assessment. The results in this article are from Eco-LCA, which accounts for the role of ecosystem services in LCA (12). Currently, it accounts for many provisioning and supporting services and some regulating services by representing their flows in physical units. Existing LCA methods do account for some ecological resources, mainly provisioning services, and suggest aggregation schemes discussed in section 1. EcoLCA considers a much larger variety of ecological resources and develops a hierarchy of metrics according to the degree of aggregation as illustrated in section 3. The methods discussed in section 2.1 appear as special cases of the EcoLCA approach, as illustrated in section 3.3. At the economy scale, an Eco-LCA model based on the 1997 U.S. economic input-output (EIO) model has been developed and is available as user-friendly software (22). It combines the EIO model with information about the reliance of economic sectors on specific ecosystem services. This model may be used to easily calculate the ecosystem services needed to support the final demand from any sector. This approach is similar to that of EIO-LCA (23) but with some significant differences due to the inclusion of ecosystem services and the use of thermodynamic and hierarchical metrics in Eco-LCA (12). The results in this article are based on a tiered hybrid Eco-LCA model developed by combining data about industrial processes and their supporting ecosystem services with the economy-scale Eco-LCA model.

3. Application to Transportation Fuels 3.1. Data Sources. This work relies on detailed data about the life cycles of gasoline, corn ethanol, diesel, and soybean biodiesel (see Supporting Information). Relevant exergy and transformity numbers were obtained from the literature (10, 13, 18) for inputs from ecosystems to processes in the life cycle and to economic sectors. Such information has been used to calculate ratios of physical to monetary flow for each economic sector (12), which were used in this work for calculating the cumulative energy, ICEC, and ECEC values 802

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 2, 2010

for resources from economic sectors entering the life cycle of each fuel. The ECEC values of resources were combined to avoid double counting by considering appropriate allocation rules (18, 19). In general, the ECEC of most coproduced resources may not be added, while that of resources produced over different time frames or obtained by splitting a resource may be added. Examples of coproduced resources include sunlight, biomass, wind, rain, and hydropotential since they are driven by solar input and produced concurrently. ECEC of soil may be added to that of sunlight since soil may be considered to be the product of energy available in the past. Also, the ECEC values of various nonrenewable resources may be added since their transformities are calculated by allocating the inputs in proportion to the quantity of each resource. A similar approach has also been used for calculating CEENE (17). To determine the wellto-wheel efficiency, distance traveled per unit of ICEC and ECEC were calculated for a Chevrolet Impala 2006 and a light-duty truck (LDT2). ICEC was allocated among coproducts in proportion to their mass and market values to evaluate the effect of the allocation method, but the results in this section are mainly from mass-based allocation. Market-value-based allocation assigns more weight to fuels than their corresponding coproducts, causing the ICEC metrics to be larger. However, the overall trend between different fuels remains unchanged, as shown in Supporting Information. Following the emergy algebra rules, ECEC was not allocated since coproducts have different physical and chemical characteristics and hence should have different transformities (18). Sensitivity of ICEC and ECEC was analyzed by varying crop yields per hectare. 3.2. Disaggregated Normalized Data. For comparing the use of multiple resources, Figure 1 shows the results from the hybrid Eco-LCA models for corn ethanol and gasoline after the life cycle consumption of each resource was normalized with the corresponding total U.S. consumption or flow. This graph is for a future scenario in which corn ethanol would substitute 12% of motor gasoline demand in the US: 64.4 billion liters (17 billion gallons) in 2006 (8), which amounts to 92.4 billion liters (24.4 billion gallons) of corn ethanol based on the capacity of the fuel to drive a vehicle for a kilometer. The normalized values indicate the limiting factors and vulnerabilities to loss of ecological resources of each fuel. As shown, soil erosion, cropland use, and nitrogen and phosphorus from mineralization would account for 5%, 8%, and 4% of the national consumption, respectively. These numbers are for mass-based allocation. Without allocation, they will double since mass-based allocation assigns about 50% share to corn ethanol. Among the resources considered, consumption of nitrogen and phosphorus from mineralization, soil erosion, and cropland use can vary appreciably depending on corn and soybean yields, as shown by the sensitivity bars in Figure 1. 3.3. Disaggregated Thermodynamic Data. Figure 2 depicts the relative contribution of each resource in terms of ICEC, ICEC with metabolized sunlight, and ECEC of various fuels. As can be seen from Figure 2a, sunlight emerges as the predominant contributor to ICEC, even for gasoline and diesel. Although sunlight is not directly involved in the production of petroleum-based fuels, it is consumed indirectly via the supporting economic activities. In ICEC, solar exergy is based on the large amount of sunlight irradiated to forestry and agriculture per year, and even a small fraction transferred to other economic sectors from the forestry and agriculture sectors translates into very high ICEC in comparison to the exergy consumed via other resources. Sunlight is a very dilute and low-quality resource as compared to crude oil and other resources, but this difference is ignored

FIGURE 1. Normalized consumption of ecological resources for corn ethanol (92.4 × 109 L) and equivalent amount of gasoline (64.4 × 109 L) when corn ethanol meets 12% of 2006 gasoline demand in the United States. Normalized data were derived by dividing the life cycle consumption by U.S. consumption or flows. Error bars represent the range obtained from sensitivity analysis. in ICEC analysis, causing this approach to be of limited use for such holistic analysis. The result of an approach analogous to CEENE (17) based on our data is shown in Figure 2b. Here, only 2% of the sunlight is included. Now, the exergy of nonrenewables, primarily crude oil, dominates for gasoline and diesel, as is generally expected. However, solar exergy still contributes 4%. For ethanol and biodiesel, solar exergy dominates, while other resources such as soil, minerals, and fossil fuels play a relatively minor role, which is counterintuitive. As discussed in section 2, this approach still does not account for substitutability and quality differences between resources. It is interesting to note that photovoltaic cells are much more efficient than plants since they can routinely convert 7-17% of incident sunlight into electricity. However, these numbers ignore the life cycle aspects and the fact that plants are ecologically optimized via evolutionary pressures, while photovoltaics are very far from that state. Representing the resources in terms of emergy or ECEC results in the plot shown in Figure 2c. Here, nonrenewable resources contribute more to total ECEC since they require more ecosystem work, causing their transformities to be relatively large. Unlike Figure 2a,b, sunlight is nearly invisible, while nonrenewables such as fossil fuels, minerals, and ores are relatively large contributors, particularly to ethanol and biodiesel. Other resources such as soil, detrital matter, and nutrients from mineralization are much more prominent for biofuels than for fossil fuels. Detrital matter refers to litter decomposition in the field, which provides services such as soil formation, improvement in soil productivity, water retention, etc. Biofuels are also seen to capture CO2 via photosynthesis for the selected national boundary and static snapshot provided in this work. Comparing Figure 2 panels b and c shows that contribution of detrital matter in terms of ICEC is 2.2 times larger than that of soil for corn ethanol, but its contribution in terms of ECEC is 1.7 times smaller than that of soil. Soil consumption refers to soil erosion, which mainly occurs in cropland, timberland, pasture, and construction activities. The difference reflects the fact that detrital matter requires less ecosystem work than soil, which may indicate soil’s relative scarcity and importance (18), which is recognized via ECEC analysis. The contributions of soil and detrital matter can vary appreciably depending on crop yields, as

indicated by error bars in Figure 1. As seen from Figure 2c, correcting qualities of resources via ECEC analysis provides intuitive results, but as mentioned in section 2, the uncertainties in the underlying ecological processes and associated transformities need to be better understood (11). Difficulties in representing resources such as cropland and CO2 in thermodynamic units prevent their inclusion in some aggregation schemes in Figure 2. The details in Figure 1 in the proposed hierarchy of metrics prevent ignorance of such resources. 3.4. Aggregate Thermodynamic Metrics. 3.4.1. Return on Investment. Five different return on investment (ROI) metrics are plotted in Figure 3: conventional energy (rE), monetarily weighted energy (rE$), ICEC (rEx), ICEC with metabolized sunlight (rEx(2%)), and ECEC (rEm). Detailed definitions of these metrics are given in Supporting Information. Returns on energy investment (rE) of biofuels studied are lower than those of gasoline and diesel, indicating that biofuels are energetically less competitive. However, monetarily weighted ROI (rE$) for biofuels, with crude oil as the basis price, pb (11), are significantly larger than the traditional ROI (rE). The difference is larger for biofuels, particularly for biodiesel, due to their higher 1997 prices. These higher prices may reflect people’s willingness to pay for the perceived benefits of biofuels even though there is not much difference in fuel quality in terms of doing work. With increasing crude oil prices, trends of rE$ may be reversed. The exergy return on investment (rEx), calculated by including the ICEC of all resources needed for processing the feedstock into fuel, is shown as the fourth bar in Figure 3, by use of mass-based allocation. Processing ICEC accounts for materials and fuels including indirect sunlight used in the process but does not include ICEC of feedstock exergy and direct sunlight. Due to the dominance of indirect sunlight, rEx is almost invisible and less than 1 for all fuels. This would imply that all fuels are exergetically unfeasible. This misleading result is due to ignoring the quality of various resources and the resulting dominance of solar exergy. Considering only the metabolized fraction of sunlight results in rEx(2%), shown as the third bar for each fuel in Figure 3. This ratio is still lower than 1 for biofuels due to large indirect sunlight consumption and implies that biofuels are not feasible. However, fossil fuels look good with rEx(2%) for gasoline and diesel being 6.7. VOL. 44, NO. 2, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

803

FIGURE 2. Relative contribution of various resources in terms of different thermodynamic units. (a) Conventional ICEC. (b) ICEC with extracted resources (17). (c) ECEC. Only major resources are shown here. Electricity not including coal means electricity produced from hydropower, wind, and geothermal. The ROI based on ECEC (rEm) was obtained by dividing ECEC of the output by ECEC from the economy, which includes materials and fuels needed for processing the feedstock but not the feedstock itself. As shown in Figure 3, rEm of biodiesel and corn ethanol are smaller than those of gasoline and diesel. This trend suggests that fossil fuels require less effort from economic activities for converting them into transportation fuels. Another interpretation is that the fossil feedstock is of higher quality and easier to transform into fuel than biomass feedstock because, for the latter resource, human activity is needed to do the type of work that nature has already done for producing fossil fuels. Except for the monetarily weighted ROI, all other ROI metrics indicate that corn ethanol and biodiesel consume 804

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 2, 2010

more resources for processing the feedstock into fuel as compared to gasoline and diesel, implying that the latter fuels are thermodynamically superior. It also implies that because the processing energy and exergy reflect inputs from the economy, minimizing these inputs would increase the ROI of biofuel production. ROI metrics for biodiesel are more favorable than those for ethanol, which is partly due to lower fossil fuel consumption. Even though trends of rE, rEx, and rEm are similar for the fuels studied, they reflect different types of returns, with rE focusing only on energy investment from the economy, rEx focusing on materials and energy investment from both the economy and ecosystem, and rEm focusing on quality-adjusted materials and energy investment from the economy. For example, increasing water consump-

FIGURE 3. Various thermodynamic return on investment metrics for selected fuels. tion in a production process lowers rEx and rEm whereas rE remains the same. For the fuels studied, rE and rEm provide more intuitive insight. 3.4.2. Renewability Indicator. A renewability indicator quantifies the relative contribution of renewable resources to total resource consumption. Percent renewability, R, is defined as follows: R)

xj ren × 100 xj

(5)

where xjren is the aggregate quantity of renewable resources and xj is the total resource. In general, resources considered to be nonrenewable include goods and services from the lithosphere such as metallic ores, nonmetallic minerals, fossil fuels, etc., whereas renewable resources include contributions from the atmosphere, hydropower, geothermal, wind power, sunlight, soil, etc. In this study, resources that can be regenerated within 50 years have been classified as renewable. Soil was treated as a renewable source, but even otherwise, the overall conclusions remain unchanged. Degradation of ecosystems and resources needed for restoration were not considered. Other approaches based on considering stocks and rate of use of resources could also be used for quantifying renewability. Figure 4 shows the renewability index based on the studied aggregation schemes. Generally, energy analysis does not provide such a metric due to its traditional focus only on nonrenewable fuels. In terms of ICEC and ECEC, both biodiesel and corn ethanol are indicated to be more renewable than gasoline and diesel. In the case of ICEC, however, the degree of renewability (REx) is large for all fuels due to the high contribution of solar exergy that overwhelms even fossil fuel consumption for gasoline and diesel. ICEC with metabolized sunlight indicates less renewability (REx(2%sunlight)) but still suggests gasoline and diesel to be about 5% renewable due to the large contribution of low-quality sunlight to their life cycles. This approach indicates more than 90% renewability of ethanol and biodiesel. In contrast, ECEC indicates renewability (REm) of corn ethanol and biodiesel to be 25% and 49%, respectively, while gasoline and diesel have a renewability index of nearly zero. This is because ECEC accords less weight to low-quality sunlight and hence may provide a more balanced measure of renewability. Therefore, renewability index based on qualityadjusted resource consumption can be a metric of choice if renewability of the various alternatives is to be compared. Other thermodynamic metrics related to renewability include the breeding factor (9) and emergy loading ratio (18). Again, the results from ECEC are closer to the intuition developed in other studies. 3.4.3. Efficiency Measures. Efficiency metrics provide a picture of the overall resource consumption and can be useful for identifying improvement opportunities. Unlike ROI, efficiency includes the feedstock in the denominator. Since

the fuels under study are specifically used for transportation, it is relevant to measure efficiency in terms of kilometers traveled per unit of resource consumed as ICEC (J) and ECEC (sej): the larger the value, the higher the well-to-wheel efficiency. As shown in Figure 5, efficiency expressed as km/ICEC (J) decreases from fossil to biomass fuels. Since sunlight accounts for 99% of ICEC, it indicates that utilization of plants with better photosynthetic efficiency or higher biofuel yield such as algae may improve exergetic efficiency. Reducing the use of sunlight could also reduce reliance on land area and chemical, material, and labor inputs by producing more output per unit area. Nevertheless, since ICEC ignores the low-quality nature of sunlight, these results are of limited use. Efficiency information obtained from traditional energy analysis is incomplete due to its focus on fossil resources. Thus, fuels consumed in producing fertilizers are included but fertilizer consumption is not. This can make a difference for products that utilize appreciable amounts of nonfuel materials. For diesel and gasoline, this contribution is minuscule compared to that of fossil fuels and electricity. However, nonfuel materials have a significant contribution to biodiesel and corn ethanol production in terms of ECEC and ICEC with metabolized sunlight. Ignoring the contribution of nonfuels could hide opportunities for potential efficiency improvements. For example, the largest contribution to nonfuel ECEC for biofuels comes from detrital matter, water, and soil. Minimizing their consumption could, therefore, increase the overall system efficiency. Another metric of interest is km/nonrenewable ICEC (J), which increases from fossil to biomass fuels. E85 and BD100 reduce nonrenewable ICEC over gasoline by factors of 2 and 5 per kilometer traveled, respectively. Except for crude oil, almost all other kinds of nonrenewable resources considered in this study are consumed in larger quantities by biofuels in comparison to petroleum-based fuels. Still the larger km/ nonrenewable ICEC values for biofuels than fossil fuels are due to the large nonrenewable ICEC of crude oil for gasoline and diesel. The km/ECEC (sej) metric increases from fossil to biomass fuels. This implies that corn ethanol and biodiesel utilize the ecological resources included in ECEC more efficiently than gasoline and diesel. The smaller cradle-to-wheel ecological efficiencies of gasoline and diesel are attributed to their reliance on nonrenewable crude oil as feedstock, which requires more ecosystem work and is of higher quality. This suggests that ecological efficiency may be enhanced by increasing the relative use of renewables. The higher ecological efficiencies (km/ECEC) of biofuels stand in contrast to their lower energy and exergy returns on investment, indicating a trade-off.

4. Discussion Thermodynamic methods have been increasingly popular for life cycle evaluation due to their scientific rigor and ability to account for a large variety of resources in terms of a common unit. However, like other aggregation methods, thermodynamic aggregation also involves assumptions about the substitutability of resources, and without careful aggregation, it is possible to get misleading results. This work provides insight into the ability of various thermodynamic methods to compare and aggregate resource use over a life cycle, which is illustrated via application to transportation fuels. In terms of the return on investment metric for the fuels studied, the trend for all methods except monetary weighting is similar and shows the superior ROI of fossil fuels over biofuels. Despite the similar trend, we show that energy analysis is appropriate for calculating ROI only when the VOL. 44, NO. 2, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

805

FIGURE 4. Renewability index of fuels based on ICEC, ICEC with extracted resources, and ECEC analysis.

FIGURE 5. ICEC- and ECEC-based efficiency metrics for various fuels. resources used in the life cycle are primarily fossil fuels. For capturing materials and energy use, exergy-based methods are better, but the weighting scheme for aggregation is important. If the resources considered are diverse, then ROI based on ICEC is of limited use due to solar dominance, which makes all fuels seem unattractive. ICEC with 2% sunlight does better, but ECEC provides the most intuitive results. Monetarily adjusted ROI is sensitive to fluctuations in market prices and ignores ecological costs and may not be appropriate, despite the claim made by Cleveland et al. (11) in its favor. For the renewability index, because energy analysis excludes the direct contribution of sunlight in most cases, this index based on energy may not be appropriate. This index from exergy-based methods is again most intuitive for ECEC. As in ROI, ICEC is dominated by solar exergy even for fossil fuels, and ICEC with metabolized sunlight also gives quite high renewability due to ignoring quality differences. Although ECEC seems to be most attractive among the aggregation methods considered in this work, it suffers from many shortcomings due to its reliance on highly uncertain data about ecosystems, inability to account for many essential ecosystem services such as pollination and pest regulation, confusing allocation schemes, etc. (21). Nonetheless, this study shows that by quickly processing low-quality sunlight, albeit inefficiently, into high-quality energy, biofuels offer an advantage over fossil fuels due to needing less work from ecosystems in producing various resources. However, most of the ecosystem work used by fossil fuels was done in the past, while biofuels rely more on 806

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 2, 2010

ecosystem work done in the present time. Therefore, the industrially based life cycles of biofuels are associated with significant environmental impacts such as high processing energy as reflected by their lower rE, higher water and land use, eutrophication, aquatic toxicity, and other emissions (8, 24). This indicates that for biofuels to be environmentally attractive, minimizing reliance on economic inputs and environmental impacts and increasing the reliance on “free” ecosystem goods and services are critical. Accounting for the wide range of ecosystem goods and services and analyzing the results via the suggested aggregation schemes may help in guiding the development of new fuels toward sustainability. More research is needed for further evaluation and development of aggregation schemes to result in standardized hierarchical metrics analogous to those for life cycle impact assessment. When the wide variety of uses of natural resources is considered, such standardization is a formidable challenge. Until this challenge is met, this work shows that thermodynamic aggregation methods should be used carefully with full cognizance of the underlying assumptions to avoid misleading claims.

Acknowledgments This work was supported by the National Science Foundation (ECS-0524924) and the Environmental Protection Agency.

Supporting Information Available Background data and calculations. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Bare, J. C.; Norris, G. A.; Pennington, D. W.; McKone, T. TRACI: The tool for the reduction and assessment of chemical and other environmental impacts. J. Ind. Ecol. 2002, 6 (3-4), 49–78. (2) Goedkoop, M.; Spriensma, R. Eco-Indicator 99: A Damage Oriented Method for Life Cycle Impact Assessment; Pre Consultants: Amersfoort, The Netherlands, 2000. Available at http:// www.leidenuniv.nl/cml/ssp/projects/lca2/report_abiotic_depletion_web.pdf (accessed on July 2, 2009). (3) Van Oers, L.; De Koning, A.; Guine´e, J. B.; Huppes, G. Abiotic Resource Depletion in LCAsImproving Characterisation Factors for Abiotic Resource Depletion as Recommended in the New Dutch LCA Handbook; Report DWW-2002-061; RWS-DWW, Road and Hydraulic Engineering Institute, Public Works Department, Delft, The Netherlands, 2000. (4) Ewing, B.; Reed, A.; Rizk, S. M.; Galli, A.; Wackernagel, M., Kitzes, J. Calculation Methodology for the National Footprints Accounts, version 1.1; Global Footprint: Oakland, CA, 2008. Available at http://www.footprintnetwork.org/en/index.php/GFN/page/ ecological_footprint_atlas_2008 (accessed on July 2, 2009). (5) Schulz, W. The costs of biofuels. Chem. Eng. News 2007, 85 (51), 12–16. (6) Pimentel, D.; Patzek, T. Ethanol production using corn, switchgrass, and wood; biodiesel production using soybean and sunflower. Nat. Resour. Res. 2005, 14 (1), 65–76. (7) Farrell, A. E.; Plevin, R. J.; Turner, B. T.; Jones, A. D.; O’Hare, M.; Kammen, D. M. Ethanol can contribute to energy and environmental goals. Science 2006, 311 (5760), 506–508. (8) Hill, J.; Nelson, E.; Tilman, D.; Polasky, S.; Tiffany, D. Environmental, economic, and energetic costs and benefits of biodiesel and ethanol biofuels. Proc. Natl. Acad. Sci. U.S.A. 2006, 103 (30), 11206–11210. (9) Dewulf, J.; Van Langenhove, H.; Van De Velde, B. Exergy-based renewability assessment of biofuel production. Environ. Sci. Technol. 2005, 39 (10), 3878–3882. (10) Patzek, T. W. Thermodynamics of the corn-ethanol biofuel cycle. Crit. Rev. Plant Sci. 2004, 23 (6), 519–567. (11) Cleveland, C. J.; Kaufman, R. K.; Stern, D. I. Aggregation and the role of energy in the economy. Ecol. Econ. 2000, 32 (2), 301–317.

(12) Zhang, Y., Baral. A., Bakshi, B. R. Toward accounting for ecological resources in life cycle assessment. Environ. Sci. Technol. (accepted). (13) Szargut, J.; Morris, D. R.; Steward, F. R. Exergy Analysis of Thermal, Chemical and Metallurgical Processes, 1st ed.; Hemisphere Publications: New York, 1988. (14) Cu ¨ engel, Y. A.; Boles, M. A. Thermodynamics, an Engineering Approach; McGraw-Hill Inc: New York, 1994. (15) Berthiaume, R.; Bouchard, C.; Rosen, M. A. Exergetic evaluation of the renewability of a biofuel. Exergy Int. J. 2001, 1 (4), 256– 268. (16) Ukidwe, N. U.; Bakshi, B. R. Industrial and ecological cumulative exergy consumption of the United States via the 1997 inputoutput benchmark model. Energy 2007, 32, 1560–1592. (17) Dewulf, J.; Bosch, M. E.; De Meester, B.; Van der Vorst, G.; Van Langenhove, H.; Hellweg, S.; Huijbregts, M. A. J. Cumulative exergy extraction from the natural environment (CEENE): a comprehensive life cycle impact assessment method for resource accounting. Environ. Sci. Technol. 2007, 41 (24), 8477–8483. (18) Odum, H. T. Emergy and Environmental Decision Making; John Wiley & Sons: New York, 1996. (19) Hau, J. L.; Bakshi, B. R. Expanding exergy analysis to account for ecosystem products and services. Environ. Sci. Technol. 2004, 38 (13), 3768–3777. (20) Giampietro, P. Comments on “The energetic metabolism of the European Union and the United States” by Haberl and colleagues. J. Ind. Ecol. 2006, 10 (4), 173–185. (21) Hau, J. L.; Bakshi, B. R. Promise and problems of emergy analysis. Ecol. Model. 2004, 178 (1-2), 215. (22) Center for Resilience and Process Systems Engineering Group. Ecologically-Based Life Cycle Assessment (Eco-LCA). Available at resilience.osu.edu/ecolca-ui/ (accessed on April 10, 2009). (23) Green Design Institute. EIOLCA.net, Carnegie Mellon University. Available at http://www.eiolca.net/about.html (accessed on April 2009). (24) Kim, S.; Dale, B. Environmental aspects of ethanol derived from no-tilled corn grain: nonrenewable energy consumption and greenhouse gas emissions. Biomass Bioenergy 2005, 28, 475–489.

ES902571B

VOL. 44, NO. 2, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

807