In the workshop’s sixth session—and the last one on the first day—two speakers spoke about air and land issues related to the increasing use of biofuels. The focus of the first presentation was on determining in detail the way that biofuels-related emissions are distributed in space and time as a way of understanding which groups of people are likely to be most affected by increasing use of biofuels. The second presentation examined how the U.S. forestry industry might be affected by the push to use more biofuels and bio-based products in general.
In the session’s first presentation, Jason Hill, assistant professor in bioproducts and biosystems engineering at the University of Minnesota, discussed the life-cycle impacts of transportation biofuels on land and air quality.
Hill began by noting the wide variation that appears in reports that carry out life-cycle assessments (LCAs) of the various impacts of biofuels production. For example, he displayed the results of seven studies that looked at the life-cycle greenhouse gas emissions of ethanol produced from corn (NRC, 2011). The studies, all published between 2006 and 2011, produced estimates of the amount of carbon dioxide equivalent released per megajoule of energy in the production of corn-based ethanol. The figures ranged from a low of 52 grams of carbon dioxide per megajoule to a high of 177 grams per megajoule. He mentioned that gasoline is approximately 95 grams of carbon dioxide per megajoule, meaning that some investigators have concluded that corn ethanol is a higher emitter
of greenhouse gases and others have found it to be lower. Much of the variation is caused by the studies’ underlying assumptions, he said.
At first glance, this variation seems somewhat surprising, he said, because LCA is standardized by the International Organization for Standardization (ISO). He noted, however, that the ISO standard itself is more concerned with standardizing terminology and reporting than it is in the particular steps one takes when conducting an LCA. To offer a specific, detailed example of how and why LCAs vary, he described how studies of the same phenomenon—ethanol production—might be carried out from three very different perspectives.
Three Approaches to Life-Cycle Assessments
Hill noted that the three approaches he would describe are summarized from a lengthier discussion in a National Research Council report, Renewable Fuel Standard: Potential Economic and Environmental Effects of U.S. Biofuel Policy (NRC, 2011). The approaches examine how three different people might use an LCA, approaching the subject from the point of view of an ethanol plant owner, an ethanol industry analyst, and a government regulator.
To begin with, these individuals might be interested in very different physical volumes of ethanol, Hill said. “An ethanol plant owner may be wondering about the greenhouse gas profile of the fuel being produced in his or her facility. The ethanol industry analyst may be wondering about the average corn ethanol produced in the United States in a given year. And the regulator may be interested in the marginal increase in ethanol production that occurs as a result of a policy, say RFS2 [Renewable Fuel Standard, version 2.0],” he said.
Furthermore, each of the individuals would likely choose different data as model inputs. Consider yield (mass of crop per acres harvested), for example. The ethanol plant owner might choose the yield of the corn that feeds into his or her facility. The ethanol industry analyst might choose the average yield for all U.S. corn production. The government regulator might focus on the future yield that is expected as a result of a certain policy. “All of those methods are correct, but they need to be used only for their intended purposes,” Hill said. “Otherwise, you run into trouble.”
The three different individuals might also differ in which “flavor” of LCA they chose. Attributional LCA looks at the overall impact of a production process—the total emissions, for example—and divides that impact among the different inputs and processes that went into the
production. It requires, for example, a careful examination of the supply chain for a process.
By contrast, consequential LCA seeks to estimate the change in environmental impact resulting from running a given process. It asks, What is the change in total emissions as a result of a marginal change in the production or consumption of a product? This approach includes things such as market-mediated effects that may occur as a result of changes in supply and demand of different products. The two different approaches to LCAs can lead to very different results, Hill said.
Challenges with Life-Cycle Assessments
Although the intentions behind LCAs are good, challenges arise in how they are constructed, applied, and interpreted, Hill said. As an example, he discussed how the U.S. Environmental Protection Agency (EPA) concluded that corn ethanol reduces greenhouse gas emissions 21 percent relative to gasoline. Corn ethanol needed to reduce greenhouse gas emissions by at least 20 percent in order to meet the EPA standard, he noted, although he added, “Most of it was grandfathered in, so it is basically an academic question.” Still, it is valuable to examine how the EPA reached its determination.
Table 6-1 shows the EPA’s estimates of the relative greenhouse gas emissions of corn ethanol compared to gasoline. The EPA estimated that by 2022 corn ethanol produced with energy supplied by natural gas would emit either 17 or 27 percent fewer greenhouse gases than gasoline, depending on whether the process had wet or dry co-products. The EPA performed a weighted average of the 2022 estimates, based on how it believed the industry would be operated over time, and came up with a 21 percent decrease in emissions.
What does not appear explicitly in the numbers, Hill said, is how the estimates for the greenhouse gas emissions associated with corn ethanol production went from 21 to 33 percent greater than gasoline in 2012 to 17 to 27 percent less than gasoline in 2022. What happened, he explained, is both the result of an anticipated increase in industry efficiency and the result of the particular method of accounting for land-use change that the EPA used in their analysis. Specifically, a large contributor to the greenhouse gas emissions of corn ethanol is the change in land use, with forests or grasslands being converted to agriculture either directly or indirectly, with an associated increase in greenhouse gas emissions. In the EPA’s analysis, most of these emissions are allocated
TABLE 6-1 Greenhouse Gas Emissions from Corn-Grain Ethanol Relative to Gasoline as Determined by the EPA
|Biorefinery Heat Source||Dried Distillers Grain with Solubles||2012||2017||2022|
NOTE: * = meets 20 percent reduction in greenhouse gas emissions for corn ethanol; + = higher greenhouse gas emissions than gasoline;-=lower greenhouse gas emissions than gasoline. Dried distillers grains with solubles is a feed co-product produced in wet and dry forms as a result of ethanol production.
SOURCE: EPA, 2010.
to the 2012 and 2017 estimates of life-cycle greenhouse gas emissions, which makes the apparent cost of land use appear artificially low in the 2022 estimates. As a second example of the sorts of challenges that can arise in LCAs of biofuels, Hill discussed the uncertainty in future biomass availability. He described projections from three federal agencies of how the 2022 EPA targets for cellulosic biofuels and other advanced biofuels could be met. As had been noted several times earlier in the workshop, the mandate calls for 21 billion gallons of advanced biofuels to be produced by 2022, of which at least 16 billion gallons must be advanced cellulosic biofuels, in addition to 15 billion gallons of ethanol produced from corn.
Figure 6-1 shows how models from the EPA, the U.S. Department of Energy (DOE), and the U.S. Department of Agriculture (USDA) each project that the 2022 mandate will be met. Although all agree that 15 billion gallons of corn-based ethanol will be produced by 2022, they agree on little else. The EPA model, for example, expects that ethanol made from perennial grasses will make up a major portion of the advanced biofuels mandate—nearly 8 billion gallons—while DOE’s model sees perennial grasses contributing barely more than 3 billion gallons. The USDA model expects more than 7 billion gallons of ethanol per year to be produced from sweet sorghum and other annual energy crops, while DOE’s and the EPA’s models predict very little, if any, ethanol to come from this source. The EPA’s model expects almost 5 billion gallons of biofuels from municipal solid wastes and imported sources, while those sources are almost nonexistent in the USDA’s and
FIGURE 6-1 The types of biomass expected to be produced by 2022.
NOTE: DOE = U.S. Department of Energy, EPA = U.S. Environmental Protection Agency, MSW = municipal solid waste, USDA = U.S. Department of Agriculture.
SOURCE: Keeler et al., 2013. Reprinted with permission. Copyright 2013 American Chemical Society.
DOE’s model projections. In short, Hill said, there are very large differences in how the industry is projected to develop, even among federal agencies.
Similarly, the three agencies differ greatly in their projections of where biomass will be grown for use in fuels. In the USDA’s estimates for corn stover, for instance, Minnesota will be a large producer, while in the EPA’s estimates it will produce none. In the case of perennial grasses, the USDA’s scenario assumes that Conservation Reserve Program land will be open to production, which leads it to predict many acres of perennial grasses being grown in Iowa, Minnesota, and North and South Dakota, while the EPA’s and DOE’s projections include no such use in those states at all. While DOE’s estimate relies almost exclusively on Kansas, Oklahoma, and Texas for the production of switchgrass in 2022, the USDA’s estimate has the production shifted further north. And that, Hill noted, means that the agencies predict very different futures for things such as the change in ecosystem services and the change in water quality. “So, if you’re looking to invest any sort of resources into the development of the industry, you have very different potential outcomes depending on which of these projections you are listening to,” Hill said.
These differing expectations matter, he said, because the estimates are used to determine the different agencies’ plans. The expectations will influence where DOE funds pilot and commercialization facilities, for example, and where the USDA invests its dollars in research programs and how the EPA will institute its final rule for different biofuels.
In assessing the environmental impacts of biofuels, Hill said, researchers have typically relied on two types of studies. The first is the LCA. It uses broad characterization factors and typically provides little detail about where resources are used and where emissions occur, let alone where the impacts occur.
Landscape-level studies, in contrast, typically provide a substantial amount of detail, but often miss the impacts that occur beyond a particular landscape. In particular, they miss impacts that may occur in the supply chain or as a result of market-mediated effects.
Hill then described an approach that he and colleagues have developed that is intended to combine the advantages of the two types of studies and to produce spatially and temporally explicit life-cycle inventories and impact analyses. In his description of the new approach, he focused on air quality, but it can be used to analyze any sort of impact.
In a typical LCA, one sums up the resources used and emissions released in the stages of ethanol production and compares them to the resources used and emissions released using gasoline. With this bottom line in hand, one can then use this information to decide which direction to take.
“There are all sorts of problems with simply using that approach,” Hill said, “but what I want to focus on is how we can improve that.” To illustrate, he showed a graph of the different emissions from gasoline and various biofuels (see Figure 6-2). The numbers come from the GREET (Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation) model from Argonne National Laboratory.
One thing to notice from the graph, he said, is that for all the different types of gases, the tailpipe emissions—labeled as “Use” in the graph—are pretty much the same no matter whether the fuel is gasoline, corn ethanol, cellulosic ethanol from stover, or cellulosic ethanol from switchgrass. For volatile organic compounds (VOCs), a large percentage of the total emissions are tailpipe emissions, so the best approach to decreasing those emissions would be to capture the emissions from the vehicles.
However, in the case of the sulfur oxides, the nitrogen oxides, and ammonia, the vast majority of the emissions come from the production end (except for ammonia, which is a minor production byproduct when making gasoline). “So, if you’re just looking at tailpipe emissions,” Hill said, “you’re missing most of the story of the impact of those different
FIGURE 6-2 Spatial and temporal distributions of emissions from gasoline and various biofuels for six of the highest emitting processes in the life-cycles.
NOTE: NH3 = ammonia, NOx = nitrogen oxide, PM2.5 = particulate matter less than 2.5 microns in diameter, PM10 = particulate matter less than 10 microns in diameter, SOx = sulfur oxide, VOC = volatile organic compound.
SOURCE: Tessum et al., 2012. Reprinted with permission. Copyright 2012 American Chemical Society.
fuels.” The lesson is that it is very important to look at the entire life-cycle when developing impact assessments. “Certainly it’s important to understand the tailpipe emissions, but you need to know a lot more.”
Understanding exactly where those emissions come from can be extremely complicated, Hill said. “There are actually hundreds of different emission sources in the supply chain of the production and use of these different fuels.”
What Hill and his colleagues did was to examine each of the different sources and track the levels of emissions geographically and through time. It involved tracking a number of different pollutants; for example, VOCs consist of many different compounds, each of which may vary from the others in its spatial and temporal profiles. The result was a dataset showing where and when various emissions of various compounds took place in the life-cycle of various fuels. Figure 6-2 shows the sort of results that were produced.
The six sections of the figure show the geographic distribution of six different processes: vehicle tailpipe emissions, ethanol plant evaporation and non-combustion, coal mining with non-combustion included, fertilizer nitrification, gasoline refining, and sulfuric acid production. The spatial emission profiles vary according to which fuel is under consideration: gasoline, corn ethanol, or cellulosic biofuel from corn stover. And there are actually more than six processes, Hill noted, with many hundreds in the production of these fuels alone.
Furthermore, the emission profiles vary in time, both month to month and hour to hour within a day. For example, the profile for vehicle tailpipe emissions in section (a) has two peaks, corresponding to the morning and evening rush hours.
Hill described the process he uses to get complete profiles for emissions from gasoline and biofuels: “You take all those individual profiles and you layer them, one on top of the other. Hundreds of those layer on top of the other for each species for each fuel. What you get is a spatially explicit, temporally explicit inventory of fuels. These are done at hourly time steps over the whole year as well.”
The result is a set of figures like Figure 6-3, which shows the distribution of ammonia emissions around the country for gasoline versus corn ethanol and stover cellulosic ethanol. As can be seen, the emissions are by far the highest in the Midwest because that is where ethanol production is centered. The emissions due to gasoline are far smaller and are concentrated in the large metropolitan areas.
FIGURE 6-3 Cumulative spatial distributions of ammonia emissions from gasoline, corn ethanol, and stover cellulosic ethanol.
NOTE: NH3 = ammonia.
SOURCE: Tessum et al., 2012. Reprinted with permission. Copyright 2012 American Chemical Society.
Hill and his colleagues have also produced cumulative temporal emission profiles that are done hour to hour throughout the day, day to day throughout the week, and month to month throughout the year. The emission of volatile organic compounds and particulate matter dips on the weekend, for example, while ammonia emissions jump during the spring when plants are being fertilized most heavily.
It is not enough to simply look at when and where these emissions occur, Hill said. It is important to understand the ultimate implications for health. “So, we’re looking at advanced dispersion modeling, air quality impacts, and then finally environmental justice.”
He then showed some preliminary results from those sorts of analyses. In particular, he and his colleagues examined the health costs of meeting the mandates set in RFS2, the Renewable Fuel Standard, version 2.0, above and beyond meeting the mandates set in RFS1. To get from RSF1 to RFS2 requires producing an additional 7.5 billion gallons per year of corn ethanol. They also assumed that the mandate will lead to an increase of 5 billion gallons per year of stover cellulosic ethanol. This is not enough to meet the current mandate, but most people do not think that mandate is realistic, so Hill and his colleagues took 5 billion gallons
as a more realistic result, and they choose stover cellulosic ethanol because stover was likely to be the lowest-cost feedstock. “So, in our scenarios, we ran 5 billion gallons a year of stover cellulosic ethanol and 7.5 billion gallons a year of corn ethanol and summed those two and compared them to the energy-equivalent volumes of gasoline that could have been used in their stead,” Hill said.
They produced spatial distribution maps of the various emissions due to the increase in the use of biofuels resulting from RFS2. Not surprisingly, the emissions were concentrated in the Midwest, where the vast majority of biomass production and biofuels plants are located. From there it was possible to estimate the health effects by looking at the geographic distribution of the emissions, combining that with the geographic distribution of the population, and using what is known about how increases in various atmospheric pollutants affect health.
Hill and his colleagues used the data to estimate the number of deaths caused by emissions from the biofuels versus gasoline. Replacing approximately 8 billion gallons of gasoline per year with 7.5 billion gallons of corn ethanol and 5 billion gallons of stover cellulosic ethanol could be expected to lead to an additional 260 deaths per year in the United States, they found. (Ethanol has approximately two-thirds the energy density of gasoline such that the miles that can be driven on 12.5 billion gallons of ethanol is approximately equal to those that can be driven on approximately 8 billion gallons of gasoline.)
One of the more interesting aspects of the study, Hill said, is that it makes it possible to look at the demographic distribution of the health effects—to see, for example, how rural populations are affected by the changes versus urban populations, or low-income populations versus high-income. Preliminary results indicated that in the case of income groups, low-income individuals are slightly more affected by gasoline emissions than high-income individuals, and there is very little difference between the groups for biofuels. In terms of race, nonwhite groups are much more affected by gasoline emissions than white groups, while the difference for biofuels is much smaller, although it is still apparent. Not surprisingly, people in urban areas are approximately twice as likely to die as those from rural areas from the health effects of emissions from the use of gasoline—cities have a far greater concentration of cars. Switching to biofuels evens that out somewhat, as those in rural areas—where biomass production and ethanol production plants are most likely to be situated—are much more affected by the Diofuels-related emissions than are those in cities.
So, in terms of environmental justice, switching from gasoline to biofuels may likely lead to increased mortality, but may somewhat blunt the disparate health impacts of the emissions related to transportation fuels, reducing the current disadvantage experienced by low-income individuals, nonwhite individuals, and people living in urban areas by, in essence, switching to a type of fuel whose related emissions are more likely to affect those who are high-income, white, or living in rural areas.
The session’s second presenter was Daniel Cassidy, a senior adviser for renewable energy and natural resources at the USDA, who spoke on forest management and forest-based bioenergy initiatives.
He began by showing a map illustrating which parts of the United States have forests. The majority of forestland, he noted, is in the Southeast, the Northeast, and the Pacific Northwest. There are about 750 million acres of forest land in the country—roughly the same amount as a century ago—and about 80 percent of it is government-owned. The 20 percent that is privately owned supplies 60 percent of the wood fiber to the forest products industry. About 328 million dry tons of forest products are harvested each year.
The woody biomass that could be used for biofuels consists of the parts of the tree left over after harvesting—stumps, limbs, and branches. Cassidy noted that it had been referred to as wood waste earlier, but he suggested avoiding that term “because if it’s a waste, then it means it’s worth nothing to us.” And some of the woody biomass used for biofuels is wood grown specifically for that purpose, he noted.
If the wood residuals are to be used for biofuels, it will be important to have safe and efficient ways to collect and transport them. The smaller branches and debris—the “slash”—can be put into piles that are then formed into log-shaped assemblies using a material similar to chicken wire; these assemblies are called slash logs. But the larger pieces, Cassidy said, will require the development of equipment to collect and package them for transport, and safety will be an important consideration in the design of that equipment since the larger pieces are heavy enough to cause serious injury.
In the Southeast, pine trees are planted and grown on pine plantations, with the trees evenly spaced in long rows. “It’s our version
of row cropping,” Cassidy commented. The trees take several decades to reach the size at which they are harvested, and while they are growing, he suggested, it would be possible to plant switchgrass between the trees “so you can have an annual harvest of switchgrass and then you can harvest the trees when they reach the size that you need for your commercial operations.”
Why Wood-Based Biofuels Production Could Help the Forestry Industry
The forestry industry supports 1.6 million workers with $50 billion of annual income, he said, but in recent years a number of the pulp mills and paper mills have closed. “So, this is an industry that needs some type of support, and we think that bio-based products and biofuels—or, as I like to call it, the bio-economy—can really fit this need,” Cassidy said.
A lot of the country’s forest stands now are overstocked because they are not getting harvested, he said. “There are just no markets.” And that creates another set of problems in addition to the economic needs of the workers. One threat comes from forest fires. “There have been over 200,000 acres burned just this month alone in the United States—just from January 1 to today [January 24],” Cassidy said. The smoke from the fires creates both air pollution and greenhouse gases. It would be good for the overstocked forests if some of their trees could be harvested and used in the bio-economy.
Furthermore, large sections of forest are filled with trees that have been killed by insects. “In British Columbia, they’ve lost 43 million acres to a bug about the size of a grain of rice,” he said, noting that 43 million acres is about the size of Uruguay. This leaves a tremendous amount of standing dead wood that can’t be used for lumber. “We can’t use it for anything except for the bio-economy, which would be perfect for it,” Cassidy said.
Finally, Cassidy noted that forest areas are regularly being cleared to build new homes and subdivisions. “They clear out forestland because developers can pay a landowner more for it than the landowner can maintain it for.” This offers another potential source of wood for the bio-economy.
The bio-economy, Cassidy said, refers to the production of products based on biological products. Many chemicals are bio-based, he said.
“The global chemical industry is expected to grow by 6 percent annually between now and 2025, and over 20 percent of that growth can come from biomass and bio-based products.” The USDA has a program, the Biopreferred Program, that calls for purchasing, whenever possible, products that are safe and green and come from biomass.
A specific example of a product in the bio-economy is the wood pellets that are now being used in many places to provide heat. There is increasing demand for the pellets, Cassidy said, and much of that growth is being driven by the demand from the European Union, which is moving away from coal and toward using wood products. The pellets are being used to heat an elementary school in Darby, Montana, as part of a Fuels for Schools Program sponsored by the U.S. Forest Service. “We were able to put in a system that would use wood pellets to heat the entire school,” he said. “It’s saving them $62,000 on annual heating costs. In a school system as small as Darby, Montana, that’s a significant amount of money.” He added that the USDA is also working to create bio-based aviation fuels.
The USDA is only interested in bio-based products that can be sustainable, Cassidy said, and they should be sustainable in several ways. They should be economically sustainable, and “the bottom line will guide the path forward.” They should also be environmentally sustainable, which involves a variety of factors: water, air, soil, biodiversity, and carbon sequestration, among others. And they should be socially sustainable. “Is this something that we want in our backyards? Do our communities feel that this is safe and healthy? Do we have the infrastructure to manage these types of facilities?”
There are a variety of opportunities for people interested in developing uses for woody biomass, Cassidy said. The Forest Service has woody biomass research grants, for example. “They are looking at the logistics and engineering plans and safety plans of all the equipment and how we can go out and harvest these wood residues.”
Cassidy explained that he is working on detail as a senior adviser but is also a National Program Leader with the National Institute of Food and Agriculture, “so I’d like to talk a little bit about the opportunities there.” The goal at the institute is to develop regionally appropriate biofuels and bio-based product programs, he said, and one way it is approaching this is with the use of CAPs, or coordinated agriculture programs. The CAPs
bring together academic researchers with people from industry and landowners. “We’re building a large alliance to help support the regional sustainable biofuels systems,” Cassidy said.
The USDA’s efforts to develop bio-based products and biofuels cover a spectrum of approaches, from working to develop superior genotypes of energy crops to improving feedstock production practices, developing conversion technologies that can accept a broad range of feedstocks, and working on regional sustainability analyses and decision-making tools.
The National Institute of Food and Agriculture is currently working with the Agricultural Research Service’s National Agricultural Library to develop an open-source, LCA database for the United States so that “all this great information that universities are producing” will be made available to industry so that companies can do a better job of analyzing their options, Cassidy said.
Other USDA programs are the Agriculture and Food Research Initiative and the Biomass Research and Development Initiative. The latter is working with the Global Bioenergy Partnership, which has come up with a set of criteria for developing bioenergy, including criteria concerning the health of the worker involved.
A brief discussion period followed Cassidy’s presentation. The first question concerned the role that recycling has played in the recent economic woes of the forestry industry. “I think that recycling plays a major role in the paper industry,” Cassidy said, “but I don’t believe it’s what killed the paper industry. I believe what killed it was finding ways to do business cheaper overseas.”
Stephen Reynolds from Colorado State University noted that forestry remains one of the most dangerous occupations in the United States, and the costs to the industry of the occupational hazards are not insignificant. “There hasn’t been a lot of work on this, but Paul Lee at University of California, Davis, estimated that the cost of occupational injuries and fatalities in the United States is around of $250 billion per year. A good proportion of that derives from agriculture and forestry in particular.” This issue of workforce risks and the associated costs should be taken into account in any efforts to develop biofuels from wood products, he suggested.
Bernard Goldstein of the University of Pittsburgh asked whether the database being developed by the National Agricultural Library will have quality standards. In particular, he mentioned Hill’s comment that LCAs are sometimes in need of improvement. Cassidy responded that he believed there would be standards and that they would probably be applied in a way that encouraged careful LCAs. The USDA will probably also use its power as a funding agency to shape the reporting of findings, he suggested. “One of the great things about being a granting agency is … we can basically tell you, ‘If you want this money, this is the standard of reporting you’re going to have to meet.’” Hill reinforced that comment, saying, “From the point of view of somebody who is going to be submitting to that database, they have made very good efforts to engage us in the process—everything from how the databases would be set up to how our reporting could be done so as to be consistent with other institutions.”
Responding to a question from Jack Spengler of Harvard University on the subject of environmental justice and how big a role it is likely to play in decisions concerning biofuels, Hill commented that one of the advantages of LCA is that it makes it possible to quantify some of the impacts of decisions on the different groups or populations that are affected. “If we’re shifting different parts of the burden of these different energy sources or agricultural products … from one part of the supply chain to another, there are going to be different impacts. And if we’re producing large amounts of these biofuels in the upper Midwest and displacing fuels that would otherwise be produced in Texas and North Dakota and California, there are going to be differences there as well. What this lets us do is understand the magnitude of those differences and whether they matter… . It then allows us to target our mitigation efforts more effectively.”
EPA (U.S. Environmental Protection Agency). 2010. Fuel-specific lifecycle greenhouse gas emissions results. Available at http://www.regulations.gov/#!documentDetail;D=EPA-HQ-OAR-2005-0161-3173 (accessed November 15, 2013).
Keeler, B., B. Krohn, T. Nickerson, and J. Hill. 2013. U.S. federal agency models offer different visions for achieving Renewable Fuel Standard (RFS2) biofuel volumes. Environmental Science & Technology 47(18):10095–10101.
NRC (National Research Council). 2011. Renewable fuel standard: Potential economic and environmental effects of U.S. biofuel policy. Washington, DC: The National Academies Press.
Tessum, C. W., J. D. Marshall, and J. D. Hill. 2012. A spatially and temporally explicit life-cycle inventory of air pollutants from gasoline and ethanol in the United States. Environmental Science and Technology 46(20): 11408–11417.