New perspectives on the energy return on (energy) investment (EROI) of corn ethanol
- First Online:
- Cite this article as:
- Murphy, D.J., Hall, C.A.S. & Powers, B. Environ Dev Sustain (2011) 13: 179. doi:10.1007/s10668-010-9255-7
- 687 Downloads
Research on corn ethanol is overly focused on whether corn ethanol is a net energy yielder and, consequently, has missed some other fundamental issues, including (1) whether there is significant error associated with current estimates of the EROI of corn ethanol, (2) whether there is significant spatial variability in the EROI of corn ethanol production, (3) whether yield increases will translate linearly to increases in EROI, (4) the extent to which assumptions about co-product credits impact the EROI of corn ethanol, and (5) how much of the ethanol production from biorefineries is net energy. We address all of these concerns in this research by: (1) performing a meta-error analysis of the calculation of EROI, (2) calculating the EROI for 1,287 counties across the United States, and (3) performing a sensitivity analysis for the values of both yield and co-products within the calculation of EROI. Our results show that the average EROI calculated from the meta-error analysis was 1.07 ± 0.2, meaning that we are unable to assert whether the EROI of corn ethanol is greater than one. The average EROI calculated across 1,287 counties in our spatial analysis was 1.01, indicating that the literature tended to use optimal values for energy inputs and outputs compared to the average conditions across the Unites States. Increases in yield had a trivial impact on EROI, while co-product credits had a large impact on EROI. Based on our results from the spatial analysis and the location of biorefineries across the United States, we conclude that the net energy supplied to society by ethanol is only 0.8% of that supplied from gasoline. Recent work indicates that only energy sources extracted at EROIs of 3:1 or greater have the requisite net energy to sustain the infrastructure of the transportation system of the United States. In light of this work, we conclude that production of corn ethanol within the United States is unsustainable and requires energy subsidies from the larger oil economy.
KeywordsCorn Ethanol EROI Net energy Gradients Spatial analysis
Over the past decade, there has been considerable debate on corn ethanol, most focused on whether it is a net energy yielder. The argument is generally that “if the Energy Return on Investment (EROI) of corn ethanol is positive then it should be pursued.” On one side are Pimentel (2003) and Patzek (2004) who claim that corn ethanol has an EROI below one energy unit returned per energy unit invested, and on the other side are a number of studies claiming that the EROI is positive, reported variously as between 1.08 and 1.45 (Wang et al. 1997, 2007; Wang 2001; Shapouri et al. 2002; Graboski 2004; Shapouri et al. 2004; Oliveira et al. 2005; Farrell et al. 2006). Even with numerous publications on this issue, disagreement remains as to whether corn ethanol is a net energy yielder.
We believe that focus within the literature on whether or not corn ethanol yields a positive net energy gain has diverted attention from other fundamental issues. The following is a brief description of some of these issues and how we addressed them in this research.
First, none of the major studies of the EROI of corn ethanol account for statistical error within their analysis. Error is associated with all measurements, and we should expect there to be error associated with EROI as well. Yet each of Farrell et al. (2006), Wang et al. (2007), Patzek (2004), Pimentel (2003), and Shapouri et al. (2002) fail to report even general error statistics associated with their calculation of EROI. Considering that the range of published values for the EROI of corn ethanol is so small (from 0.8 to 1.5), one would expect that even a relatively small amount of error could be meaningful. In response to these concerns, we performed an error analysis for the calculation of the EROI of corn ethanol.
Second, most analyses to date, including those referenced previously, use optimal (i.e. Iowa) values for corn yield, fertilizer, and irrigation, despite the fact that each of these have large geographical (as well as other) variation. Because of this, they fail to represent the variable nature of corn production across space, and by extension, the subsequent variability in the EROI of corn ethanol. Our spatial analysis addressed this issue by examining the impacts of the natural geographic variability of corn inputs and yields on the EROI of corn ethanol production within the United States.
Third, the assumption about increasing corn yields on the EROI of corn ethanol has resulted in much confusion. For example, Wang et al. (2007) report that yield levels could reach 11,000 kg/ha (180 Bu/Ac) by 2015, which is roughly 25% higher than the average 2005 level. Yet they do not indicate how this will impact the EROI of corn ethanol or what increases in fertilizer, pesticides, etc. will be required to reach these elevated yield levels. Although it is clear that increasing corn yields will increase the gross output of corn per unit area, its effect on the EROI of the entire corn ethanol process is less clear because the corn itself becomes just one of many intermediate inputs. The effect of corn yields on EROI depends upon its fraction of the total energy input to corn ethanol production.
Fourth, the debate over whether the co-products of ethanol production, e.g. Distiller’s Dry Grains, deserve an energy credit warrants exploration. On one side, Patzek (2004) believes that the co-products must be returned to the field to replenish soil humus. On the other side, Wang et al. (1997, 2007), Shapouri et al. (2002), and Farrell et al. (2006) consider the co-product a valuable output of the corn ethanol production process and assign it an energy credit. Unlike yield, the energy content of the co-products is added directly to the energy content of the ethanol in the calculation of EROI. As a result, energy credits for co-products can have a large impact on the EROI of corn ethanol. To address the concerns about the impacts of both yield increases and co-product credits on EROI, we performed a sensitivity analysis to gauge how EROI will change given changes in either input.
Fifth, Mulder and Hagens (2008), Mearns (2008) and Hall et al. (2009) discuss some of the more comprehensive implications of net energy analysis, including (1) using EROI to relate gross and net energy, (2) the concept of “the minimum EROI for a sustainable society,” and (3) “The Net Energy Cliff”. Below we discuss how both of these concepts relate to our EROI analysis.
The objectives of this paper are to: (1) assess the error associated within the calculation of the EROI of corn ethanol, (2) calculate the EROI of corn ethanol for most corn-producing counties across the United States, (3) perform a sensitivity analysis to estimate how changes in both yield and co-product credits impact the EROI of corn ethanol, (4) calculate how much net energy was produced from biorefineries in 2009, and (5) assess whether the production of corn ethanol meets the minimum EROI for the transportation system of the United States.
1.2 EROI definitions and concepts
Hall et al. (2009) and Murphy and Hall (2010) summarize the present state of the definition, theory, and application of EROI research. The most important concepts are summarized below and can be found in greater detail in those publications.
In this study, we examine how the EROI of corn ethanol varies across the conterminous United States. This is a new endeavor, except for the work of Persson et al. (2009), who examined how the net energy of corn ethanol varied across the southeastern United States, a marginal area for corn growth.
1.3 Mass–energy balance and EROI
The value of EROI can also be derived by accounting for the laws of conservation of mass/energy within a system. The conservation laws state that mass and energy cannot be created or destroyed, which is a useful concept when examining a system with numerous inputs and outputs, such as the production of corn ethanol. Determining which inputs and outputs to consider in an EROI analysis depends entirely on the designation of the system boundary. The boundary provides a starting point (energy/mass units enter the system) and an exit point (energy/mass units exit the system). Using these entry and exit points, mass/energy balance equations can be used to account for all mass and energy that enter the system. The equation can be written simply as the sum of inputs per unit area and time must equal the sum of outputs per unit area and time (Patzek 2007). For complex systems, such as the production of corn ethanol, mass/energy balance equations aid in tracking inputs (e.g. corn kernel) from the field to multiple outputs (e.g. ethanol, greenhouse gas emissions and co-products).
Patzek (2007) uses a mass/energy conservation approach to assess the energy balance of the corn ethanol process, explicitly defining the boundaries of analysis. Much of the literature does not use this mass/energy conservation approach, which has resulted in the use of different boundaries of analysis when calculating the EROI of corn ethanol. Mulder and Hagens (2008) echoed this issue, indicating that of the four ethanol studies they examined, three used different boundaries when performing their EROI analysis. In this analysis, we adopt the boundaries used by Patzek (2004) so that our results agree with the principles of conservation of mass/energy.
1.4 Natural gradients of corn and corn ethanol production
As the economist David Ricardo pointed out long ago, farmers tend to raise crops where those crops grow best (Ricardo 1821). This “best first” principle is a well-established principle in economics and resource science; humans will tend to use the best resources first, and then subsequently lower-quality resources. For example, in the early 1900s, the United States was mining copper ore that was about 4 percent copper on average. By 1969, the average yield had dropped to about 0.5 percent copper, because the best resources had been mined first (Lovering 1969).
Ricardo’s principle, derived for agriculture, applies to American farmers. There is a definite hierarchy of corn productivity by state. For example, in 2005, an average of 173 bushels per acre (10859 kg/ha) were harvested in Iowa, while an average of only 113 bushels per acre (7,093 kg/ha) were harvested in Texas. This is consistent also with the general principal of gradient analysis in ecology, which states that individual plant species grow best near the middle of their gradient space; that is near the center of their range in environmental conditions such as temperature and soil moisture (Whittaker 1956; Hall et al. 1992).
The climatic conditions in Iowa are clearly at the “center of corn’s gradient space” and, at least until recent years, most corn raised for alcohol came from this area. What is less well understood, and what we quantify in this research, is that corn production is also less energy intensive (both per kilogram and per hectare) at or near the center of corn’s gradient space, and as a result, the EROI of the corn ethanol process varies across space. This is increasingly important as more ethanol plants are constructed in less optimal areas for corn growth, i.e. outside the states of Iowa, Minnesota, Nebraska, and Illinois.
1.5 Newer concepts within EROI literature
Recent work by Mulder and Hagens (2008), Mearns (2008), and Hall et al. (2009) have brought three new concepts to the discourse on EROI, respectively: (1) how much gross energy must be extracted to deliver one unit of net energy to society, (2) the Net Energy Cliff, and (3) the minimum EROI for fuels to sustain current society. We discuss each of these in turn with respect to corn ethanol.
Work by Mulder et al. (2010) illuminates another issue with the use of low EROI fuels; the consumption of non-energy inputs. They found that the production of ethanol from corn consumes four orders of magnitude more water (both direct and indirect) than the production of diesel fuel from conventional petroleum.
Hall et al. (2009) analyzed the energy requirements of the transportation industry within the United States. They analyzed costs, from the extraction of oil through the distribution of that oil, and even the energy required to maintain roads and bridges. Using these energy costs, they calculated that fuels must have EROIs of at least 3:1 to pay for all the energy costs associated with the transportation system. Fuels that have EROIs below 3:1 act as an energy sink on the transportation system, as they provide insufficient energy for the whole transportation system. In this analysis, we will compare our calculation of the EROI of corn ethanol with the minimum EROI for the transportation system of the United States as calculated in Hall et al. (2009).
We performed four major analyses in this research. The first was a meta-error analysis, in which we quantified the error associated with the calculation of EROI of corn ethanol based on various estimates of the energy inputs and outputs found in the literature. The second was a spatial analysis of the EROI of corn ethanol. The third was a sensitivity analysis; wherein we assess the degree to which corn yields and co-product credits impact the EROI of corn ethanol. Fourth, we combined the results of our EROI analysis with the data of biorefinery production to assess how much net energy was delivered to society by ethanol in 2009. We discuss the details of each analysis in turn.
2.1 Meta-error analysis
The main objective of this analysis was to use average values from the literature to calculate and average EROI with an estimation of error. First, we describe how we chose our variables for the average EROI calculation, and then we describe how we propagate the error through the calculation of EROI.
2.1.1 Variables for meta-analysis
Parameter values required for analysis, with references
Farrell et al. (2006)
Corn energy content
Biorefinery efficiency for corn ethanol
0.40 (L EtOH/kg Corn)
Farrell et al. (2006)
Ethanol energy content
Farrell et al. (2006)
Gasoline energy content
Energy content of a barrel of oil equivalent
Farrell et al. (2006)
Non-spatial energy inputs consumed during the agricultural phase of the corn ethanol process
Wang et al. (1997)
Shapouri et al. (2002)
Farrell et al. (2006)
Value used in Our Study
Machinery and infrastructurea
The five major studies report values that are similar for the following non-farm variables: biorefinery yield, ethanol energy content, and the cost of the biorefinery phase (note: “cost” in this study refers to energy cost, unless otherwise noted, Table 1). We used the average across all studies for each of these variables. There is little agreement among these studies in regards to the energy credits for co-products. The values range from 0.0 (Patzek 2004) to 5.89 MJ/L (Shapouri et al. 2002). We decided to use the average across all studies (3.46 MJ/L) for our calculation of average EROI. We deemed this appropriate because we estimated the effect of using different values for co-products credits within the sensitivity analysis.
2.1.2 Propagation of error
Using this technique of error propagation assumes that the input factors are independent and that all variables are normally distributed. We think these assumptions are acceptable because there is no a priori reason to assume that the errors among the input variables are correlated or that the variables have a non-normal distribution.
2.2 Spatial analysis
We performed a spatial analysis of the energy gains and costs of corn ethanol for the entire United States using spatially variable data for corn yield, fertilizer, and irrigation. Yield data was available at the county-level, while fertilizer and irrigation data were available at the state-level (USDA 2009). Yield, fertilizer, and irrigation data were converted to energy units using conversion ratios provided in Table 1, and assuming an application rate of 20 cm of water per hectare for irrigation (Patzek 2004). The energy value for corn, fertilizer, and irrigation were multiplied by the per county yield (for corn) or per state usage (for fertilizer and irrigation) to attain the total energy input for each item. All data are for 2005. The spatial data for yield, fertilizer, and irrigation were merged with a county and state boundary map attained from the United States Census Bureau using the ArcGIS software program and Federal Information Processing Standards codes (USCB 2007; ESRI 2007). This merge allowed us to view all data spatially. The values used in this study for non-spatial variables were the average of the values listed in five other studies except for seeds and machinery (Tables 1 and 2).
2.3 Sensitivity analysis
To address the impact of possible future higher yields on the EROI of corn ethanol, we calculated EROIs for various scenarios using yield levels that were up to three times greater than the average yield in 2005. We do not expect that average corn yields will reach a level three times greater than the 2005 average; rather we include them to serve as a theoretical maximum to show the trend in EROI given changes in yield. Although increasing yields would certainly require increases in the use of at least some fertilizers, lime, and/or irrigation, for simplicity’s sake, we increased yield levels only, keeping other numbers in the EROI calculation constant.
2.3.2 Co-product credits
To assess how sensitive the calculation of EROI is to changes in co-product credits, we performed three calculations. We first calculated the EROILIT based on the average co-product credits calculated across all five studies (3.46 MJ/L). Then, we calculated the EROI without co-product credits, called the “Patzek Case.” Lastly, we calculated the EROI using a co-product credit of 5.89, called the “Shapouri Case.”
2.4 Gross versus net energy analysis
We used Eq. 3 to estimate how much of the current ethanol production is "gross" and how much is "net" energy. To do this, we overlaid a map of 180 biorefineries onto the map of EROI (RFA 2009). Of the 180 biorefineries that we overlaid on the map, a number of them were excluded from our analysis for one of three reasons: (1) they were in a county for which we had no data, (2) they were ethanol facilities under construction and not producing ethanol in 2009, or (3) corn was not the sole feedstock. Our final list included 127 biorefineries that produced 31.6 billion liters of ethanol in 2009—approximately 93% of total US ethanol production. The merged biorefinery and corn production data (including EROIRG) are included in Appendix 1. We assumed that the EROIRG for the county in which the biorefinery is located is an accurate measure of the EROIRG for corn ethanol produced from that biorefinery. Then, we input the EROIRG and production data for each biorefinery into Eq. 3 to figure out how much net energy is produced from our current ethanol infrastructure. For perspective, we used the biorefinery data to compare gross and net energy production from both corn ethanol and gasoline.
3 Results and discussion
The main results from our study were as follows. (1) The average EROILIT calculated using the literature values for inputs was 1.07 ± 0.2. (2) The average EROIRG value calculated across all counties in the spatial analysis was 1.01. (3) The net energy produced from ethanol in 2009 was only 0.8% of that produced from gasoline in the US. (4) Increasing yields have a trivial impact on EROI, while co-product credits have a strong influence on EROI. Details of these results follow.
The results from our meta-error analysis indicated that the average EROI for corn ethanol was 1.07 with a standard error of 0.1. The 95% confidence interval was 1.07 ± 0.2. This result is interpreted as follows: there is a 95% chance that the true value of the EROI of corn ethanol is contained within 0.2 of 1.07. Alternatively, this calculation means that we are unable to assert whether the true value of the EROI of corn ethanol is greater than one.
Quantity of energy used and produced in the ethanol process reported in various publications (adapted from Patzek 2004)
Agricultural phase (MJ/Ha)
Corn yield (GJ/ha)
Biorefinery phase (MJ/L)
Co-product credits (MJ/L)
Ethanol yield (L/ha)
Wang et al. (1997)
Shapouri et al. (2002)
Farrell et al. (2006)
Summary statistics of the costs and gains of the agricultural phase of corn ethanol production for states that produced at least 1% of the 2005 corn harvest in the United States, ranked by decreasing EROIRG
Agricultural phase inputs (MJ/Ha)
According to Eq. 2, to deliver one liter of ethanol as net energy at an EROI of 1.18 (max found in the spatial analysis), 7.5 liter of ethanol must be produced; 1 liter as net energy and 6.5 liter (or its energy equivalent) to be reinvested to produce more ethanol. If we assume that the average we calculated across all counties (1.01) was the actual value for EROI, then producing ethanol is virtually a zero sum game; i.e. energy produced equals energy consumed.
Applying Eq. 2 to our spatial analysis reveals other interesting results. Eight liters of ethanol must be produced to deliver one unit of net energy in Minnesota, using an EROI of 1.14. Another way, only 13% of the ethanol produced in Minnesota is net energy because the energy equivalent of 87% of the ethanol produced must be reinvested to produce more ethanol. The energy reinvested is in many forms, including, but not limited to, the fossil energy required to generate corn, fertilizer, lime, gasoline, natural gas, diesel, etc. For states with an EROI below 1.0 (Texas and Missouri), the production of ethanol is acting as a drain on the energy system, requiring more energy to produce ethanol than the energy contained in the ethanol product.
The EROI values for counties with biorefineries ranged from 0.64 in Stark, North Dakota, to 1.18 in Phillips, Kansas. Our analysis of 127 biorefineries indicated that of 31.6 billion liters of ethanol produced in the United States, only 1.6 billion liters were net energy (roughly 5%). As a point of comparison, of the 136 billion liters of gasoline consumed in 2009, roughly 122 billion liters (90%) were net energy, assuming that the 136 billion liters were produced at an EROI of 10 (Cleveland 2005). Adjusting for the lower energy content of ethanol (21.46 MJ/L etoh vs. 34.56 MJ/L gasoline = 0.62), we calculated that the net energy from ethanol is roughly 0.99 billion “gasoline-equivalent” liters. Dividing the net energy supplied to society from ethanol by that from gasoline, we calculated that the supply of net energy to society from ethanol is only 0.8% of that from gasoline (0.99/122 = 0.8%). Thus comparing simply the gross production of gasoline-equivalent liters of both ethanol and gasoline is misleading, as one would conclude that the US production of ethanol is 14% of gasoline consumption (19.6/136 = 14%).
3.1 Sensitivity analysis
EROI analysis is highly sensitive to co-product credits. When using the “Patzek Case” (energy credit = 0), the mean US EROI of corn ethanol decreases from 1.07 to 0.91, but when using the “Shapouri Case” (energy credit = 5.89), the EROI increases from 1.07 to 1.17. Thus, the co-product credit alone can determine whether the EROI is less than or greater than one. This contradicts Shapouri et al. (2002) who claimed that the EROI is greater than one before accounting for co-product credits. Using an alternative weighting mechanism, such as price, may ameliorate some of the sensitivity of the EROI statistic to co-product credits.
Fundamentally, the disagreement over the value of co-product credits hinges on one’s attitude toward the science of nutrient cycling and erosion. Those who believe that corn yields are maintained without spreading the nutrients contained in the co-products back onto the field will generally assign a co-product credit in the EROI calculation. Those who believe that the science is unclear will generally assign a conservative co-product credit or even omit the credit altogether. We believe that until a clear consensus emerges, the precautionary principle should apply, and one should be very cautious in assigning coproduct credits.
The debate over the EROI of corn ethanol has been concerned mostly with whether it is a net energy yielder. As such, the dialogue has veered away from many of the larger implications of EROI analyses. Our results indicate that the EROI of corn ethanol is statistically inseparable from one energy unit returned per energy unit invested, and it is likely that much of our ethanol production is acting as an energy sink, requiring more energy for production than that contained in the ethanol product. This conclusion was confirmed in our spatial analysis, where the average EROIRG was 0.06 lower than the average calculated from the literature.
Increasing yields is oft-touted as a way to increase the EROI of corn ethanol, but our analysis indicates that the gains in EROI are small even when the average yield from 2005 was tripled. Co-product credits, on the other hand, have a large influence on the EROI from corn ethanol. There is no consensus within the literature regarding an appropriate co-product value, and until one emerges (one way or another), we should err on the side of caution when applying credits to co-products. Finally, the analysis of ethanol production from biorefineries supports our conclusion from the spatial analysis: the EROI is too low in too many locations to make an impact on our gasoline consumption. Our best estimate is that the net energy provided from ethanol accounts for only 0.8% of the net energy provided by gasoline.
The evidence provided in this research is clear: we do not know the exact EROI of ethanol, but even if we are remotely close (± 0.2), we are still, in the best case scenario, gaining an insignificant amount of net energy. Furthermore, Hall et al. (2009) estimated that only fuels with an EROI greater than 3:1 provide the requisite net energy to provide a fuel source and to maintain the infrastructure associated with the current U.S. transportation system. Fuels that have an EROI below 3:1 require subsidies from other energy sources to pay for all of the infrastructure associated with the transportation system of the US. The EROI of corn ethanol that we calculated is lower than the 3:1 threshold, indicating that corn ethanol requires large subsidies from the general fossil fuel economy, and as a result, drains energy from the US transportation system.
The authors would like to thank the Santa Barbara foundation for financial support. We would also like to thank 3 anonymous reviewers for many helpful comments.