Several studies have been done on energy life-cycle analysis of soybean biodiesel. These include:
All of these studies varied in their conclusions. The variations were due to different assumptions about energy inputs. This article addresses the varied energy inputs for soybean agriculture.
The energy input for soybean agriculture varied from as low as 4,032 megajoules per hectare (MJ/ha) in Ahmed's model, to 15,506 MJ/ha in Pimentel's model (Table 1). Considering average biodiesel production from soybean agriculture of 497 liters per hectare (L/ha) (201.4 L/acre, Peterson (2005)) and biodiesel energy content of 32.5 MJ/l (Mittelbach and Remschmidt (2005)), the biodiesel energy produced from one hectare of soybeans is 16,152.5 MJ. It should be noticed that this number depends on the assumed value of the yield of soybeans per acre. The difference in agricultural energy input alone was about 71 percent of the total energy input to biodiesel.
Analysis showed that the main differences were the assumed energy equivalence for seed, machinery, labor and lime. These four inputs accounted for 69 percent of the agricultural energy in Pimentel's model versus 4.13 percent in NREL and 0 percent in the Ahmed and GREET models. Other energy inputs such as fuel use and bean transportation were also significantly different.
Pradhan et al. (2009) updated the study conducted by Sheehan et al. (1998) (also known as the NREL report) with more recent data, and the new data showed that the actual energy input for agriculture has been reduced. One of the probable reasons was the reduction in agriculture inputs because of the adoption of Roundup Ready soybeans and minimum tillage practice.
Table 1 : Energy use in soybean farming (MJ/ha except for energy allocation)
|Inputs||Ahmed et al. (1994)||ANL (2006)||Sheehan et al. (1998)||Pimentel and Patzek (2005)||Pradhan et al. (2009)|
|Share of energy to Oil(%)||18.803||62.10||18.4||80.362||20.6|
1 Includes both herbicide and insecticide.
2 This included seed, on farm electricity and lime.
3 Back calculated from presented results for comparison.
The energy equivalent of different agricultural inputs varied among models (Table 2). For instance, the energy value of seed in Pimentel's model is almost eight times higher than that assumed in the NREL model. The conversion factor used by the NREL model reflects consideration of life-cycle energy. It is interesting to see that hardly any conversion factors exactly match among models, even for standard inputs such as electricity.
The seed energy refers to the energy equivalence of soybean seed. There are two ways to assign energy equivalence to an input. One is to consider the absolute chemical or physical energy contributed by an input, which is referred to as "calorific value." For instance, diesel fuel has a calorific value of about 36 megajoules per liter (MJ/L) of fuel. Ahmed and the ANL (the GREET model) used this value (Table 2) for energy conversion. Even though this approach is simple and direct, it does not provide insight into renewability and environmental impacts.
The second method uses the energy consumed in producing a specific input as an energy equivalence of that input. This is referred to as "life-cycle energy." For instance, it takes some energy to extract and refine the mineral oil into diesel fuel. That is the reason the energy equivalent of diesel fuel is considerably higher in NREL's and Pimentel's models. The disadvantage of using this second approach is that it can be extremely complex and creates ambiguity in defining the system boundary. For instance, accounting for the energy content in the seed may be quite complex and variable if it were to include the energy consumed by a seed company.
Table 2: Farm inputs energy equivalent (Dashed line indicates the value either not considered or not reported)
|Inputs||Ahmed et al. (1994)||ANL (2006)||Sheehan et al. (1998)||Pimentel and Patzek (2005)|
|LP Gas (MJ/L)||-||23.66||30.63||31.70|
|Natural Gas (MJ/L)||-||-||0.05||-|
|Soybean Transport (MJ/tonne of beans/km)||0.28||-||1.29||1.09|
Usually, the life cycle energy of an input is higher than its calorific value, with seed as an exception. According to the NREL model, it takes 3.16 MJ of total fossil energy per kilogram (kg) of soybean seed production, whereas soybeans contain 16.8 MJ/kg (estimated from the equivalent energy of protein, carbohydrate and fat in the seed). The rest of the energy in seed comes from renewable solar energy trapped by the crop. Renewable energy input is not counted in calculating the total nonrenewable energy use. The NREL report modified the life cycle energy (3.16 MJ) of the seed by a multiplication factor of 1.5 to account for the energy requirement of handling and processing. Pimentel assumed that the energy requirement for seed was twice as much (33.45 MJ) as the calorific value of the soybean, which was almost equal to the calorific value of diesel fuel itself (Table 2).
The ISO 14040 (1998) standard for life cycle analysis (LCA) study requires careful definition of the goal and scope of every LCA. If the purpose of the study is to assess the renewability of biodiesel fuel production, then it makes sense to consider the life-cycle energy of all inputs. However, if the objective of the model is to calculate the system efficiency, it is not incorrect to consider the larger value: either calorific energy, or life-cycle energy. But, it should be kept in mind that according to the second law of thermodynamics, system efficiency is less than 100 percent for all practical fuel production systems. Since life cycle energy was used for agricultural inputs, if not for any other reason than for the sake of consistency, life cycle energy should be considered for seed as well.
The labor and machinery inputs were used only in Pimentel's model. The labor and machinery inputs were derived from previous work by Pimentel and Pimentel (1996). Looking at the original source of these numbers, it was found that the energy for machinery had been only an estimate from an unknown source. In terms of labor use, Pimentel's study used 7.1 humans per hectare, derived from Ali and McBride (1990). For the average human power output, the energy consumed by an average person during an entire year was considered, which is equivalent to human power output of 46.5 kW (62.3 hp).
Calculating labor energy in this manner has some drawbacks. First, the average per capita energy consumption may not represent a person living on a farm compared to a person who commutes 50 miles daily to work. Second, consideration of annual energy consumption by human laborers does not aid in evaluating the renewability of biodiesel, as human food consumption is independent of soybean agriculture. Third, people from other sectors of the economy use the service provided by people involved in biodiesel production through use of biodiesel, thus reducing the consumption of services from the competitive fuel industry (i.e. regular diesel fuel). This makes the accounting very ambiguous and difficult to track how much net labor was incurred or saved. Fourth, people are hired primarily for their ability to perform a task and not for their energy output. The physical energy input from human labor makes up a negligible fraction of total energy input.
Finally, the labor is not considered in other energy LCA analyses (Shapouri et al. (2006); Wu et al. (2006)). Therefore, it is recommended not to include labor as an energy input in biodiesel energy LCA.
Pimentel's model has an error in interpreting lime use data from their original source. Pimentel's model included all of the lime (4,800 kg/ha) as if it applied to one year's soybean crop, which accounted for 36 percent of the total agricultural energy input. The data were taken from Kassel and Tidman (1999), in which lime was recommended for acidic soil only once in several years (usually 5-10 years). Since lime is not applied every year, the total application amount should be spread out on a per-year basis. Crop rotation does not affect the lime allocation when the lime consumption per year is used in the calculation.
While different energy input values are major factors in causing results to vary among the studies, the biggest difference, which can completely reverse the final result, is the method used to allocate the energy use for co-products. The biodiesel process produces not only fuel, but also often seed meal, which can be sold as livestock feed. Since there is more than one final product from the biodiesel process, each product should share the input energy in some way.
The most common method of energy allocation is a mass fraction basis. In other words, energy is assigned in proportion to the mass (weight) of the various co-products. Heavier co-products are assigned more of the input energy. The NREL and GREET models assigned the energy in proportion to the mass fraction of output. One of the outstanding benefits of mass based allocation is its robustness that does not change over time (Pradhan et al., 1998) and results in comparison that is more robust. Disadvantages are that the mass does not always reflect the value of the product.
Another way to assign the energy inputs is based on the calorific value of the co-products (Shapouri et al. (2006)). Calorific value would be a good way to assign energy credits if all co-products are used for energy. In the case of biodiesel, soybean meal is not used as an energy source but as animal feed, and hence it has higher market price than biodiesel for an equivalent caloric value. Pimentel subtracted the energy contained in the meal from total input energy. Hill et al. (2006) also used the calorific value of meal as an energy credit.
Another logical way to allocate energy for different products would be according to the market value of each product (Pradhan et al., 2008). From a producer's point of view, the input cost must be justified by the economic value of the end product. High value products should absorb a bigger portion of the input cost than low value products. One of the costs in biodiesel production is the purchase of input energies, and the fraction of this input energy can be assigned to different end products according to their economic value. The market value of a product can serve as a means to weigh the quality of energy contained in a product. This approach would measure a biofuel's economic sustainability. One drawback to this approach is that market dynamics can change the allocation factor, and a good energy balance today can turn out to be a poor energy balance tomorrow (Pradhan et. al, 2008).
Shapouri et al. (2006) pointed out another method of energy allocation based on the replacement value of the primary product. For biodiesel, the replacement value is based on the energy required to produce a substitute for each co-product. Even though this is one of the scientifically preferred methods (Farrell et al. (2006); Kim and Bruce (2002)), the difficulty with this method is finding the exact substitute for soybean meal. Although dried distillers grain (DDG) or canola meals are animal feed products similar to soybean meal, they are not exact substitutes. Even if they were comparable substitutes, it may be impossible to calculate a precise replacement energy value.
Ahmed,I., Decker, J., and Morris, D. (1994) How much energy does it take to make a gallon of soydiesel? Jefferson City, MO: National Soydiesel development board.
Ali,M.B. and McBride, W. D. (1990) Soybeans: State level production costs, characteristics and input use.: USDA ERS.
ANL (2006) The greenhouse gases, regulated emissions, and energy use in transportation (GREET) model Version 1.6. U.S.Department of Energy laboratory, Argonne, LLC. Available at: http://greet.es.anl.gov/
Farrell, Alexander E., Plevin, Richard J., Turner, Brian T., Jones, Andrew D., O'Hare, Michael, and Kammen, Daniel M. (1-27-2006) Ethanol Can Contribute to Energy and Environmental Goals. Science 311(5760), 506-508.
Hill, J., Nelson, E., Tilman, D., Polasky, S., and Tifanny, D. (2006) Environmental, economic, and energetic costs and benefits of biodiesel and ethanol biofuels. PNSS 103(30), 11206-11210.
Kassel,P. and Tidman, M. (1999) Ag lime impact on yield in several tillage systems. Iowa State University, Ames IA: Integrated crop management.
Kim, Seungdo and Bruce, Dale (2002) Allocation procedure in ethanol production system from corn grain. The International Journal of Life Cycle Assessment 7(4), 237-243.
Lakomy, H. K. A. (1993) Laboratory measurement of human power output during maximum intensity exercise. Physics Education 28(6), 376-379.
Mittelbach,M. and Remschmidt, C. (2005) Biodiesel the comprehensive handbook. 2 ed.: Boersedruck Ges.m.b.H, Vienna.
Peterson,C. (2005) Potential production of biodiesel. In The biodiesel handbook AOCS press, Champaign, IL, pp. 231-238.
Pimentel,D. and Pimentel, M. (1996) Energy use in grain and legume production. In Food energy and society University press of Colorado, pp. 107-130.
Pimentel, David and Patzek, Tad W. (2005) Ethanol production using corn, switchgrass, and wood; biodiesel production using soybean and sunflower. Natural Resources Research 14(1), 65-76.
Pradhan, A., D. S. Shrestha, J. Van Gerpen, and J. Duffield. 2008. The energy balance of soybean oil biodiesel production: A review of past studies. Transactions of the ASABE. 51(1): 185-194.
Pradhan, A., Shrestha, D. S., McAloon, A., Yee, W., Haas, M., Duffield, J.A., and Shapouri, H. (2009) Energy Life-Cycle Assessment of Soybean Biodiesel. United States Department of Agriculture, Agricultural Economic Report Number 845.
Shapouri,H., Wang, M., and Duffield, J. (2006) Net energy balance and fuel-cycle analysis. In Renewable based technology John Wiley & Sons, England, pp. 73-86.
Sheehan,J., Camobreco, V., Duffield, J., Graboski, M., and Shapouri, H. (1998) Life cycle inventory of biodiesel and petroleum diesel for use in an urban bus. NREL/SR-580-24089 Golden, CO: National Renewable Energy Laboratory. U.S. Department of Energy.
Wu,M., Wang, M., and Huo, H. (2006) Fuel-cycle assessment of selected bioethanol production pathways in the United States. Reference No. ANL/ESD/06-7 Argonne, IL: Argonne National Laboratory.