CC BY-NC-ND
0Contents lists available at SciVerse ScienceDirect
International Dairy Journal
journal homepage: www.elsevier.com/locate/idairyj
Carbon footprint analysis of dairy feed from a mill in Michigan, USA
Felix Adoma,b, Charles Workman a, Greg Thomac, David Shonnarda,b*
a Department of Chemical Engineering, Michigan Technological University, 1400 Townsend Drive, Houghton, MI 49931-1295, USA b Sustainable Futures Institute, Michigan Technological University, 1400 Townsend Drive, Houghton, MI 49931-1295, USA c University of Arkansas, Ralph E. Martin Department of Chemical Engineering, 3202 Bell Engineering Center, Fayetteville, AR 72701-1201, USA
CrossMark
ARTICLE INFO
ABSTRACT
Article history: Received 21 September 2011 Received in revised form 28 May 2012
Accepted 18 September 2012
A carbon footprint analysis was conducted for a single dairy feed mill located in Michigan, USA with the aim of developing a preliminary assessment of dairy feed mill operations. The goal was to determine the greenhouse gas (GHG) emissions for 1 kg of milled dairy feed. Inputs and activities identified in this analysis included production of feed ingredients, onsite energy, and transportation of feed inputs to the milling site and mill output to dairy farms. Feed mill GHG emissions were calculated to be 0.62 and 0.93 kg CO2-eq (equivalent) kg-1 of milled dairy feed for economic and mass allocation, respectively. The highest emissions were due to the feed ingredient inputs that contributed 73—82% toward the carbon footprint, depending on the allocation method. Energy and transportation impacts together contributed between 8 and 12%. Scenarios investigated feed ingredient inputs likely to represent different USA mill locations.
© 2012 Elsevier Ltd. All rights reserved.
1. Introduction
Key business decisions should take into account environmentally-benign processes and products as a means of addressing environmental issues. It was on this premise that the USA dairy industry embarked on a project to study the greenhouse gas (GHG) emissions from the production of milk in the USA dairy industry. Findings from this dairy study were presented in a report by Thoma et al. (2010). Subsequently, Thoma et al. (2013) reported nine major stages comprising the USA dairy industry as (i) feed production stage (cultivation of grain and forage crops and other mill feed ingredients plus mill operations and all transportation steps), (ii) milk production, (iii) delivery to processor, (iv) processing, (v) packaging, (vi) distribution, (vii) retail, (viii) consumption and (ix) disposal. Analyzing each stage separately and then combining all stages provided the carbon footprint of the USA dairy milk supply chain. The analysis reported here, however, required a carbon footprint study of a USA dairy feed mill as part of the feed production stage listed above. Additionally, a detailed literature review by the authors revealed that no previous studies were found with regard to carbon footprint analysis of any animal feed mills in the USA. Shaw, Buharivala, Parnell, and Demny (1998) investigated the development of emission factors for unloading grain and loading feed at mills
* Corresponding author. Tel.: +1 906 487 3468. E-mail address: drshonna@mtu.edu (D. Shonnard).
0958-6946/$ — see front matter © 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016lj.idairyj.2012.09.008
for cattle feed yards. A recent global dairy sector GHG emissions life cycle assessment (LCA) compared impacts of fat- and protein-corrected milk production and processing for different countries and agricultural cultivation settings, but did not include an analysis of dairy feed mills (Gerber, Vellinga, Opio, Henderson, & Steinfeld, 2010). Therefore, our study makes a contribution in understanding the GHG emissions of dairy feed mills and identifies major mill inputs contribution to the carbon footprint.
The American Feed Industry Association (AFIA), which represents the USA animal feed industry, is a trade association which estimates that approximately 3,000 feed mills exist in the USA and these mills produced between 107, 000 to 112,000 million kg of animal feed over the last ten years (Batal et al., 2008). The feed mill sector is a very important part of the agricultural industry for the USA from an economic perspective because the sector directly employs about 110,000 individuals and contributes approximately $35 billion from feed sales toward the USA economy annually (International Feed Industry Federation, 2009). The mandatory reporting of GHG emissions proposed by the United States Environmental Protection Agency (USEPA) requires industrial facilities emitting more than 25 million kg of CO2 equivalents each year to report to the USEPA. This study calculates the magnitude of GHG emissions expected from a dairy feed mill, whose facilities have yet to be subject to such analysis in the USA. Specific study goals were to develop an LCA methodology applicable to the animal feed mill industry to accommodate a large number of inputs and activities
associated with dairy mill operations, and to gain an understanding of the relative importance of milled dairy feed inputs and activities on the GHG emissions of the outputs of the mill (that are themselves inputs to dairy milk production) through the application of these developed methodologies.
2. Materials and methods
2.1. Goal and scope definition
This is an analysis of a single dairy feed mill including transport of milled dairy feed to various dairy farms in Michigan. The scope of this carbon footprint analysis did not include biogenic carbon removals and emissions, emissions from employee travel to or from the mill, the impacts of manufacturing the mill itself, and other passenger vehicles used on the milling premises. The goal was estimation of GHGs emitted from feed mill operations on the basis of 1 kg of dairy feed output from the mill (kg CO2-eq kg-1 of milled dairy feed), including delivery to local dairy farms. The scope specifically included GHG emissions only (see Fig. 1). The study authors acknowledge that different formulations for dairy feed are possible depending on animal age and other factors. Indeed, the mill under study produces custom formulation of dairy feeds for specific customers. However, this analysis was meant to determine the impacts of producing dairy feed averaged over a typical year, by extrapolating the data provided over an annual cycle.
2.1.1. Audience
This study was a subsystem of a larger study undertaken for the USA dairy industry sector, yet the results are relevant to animal feed mill industry sector, the general public and federal government agencies responsible for the regulation of emissions from industrial operations.
2.1.2. Functional unit
The functional unit was 1 kg of milled dairy feed at its exit moisture content (an average feed formulation for dairy animal nutrition at this mill).
2.1.3. System boundaries
System boundaries included production and transport of feed inputs (grain crops, processed feed components, nutrients and other additives, and energy use) to the mill, for milling of the feed ingredients, to the delivery of milled feed to dairy farms. Fig. 1 shows a schematic diagram (black line indicates the system boundaries) for the stages considered in this analysis. The green
ellipses represent the various inputs at each stage while the red rounded squares represent corresponding emissions.
To the extent possible, ecoinvent™ unit processes (PRe Consultants, 2009) have been used. The ecoinvent™ data are mostly based on European conditions, whereas the geographic context of our study was the USA. This situation introduced a geographic-relevance conflict; however, technology relevance is still strong because both EU and USA manufacturers use modern production technology. For major crop and agricultural by-product inputs to this study, we have developed inventories based on our own research using USA data sources. There were many inputs for which unit processes were modeled using Open input—output (IO) data (Sustainability Consortium, 2011) and also some data were obtained from peer reviewed journal articles. Differences in system boundaries, particularly between input—output and process-based models will result in inconsistent system boundaries. This is because Open IO models in essence have no specific boundary cut-off criteria. However, in this study, a relatively small fraction of the mass of feed inputs to the mill has been modeled with the IO approach. The specific items for which IO data have been used are restricted to nutritional supplements for feed ingredients in category 3. Section 2.4.1 provides more details on the different categories of feed ingredients.
2.1.4. Geographical boundaries
This mill, located in the lower peninsula of Michigan, is the geographical context for this carbon footprint study. It is a modern milling site with the bulk of its milled animal feed being dairy feed. Results from this mill carbon footprint analysis may not be representative of other dairy feed mills in the USA. However, in an attempt to model mills from other locations in the USA, sensitivity analyses in section 4 of this article model GHG emissions of milled dairy feeds with a predominance of dry distillers grains and solubles (DDGS), soybean meal, and oats, respectively in separate scenarios.
2.1.5. Allocation procedures
The ISO guidelines were followed for co-product allocation in this carbon footprint study. Specifically, ISO standards 14040:14044 (ISO, 2006a,b) and British Standards Institute (2008) recommend the avoidance of allocation by using system expansion. However, system expansion was not possible in our study given that LCA results are not currently available to credit the non-dairy feed products from this mill. Apart from this, it has been stated in section 2.1.1 that this study was a subsystem of a larger study (Thoma et al., 2013). In the overall study, economic and mass allocations were used, and hence to be consistent we used both of these allocation approaches. An economic allocation factor of 0.90
[ CO;, CHj, N;Q, CFCs ] Fig. 1. Schematic diagram of various stages for dairy feed mill carbon footprint analysis.
was used for milled dairy feed based on consultation with the mill manager who indicated that 90% of total mill revenue generated was attributable to the sale of dairy feed output. A mass allocation factor of 0.88 was used based on the fact that 88% of the mill outputs were dairy feed while the remaining outputs were non-dairy feed products.
2.2. Collection of input data
Data collection efforts have been a combination of a survey instrument developed for the mill manager, internet searches (e.g., ISI, Google scholar, ProQuest, etc.), peer-reviewed journal articles, a mill site visit, and direct communication with the feed mill manager. Inputs such as types of feed, mass of each feed ingredient, transportation distances, as well as unit and total cost of feed ingredients were all obtained from the purchase history documents of the milling facility, provided by the mill manager. The next sections show how input data were collected and organized as well as some sensitivity analyses considered in this study.
2.3. Developing a data collection spreadsheet (survey)
The life cycle inventory (LCI) stage of this project required gathering input and output data for the milling operation. A survey instrument was created and used to collect data from the mill facility (see Appendix A of Supplementary Materials (SM)). This survey instrument can broadly be categorized into three major sections. Questions in Appendix A-l (SM) sought information on the various types of fuel used in the milling operations, types of feed produced aside from dairy feed, and the annual energy consumption for the milling processes. The main objective in Appendix A-2 (SM) of the survey instrument was to determine the kind and amount of feed that go into producing starter, lactating and dry feed for dairy cattle. In the transportation section, Appendix A-3 (SM), questions specifically targeted the transportation of feed inputs to the milling site, including modes of transportation, the kind of road vehicles used, and distances covered in transporting feed ingredients to the milling site. The data obtained were collected between March l and June 30, 2009. The feed mill manager confirmed that this dataset was representative of annual production.
2.4. Organization of input data for carbon footprint analysis
As identified in Fig. l, input data from this mill facility were organized for this carbon footprint analysis into feed ingredients, transport of feed ingredients to milling site, mill electricity and natural gas use, and milled product transportation.
2.4.1. Categories of feed ingredients and sources of inventory data
The feed ingredients were organized into three categories based on (i) specific functions, (ii) source of emission factors, and (iii) environmental impact modeling approach. The total 4 month input of feed ingredients to the mill was approximately 9,683,000 kg, and this was increased to an annual input (three-fold increase) in consultation with the feed mill manager. The mill manager confirmed that inputs equal to mill feed outputs.
The first category of mill inputs was the majority of feed ingredients on a mass-input basis (Category l). Inventory data for these ingredients were obtained primarily from unit processes in the ecoinvent™ database and also from the study by Adom et al. (2012). This first feed category was comprised mainly of soybean co-products, DDGS, and other high-mass inputs. Table l shows the individual feed components, their overall percentage contributions toward the feed mill inputs, and organizes these components into
Table 1
Major feed inputs on a 4-month basis: soybean, dried distiller grain and other co-products (Category 1).
Feed type Feed inputs (T, truck; R, rail) Units purchased Percentage
(1000 kg) (%)
Cottonseed Fuzzy cottonseed (T) 124 1.28
Dried distiller Corn gluten feed bulk (T) 578 5.97
grain Distillers bulk (T) 843 8.70
Corn gluten direct (T) 127 1.31
Direct distillers (T) 89 0.91
Soy meal Canola meal (T) 304 3.14
Heifer concentrate 35% (T) 6 0.07
Heifers edge direct (T) 27 0.28
Soybean meal 48% direct (T) 83 0.86
Chief beef finisher 36 (T) 25 0.26
Dairy beef finisher (T) 3 0.03
Bran meal 50# (T) 0.05 0.0005
Bulk 48% soy 50# (T) 1915 19.78
Heifers edge bulk (T) 46 0.48
Soy chlor 16 50# (T) 11 0.11
Soy plus bulk 50# (R) 3271 33.78
Vita soy bulk (T) 6 0.06
Sugar Dairy sugar 38(T) 53 0.54
Dairy sugar 38(T) 8 0.08
Soy hulls Direct soy hulls (T) 22 0.23
Direct soy plus (T) 21 0.22
Soy hulls bulk (T) 189 1.95
Animal meal Blood meal 50# (T) 0.005 0.0005
Fish meal 50# (T) 4 0.04
Pork and bone meal bulk (T) 108 1.12
Fat A/V blend fat bulk (T) 94 0.97
Choice white grease bulk (T) 79 0.82
Energy booster 100 50# bag (T) 30 0.31
Megalac 50# (T) 2 0.02
Molasses Dry molasses 50# (T) 7 0.07
Liquid molasses-bulk (T) 29 0.30
Molasses tub-16% (T) 1 0.01
Molasses tub-25% (T) 1 0.01
Direct molasses (T) 5 0.06
Oats Rolled oats 50# (T) 2 0.02
Urea Feed urea bag 50# (T) 41 0.43
Whey Dried whey 50# (T) 2 0.02
Totals 8158 84
major feed types for which inventory data were available. Reported feed types in both Tables l and 2 were obtained from the purchase history document obtained from the feed mill manager. The percentage composition of the individual components making up the total 4 month input were estimated by dividing their individual masses (kg) of feed types by the total (9,683,000 kg). For this particular feed mill, soybean meal-type feed alone accounted for approximately 59% of the mill inputs while DDGS contributed close to 17%. Category 1 of the feed ingredients contributed about 84% of the mill's total feed input by mass.
Miller, Ramsey, and Madsen (1988) and Siciliano-Jones, Socha, Tomlinson, and DeFrain (2008) established that trace minerals such as Zn, Mn, Cu, and Co plays a very important role in overall health of dairy animals. For example, these trace minerals help in protein synthesis, vitamin metabolism, formation of connective tissue, and immune function in animals. The second category of feed ingredients (see Table 2, Category 2) comprised mineral ingredients and other feed components, contributing approximately 12% by mass to the feed mill inputs. These are highly-processed feed ingredients. For example, dairy base mix (Hubbard Feeds, 2007) provides calcium, phosphorous, magnesium, and other trace minerals.
The largest input to Category 2 ingredients was soda powder, contributing approximately 5% toward total feed mass. In addition to serving as a source of sodium, soda powder also offers buffering qualities that help stabilize rumen pH by reducing acid conditions. Finally, feed input labeled mineral mixture contributed less that
Table 2
Feed inputs on a 4-month basis: minerals and others (Category 2).
Feed type Feed inputs (T, truck; R, rail) Units purchased (1000 kg) Percentage (%)
Gypsum Calcium sulfate bag 50# (T) 27 0.276
Lime Hydrated lime 50# bag (T) 2 0.019
Limestone Calcium carbonate bulk (T) 281 2.903
Calcium carbonate 50# (T) 13 0.131
Dical bag 50# (T) 1 0.009
Magnesium oxide Magnesium oxide bag 50# (T) 19 0.197
Magnesium sulfate Magnesium sulfate 50# 4 0.044
(MgSO4) bag (T)
Other trace 24-12 mineral 50# (T) 2 0.023
minerals Copper sulfate — fine 50# (T) 2 0.019
Copper sulfate — cryb 50# (T) 0.3 0.004
Dairy base mix bulk (T) 85 0.879
DCAD plus-potasm carb 10 0.103
50# (T)
Dical/monocal bulk (T) 49 0.505
Iodine 50 50# (T) 0.05 0.001
Manganese sulfate 50# (T) 2 0.019
Propnos mineral w/ 0.005 0.002
altosiu (T)
Minerals mixture 38 0.392
Salt (NaCl) Mixing salt bag (T) 14 0.149
Tm bocks w/sel (T) 4 0.041
Tm salt bag (T) 10 0.103
Mixing salt bulk (T) 136 1.405
White salt blocks (T) 2 0.026
White salt 50# (T) 2 0.023
TM blocks (T) 7 0.072
Soda powder Bicarb bulk (R) 445 4.596
Bicarb-bag (T) 9 0.090
Totals 1165 12
0.5% toward the feed milling input by weight even though it was comprised of 41 different ingredients (see Appendix B of SM). These ingredients contain varying concentrations of trace minerals such as selenium, copper, zinc, among others, which were grouped and referred to as minerals mixture. Inventory data for Category 2 dairy feed inputs were obtained from ecoinvent™.
The third category for the feed mill inputs (Category 3) was comprised of 66 different components with much smaller amounts on a weight basis (see Appendix C of SM). This category mainly included highly-processed ingredients like vitamins and amino acids such as lysine 98.5%, methionine, aureomycin 50, among others. This category, however, contributed approximately 4% toward the mill inputs by mass. Inventory data for Category 3 dairy feed inputs were obtained from the Open IO database because the ecoprofiles for them were not available in ecoinvent™ or any other literature sources.
Open IO is a comprehensive analytical database developed and created by staff of the Applied Sustainability Center at the Walton College of Business, University of Arkansas for the Sustainability Consortium (2011). In analyzing feed inputs in Category 3, the economic sector most closely related to these mill input ingredients was identified as "other food manufacturing" (sector-311119) and was used to complete the inventory. This sector ecoprofile was imported into SimaPro and modified to remove the contribution of Category 1 and 2 inputs, and the outputs re-normalized so that the relative contribution of all other sectors would be proportionally increased.
2.4.2. Onsite energy
For the energy analysis in this study, two major inputs were identified using data obtained from the mill operation survey: electricity and natural gas. The total electricity used (kWh) for three electricity meters was obtained for an eleven month period
Table 3
Summary of electricity inventory data for milling site from 2008 to 2009 (11 months).
Meter # kWh
10988145 21,940
7838695 42,514
83157581 38,270
(see Table 3). Electricity consumption averaged over the eleven month period was used as an estimate for the twelfth month to obtain the total annual electricity used. In the case of natural gas, annual average for natural gas used at the site for 2007 and 2008 were used in the calculations. Data for natural gas inputs are presented in Table 4.
2.4.3. Transportation
The goal for the transportation analysis was to model the GHG emissions of transportation of feed ingredients to the mill site as well as the milled products to the various local dairy farms. For this section of the analysis, the site manager provided the required data inputs for assessing both steps. Appendix D of SM shows transportation data of all the feed ingredients input to the mill facility. These data included the miles traveled, amount transported, and transportation mode. Appendices D-1, D-2 and D-3 show the transportation inputs in terms of miles traveled for feed ingredients in categories 1, 2 and 3, respectively. Using this information, ecoinvent™ ecoprofiles most closely matching transport mode were used. A 16,257—32,514 kg European road transport ecoprofile and a USA freight train ecoprofile were selected from the ecoinvent™ database. The freight train emission factor used was 3.8 10-5 kg CO2 -eq (kg km) 1, and multiplying this by the corresponding payload—distance (kg km) values for each ingredient, the total GHG emissions for each ingredient transported were estimated. Using a similar approach for a 16,257—32,514 kg capacity road transport, with emission factor of 1.7 x 10-4 kg CO2-eq (kg km)-1, the GHG results were estimated for road transport of feed ingredients.
Inputs for the transportation of milled dairy feed products using the mill fleet of trucks to local dairy farms were provided by the mill manager in terms of the diesel use. These transport inputs are summarized in Table 5. Data covered the period January 2007 to August 2009; however, the average amount of diesel used for transportation in 2007 and 2008 was used in this analysis due to the incomplete data reported in 2009. Using diesel density of 840 kg m-3 and heating value of 42.8 MJ kg-1 of diesel (Edwards, Larive, Mahieu, & Rouveirolles, 2006), the total mass (kg) as well as the total amount of energy (MJ) were estimated. Inventories of GHG emissions for production and combustion of diesel were obtained using the ecoinvent™ profile "diesel, burned in diesel-electric generating set/GLO S" (90 g CO2 MJ-1), which closely approximates diesel emissions from use in trucks.
Table 4
Summary of natural gas inventory data for milling site from 2008 to 2009 (1 year).a
Year Cubic meters
2008 125,826
2009 180,401
Total (Average) 153,115
a The ecoinvent profile used for natural gas is: heat, natural gas, at boiler modulating <100 kW RER S. The emission factor for electricity assuming state grid was modified according to the study by Deru and Torcellini (2007).
Table 5
Summary fuel usage input data (average for 2007 and 2008) for road transport of milled feed product from mill to Michigan dairy farm.
Date Diesel (m3) Total mass (kg) Total amount of energy (MJ)
1/1/2009-8/31/2009 58.94 49,512 2,119,123
1/1/2008-12/31/2008 145.11 121,903 5,217,443
1/1/2007-12/31/2007 135.28 113,648 4,864,142
Average (2007-2008) 140.19 117,776 5,040,792
2.5. Life cycle impact assessment
The IPCC GWP 100a method in SimaPro 7.3 was used to convert GHG inventory data into equivalent emissions of CO2. This method uses global warming potentials (GWPs) of 1 for CO2,25 for CH4, and 298 for N2O. In addition to these three greenhouse gases, the analysis included emissions of refrigerants and of other chemicals with high GWPs that were included in the inventory data from ecoinvent™ and the open IO model.
2.5.1. Emission factors for GHG analysis
Table 6 summarizes the GHG emission factors used in this mill analysis. The majority of GHG emission factors for inputs to the feed mill were obtained using ecoprofiles™ in the ecoinvent™ database or were generated from original crop inputs from another study (Adom et al., 2012). In the case of sugar and animal meal, emission factors for these inputs were obtained from LCA Food Database (Nielsen, Weidema, Dalgaard, & Halberg, 2003). Also, the emission factor for "other trace minerals" was a unit process comprising of all the commonly used minerals in feed input category 2. Emission factors used for electricity and natural gas from the ecoinvent™ database were 0.82 kg CO2-eq kWh-1 assuming a Michigan grid mix and 0.075 kg CO2-eq MJ-1 of natural gas.
2.6. Sensitivity analyses
Sensitivity analyses were performed to compare three major scenarios to the base case study (the MI mill inputs). In the base
case, soybean meal dominated the ingredients on a mass-input basis by contributing 59% (w/w), while DDGS from dry corn mill facility contributed 17% (w/w). In Scenario 1, we investigated the feed mill's GHG impacts when using DDGS from a wet corn mill facility as oppose to a dry mill, without changing the mass input contributions of any other feed inputs. Scenarios 2 and 3 investigated the impact of input grain crop type by modifying the major crop inputs. To investigate a DDGS dominant case, DDGS from a dry corn mill and soybean meal were assumed to contribute 59% and 17%, respectively to the total feed input in scenario 2 (the inverse of the MI mill). In scenario 3, oats was assumed to contribute 42% and DDGS (from dry mill facility) and soybean meal were assumed to each contribute 17% to the total feed input on a mass-input basis. These scenarios reflect the geographical preferences for the feed inputs. For example, DDGS is likely to be dominant over soybean and soybean meal in regions with high production of DDGS such as Iowa (scenario 2). Scenario 3 is more relevant for regions where oats is more prevalent in the local grain-crop supply, such as North and South Dakota. In section 4 of this manuscript, results obtained from the various scenarios investigated are presented.
3. Results and discussion
3.1. LCA results and discussion of base case
3.1.1. GHG impact of a dairy feed mill in Michigan, USA
In Fig. 2, the GHG footprint contributions of various inputs and activities for the base case study are presented. The pie charts compare the effect of allocation choice on the resultant carbon footprint for the mill output. For both mass and economic allocation (Fig. 2A and B), the majority of the GHG footprint of the dairy feed mill products were due to the input crops and other major ingredients to the mill (Category 1 inputs contributed approximately 84% of mill inputs by mass). Depending on allocation used, 73—82% of the total feed mill's GHG footprint was attributable to feed inputs in category 1. Category 1 impact was lower (73%) in the feed mill's GHG footprint when economic allocation
Table 6
Emission factors and mill greenhouse gas analysis.
Feed inputs Emission factors (kg CO2-eq kg MA 1 feed input)3 EA Reference
Category 1 PRé Consultants (2009)b
Cottonseed 1.27 0.39
DDGS (dry mill) 2.30 0.91 Adom et al. (2012)
DDGS (wet mill) 2.21 0.67 Adom et al. (2012)
Soy meal 0.54 0.41 Adom et al. (2012)
Sugar (cottonseed, at regional storehouse/US U)b 0.51 0.51 Nielsen, Nielsen, Weidema,
Dalgaard, and Halberg (2003)
Soy hulls 0.50 0.41 Thoma et al. (2010)
Animal meal 0.07 0.07 Nielsen et al. (2003)
Fat (tallow, at plant/CH U) 0.66 0.66 PRé Consultants (2009)
Molasses 0.11 0.11 PRé Consultants (2009)
Oats 0.58 0.58 Adom et al. (2012)
Urea (as N, at regional storehouse/RER U) 3.30 3.30 PRé Consultants (2009)
Category 2
Gypsum (mineral, at mine/CH U) 0.002 0.002 PRé Consultants (2009)
Lime (hydrated, loose, at plant/CH U) 0.75 0.75 PRé Consultants (2009)
Limestone (milled, loose, at plant/CH U) 0.013 0.013 PRé Consultants (2009)
Magnesium oxide (at plant/RER U) 1.05 1.05 PRé Consultants (2009)
Magnesium sulfate (at plant/RER U) 0.30 0.30 PRé Consultants (2009)
Other trace minerals (mixture, at factory/US U) 1.59 1.59 PRé Consultants (2009)
Sodium chloride (powder, at plant/RER U) 0.18 0.18 PRé Consultants (2009)
Soda powder (at plant/RER U) 0.44 0.44 PRé Consultants (2009)
Category 3
Supplements 1.07 1.07 Open IO database
a MA, mass allocation; EA, economic allocation. b Ecoinvent™ database. CH is data for Switzerland, U is unit process-based data, RER is average for Europe, US is United States.
Fig. 2. Relative contribution to greenhouse gas emissions of milled dairy feed (base case analysis): Panel A, mass allocation; panel B, economic allocation.
was used. This was because the emission factors (Table 6) for co-products such as cottonseed, DDGS and soybean meal on economic allocation basis were smaller given the lower value of these co-products in the market compared to those estimated using a mass allocation.
The next largest category for GHG emissions was mineral ingredients (Category 2) that contributed approximately 6-9% to total mill carbon footprint (and 12% of total feed mass). The next largest category for GHG emissions was supplements (Category 3), which contributed 4-7% of the carbon footprint depending on allocation method (approximately 4% of total feed mass). Category 3 feed input GHG impact was estimated using Open IO data, and thus has a different system boundary than other inputs, as discussed in section 2.1.3. Nonetheless, this larger GHG intensity (per unit mass of Category 3 input) was expected given that many of these inputs (e.g., amino acids) were subjected to much more processing compared to the major crop inputs (e.g., oats, soybean meal, DDGS). An analysis of all unit processes contributing to the feed mill showed that the economic IO data represents about 4% of the total mill carbon footprint, and thus system boundary
inconsistencies do not have substantial influence on the final GHG results.
On-site energy consumption at the mill contributed only about 2-3% (see Fig. 2) to the total GHG emissions depending on allocation, and natural gas for crop drying accounted for 80% of this energy impact.
All transportation, both raw material delivery and distribution of the feed to local MI dairy farms, contributed approximately 6-9% of the footprint depending on allocation method, as shown in Fig. 2. Section 3.2 provides details of the transportation impacts.
3.1.2. Discussion of base case LCA results for annual emissions
Category 1 feed inputs contributed approximately 19 and 11 million kg CO2-eq y_1 for mass and economic allocation, respectively. This was due to high mass input rate and differences in emission factors based on economic and mass allocation as previously explained in section 3.1.1. Category 2 inputs contributed 1.3 and 1.4 million kg CO2-eq y_1 for both allocation methods considered. Category 3 input contributions were approximately 1 million kg CO2-eq y_1 for both allocation methods considered. In
the final analysis, the total GHG emission of all feed inputs of this milling site was estimated to be approximately 22 and 14 million kg CO2-eq y"1 for mass and economic allocation, respectively.
A total of approximately 1.4 and 1.5 million kg CO2-eq y"1 for mass and economic allocations, respectively, was the estimated GHG emissions due to fuel inputs associated with transportation. This accounted for GHG burdens due to transport of all feed ingredients to the milling site as well as the transportation of the processed dairy feed to various dairy farms. Fig. 2 provides more details on the transportation impact. GHG burdens due to the transportation of feed ingredients to the milling site were about three times more than the impact due to the transport of milled dairy output to the various dairy farms. Transportation impact of feed ingredients (all feed categories) was estimated to be about 1 million kg CO2-eq y"1 whereas transportation to various dairy farms was estimated to be 400,000 kg CO2-eq y"1. The reason for this difference is that this milling site serves mainly the local market and is located at a distance close to customers whereas purchased mill inputs are transported much further.
Annual GHG emissions as a result of onsite energy use at this mill facility were approximately 450,000 kg CO2-eq y"1 (economic allocation). Natural gas was the largest contributor, accounting for 80% of this annual total, with electricity consumption accounting for the remaining 20%. Natural gas is used in drying corn grain, which arrives at the milling site with relatively high moisture content which is typical of a northern USA mill location. Mills in southern locations of the USA generally receive corn that is of lower moisture content and hence tend to use much less energy in drying (based on communication with a mill manager).
Cradle-to-dairy farm GHG annual emissions were approximately 16 and 24 million kg CO2-eq y"1 for the milled dairy feed product system including all inputs and transport activities using economic and mass allocations, respectively. When restricting the mill inputs to those directly consumed in mill operations, such as electricity, natural gas, and diesel fuel for transport of feed to dairy farms, annual milled dairy feed-related GHG emissions were much lower (860,000 kg CO2-eq y"1 using economic allocation). Total annual emissions from the MI feed mill, including dairy and non-dairy products are 860,000/0.90 = 950,000 kg CO2-eq y"1, where 0.90 is the economic allocation factor for this mill.
3.2. Discussion of results from sensitivity analyses
As described in section 2.6, sensitivity analyses were conducted to investigate three major scenarios for comparison with the base case GHG analysis. Fig. 3 summarizes the GHG results estimated for all the scenarios considered. Appendices E-1 through E-3 present GHG profile pie charts of the scenario results based on mass and economic allocations. In scenario 1, use of DDGS from a wet mill facility reduces the overall footprint of this mill by just 2—6% depending on allocation method (see Appendix E-1). This is because the differences in emission factor values for DDGS from a wet mill relative to those from a dry mill were minor, especially for mass allocation (see Table 6).
Scenario 2, in which DDGS (from the dry mill facility) was considered to be dominant, resulted in a substantial increase in the mill GHG emission (1.70 kg CO2-eq kg"1 dairy mill output based on mass allocation) which was about two times that of the base case using a mass allocation (see Appendix E-2). The feed mill GHG burdens increased by approximately 35%, from 0.62 to 0.84 kg CO2-eq kg"1 dairy mill output, based on economic allocation. This was due to the relatively high emission factors for DDGS as opposed to soybean meal (See Table 6).
Fig. 3. Sensitivity analysis of feed inputs to dairy feed mill greenhouse gas profile.
In scenario 3 (oats dominant), the GHG profiles for this mill were calculated to be 0.69 and 0.95 kg CO2-eq kg"1 dairy mill output for both economic and mass allocation, respectively (see Appendix E-3). This resulted in a small increase relative to the base case of between 2 and 11% in the feed mill's GHG profile, depending on allocation method. This was not surprising given that the emission factor for oats reported in Table 6 is comparable with the base case in which soybean meal is the dominant feed ingredient.
These scenario analyses demonstrate that geographic differences in dairy feed mill GHG impacts can be substantial, especially for mill locations that predominantly process GHG-intense ingredients such as DDGS.
4. Conclusions and recommendations
The goals of this carbon footprint study were to (i) develop an LCA methodology applicable to the animal feed mill industry to accommodate a large number of inputs and activities associated with dairy mill operations, and (ii) gain an understanding of the relative importance of milled dairy feed inputs and activities on the GHG emissions of the outputs of the mill (which are themselves inputs to dairy milk production) through the application of these developed methodologies. Our methods were able to accommodate a very large number of system inputs using a variety of inventory data sources, including existing databases, new LCA results for USA crops and agricultural co-products, and industry sector 1O data on highly processed ingredients for which no ecoprofiles currently exist.
GHG emission values of 0.62 and 0.93 kg CO2-eq kg"1 milled dairy feed were calculated based on economic and mass allocations, respectively. Overall, the highest contributors to the mill feed carbon footprint were agricultural co-product feed inputs (e.g., DDGS, soybean meal), contributing between 88 and 92% of the carbon footprint depending on the allocation method (see Fig. 2). Mill energy use and transportation of mill inputs and of mill products together contributed 8-12%. In the final analysis, this mill facility emits approximately 16-24 million kg CO2-eq y"1 (depending on allocation method) assuming the study system boundary of cradle-to-dairy farm gate. Annual GHG emissions directly attributable to dairy and non-dairy feed mill activities, including on-site electricity use, process heat demands, and road transport of mill feed to local farms, totals 950,000 kg CO2-eq y"1. It is very clear from scenarios 2 and 3 that the type of feed crop greatly affects the feed mill GHG emissions. Crop inputs are likely to vary from USA region depending on local supply of feed crops.
This study is of a single dairy feed mill, and therefore further study is required to investigate location-specific differences in dairy feed mill inputs and resulting effects these differences have on GHG emissions for the mill feed products. Mill site energy consumption and transportation fuel emissions are under the control of mill operators. Suggested measures to reduce dairy feed mill GHG emissions will center on the use of cleaner sources of electricity and low carbon fuels, such as biodiesel and renewable hydrocarbon diesel from biomass. It is also recommended that further studies be conducted to increase the mill sample size and to include several facilities from southern USA locations. Finally, given the large number of ingredients to the mill, we also recommend further studies of other highly processed supplements to help improve the accuracy of estimating the GHG burdens of milled dairy feeds.
Acknowledgments
The authors would like to express their gratitude to the Innovation Center for U.S. Dairy for the financial support provided for this study. The Innovation center also helped frame the goals of the research and worked with the publisher in the establishment of this special issue.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.idairyj.2012.09.008.
References
Adom, F., Maes, A., Workman, C., Clayton-Nierderman, Z., Thoma, G., & Shonnard, D. (2012). Regional carbon footprint analysis of dairy feeds for milk production in the USA. International Journal of Life Cycle Assessment, 17, 520—534. Batal, A., Dale, N., Waller, J., Pettigrew, J. E., Hollis, G. R., Baker, D. H., et al. (2008). Feedstujfs 2009: Reference issue and buyers guide. Minnetonka, MN, USA: Sarah
Muirhead, URL. http://fdsmagissues.feedstuffs.com/fds/Reference_issue_2010/ Reference_issue_2009/Section1_2008.pdf.
British Standards Institute. (2008). Specification for the assessment of the life cycle greenhouse gas emissions of goods and services (PAS 2050: 2008). BSI.
Deru, M. P., & Torcellini, P. A. (2007). Source energy and emission factors for energy use in buildings (NREL/TP-550-38617). Golden, CO, USA: National Renewable Energy Laboratory.
Edwards, R., Larive, J. F., Mahieu, V., & Rouveirolles, P. (2006). Well-to-wheels analysis of future automotive fuels and powertrains in the European context: Well-to-wheels report. Version 2b (online report). URL. http://ies.jrc.ec.europa.eu.
Gerber, P., Vellinga, T., Opio, C., Henderson, B., & Steinfeld, H. (2010). Greenhouse gas emissions from the dairy sector, a life cycle assessment. Rome, Italy: FAO Food and Agricultural Organisation of the United Nations Animal Production and Health Division.
Hubbard Feeds. (2007). Hubbards feeds www page. URL. http://www.hubbardfeeds. com/Default.aspx.
International Feed Industry Federation. (2009). International feed industry federation annual report 2009. Brussels, Belgium: International Feed Industry Federation.
ISO. (2006a). 14040: 2006 environmental management—life cycle assessment—principles and framework. Geneva, Switzerland: International Standardisation Organisation.
ISO. (2006b). 14044: 2006 environmental management—life cycle assessment— requirements and guidelines. Geneva, Switzerland: International Standardisation Organisation.
Miller, J. K., Ramsey, N., & Madsen, F. C. (1988). The trace elements. In D. C. Church (Ed.), The ruminant animal-digestive physiology and nutrition (pp. 342—401). Englewood Cliffs, NJ, USA: Prentice-Hall.
Nielsen, P. H., Nielsen, N. A., Weidema, B. P., Dalgaard, R., & Halberg, N. (2003). LCA food database (online database). URL. www.lcafood.dk.
PRé Consultants. (2009). SimaPro 7.1 LCA software (computer software). URL. http:// www.pre.nl/.
Shaw, B., Buharivala, P., Parnell, C., Jr., & Demny, M. (1998). Emission factors for grain receiving and feed loading operations at feed mills. Transactions of the ASAE, 41, 757—765.
Siciliano-Jones, J., Socha, M., Tomlinson, D., & DeFrain, J. (2008). Effect of trace mineral source on lactation performance, claw integrity, and fertility of dairy cattle. Journal of Dairy Science, 91,1985—1995.
Sustainability Consortium. (2011). Open-IO version beta 1.5 (online database). URL. http://www.sustainabilityconsortium.org/open-io/.
Thoma, G., Popp, J., Nutter, D., Shonnard, D., Ulrich, R., Matlock, M., et al. (2010). Regional analysis of greenhouse gas emissions from milk production practices in the United States. Rosemont, IL, USA: Innovation Centre for US Dairy.
Thoma, G., Popp, J., Nutter, D., Shonnard, D., Ulrich, R., Matlock, M., et al. (2013). Greenhouse gas emissions from milk production and consumption in the USA: a cradle to grave life cycle assessment circa 2008. International Dairy Journal, 31, S3—S14.
