Greenhouse Gas Life Cycle Assessment of Canola-Derived Hefa Biojet Fuel in Western

Greenhouse gas life cycle assessment of canola-derived hefa biojet fuel in Western

Canada

Debrah Zemanek

A thesis submitted to the Department of Civil Engineering in conformity with the requirements for the degree of Master of Applied Science

Queen’s University Kingston, Ontario, Canada July 2018

Biojet fuel represents the best short-term option for reducing greenhouse gas (GHG) emissions from the aviation industry. In the near term, biojet fuel is most likely to be produced via the hydroprocessed esters and fatty acids (HEFA) pathway. Canola is one of the most readily available feedstocks for HEFA fuel production in Canada, and interest has been shown in developing a domestic supply chain for biojet fuel. However, there is substantial variation in the reported emissions reductions that HEFA biojet from crop-based feedstocks can oﬀer. The primary concern surrounds the inconsistent inclusion of land use change emissions. In addition to infrequently including land use change emissions, previous life cycle assessments (LCAs) of canola biojet fuel have not used Canadian production data, or have failed to examine the sensitivity of calculated emissions intensity to variation in many LCA parameters. In this thesis, the emissions intensity of canola biojet was calculated across four co- product allocation methods both with and without land use change emissions. Land use change emissions from potential grassland conversion on the Canadian Prairies were estimated using a soil carbon model. Finally, a sensitivity analysis on life cycle inputs and co-product allocation parameters was performed. Results suggested that without land use change, canola biojet fuel may provide emissions reductions of 35% to 54% compared to petroleum jet fuel. Land use change

i has an overwhelming inﬂuence on biojet fuel emissions intensity, and its inclusion results in negative reductions in most cases. Emissions intensity and the sensitivity to parameter changes varied across allocation methods. This research points to the importance of standardized LCA methodology, and to the development of advanced biojet fuel production pathways that do not compete for arable land.

ii Acknowledgments

I would like to thank my supervisors, Dr. Pascale Champagne and Dr. Warren Mabee, for their guidance throughout my research. Their relevance in the renewable energy field is inspiring and I am grateful to have had their support and encouragement. I would like to acknowledge the financial support of Green Aviation Research and Development Network, Biofuel Net, and the Natural Sciences and Engineering Research Council of Canada. I am grateful to have been given the opportunity to participate in the Canadian Biojet Supply Chain Initiative, which has allowed me to gain an appreciation for the partnerships between academia, government, and industry that move great ideas forward. I would also like to thank the many people who helped my in my research through providing access to data and information on the canola biojet life cycle. Warren Ward from the Canadian Canola Council; Elwin Smith with Agriculture and Agri- food Canada; James Andersen from Honeywell UOP; Oskar Meijerink from SkyNRG and Jamie Stephen from Torchlight Biofuels. I’d also like to thank Dr. Neil Scott for his help and feedback on soil carbon modeling. From my frisbee throwing partners to my office coffee crew, I feel lucky to have made the friends I did in the Civil Engineering Department. I’m especially grateful to Shuang for knowing exactly when I needed a snack break, and for being my friend

iii since day one of my time at Queen’s. To my friends and family, I appreciate your love and support. And thank you, J.P., for giving me the reason I needed to pursue a Masters at Queen’s.

iv Statement of Co-Authorship

This thesis is based on the following two manuscripts:

Chapter 3 Zemanek, D., Champagne, P., Mabee, W. (2018). Review of Life Cy- cle Greenhouse Gas Emissions Assessments of Hydroprocessed Renewable Fuel (HEFA) from Oilseeds. Planned submission to Biofuels, Bioproducts, and Biore- ﬁning.

Chapter 5 Zemanek, D., Champagne, P., Mabee, W. (2018). Greenhouse Gas Life Cycle Assessment of Canola-Derived HEFA Biojet Fuel in Western Canada. Planned submission to Environmental Science and Technology.

I was responsible for data collection and the design and implementation of the analysis, as well as the writing of all manuscripts. P. Champagne and W. Mabee assisted with the scoping of Chapter 3 and Chapter 5. All authors provided valuable assistance with manuscript editing.

v Contents

Abstract i

Acknowledgments iii

Statement of Co-Authorship v

Contents vi

List of Tables ix

List of Figures xi

Glossary xiii

Chapter 1: Introduction 1 1.1 Background information ...... 1 1.1.1 Emissions from aviation ...... 1 1.1.2 Alternatives to petroleum-derived jet fuel ...... 3 1.1.3 Biojet fuel production in Canada ...... 6 1.1.4 Life cycle assessment ...... 8 1.2 Problem statement ...... 9 1.2.1 Land use change emissions ...... 9 1.2.2 Co-product allocation method ...... 10 1.2.3 Input data ...... 10 1.3 Thesis objectives and organization ...... 11 Bibliography ...... 13

Chapter 2: Background on HEFA 19 2.1 Introduction to the HEFA pathway ...... 19 2.1.1 Hydroprocessing reactions ...... 21 2.1.2 Pre-treatment ...... 24 2.2 Global status of HEFA production ...... 25

vi 2.2.1 hydroprocessed esters and fatty acids (HEFA) technology providers 30 2.2.2 Other BioSPK conversion pathways ...... 30 2.2.3 Summary of biojet production pathways ...... 32 2.3 Biojet adoption in Canada ...... 36 2.3.1 Feedstock availability in Canada ...... 36 2.3.2 Canadian Biojet Supply Chain Initiative (CBSCI) ...... 37 2.4 Conclusion ...... 38 Bibliography ...... 38

Chapter 3: Literature review 48 3.1 Introduction ...... 48 3.2 Review methodology ...... 49 3.2.1 Scope ...... 49 3.2.2 Analysis ...... 51 3.2.3 Parameters analyzed ...... 53 3.3 Results and Discussion ...... 56 3.3.1 Distribution of results across all scenarios ...... 56 3.3.2 Feedstock ...... 56 3.3.3 Co-product allocation method ...... 59 3.3.4 Land use change ...... 62 3.3.5 HEFA technology ...... 64 3.4 Conclusion and recommendations ...... 65 Bibliography ...... 67

Chapter 4: Land use change 73 4.1 Introduction ...... 73 4.2 Model description ...... 77 4.2.1 Data sources ...... 78 4.2.2 Carbon ﬂows ...... 80 4.2.3 Model evaluation ...... 82 4.2.4 Results and Discussion ...... 84 4.2.5 Conclusion ...... 87 Bibliography ...... 88

Chapter 5: Life cycle assessment 96 5.1 Introduction ...... 96 5.2 Methods ...... 99 5.2.1 Life cycle assessment (LCA) in SimaPro ...... 99 5.2.2 Scope ...... 101 5.2.3 Co-product allocation ...... 106 5.2.4 Sensitivity analysis on inputs ...... 109 vii 5.2.5 Sensitivity analysis on co-product allocation parameters . . . . 110 5.3 Results and Discussion ...... 111 5.3.1 Contribution analysis ...... 113 5.3.2 Sensitivity analysis ...... 114 5.4 Conclusion ...... 119 Bibliography ...... 120

Chapter 6: Conclusion 126 6.1 Thesis summary ...... 128 6.1.1 Recommendations and future work ...... 130 6.1.2 Contribution to engineering ...... 132 Bibliography ...... 133

Appendix A: Table of literature reviewed 136 Bibliography ...... 147

Appendix B: Soil carbon model 151 B.1 Decay of SOC pools ...... 151 B.1.1 Rate-modifying factors ...... 152 Bibliography ...... 154

Appendix C: Data sources and allocation methodology 155 C.1 Data sources for life cycle inputs ...... 155 C.1.1 Canola farming ...... 156 C.1.2 Oil extraction and upgrading ...... 160 C.1.3 Reﬁning ...... 161 C.1.4 Transportation and distribution ...... 162 C.1.5 Combustion ...... 163 C.1.6 Displaced soymeal ...... 163 C.1.7 Displaced fuels ...... 168 C.2 Application of co-product allocation to calculation of emissions intensity169 Bibliography ...... 170

viii List of Tables

2.1 Physical and chemical properties of diesel and jet fuels ...... 20 2.2 Global operational and proposed HEFA facilities ...... 27 2.3 Biojet conversion pathways in the American Society for Testing and Materials (ASTM) review process ...... 33

3.1 Seed oil content by weight of oilseed feedstocks reported in the literature 53 3.2 Co-product allocation methods applied in GHG-LCAs of HEFA fuels 59

4.1 Inputs into soil carbon model ...... 79 4.2 Initial size and decay constants of soil organic carbon (SOC) pools . . 81 4.3 Results of the soil carbon model across ﬁve amortization periods . . . 84

5.1 Inputs and yields of the canola HEFA biojet life cycle ...... 102 5.2 Co-product allocation methods compared in this study ...... 109

A.1 GHG-LCAs of HEFA fuels from oilseeds ...... 137

B.1 Monthly parameters in SOC model ...... 152

C.1 Properties of Western Canadian petroleum fuel inputs and jet fuel baseline ...... 159

ix C.2 Parameters of soybean and canola oil extraction and their co-product relationships ...... 165 C.3 Properties of petroleum fuels displaced in reﬁning stage ...... 168

x List of Figures

1.1 Illustrative schematic of global emissions from aviation from 2005 to 2050 and the three environmental targets of the aviation industry . . 3

2.1 Reactions involved in hydroprocessing canola oil feedstock ...... 21

3.1 Number of GHG-LCAs of HEFA jet or diesel from crop-based oilseeds by year of publication ...... 50 3.2 Well-to-wake (WTW) scope and associated life cycle stages of HEFA life cycle assessments (LCAs) ...... 51 3.3 The distribution of reported emissions intensities for HEFA diesel (n=69) and HEFA biojet (n=55) from scenarios that were reviewed within the literature...... 57 3.4 Studies comparing the emissions intensity of HEFA fuels from different feedstocks ...... 58 3.5 Studies comparing the emissions intensity of HEFA fuels when different co-product allocation methods are applied ...... 60 3.6 Studies comparing the emissions intensity of HEFA fuels under different land use change assumptions ...... 63 3.7 Studies comparing the emissions intensity of HEFA fuels from different refining technologies ...... 66

xi 4.1 Canola growing regions of Canada ...... 76 4.2 Change in total soil organic carbon (SOC) and change in SOC in each of four active pools over 100 years as a result of grassland to cropland conversion ...... 85

5.1 An example of the application of ratio allocation and displacement allocation to the flow of process emissions between multiple outputs of a life cycle...... 107 5.2 Emissions intensity of canola biojet fuel across four co-product allocation methods with and without land use change ...... 112 5.3 The contribution of each life cycle stage and displacement credit to total life cycle emissions ...... 114 5.4 Sensitivity analysis showing the effect on the emissions intensity of canola biojet fuel from a ±10% change in the quantity of any input making ≥ 1% contribution to total emissions. The effect of a -10% change is shown in red and the effect of a +10% is shown in blue. . . 115 5.5 Change (%) in the emissions intensity of canola biojet fuel from a ±10% change in the size of allocation parameters ...... 117

xii Glossary

AAFC Agriculture and Agri-food Canada.

ASTM American Society for Testing and Materials.

BL billion litres.

boe barrels of oil equivalent.

CBSCI Canadian Biojet Supply Chain Initiative.

CCC Canola Council of Canada.

CH4 methane.

CHJ catalytic hydrothermolysis jet.

CO2 carbon dioxide.

CO2eq carbon dioxide equivalents.

COPA Canadian Oilseed Producers Association.

CORSIA Carbon Oﬀsetting and Reduction Scheme for International Aviation.

DCO decarboxylation. xiii dLUC direct land use change.

DPM decomposable plant matter.

EF emission factor.

FFAs free fatty acids.

FT Fischer-Tropsch.

GARDN Green Aviation Research and Development Network.

GHG greenhouse gas.

GHG-LCA greenhouse gas life cycle assessment.

GREET Greenhouse Gases, Regulated Emissions, and Energy Use in Transporta- tion.

GtL gas-to-liquids.

GWP global warming potential.

HDCJ hydro-depolymerized cellulosic jet.

HDO hydrodeoxygenation.

HEFA hydroprocessed esters and fatty acids.

HHV high heating value.

IATA International Air Transport Association.

xiv ICAO International Civil Aviation Organization.

iLUC indirect land use change.

IOM inert organic matter.

IPCC Intergovernmental Panel on Climate Change.

ISO International Standard Organization.

ITAKA Initiative Towards sustAinable Kerosene for Aviation.

LCA life cycle assessment.

LHV low heating value.

MJ mega-joule.

Mt million metric tonnes.

N2O nitrous oxide.

NRC Natural Resources Canada. otd oven-dry tonnes.

PET potential evapotranspiration. ppm parts per million.

RED Renewable Energy Directive.

RothC Rothamstead soil carbon model. xv RPM resistant plant matter.

RSB Roundtable on Sustainable Biomaterials.

SOC soil organic carbon.

UCO used cooking oil.

WTW well-to-wheels.

WTWa well-to-wake.

xvi Chapter 1

Introduction

1.1 Background information

1.1.1 Emissions from aviation

The aviation industry is a small but persistent contributor to global warming. The combustion of petroleum-derived jet fuel releases carbon dioxide (CO2), methane

(CH4), and other greenhouse gases (GHGs) that result in a net increase in the radiative forcing effect, or heat balance, of the earth and its atmosphere. Roughly 4% of the overall global radiative forcing effect is attributable to GHGs from the aviation industry (Prather et al., 1999). Although the aviation industry currently emits fewer GHGs than most other transportation industries (Environment Canada, 2015), emissions are expected to increase in coming years. Globalization and an expanding middle class have made air traffic more prevalent and more accessible worldwide. In Canada alone, emissions from domestic and international air travel combined grew 43% between 2005 and 2015 (Transport Canada, 2016). Globally, a projected 5% annual growth rate in air traffic could result in a steady increase in the contribution of the aviation industry to global

1 1.1. BACKGROUND INFORMATION

radiative forcing (IATA, 2015). To minimize the negative environmental consequences of expanding air traﬃc, the International Air Transport Association (IATA) adopted three environmental targets in June 2009 (IATA, 2009):

1. 1.5% average fuel eﬃciency improvement from 2009 to 2020

2. Carbon-neutral growth from 2020 onwards

3. 50% reduction on 2005 emission levels by 2050

Figure 1.1 depicts the projected emissions from aviation from 2005 to 2050, with the expected contribution that various emissions reductions measures are expected to make towards achieving each target. Progress towards the first two targets has been made as a result of ongoing tech- nological development and new economic policy. Incremental gains in fuel efficiency over time (1 in Figure 1.1) have been made possible due to improvements in aircraft design and flight logistics. The International Civil Aviation Organization (ICAO) has agreed upon an international emissions trading system, the Carbon Offsetting and Reduction Scheme for International Aviation (CORSIA), that will be implemented starting in 2021 to address any increase in aviation emissions above 2020 levels (IATA, 2018) (2 in Figure 1.1). However, a 50% reduction on 2005 emission levels by 2050 (3 in Figure 1.1) will require solutions that are outside the scope of these current emissions reductions measures. To achieve the third environmental target, additional technologies must be found, namely alternatives to conventional, petroleum-derived jet fuel that are less emissions-intensive.

2 1.1. BACKGROUND INFORMATION

Figure 1.1: Illustrative schematic of global emissions from aviation from 2005 to 2050 and the three environmental targets of the aviation industry (ATAG, 2013).

1.1.2 Alternatives to petroleum-derived jet fuel

Alternative technologies

The ‘more-electric aircraft’ concept has been discussed for many years as a way to reduce jet fuel consumption and increase aircraft reliability (Emadi et al., 2000; Quigley, 1993). Many aircraft today have certain systems that are powered electrically instead of mechanically, such as braking and fuel pumping (Huang, 2015). Full electric flight, however, would require radical changes to aircraft propulsion systems. Current jet propulsion combusts high density, low freezing-point liquid kerosene to drive a turbine or turbofan engine. Full-electric flight would use power derived from batteries or fuel cells to drive a turbine engine instead. Although demonstration full-electric flights

3 1.1. BACKGROUND INFORMATION

have already been made on small aircraft, electric flight is still decades away from commercialization. Due to present-day limits in energy storage, electric flight will likely be restricted to regional flights or hybrid systems for many decades to come. Other alternatives to conventional jet fuel include liquid hydrogen and natural gas. In the past, these fuels have been mainly proposed as a method of coping with a potential liquid petroleum (crude oil) shortage (Birch, 2000). Although the combustion of hydrogen and natural gas is cleaner than the combustion of petroleum (Birch, 2000), there exists barriers to the adoption of these fuels by the aviation industry. Natural gas can be converted to kerosene, the main component of jet fuel, via the Fichser-Tropch gas-to-liquids (GtL) pathway. However, the GtL pathway is costly and the advantages to using it are limited. The life cycle GHG emissions from GtL jet fuel are only marginally lower than those of conventional jet fuel, and only if carbon-capture technology is applied (Stratton et al., 2010). Liquid hydrogen is even less likely to substitute conventional jet fuel. Hydrogen poses a challenge to its use in aviation due to its low density, which makes it difficult to store on aircraft, and its incompatibility with existing airport infrastructure.

Drop-in biojet fuel

The most feasible alternatives to petroleum-derived jet fuel in the near term are drop-in aviation fuels from renewable feedstocks. Drop-in aviation fuels are fuels that can be used in existing engines and fueling infrastructure without modiﬁcation.

Although CO2 is being researched as a potential feedstock for drop-in renewable fuel production (IATA, 2015), all existing pathways for drop-in fuel production use solid or liquid biomass as a raw feedstock. Drop-in aviation fuel from biomass is called

4 1.1. BACKGROUND INFORMATION

biojet fuel. Biojet fuels from six different production pathways have been approved for commercial use in aircraft, with more in various stages of the approval process (Table 2.3). Biojet production pathways can be categorized as oleochemical, thermochemical, or biochemical, based on the feedstock and the mechanism by which biomass conversion takes place. Oleochemical pathways use biologically-derived lipid feedstocks. The conversion of lipids to hydrocarbons requires few steps, and the approach and technologies required to do so are relatively mature, considering the same processes are used in petroleum refineries to refine crude oil. In fact, several existing petroleum refineries have been re- purposed for the production of drop-in fuels from the oleochemical pathway. However, lipid feedstocks are more expensive than most cellulosic feedstocks, and feedstock purchase makes up the largest portion of operating costs for oleochemical pathways (de Jong et al., 2015). Biochemical pathways use cellulosic feedstocks. Sugars contained in cellulosic feedstock are fermented to produce alcohols, which can be further converted to drop-in fuels. Biochemical pathways are more complex than oleochemical or thermochemical pathways, as the microorganisms involved in fermentation can be somewhat delicate, requiring relatively precise conditions and pre-processed feedstock. Lignocellulosic feedstocks, such as wood residues, are mostly unsuitable for biochemical conversion due to the difficulty in separating cellulose, which can be fermented, and lignin, which cannot. Biochemical conversion has already been well-established for ethanol production via the fermentation of sugars from corn and sugarcane. Thermochemical pathways are more feedstock-flexible than other pathways and

5 1.1. BACKGROUND INFORMATION

can process a range of feedstocks, including lignocellulosic biomass and municipal solid waste. In thermochemical conversion, solid biomass is converted to gas (gasiﬁcation) or liquid products (hydrothermal liquifaction or pyrolysis) under high temperature and pressure conditions (Goyal et al., 2006). The products of thermochemical conversion vary depending on the heat, pressure, and biomass retention time in the reactor. Like products from biochemical pathways, products from thermochemical pathways must undergo a secondary conversion step to produce liquid drop-in fuels, which could increase the cost and complexity of the pathway. Although the pathways employing thermochemical or biochemical conversion are more numerous (Table 2.3) and the feedstocks are available at a lower price, oleochemical conversion pathways are still most likely to achieve wide-scale production of drop-in fuels in the short term. In a pioneer plant analysis, de Jong et al. (2015) found that although biojet fuel from two thermochemical conversion pathways, hydrothermal liquefaction and pyrolysis, achieved a lower minimum fuel selling price than fuel from an oleochemical pathway, the oleochemical pathway remained the most favorable due to its commercial status. The oleochemical pathway HEFA is the only drop-in fuel production pathway to have achieved a consistent commercial-scale production of biojet fuel Karatzos et al. (2017).

1.1.3 Biojet fuel production in Canada

Due to the maturity of the HEFA pathway, it is likely that biojet fuel will be produced from lipid feedstocks in the short term (Karatzos et al., 2017). In Canada, the availability of lipid feedstocks consists of a small amount of by-products of the animal and vegetable oil processing industries and a larger supply of crop-based oilseeds like

6 1.1. BACKGROUND INFORMATION

canola (Transport Canada, 2015). Globally, by-product lipids like tallow and used cooking oil (UCO) are the primary feedstock for HEFA production. By-product lipids are inexpensive and considered to have no alternative uses, although supply may limit biofuel production levels. Emerging concerns surrounding ecology and food security have incited a global shift away from encouraging the production of ‘first generation’ biofuels, despite high feedstock availability and well-established conversion pathways. Although biofuels have previously been classified as first, second, and third generation according to the commercial status of the relevant conversion technologies (International Energy Agency, 2008), for the purpose of this study, generations of biofuels are classified by feedstock, as follows (Saladini et al., 2016):

• First generation biofuels are from food-crop feedstocks, like corn, sugarcane, and soybeans

• Second generation biofuels are from non-food feedstocks, like UCO, switchgrass, and municipal solid waste (MSW)

• Third generation biofuels are from aquatic feedstocks like microalage

First generation biofuels are considered to be less sustainable than second or third generation biofuels as they are produced from feedstocks that compete with the global food supply. Here, ’sustainable’ refers to the ability to meet the needs of the present without compromising the ability of future generations to meet their own needs (WCED, 1987). Although the sustainable production level of ﬁrst generation biofuels may be lower, the commercialization of a biojet fuel production pathway from second or third generation feedstocks has not yet been achieved (Karatzos et al.,

7 1.1. BACKGROUND INFORMATION

2017). Until biojet fuel can be produced on a commercial scale from second and third generation feedstocks, ﬁrst-generation biojet from the oleochemical HEFA pathway could help bridge the transition away from petroleum-dervied jet fuel. Higher biojet production levels than those that would be possible from by-product lipids alone could lead to lower costs through supply chain eﬃciencies, and result in more widespread adoption of biojet fuel. If sustainability concerns could be allayed, biojet from oilseeds could be a major contributor to the decarbonization of the aviation sector. To address these concerns, the life cycle environmental impact of oilseed biojet must be assessed.

1.1.4 Life cycle assessment

Life cycle assessment (LCA) is the primary tool used to assess the potential environmental impact of a product. LCA is deﬁned by International Standard Organization (ISO) standard 14040 as ‘a compilation and evaluation of the inputs, outputs and the potential environmental impacts of a product system throughout its life cycle’ (ISO, 2006). One of the major strengths of LCA is as a method to compare the life cycle of two or more alternative products. Although modern LCA tools can be used to evaluate the potential impacts of a product to multiple aspects of the natural environment, such as resource depletion, biodiversity, and water quality, LCA is frequently focused on GHG emissions. Green- house gas life cycle assessment (GHG-LCA) will be used herein to refer to an LCA that only quantiﬁes GHG emissions from a product system, and does not assess any other potential impacts to the natural environment. GHG-LCA is recommended by the International Energy Agency for comparing bioenergy systems to energy systems

8 1.2. PROBLEM STATEMENT

based on petroleum-derived fuels (Bird et al., 2011). To assess possible GHG reductions from a biofuel, the emissions intensity of a fuel in grams of GHGs per unit of fuel can be calculated and compared to a reference petroleum-derived fuel.

1.2 Problem statement

To develop a biojet supply chain in Canada, suﬃcient GHG reductions must be demonstrated for canola-based biojet fuel compared to petroleum-derived jet fuel. GHG-LCA is a powerful tool for performing comparisons between alternative fuels, but large ranges for estimated GHG reductions from biofuels have been reported in the literature (Bird et al., 2011). Variation in input data, life cycle assessment methodology, and product system boundaries can all aﬀect emissions intensity, and consequentially, possible GHG reductions. Because the calculation of life cycle emissions is so dependent on assumptions, it has been recommended that decision makers be given a range of potential GHG reductions that could result from the adoption of an alternative fuel (Stratton et al., 2011). Three important assumptions in LCA are land use change emissions, co-product allocation method, and input data.

1.2.1 Land use change emissions

Land use change emissions are released when land is converted from one use to another. For ﬁrst-generation (food crop-based) biofuels, land use change emissions are a special concern. If farmers are incentivized to grow crops for biofuels instead of food, maintaining the food supply would require agricultural intensiﬁcation or the establishment of additional cropland. The inclusion of land use change emissions assumes that the establishment of additional cropland is part of the biofuel product system, or

9 1.2. PROBLEM STATEMENT

life cycle. Although estimates of land use change are often intended to be illustrative rather than quantitative, several inﬂuential studies addressing land use change have contributed to the perception that crop-based biofuels may actually be worse for the environment than fossil fuels (Fargione et al., 2008; Hertel et al., 2010; Searchinger et al., 2008). Despite this, fewer than 25% of LCAs of HEFA fuel from oilseeds have included estimates of land use change emissions. When land use change emissions are included, estimations are usually based on generic emission factors, which can fail to reﬂect local conditions (Goglio et al., 2015).

1.2.2 Co-product allocation method

The method by which emissions from a life cycle are allocated to multiple outputs of the life cycle is referred to as the co-product allocation method. The choice of co-product allocation method is known to be a signiﬁcant factor behind variability in calculated emissions intensity (Czyrnek-Deletre et al., 2017; Gnansounou et al., 2009). A comparison of co-product allocation methods is commonly performed in biofuel LCAs, but a sensitivity analysis on the size of co-product allocation parameters is rarely performed.

1.2.3 Input data

Although Canada is one of the world’s top producers of canola and a 2011 survey has made high quality canola production data available (Smith and Barbieri, 2012), no GHG-LCA of canola-derived biojet using Canadian production data has been published to date. Although Miller and Kumar (2013) performed an LCA of canola- derived HEFA diesel produced in Canada, the range of co-product methods across

10 1.3. THESIS OBJECTIVES AND ORGANIZATION

which possible GHG emissions reductions were calculated was limited. In addition, although variation in the quantity of one or two inputs is often analyzed in biofuel LCA, a sensitivity analysis on the quantities of all life cycle inputs has not been performed for biojet fuel from oilseeds.

1.3 Thesis objectives and organization

Here, a range of possible GHG reductions from canola-derived HEFA biojet was calculated using Canadian production data. The emissions intensity of canola biojet was compared across four co-product allocation methods, with and without the inclusion of a model-derived, site-speciﬁc estimate of land use change. The sensitivity of emissions intensity to variation in the quantities of life cycle inputs and to the size of co-product allocation parameters were analyzed in sensitivity analyses. The primary objective of this thesis is to inform the establishment of a domestic HEFA biojet supply chain in Canada through a tailored life cycle assessment. This objective will be achieved in four ways:

• Research the commercial status of HEFA drop-in fuel production worldwide

• Critically review the existing GHG-LCA literature on HEFA drop-in fuel production from oilseeds

• Apply a soil carbon model to estimate land use change emissions from canola production in Western Canada

• Assess the emissions intensity of canola-derived HEFA biojet fuel in a Canadian context across diﬀerent methodological assumptions

11 1.3. THESIS OBJECTIVES AND ORGANIZATION

Chapter 2 provides a background on the HEFA fuel production pathway. This chapter describes the chemistry and economics involved in pre-treating and hydroprocessing lipid feedstock, reviews the commerical status of HEFA drop-in fuel production globally, and summarizes other pathways for biojet fuel production. Research for this chapter helped inform the Phase II report of the Canadian Biojet Supply Chain Ini- tiative (CBSCI) titled ‘The HEFA Process and IATA Meta-Standard’, which will be made publicly available at www.cbsci.ca after it is approved by the CBSCI Steering Committee. Chapter 3 is a comprehensive review of twenty GHG-LCAs on HEFA drop-in fuels from oilseeds that can be grown in Canada. The study intended to analyze possible sources of methodological variability in the emissions intensity of HEFA fuels reported in the literature, and inform the scope of the LCA in Chapter 5. The review addresses feedstock, land use change inclusion, co-product allocation methodology, and refining technology. The relationship between these methodological assumptions and the emissions intensity of HEFA fuel is analyzed. This review, co-authored by Dr. Pascale Champagne and Dr. Warren Mabee, will be submitted to the journal Biofuels, Bioproducts, and Biorefining for publication under the title ‘Review of Life Cycle Greenhouse Gas Emissions Assessments of Hydroprocessed Renewable Fuel (HEFA) from Oilseeds’. Chapter 4 describes the application of a soil carbon model to estimate land use change emissions from canola farming on the Canadian Prairies. The data sources, model evaluation, and model results are summarized in this chapter. Chapter 5 describes the GHG-LCA of canola-derived HEFA biojet in a Canadian context. To increase the relevance of the findings to Canada, Western Canadian input

12 BIBLIOGRAPHY

data is used for as much of the product system as possible. Possible GHG reductions are reported across four co-product allocation methods both with and without land use change. Additionally, a sensitivity analysis is performed on the quantities of life cycle inputs and the size of co-product allocation parameters. Some information from Chapter 4 will be combined with Chapter 5 and submitted for publication to the journal Environmental Science and Technology under the title ‘Greenhouse Gas Life Cycle Assessment of Canola-Derived HEFA Biojet Fuel in Western Canada’, with co-authors Dr. Pascale Champagne and Dr. Warren Mabee. Chapter 6 summarizes the ﬁndings of the research and provides recommendations for future work. The contributions to engineering include a life cycle assessment of which many aspects are novel, including the application of a simple soil carbon model to assess land use change and a sensitivity analysis of co-product allocation parameters. Appendix A contains a table summarizing all literature reviewed in Chapter 3. Appendix B provides calculations for components of the land use change model. Appendix C provides detailed information on the sources of input data for each life cycle stage.

Bibliography

ATAG. Reducing emissions from aviation through carbon neutral growth from 2020. Technical report, Air Transport Action Group, 2013.

Birch, N. T. 2020 Vision: the Prospects for Large Civil Aircraft Propulsion. Aero- nautical Journal, 104(1038):347–352, 2000. ISSN 00019240.

Bird, N., Cowie, A., Cherubini, F., and Jungmeier, G. Using a Life Cycle Assessment 13 BIBLIOGRAPHY

Approach to Estimate the Net Greenhouse Gas Emissions of Bioenergy. Technical report, IEA Bioenergy, 2011.

Czyrnek-Deletre, M. M., Smyth, B. M., and Murphy, J. D. Beyond carbon and energy: The challenge in setting guidelines for life cycle assessment of biofuel systems. Renewable Energy, 105:436–448, 2017. ISSN 18790682. doi: 10.1016/j.renene.2016. 11.043. de Jong, S., Faaij, A., Slade, R., Mawhood, R., and Junginger, M. The feasibility of short term production strategies for renewable jet fuels- a comprehensive techno- economic comparison. Biofuels, Bioproducts and Bioreﬁning, 6(3):246–256, 2015. ISSN 1932104X. doi: 10.1002/bbb.

Emadi, K., Ehsani, M., Emadi, A., and Ehsani, M. Aircraft power systems: technology, state of the art, and future trends. Aerospace and Electronic Systems Magazine, IEEE, 15(1):28–32, 2000. ISSN 0885-8985. doi: 10.1109/62.821660.

Environment Canada. National Inventory Report 1990-2013. Greenhouse gas sources and sinks in Canada. The Canadian Government’s Submission to the UN Frame- work Convention on Climate Change, page 199, 2015. ISSN 17190487. doi: ISSN:1719-0487.

Fargione, J., Hill, J., Tilman, D., Polasky, S., and Hawthorne, P. Land Clearing and the Biofuel Carbon Debt. Science, 319(1235):1235–1237, 2008. ISSN 15829596. doi: 10.1126/science.1152747.

Gnansounou, E., Dauriat, A., Villegas, J., and Panichelli, L. Life cycle assessment

14 BIBLIOGRAPHY

of biofuels: Energy and greenhouse gas balances. Bioresource Technology, 100(21): 4919–4930, 2009. ISSN 09608524. doi: 10.1016/j.biortech.2009.05.067.

Goglio, P., Smith, W. N., Grant, B. B., Desjardins, R. L., McConkey, B. G., Campbell, C. A., and Nemecek, T. Accounting for soil carbon changes in agricultural life cycle assessment (LCA): A review. Journal of Cleaner Production, 104:23–39, 2015. ISSN 09596526. doi: 10.1016/j.jclepro.2015.05.040.

Goyal, H. B., Seal, D., and Saxena, R. C. Bio-fuels from thermochemical conversion of renewable resources: A review. Renewable and Sustainable Energy Reviews, 12 (2008):504–517, 2006. ISSN 13640321. doi: 10.1016/j.rser.2006.07.014.

Hertel, T. W., Golub, A. A., Jones, A. D., O’Hare, M., Plevin, R. J., and Kammen, D. M. Eﬀects of US Maize Ethanol on Global Land Use and Greenhouse Gas Emissions: Estimating Market-mediated Responses. BioScience, 60(3):223–231, 2010. ISSN 0006-3568, 1525-3244. doi: 10.1525/bio.2010.60.3.8.

Huang, H. Challenges in More Electric Aircraft (MEA). IEEE Transportation Elec- triﬁcation Newsletter, 2015.

IATA. Aviation and Climate Change: Pathway to carbon-neutral growth in 2020. Technical report, International Air Transport Association, Switzerland, 2009.

IATA. IATA Sustainable Aviation Fuel Roadmap- 1st edition. Technical report, International Air Transport Association, Montreal-Geneva, 2015.

IATA. Carbon oﬀsetting for international aviation. Technical report, International Air Transport Association, 2018.

15 BIBLIOGRAPHY

International Energy Agency. Energy technology perspectives: Scenario and strategies to 2050. Paris, 2008. ISBN 9789264041424. doi: 10.1787/9789264041431-en.

ISO. Environmental management - Life cycle assessment - Principles and framework (ISO 14040:2006). Technical report, International Standards Organization, Geneva, 2006.

Karatzos, S., van Dyk, J. S., McMillan, J. D., and Saddler, J. Drop in biofuel production via conventional (lipid/fatty acid) and advanced (biomass) routes. Part 1. Biofuels, Bioproducts and Bioreﬁning, pages 344–362, 2017. ISSN 1932104X. doi: 10.1002/bbb.

Miller, P. and Kumar, A. Development of emission parameters and net energy ratio for renewable diesel from Canola and Camelina. Energy, 58:426–437, 2013. ISSN 03605442. doi: 10.1016/j.energy.2013.05.027.

Prather, M., Sausan, R., Grossman, A., Haywood, J., Rind, D., and Subbaraya, B. Potential Climate Change from Aviation. In Penner, J., Lister, D., Griggs, D., Dokken, D., and McFarland, M., editors, Aviation and the Global Atmosphere, chapter 6. Cambridge University Press, Cambridge, UK and New York, NY, USA, 1999.

Quigley, R. E. More Electric Aircraft. In Eighth Annual Applied Power Electronics Conference and Exposition Conference Proceedings, 1993. ISBN 0780309820.

Saladini, F., Patrizi, N., Pulselli, F. M., Marchettini, N., and Bastianoni, S. Guide- lines for emergy evaluation of ﬁrst, second and third generation biofuels. Renew- able and Sustainable Energy Reviews, 66(September 2015):221–227, 2016. ISSN

16 BIBLIOGRAPHY

18790690. doi: 10.1016/j.rser.2016.07.073.

Searchinger, T., Heimlich, R., Houghton, R. A., Dong, F., Elobeid, A., Fabiosa, J., Tokgoz, S., Hayes, D., and Yu, T.-H. Use of U. S. Croplands for Biofuels Increases Greenhouse Gases Through Emissions from Land Use Change. Science, 319(2008): 1238–40, 2008. ISSN 1095-9203. doi: 10.1126/science.1151861.

Smith, E. G. and Barbieri, J. Summary of inputs and production practices used by canola growers in 2011. Technical report, Lethbridge Research Centre, Lethbridge, 2012.

Stratton, R., Wong, H., and Hileman, J. Life Cycle Greenhouse Gas Emissions from Alternative Jet Fuels v 1.2. Technical report, Partnership for Air Transportation Noise and Emissions Reduction, Cambridge, MA, 2010.

Stratton, R. W., Wong, H. M., and Hileman, J. I. Quantifying Variability in Life Cycle Greenhouse Gas Inventories of Alternative Middle Distillate Transportation Fuels. Environmental Science and Technology, pages 4637–4644, 2011.

Transport Canada. Feasibility study of Canadian biojet fuel supply chain- Phase 1: Preliminary Assessment. Technical report, Transport Canada, Clean Transporta- tion Initiative, Aviation Sector Research and Development, 2015.

Transport Canada. Canada’s Action Plan to Reduce Greenhouse Gas Emissions from Aviation: 2015 Annual Report. Technical report, Government of Canada, Ottawa, ON, 2016.

WCED. Our Common Future: Report of the World Commission on Environment

17 BIBLIOGRAPHY

and Development. Technical report, World Commission on Environment and De- velopment, Oxford, 1987.

18 Chapter 2

Background on HEFA

2.1 Introduction to the HEFA pathway

The HEFA oleochemical drop-in fuel production pathway catalytically refines fats and oils into hydrocarbon molecules that closely resemble their petroleum-derived counterparts. The HEFA pathway can be used to produce biojet fuel, also called hydroprocessed renewable jet fuel (HRJ), as well as renewable diesel, or HRD. In 2011, the HEFA pathway received ASTM certification for the commercial use of HEFA biojet. Drop-in fuels like HEFA are biofuels that can be used in applications like aviation where other biofuels, like biodiesel and ethanol, are unsuitable. Table 2.1 compares the fuel properties of biodiesel, conventional petroleum-derived fuels, and HEFA fuels. The absence of oxygen allows HEFA fuels to be used in engines and pipelines without modifications, and a lower freezing-point allows for utilization under cold climate or high altitude (Karatzos et al., 2017). However, the blend ratio of HEFA biojet to conventional jet fuel is only certified up to 50% due to the absence of aromatics, which affect fuel lubricity and seal swell in aircraft engine systems (Corporan et al.,

19 2.1. INTRODUCTION TO THE HEFA PATHWAY

Table 2.1: Physical and chemical properties of diesel and jet fuels (Hoekman et al., 2009; NREL, 2013; Pearlson, 2011; (S&T)2 Consultants Inc., 2013; UOP, 2018a)

Diesel Jet Property Petroleum Biodiesel HEFA Petroleum HEFA Carbon chain length C18-C20 C18-C20 C18-C20 C9-15 C9-C15 Oxygen (% wt.) 0 11.2 0 0 0 Speciﬁc gravity 0.84 0.88 0.78 0.75-0.84 0.73-0.77 Cetane 40-52 45-55 70-90 <40 <80 Cloud Point (◦C) -5 -25 to 2 -30 to -5 -47 to -40 -57 Speciﬁc energy (MJ 40.03 39 44.18 43.04 44.10 LHV/kg) Aromatics (vol %) <12 0 0 <25 0

2011). As of July 2018, HEFA was one of six biojet production pathways to have received ASTM certification for commercial use. The most recent pathway, HEFA co-processing, was certified in early 2018 (Table 2.3). Fuels are certified under ASTM D7566, Standard Specification for Aviation Turbine Fuels Containing Synthesized Hydrocarbons. After blending with petroleum fuel, the fuel can be recertified under ASTM D1655, Standard Specification for Aviation Turbine Fuels, and used without restrictions in aircraft engines. The HEFA pathway has been commercially operational to produce renewable diesel for road transportation since 2007 (Table 2.2). Producing HEFA biojet is a modification of the process conditions required for HEFA diesel production. Most HEFA fuel suppliers view HEFA biojet and HEFA diesel production as separate processes, as biojet fuel production requires additional infrastructure investments (Mei- jerink, 2017; Radich, 2015).

20 2.1. INTRODUCTION TO THE HEFA PATHWAY

Figure 2.1: Reactions involved in hydroprocessing canola oil feedstock. Adapted from Pearlson (2011).

2.1.1 Hydroprocessing reactions

The same hydroprocessing and distillation processes that are involved in petroleum reﬁning are used in the HEFA pathway. An overview of the reactions involved in hydroprocessing can be seen in Figure 2.1. Hydroprocessing of petroleum- or biologically-derived lipids requires a hydrogen-rich environment under moderately high temperatures and pressures [315-370 ◦C, 41-101 bar] combined with a solid catalyst to drive chemical reactions and favour the formation of desired molecules (Chaudhuri, 2011; NREL, 2006). Any biologically-derived lipid can be converted in this manner, although pre-processing requirements and energy demand vary from feedstock to feedstock (UOP, 2013). Biologically-derived lipids are composed of triglyceride molecules, usually with some additional free fatty acids (FFAs). Triglycerides are composed of a three-carbon glycerol backbone and three long-chain carboxylic (fatty) acids, typically ranging in

21 2.1. INTRODUCTION TO THE HEFA PATHWAY

carbon chain length from C14-C20. Canola oil has an average fatty acid composition of 95% C18 molecules (Ackman, 1990), and is 94.4-99.1% triglycerides and 0.4-1.2% FFAs (Przybylski, 2011).

Hydrogenation

In a HEFA reﬁnery, excess hydrogen is supplied to lipid feedstock in a pressurized, heated catalytic reactor (Kalnes et al., 2009). The hydrogen reacts with any unsaturated fatty acids (double bonds present) in the triglycerides. Substantially more hydrogen is consumed when processing triglycerides with a high unsaturated fatty acid (Kalnes et al., 2008); based on feedstock chemistry and previous process modeling (Pearlson, 2011), Han et al. (2013) found that camelina oil could require 34% more hydrogen then palm oil and 11% more than canola oil. Canola oil contains approximately 94% unsaturated fatty acids (Przybylski, 2011).

Propane removal

Following saturation, the glycerol backbone is separated from the triglyceride molecule, and through a series of reactions, converted to propane (Donnis et al., 2009). One propane molecule and three fatty acids are produced for every triglyceride. Renewable propane can be either sold or used on site for heat, electricity, or hydrogen production (Neste, 2012).

22 2.1. INTRODUCTION TO THE HEFA PATHWAY

Deoxygenation

Oxygen present in the carboxyl group of the fatty acids is removed via two simulta- neous reactions (Figure 3.7). The extent of each deoxygenation reaction can be con- trolled through the choice of catalyst and manipulation of process conditions (Neste, 2012). In the hydrodeoxygenation (HDO) reaction, hydrogen replaces oxygen, releas- ing H2O as a byproduct. In the decarboxylation (DCO) reaction, oxygen is eliminated from the fatty acids as CO2, which can react with excess hydrogen and produce low- value methane (Pearlson, 2011). HDO has a higher demand for hydrogen, but DCO produces a shorter hydrocarbon product and hence reduces fuel yield. The favoured reaction will depend on the prices of hydrogen and the desired ﬁnal fuel product (Don- nis et al., 2009). The products of deoxygenation are straight-chain alkanes, water, and CO2.

Isomerization and hydrocracking

Alkane molecules are hydroprocessed once again in a catalytic hydroisomerization and hydrocracking reactor to form branched-chain alkanes with a desired freezing point and carbon chain length. Unlike deoxygenation, hydroisomerization is a very selective process, so less hydrogen is consumed (Kalnes et al., 2008). Because most biologically-derived lipids have a high proportion of C20 fatty acids (Przybylski, 2011), diesel (C18-C20) can be produced from straight-chain alkanes with little further hydrocracking. Hydroprocessing yields between 85% and 95% diesel- length molecules along with a small amount of naphtha (C8) and jet fuel (C9-C15) (UOP, 2013). Under conditions where diesel production is maximized, jet-length molecules are rarely removed from the diesel, as an additional distillation step would

23 2.1. INTRODUCTION TO THE HEFA PATHWAY

be required (Meijerink, 2017). To maximize jet fuel production, the intensity of hydrocracking can be increased, to produce 50-70% jet fuel and 15-25% renewable diesel and naphtha (UOP, 2013). Although it is technically possible to convert all alkanes to jet fuel, a larger amount of shorter naphtha molecules would be generated as a co-product (Pearlson et al., 2013). Because shorter molecules are less valuable, process economics dictate that hydrocracking is not carried out to the fullest extent, even under a jet-fuel maximizing process (Pearlson et al., 2013).

Distillation

Final fuel products are separated in a conventional distillation column. Like propane, renewable naphtha can be sold as a renewable fuel, blend component, or recycled on-site. Before being certiﬁed for commercial use, HEFA diesel and biojet undergo downstream processing, such as blending and the inclusion of additives to improve fuel handling and performance (Hemighaus et al., 2006).

2.1.2 Pre-treatment

The ability of a refinery to proces different feedstocks is dependent on the quality of the feedstock available and the pre-treatment capacity of the production facility. HEFA production facilities are configured to accept feedstocks with specific impurity levels (Meijerink, 2017). Crude vegetable oils are cheaper than highly refined, edible quality oils, but they can contain unsuitably high levels of impurities like phospholipids. Crude unprocessed canola oil can contain up to 2.5% phospholipids, or 1190 parts per million (ppm) phosphorus (Przybylski, 2011).

24 2.2. GLOBAL STATUS OF HEFA PRODUCTION

To remove impurities, crude vegetable oil or animal fat is typically refined using water or acid degumming followed by bleaching with adsorbent clay (Marker, 2005). These processes, in combination with steam deodorization, are used in oilseed crushing facilities to produce edible quality oil. Degumming and bleaching can reduce the phosphorus content of canola oil to 0.2 ppm (Przybylski, 2011). Some refineries can adjust the intensity of degumming depending on the type and quality of feedstock to be treated (Neste, 2012). Pre-treatment is an additional cost for HEFA fuel producers. Installing pre- treatment facilities increases the operating cost of HEFA refineries, and could hinder the development of commercial HEFA production capacity. The capacity to refine vegetable oils to a quality suitable for hydroprocessing already exists in the edible oil industry. Using this existing, decentralized pre-treatment capacity has been suggested as a way to increase supply chain efficiencies when scaling up biofuel production (Richard, 2010).

2.2 Global status of HEFA production

Actual global production of HEFA fuels is estimated to have been 5 billion litres (BL), or over 4 million metric tonnes (Mt), in 2016 (U.S. Energy Information Administra- tion, 2015). In energy terms this is equal to 31.5 million barrels of oil equivalent (boe). Most of the fuel produced was HEFA diesel for road transport. This is likely due to its higher yield and the established demand for renewable diesel via biofuel policies currently in place, such as Canada’s Renewable Fuel Regulations (Government of Canada, 2013). As described above, biojet can be produced as a main product of

25 2.2. GLOBAL STATUS OF HEFA PRODUCTION

HEFA reﬁning should market pricing or targeted policies justify this additional processing step. Currently, only AltAir Fuels in California operates their HEFA facility with the capability to simultaneously produce HEFA diesel and biojet (Table 2.2). Other companies, such as Neste, have produced HEFA biojet on a contract basis for projects like Initiative Towards sustAinable Kerosene for Aviation (ITAKA), (Neste, 2012) but this requires an operational shutdown (Meijerink, 2017). Other proposed HEFA producers have announced their intention to produce biojet, though these are currently non-operational (Table 2.2).

26 2.2. GLOBAL STATUS OF HEFA PRODUCTION Altair Fuels (2017); Califor- nia Energy Com- mission (2018); Csonka (2017a); UOP (2018a) Dar Pro Bio En- ergy (2018); UOP (2018a); Valero Energy (2017) F.O. Licht (2017); UOP (2018a) Neste (2012, 2017, 2018) Neste (2011, 2018) 1 Product Reference HRD,HRJ UOP HRD ogy UOP UOP HRD Neste HRD Neste HRD 2 Feedstock Technol- VO VO UCO VO VO ML/yr Capac- ity Continued on next page up 2021 2080 2020+ 718 2022 2000 Table 2.2: Global operational and proposed HEFA facilities Louisiana, US 2013 1200 Tallow, Netherlands 2011 1000 Tallow, 30% biojet; also marketingvegetable navy oil distillate Phase II Phase II Phase II 1 2 Diamond Green Diesel EniNeste Italy Singapore 2014 2010 513 1000 Palm oil, Tallow, ProducerOperational LocationAltAir Fuels Start California, US 2015 163 Tallow,

27 2.2. GLOBAL STATUS OF HEFA PRODUCTION IATA (2015b); Neste (2012, 2018) Kotrba (2018); REG (2014) UPM (2012) EKAE (2014); Kotrba (2018); Pearson Fuels (2017) 3 Product Reference HRD HRD ogy Neste HRD - HRD Axens HRD Total (2018) troleum Topsoe 4 Feedstock Technol- VO oils ML/yr Capac- ity Continued on next page Table 2.2: Continued from last page up TBD 460 Finland 2007 600 Tallow, Kansas, US 2017 77 CDO Louisiana, US 2010 284 Tallow Syn- HRJ has been producedcorn as distillers a oil batch for projects like ITAKA Phase II 3 4 Producer Location Start Total operational annual capacity (2017)Under Construction TotalAnnounced France 2018 627 Fats and 4957 ML Renewable En- ergy Group UPM Finland 2015 120 Tall oil Haldor East Kansas Agri-Energy

28 2.2. GLOBAL STATUS OF HEFA PRODUCTION SG Preston Ewnig and Han- niﬁn (2011); Har- rington (2016) UOP (2018b) SG Preston (2015, 2017) (2015, 2017) SG Preston (2015, 2017) Product Reference ogy UOPUOP HRD,HRJ HRD,HRJ UOPUOP HRD,HRJ HRD,HRJ UOP HRD,HRJ Feedstock Technol- oils VO oils oils oils ML/yr Capac- ity Table 2.2: Continued from last page up UAE 2020+ 1250 Fats and Texas, US 2020+ 357 Tallow, Indiana, USOhio, US 2020 450 2020 Fats and 450 Fats and Petrixo Oil and Gas Producer LocationEmerald Biofu- els Start Total planned annual capacity (2020+) 10,388 ML SG Preston Ohio, US 2020 450 Fats and

29 2.2. GLOBAL STATUS OF HEFA PRODUCTION

2.2.1 HEFA technology providers

The largest commercial provider of HEFA processing technology is the petroleum technology corporation Honeywell UOP. Their HEFA technology ‘Ecofining’ is patented in partnership with the Italian oil and gas company Eni. Ecofining is licensed for the production or expected production of HEFA biojet by six renewable fuel producers throughout Europe and North America (Table 2.2). Neste’s proprietary HEFA process ‘NExBTL’, or Next Generation Biomass to Liq- uid, is employed at their three renewable diesel refineries in Singapore, the Nether- lands, and Finland. As of January 2017, Neste’s renewable diesel product line has been rebranded as Neste MY (Neste, 2017). In Scandinavia, the Danish catalysis company Haldor-Topsoe licenses their ‘Hydroflex’ process to the HEFA refinery UPM and the petroleum refinery/co-processing operation Preem. Other providers of HEFA processing technology are American technology devel- oper Syntroleum, now owned by North American biofuels producer Renewable Energy Group (REG), and international oil and gas technology provider Axens. REG employs the Syntroleum HEFA process ‘Biosynfining’ in their HEFA diesel plant in Lousisana. Axens has licensed their HEFA technology ‘Vegan’ to French oil and gas company Total for their La Mede petroleum refinery retrofit, expected to be operational to produce HEFA diesel in 2018 (Table 2.2).

2.2.2 Other BioSPK conversion pathways

Oleochemical pathways

In addition to HEFA, there are ﬁve other oleochemical pathways at various stages of seeking ASTM certiﬁcation for the production of biojet fuel (Table 2.3).

30 2.2. GLOBAL STATUS OF HEFA PRODUCTION

Chevron Lummus Global and Applied Research Associates (ARA) have collab- orated on a patented catalytic hydrothermolysis jet (CHJ) technology called Biofu- els Isoconversion, which combines fatty acids with water to achieve hydrothermal cracking and cyclization prior to hydroprocessing (ARA Inc., 2013). This creates an aromatic biojet fuel, avoiding the need to blend with petroleum-derived jet fuel. In 2012, the worlds first 100% biojet flight was fueled with CHJ from a Canadian-grown alternative oilseed crop called Brassica carinata in partnership with the National Re- search Council of Canada (National Research Council, 2012). North American biofuels company Aemetis has licensed Isoconversion technology for proposed renewable diesel production (Aemetis, 2017). Blue Sun Energy licensed Isoconversion technology since 2013 for their demonstration scale biojet facility in Missouri (ARA Inc., 2015). Co-processing is the most recent oleochemical conversion pathway to have attained ASTM certification. Co-processing hydroprocesses a blend of biologically-derived lipids and crude petroleum oil to produce a typical range of refinery products that contain a partially renewable content. Co-processing is practiced at the Preem refinery in Goteburg, Sweden using a 50:50 blend of tall oil biodiesel and petroleum (Preem, 2015). Co-processing operations have previously been established in Ireland, Spain, Portugal Brazil, and Australia, although no production has been reported re- cently (EcoRessources Consultants, 2012). Co-processing was developed by Petrobras in 2007 but worldwide activity has slowed, possibly due to low oil prices. The HEFA Plus pathway has been developed by Boeing, in collaboration with HEFA fuel producers, as a renewable jet fuel blend component (Boeing, 2017). HEFA Plus is HEFA diesel with a higher content of low-freezing point molecules, to enable a

31 2.2. GLOBAL STATUS OF HEFA PRODUCTION

blend of up to 10% with petroleum-derived jet fuel (IATA, 2015a). Should the HEFA Plus speciﬁcation be approved, it could lower the cost of transitioning away from conventional jet fuel, as there is already established HEFA diesel production capacity (IATA, 2015a). Forge Hydrocarbons and SBI Bioenergy are two Edmonton-based technology companies who are developing oleochemical drop-in fuel production pathways. These processes diﬀer from the HEFA process in that hydrogen is not used for the upgrading of lipid feedstocks (SBI BioEnergy Inc., 2016; TEC Edmonton, 2016). Both companies have built pilot plants with plans to scale up production of a variety of renewable hydrocarbons, but have not yet sought ASTM approval for biojet fuel (Csonka, 2017b; Metella, 2017). In March 2017, Shell acquired exclusive rights to develop and license SBI’s technology (Shell Canada Ltd., 2017).

2.2.3 Summary of biojet production pathways

Table 2.3 summarizes all oleochemical, biochemical, and thermochemical biojet production pathways that have been approved or are in the process of seeking approval under ASTM standard D7566 for biojet fuel production.

32 2.2. GLOBAL STATUS OF HEFA PRODUCTION Cobalt, Honey- well UOP, Lan- zatech, Swedish Biofuels Technology de- velopers Axens, Syn- troleum, Haldor Topsoe, Neste Sasol, Shell, Syntroleum Pilot plant Amyris Demonstrative Sasol Demostrative GEVO, Byogy, Operational Honeywell UOP, tus Operational, planned for jet Fermentation of sugars to farne- sane; hydroprocessing FT synthesis, alkylation of ben- zene to produce aromatic rings Fermentation of sugars to isobu- tanol or ethanol; dehydration and hydrogenation Hydroprocessing Gasification; catalytic conversion to liquid fuel Continued on next page Biochemical Thermochemical Biochemical Oleochemical Thermochemical 30% blend, 2018 50 % and ethanol Approved 2015 100% blend Approved 2014 10% blend Approved 2011 50% blend Approved 2009 50% blend 2015c; Radich, 2015; SkyNRG, 2010) (Fischer- (Fischer- 6 5 Synthetic Paraffinic Kerosene Synthetic Kerosene with Aromatics 5 6 Table 2.3: Biojet conversion pathways in the ASTM review process (CAAFI, 2018a,b,c; Csonka, 2017b; IATA, FT-SKA AtJ (Alcohol-to-Jet) Approved 2016 Tropsch with Aro- matics) SIP (Synthesized Iso- Paraffins) PathwayFT-SPK HEFA-SPK (Hy- ASTM Statusdroprocessed Esters and Fatty Type Acids) Description Commercial sta- Tropsch)

33 2.2. GLOBAL STATUS OF HEFA PRODUCTION Licella, Kior Technology de- velopers Petrobras, Hal- dor Topsoe Pilot plantDemonstrative Chevron, ARA Virent Demonstrative Virent Demonstrative Boeing Demonstrative GEVO Demonstrative Honeywell UOP, Operational for HRD tus Hydrothermoly- sis; hydroprocessing Deoxygenation; condensation and hydroprocessing APR plus generation of aromatics Hydroprocessed wide cut diesel for aviation ation of aromatics Hydrothermal liquifaction or pyrolysis to bio- crude; hydroprocessing Co- hydroprocessing lipids and petroleum Continued on next page Oleochemical Table 2.3: Continued from last page Under review Thermochemical Under review Thermochemical Early review BiochemicalEarly review AtJ plus gener- Thermochemical Under review Oleochemical 5% blend APR-SPK (Aqueous Phase Reforming) APR-SKA (Aque- ous Phase Reforming with Aromatics) HEFA PlusAtJ-SKA (Alcohol-to- Jet with Aromatics) Under reviewHDCJ (Hydrotreated Depolymerized Oleochemical Cellu- loic Jet) CHJ (Catalytic Hy- drothermolysis Jet) PathwayCo-processing ASTM Status Approved 2018 Type Description Commercial sta-

34 2.2. GLOBAL STATUS OF HEFA PRODUCTION Technology de- velopers Institute, Shell gies Pilot plantDemostrative Gas Technology Global Bioener- tus tion of sugars to isobutene; oligomerization and hydrogenation dropyrolysis Table 2.3: Continued from last page Early review Thermochemical Moderate hy- 2 Isobutene SPK Early review Biochemical Fermenta- PathwayIH ASTM Status Type Description Commercial sta-

35 2.3. BIOJET ADOPTION IN CANADA

2.3 Biojet adoption in Canada

2.3.1 Feedstock availability in Canada

In the short term, most of the feedstock for HEFA biojet production in Canada consists of oilseeds like canola and soybean. The supply chains for oilseed production and processing are well established, and some sources suggest that there is suﬃcient annual canola carry-over (seed production that is neither exported or used domestically) to support some additional demand from a biojet sector (Goodwin, 2006; Transport Canada, 2015). However, there is still the risk of displacing cropland that is currently used for food production. Camelina and carinata have been considered as promising alternative oilseeds for a Canadian biojet fuel supply chain (Ag-West Bio Inc., 2012). These oilseeds have yields and seed oil contents that are comparable to canola (Blackshaw et al., 2011), but avoid some of the disadvantages of canola production, including competition with edible oil supply. Camelina and carinata have lower water requirements than canola, and have the potential to be grown as winter cover crops (Moser, 2010; Rakow and Getinet, 1998). UCO, tallow, and other by-product lipids are also available for drop- in fuel production in Canada, though in more limited quantities. In the long term, lignocellulosic biomass has been proposed as a more sustainable feedstock for drop-in fuel production. Biofuels from lignocellulosic biomass are considered to be second generation, as they do not complete with food supply (Chapter 1). Lignocellulosic biomass can be converted to drop-in fuels through thermochemical pathways including hydro-depolymerized cellulosic jet (HDCJ) and Fischer-Tropsch (FT) synthesis.

36 2.3. BIOJET ADOPTION IN CANADA

2.3.2 Canadian Biojet Supply Chain Initiative (CBSCI)

Projects like the Canadian Biojet Supply Chain Initiative (CBSCI) have demonstrated Canada’s interest in developing a domestic biojet fuel supply chain. CBSCI is a collaboration between academic and industrial partners to demonstrate the operational feasibility of integrating biojet fuel in Canadian airports and to support carbon neutral growth of the Canadian aviation sector. The operational side of the project involved the delivery of 230,000 litres of biojet fuel to the Toronto Pearson International Airport. The fuel, derived from UCO, was blended 30:70 with petroleum-derived jet fuel. The fuel was purchased from the AltAir Fuels refinery in Paramount, California and delivered by rail to the airport’s aviation fuel storage facility. This project marked the first introduction of biojet fuel directly into a Canadian airport’s shared fueling system. Previous flights with biojet fuel have required aircraft to be fueled directly from a tanker truck (Air Canada, 2016). Emissions credits from the use of biojet fuel in this project were allocated to Air Canada flights. An approximate GHG reduction of 160 tonnes of CO2, or 70 grams CO2 per megajoule of neat biojet fuel, was estimated by project managers (Air Canada, 2018). CBSCI is one of several projects funded by Green Aviation Research and Develop- ment Network (GARDN), a non-profit collaboration between the federal government of Canada and the Canadian aerospace industry that is dedicated to increasing the competitiveness of the Canadian aerospace industry through research and development of technologies to reduce the environmental impact of aircraft (GARDN, 2017).

37 2.4. CONCLUSION

2.4 Conclusion

HEFA is the most feasible pathway for the production of biojet fuel from oleochemical feedstocks in the short term. Among the advantages of the pathway are the technology overlap with petroleum reﬁning and its operational status worldwide. The quantities of biojet fuel produced via the HEFA pathway are currently small, but HEFA production capacity is projected to double by 2020+ if all announced facilities and facility expansions are completed. In Canada, the availability of oilseed feedstocks like canola has encouraged discussion on the development of a domestic HEFA biojet fuel supply chain. However, there are many other biojet production pathways using both oleochemical and cellulosic feedstocks that are seeking ASTM certiﬁcation, and have the potential to be developed in the future.

Bibliography

Ackman, R. G. Canola Fatty Acids—An Ideal Mixture for Health, Nutrition, and Food Use. In Shahidi, F., editor, Canola and Rapeseed: Production, Chemistry, Nutrition and Processing Technology, pages 81–98. Springer US, Boston, MA, 1990. ISBN 978-1-4615-3912-4. doi: 10.1007/978-1-4615-3912-4 6.

Aemetis. The Aemetis Bioreﬁnery: Corporate Presentation, 2017. URL http://www.aemetis.com/wp-content/uploads/2012/04/Aemetis-Corporate- Presentation{\ }FINAL{\ }8-3-17.pdf.

Ag-West Bio Inc. Economic evaluation of drop-in hydrotreated renewable jet (HRJ) fuel production from Camelina sativa (Camelina) and Brassica carinata (Carinata)- Executive Summary. Technical report, Ag-West Bio Inc., 2012.

38 BIBLIOGRAPHY

Air Canada. Air Canada and Biojet Supply Chain Initiative Choose Montr´eal- Trudeau Airport for Aviation Biofuel Project, may 2016. URL https://aircanada. mediaroom.com/index.php?s=22103{\&}item=137466.

Air Canada. Air Canada to Save 160 Tonnes of Carbon on Earth Day through Innovative Biojet Fuel Project at Toronto Pearson Airport, apr 2018. URL https://aircanada.mediaroom.com/2018-04-19-Air-Canada-to-Save-160-Tonnes- of-Carbon-on-Earth-Day-through-Innovative-Biojet-Fuel-Project-at-Toronto- Pearson-Airport.

Altair Fuels. Presentation to the Department of Energy: Wet and Gaseous Feedstocks : Barriers and Opportunities, 2017. URL https://www.energy.gov/sites/prod/ﬁles/ 2017/07/f35/BETO{\ }2017WTE-Workshop{\ }BryanSherbacow-AltAir.pdf.

ARA Inc. Next generation jet and diesel fuels, 2013. URL http://www.readifuels. com/.

ARA Inc. ARA and Blue Sun Performing on Navy Certiﬁcation Contract for 100% Drop-in Fuels, may 2015. URL https://www.ara.com/news/ara-and-blue-sun- performing-navy-certiﬁcation-contract-100-drop-fuels.

Blackshaw, R., Johnson, E., Gan, Y., May, W., McAndrew, D., Barthet, V., McDon- ald, T., and Wispinski, D. Alternative oilseed crops for biodiesel feedstock on the Canadian prairies. Canadian Journal of Plant Science, 91(5):889–896, 2011. ISSN 0008-4220. doi: 10.4141/cjps2011-002.

Boeing. Build Something Cleaner: The Boeing Company 2017 Environmental Report. Technical report, The Boeing Company, Chigaco. IL, 2017.

39 BIBLIOGRAPHY

CAAFI. Co-Processing Provision Approved and Added to ASTM 1655 Annex A1, Enables Renewable Feedstocks in Jet Fuel, apr 2018a. URL http://www.caaﬁ.org/ news/News.aspx.

CAAFI. Ethanol Added as Approved Feedstock and Allowable Blend Levels Increased for ASTM Approved Alternative Jet Fuel Annex, apr 2018b. URL http://www. caaﬁ.org/news/News.aspx.

CAAFI. Global Bioenergies and SkyNRG to Collaborate on ASTM Certiﬁcation of New SAJF Pathway, apr 2018c. URL http://www.caaﬁ.org/news/News.aspx.

California Energy Commission. Powering the US Navy with biofuel, 2018. URL http://www.energy.ca.gov/transportation/tour{\ }additional/altair/index.html.

Chaudhuri, U. R. Fundamentals of Petroleum and Petrochemical Engineering. CRC Press Taylor & Francis Group, Boca Raton, FL, 2011. ISBN 1439851611. doi: 10.1201/b10486.

Corporan, E., Edwards, T., Shafer, L., Dewitt, M. J., Klingshirn, C., Zabarnick, S., West, Z., Striebich, R., Graham, J., and Klein, J. Chemical, thermal stability, seal swell, and emissions studies of alternative jet fuels. Energy and Fuels, 25(3): 955–966, 2011. ISSN 08870624. doi: 10.1021/ef101520v.

Csonka, S. Presentation: Sustainable Alternative Jet Fuel Development & Commer- cialization. Technical report, Commercial Aviation Alternative Fuels Initiative, Arlington, Virginia, 2017a.

40 BIBLIOGRAPHY

Csonka, S. Presentation: The ’real potential’ of technologies under development: Sus- tainable Alternative Jet Fuel. Technical report, Commercial Aviation Alternative Fuels Initiative, Alexandria, Virginia, 2017b.

Dar Pro Bio Energy. Diamond Green Diesel, 2018. URL https://www.darpro- bioenergy.com/solutions/diamond-green-diesel/.

Donnis, B., Egeberg, R. G., Blom, P., and Knudsen, K. G. Hydroprocessing of bio- oils and oxygenates to hydrocarbons. Understanding the reaction routes. Topics in Catalysis, 52(3):229–240, 2009. ISSN 10225528. doi: 10.1007/s11244-008-9159-z.

EcoRessources Consultants. Study of hydrogenation derived renewable diesel as a renewable fuel option in North America. Technical report, Natural Resources Canada, Montreal, QC, 2012.

EKAE. EKAE announces renewable diesel project, jul 2014. URL http://ekaellc. com/category/news-releases/.

Ewnig, D. and Hanniﬁn, P. Alternative Seals for Alternative Fuels. Aerospace Engi- neering, pages 27–30, 2011. URL aero-online.org.

F.O. Licht. Italy- ENI to expand Venice HVO plant, aug 2017. URL https://www.agra-net.com/agra/world-ethanol-and-biofuels-report/biofuel- news/biodiesel/italy---eni-to-expand-venice-hvo-plant--1.htm.

GARDN. Act For Tomorrow- Annual Report 2016-2017. Technical report, Green Aviation Research and Development Network (GARDN), Montreal, QC, 2017.

Goodwin, M. Three potential sources for increased canola oil production in the

41 BIBLIOGRAPHY

Canadian prairies to meet the needs of biodiesel demand. Technical report, Canola Council of Canada, 2006.

Government of Canada. Renewable Fuel Regulations, 2013.

Han, J., Elgowainy, A., Cai, H., and Wang, M. Q. Life-cycle analysis of bio-based aviation fuels. Bioresource Technology, 150:447–456, 2013. ISSN 09608524. doi: 10.1016/j.biortech.2013.07.153.

Harrington, K. US green Navy ﬂeet makes biofuels the new normal. AIChE ChEnected, jan 2016. URL https://www.aiche.org/chenected/2016/01/us-navy- green-ﬂeet-makes-biofuels-new-normal.

Hemighaus, G., Boval, T., Bacha, J., Barnes, F., Franklin, M., Gibbs, L., Hogue, N., Jones, J., Lesnini, D., Lind, J., and Morris, J. Aviation Fuels Technical Re- view. Technical report, Chevron Corporation, 2006. URL https://skybrary.aero/ bookshelf/books/2478.pdf.

Hoekman, S. K., Gertler, A., Broch, A., and Robbins, C. Investigation of biodistillates as potential blendstocks for transportation fuels. Technical Report January 2017, Coordinating Research Council, 2009.

IATA. IATA Guidance Material for Biojet Fuel Management- 2nd edition. Technical report, International Air Transport Association, Montreal-Geneva, 2015a.

IATA. IATA 2015 Report on Alternative Fuels- 10th edition. Technical report, Inter- national Air Transport Association, Montreal-Geneva, 2015b.

IATA. IATA Sustainable Aviation Fuel Roadmap- 1st edition. Technical report, International Air Transport Association, Montreal-Geneva, 2015c. 42 BIBLIOGRAPHY

Kalnes, T. N., Marker, T., Shonnard, D. R., and Koers, K. P. Green diesel production by hydroreﬁning renewable feedstocks. Technical Report 4, UOP, 2008.

Kalnes, T. N., Koers, K. P., Marker, T., and Shonnard, D. R. A Technoeconomic and Environmental Life Cycle Comparison of Green Diesel to Biodiesel and Syn- diesel. Environmental Progress & Sustainable Energy, 28(1):111–120, 2009. ISSN 19447450. doi: 10.1002/ep.

Kotrba, R. Against the Odds. Biodiesel Magazine, jan 2018. URL http://www. biodieselmagazine.com/articles/2516232/against-the-odds.

Marker, T. Opportunities for biorenewables in oil reﬁneries. Technical Report 1-4, United States Department of Energy, 2005.

Meijerink, O. Personal communication, 2017.

Metella, H. Construction of Forge Hydrocarbon’s biofuel production plant set to start, 2017. URL https://www.ualberta.ca/agriculture-life-environment- sciences/alesnews/2017/march/construction-of-forge-hydrocarbons-biofuel- production-plant-set-to-start?cmp=1.

Moser, B. R. Camelina (Camelina sativa L.) oil as a biofuels feedstock: Golden opportunity or false hope? Lipid Technology, 22(12):270–273, 2010. ISSN 0956666X. doi: 10.1002/lite.201000068. 43 BIBLIOGRAPHY

National Research Council. NRC to Fly World’s First Civil Aircraft Powered by 100% Bio Jet Fuel, sep 2012. URL https://www.nrc-cnrc.gc.ca/eng/news/releases/2012/ energy{\ }biofuel.html.

Neste. Neste Oil starts up Europe’s largest renwable diesel plant in Rotterdam, sep 2011. URL https://www.neste.com/en/neste-oil-starts-europes-largest-renewable- diesel-plant-rotterdam.

Neste. NExBTL R Renewable Diesel Singapore Plant- Tallow Pathway Description. Technical report, Neste, 2012. URL http://www.arb.ca.gov/fuels/lcfs/2a2b/apps/ neste-aus-rpt-031513.pdf.

Neste. Neste considers expanding renewable diesel capacity in Singapore. Biomass Magazine, dec 2017. URL http://biomassmagazine.com/articles/14906/neste- considers-expanding-renewable-capacity-in-singapore.

Neste. Who we are: production operations in four countries, 2018. URL https: //www.neste.com/sg/en/about-neste/who-we-are/production.

NREL. Biodiesel and Other Renewable Diesel Fuels (Fact Sheet). Technical report, U.S. Department of Energy National Renewable Energy Laboratory, 2006.

NREL. The Feasibility of Producing and Using Biomass-Based Diesel and Jet Fuel in the United States. Technical report, National Renewable Energy Laboratory, United States Department of Energy, 2013.

Pearlson, M., Wollersheim, C., and Hileman, J. A techno-economic review of hydroprocessed renewable esters and fatty acids for jet fuel production. Biofuels, Bio- products and Bioreﬁning, 6(3):246–256, 2013. ISSN 1932104X. doi: 10.1002/bbb. 44 BIBLIOGRAPHY

Pearlson, M. N. A techno-economic and environmental assessment of hydroprocessed renewable distillate fuels. Masters thesis, Massachusetts Institute of Technology, 2011.

Pearson Fuels. EKAE, Pearson Fuels announce exclusive marketing agreement. Biomass Magazine, dec 2017. URL http://biomassmagazine.com/articles/14892/ ekae-pearson-fuels-announce-exclusive-marketing-agreement.

Preem. Annual Report 2015. Technical report, Preem, 2015.

Przybylski, R. Canola Oil: Physical and Chemical Properties. Technical report, 2011.

Radich, T. Working Paper Series: The Flight Paths for Biojet Fuel. Technical report, United States Energy Information Administration, Washington, DC, 2015.

Rakow, G. and Getinet, A. Brassica Carinata an oilseed crop for Canada. ISHS Acta Horticulturae, 459:419–428, 1998. doi: 10.17660/ActaHortic.1998.459.50.

REG. REG begins production and sales of renewable diesel from Louisiana Bioreﬁnery, oct 2014. URL https://www.regi.com/news/2014/10/27/reg-begins- production-and-sales-renewable-diesel-louisiana-bioreﬁnery.

Richard, T. L. Challenges in scaling up biofuels infrastructure. Science, 329(5993): 793–796, 2010. ISSN 0036-8075. doi: 10.1126/science.1189139.

SBI BioEnergy Inc. SBI BioEnergy Inc., 2016. URL https://www.sbibioenergy.com/ about.

SG Preston. SG Preston and IHI E&C announce strategic agreement to engineer

45 BIBLIOGRAPHY

and construct a portfolio of advanced renewable fuels plants, oct 2015. URL https: //sgpreston.com/news-detail/24.

SG Preston. Qantas aircraft to be powered by renewable biofuel from 2020, oct 2017. URL https://sgpreston.com/news-detail/26.

Shell Canada Ltd. Shell gets exclusive rights for SBI’s biofuel technology. Canadian Biomass Magazine, jun 2017.

SkyNRG. Technology Section, 2010. URL http://skynrg.com/technology-section/.

(S&T)2 Consultants Inc. GHGenius Model 4.03 Volume 2: Data and Data Sources. Technical report, Natural Resources Canada, Ottawa, ON, 2013.

TEC Edmonton. Forge Hydrocarbons ramps up remarkable new biofuel technology, 2016. URL https://www.tecedmonton.com/forge-hydrocarbons-ramps-up- remarkable-new-biofuel-technology/.

Total. La Mede- a forward looking facility, 2018. URL https://www.total.com/en/ energy-expertise/projects/bioenergies/la-mede-a-forward-looking-facility.

UOP. UOP Renewable Jet Fuel Process. Technical report, received by email February 2017, 2013.

UOP. Honeywell Green Jet Fuel, 2018a. URL https://www.uop.com/processing- solutions/renewables/green-jet-fuel/. 46 BIBLIOGRAPHY

UOP. Honeywell Green Jet Fuel: Commerical Production, 2018b. URL https://www. uop.com/processing-solutions/renewables/green-jet-fuel/.

UPM. UPM to build the world’s first biorefinery producing wood-based diesel, 2012. URL http://www.upm.com/About-us/Newsroom/Releases/Pages/UPM- to-build-the-world’s-first-biorefinery-producing-wood-based-biodiesel-001-Wed- 01-Feb-2012-10-05.aspx.

U.S. Energy Information Administration. New biofuels elimiate need for blending with petroleum fuel, 2015. URL https://www.eia.gov/todayinenergy/detail.php? id=23692.

Valero Energy. Diamond Green Diesel considers expanding production capacity, nov 2017. URL http://biomassmagazine.com/articles/14834/diamond-green- diesel-considers-expanding-production-capacity.

47 Chapter 3

Literature review

3.1 Introduction

In the past decade, there has been an increasing number of greenhouse gas life cycle assessments (GHG-LCAs) published of HEFA drop-in fuels. GHG-LCA can be used to quantify the emissions intensity of the production and use of a biofuel, in grams of carbon dioxide equivalent emissions (CO2eq) per megajoule, and to compare it with that of a petroleum fuel over an equivalent life cycle. However, there is large variability in the reported emissions intensity of HEFA fuels, especially from oilseeds like canola. This variability hinders the adoption of HEFA fuel in Canada, one of the world’s top producers of canola. In the past ﬁve years alone, there has been at least ten LCAs published on HEFA fuels from canola or camelina, an oilseed with many of the same advantages for biofuel production as canola. This review attempts to identify the main sources of methodological variability in the LCA of HEFA fuels. Methodological variability is variability in the emissions intensity of HEFA fuels that can be attributed to variation in parameters governing

48 3.2. REVIEW METHODOLOGY

the scope or methodology of the LCA. The four parameters analyzed herein as potential sources of methodological variability are feedstock, land use change inclusion, co-product allocation method, and refining technology. In a previous review of biofuel LCAs, land use change inclusion and co-product allocation method were identified as dominant sources of variability in the reported emissions intensity of fuels (Stratton et al., 2011). The emissions intensity of biofuels across different feedstocks and production pathway assumptions has also been reviewed by Shonnard et al. (2015), but with a focus on biodiesel and ethanol. Fan et al. (2012) reviewed the methodological assumptions and data sources from some HEFA LCAs, but included a very limited number of publications. A literature search identified at least fifteen GHG-LCAs of HEFA from oilseeds that have not been included in any review to date (Han et al., 2013; Shonnard et al., 2015; Stratton et al., 2010). To address these gaps, this study reviews twenty GHG-LCAs of HEFA diesel and jet fuel from oilseeds, over half of which have been published since 2012 (Figure 3.1). Different scenarios from each study are analyzed quantitatively to report both intra-study and inter-study variability across parameters. A better understanding of methodological variability in the LCA of HEFA fuels will help guide further research on the LCA of biofuels from oilseeds, and the adoption of HEFA biojet in Canada.

3.2 Review methodology

3.2.1 Scope

This review was restricted to LCAs of crop-based oilseeds that can be grown in Canada: soybean, canola/rapeseed (Brassica napus/rapa/juncea), camelina (Camelina

49 3.2. REVIEW METHODOLOGY

3 HEFA diesel HEFA jet fuel 2.5

1.5

Number of Studies 1

0.5

0 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year of Publication

Figure 3.1: Number of GHG-LCAs of HEFA jet or diesel from crop-based oilseeds by year of publication

sativa, sunﬂower (Glycine max), carinata (Brassica carinata, and pennycress (Thlaspi arvense). Most studies reviewed were articles from peer-reviewed journals, but two LCAs were obtained from grey literature (Edwards et al., 2014; Stratton et al., 2010) and two Masters theses (Nikander, 2008; Wong, 2008) were also included. These non-peer-reviewed studies were well-cited by other studies in the literature and had noteable methodological scenarios. All LCAs considered the same eﬀective scope, beginning with the production of raw feedstock and ending with the combustion of fuel. This scope, called well-to- wheels (WTW) for road fuels and well-to-wake (WTWa) for aviation fuels, is shown

50 3.2. REVIEW METHODOLOGY

Well-to-Wake

Refining Distribution Combustion

Well-to-Pump Pump-to-Wake

Figure 3.2: Well-to-wake scope and associated life cycle stages of HEFA LCAs. WTW scope can be subdivided into well-to-pump and pump-to-wake stages. Adapted from Elgowainy et al. (2012). in Figure 3.2. If combustion emissions are not included in an LCA, the scope is referred to as well-to-tank (WTT). LCAs with both WTT and WTW/WTWa scopes are included in this review because most WTW LCAs assume that the combustion stage is negligible. Carbon dioxide makes up the majority of combustion emissions and is considered to be carbon-neutral for biofuels, as it was sequestered from the atmosphere during crop growth. Since CO2 has an atmospheric lifetime of approximately 100 years and as such is well-mixed throughout the atmosphere, there is no difference in the radiative forcing of CO2 with altitude (Braun-Unkhoff et al., 2017). LCAs of both diesel and jet fuel were included because the potential sources of methodological variability affect both fuels equally, and the life cycles are nearly identical.

3.2.2 Analysis

To examine methodological variability, 124 scenarios from twenty studies were reviewed. Scenarios were deﬁned as sensitivity analyses within a study where one or more parameters of interest were varied: feedstock, co-product allocation method, land use change inclusion, or HEFA reﬁning technology.

51 3.2. REVIEW METHODOLOGY

Nitrous oxide (N2O) emission factor (EF), nitrogen fertilizer and fertilization rate, diesel usage, crop yield, seed price, hydrogen production pathway, and refining product slate were other parameters for which sensitivity analyses existed in the literature. The relationship between these parameters and reported emissions intensity was beyond the scope of this review, but the results of these scenarios are included in Table A.1 of Appendix A. The distribution of emissions intensities calculated for HEFA fuels in the literature was reported here across a) all scenarios, grouped by HEFA product; and b) each parameter that was considered as a potential source of variability, grouped by publication. All figures that compare parameters report variation on the intra-study level. If a study analyzed variation in more than one parameter, variation in the parameter analyzed was reported across the range of the other parameters, to remove possible cofounding factors. For example, if a study analyzed variation in four co-product allocation methods for both biojet and HEFA diesel, two sets of four data-points each were plotted, one each for biojet and HEFA diesel. Calculated emissions intensities were compared in grams of carbon dioxide equivalents (CO2eq) emissions per mega-joule (MJ) of fuel. If results were reported in an alternative functional unit, such as distance (kg-km or passenger-km) (Edwards et al., 2014) or mass (kg) (Sieverding et al., 2016), they were converted to MJ based on the energy density of the fuel or average fuel consumption of a vehicle. If numerical results were not reported for a scenario, results were estimated to the nearest half integer from available figures.

52 3.2. REVIEW METHODOLOGY

Table 3.1: Seed oil content by weight of oilseed feedstocks reported in the literature

Oilseed Oil con- Most common References tent value Canola 41-46% 44% Han et al. (2013); Miller and Kumar (2013); Stratton (2010); Ukaew et al. (2016) Soy 18-21% 18% Han et al. (2013); (S&T)2 Consultants Inc. (2013); Stratton (2010) Sunﬂower 43% 43% Sieverding et al. (2016) Camelina 34-41% 34% Han et al. (2013); Li and Mupondwa (2014); Lokesh et al. (2015); Miller and Ku- mar (2013); Shonnard et al. (2010); Sieverding et al. (2016) Carinata 44% 44% Sieverding et al. (2016) Pennycress 36% 36% Fan et al. (2013)

3.2.3 Parameters analyzed

Feedstock

Feedstock was analyzed as a potential source of variability in the LCA of HEFA fuels because the product system, and thus emissions intensity, of every feedstock is diﬀerent. However, all the feedstocks examined here are oilseeds, most of which share certain agronomic and physical characteristics. Five studies included scenarios where feedstock was varied (Edwards et al., 2014; Han et al., 2013; Miller and Kumar, 2013; Sieverding et al., 2016; Stratton et al., 2010). Table 3.1 lists the oil content of the feedstocks reviewed. Where an oilseed had a range of values reported in the literature, the range is listed, as well as the most commonly reported value.

53 3.2. REVIEW METHODOLOGY

Land use change

Land use change emissions have long been identiﬁed as an important factor when considering the sustainable production of biofuels (Fargione et al., 2008; Searchinger et al., 2008). Land use change emissions refer to the release of GHGs that can occur as soil carbon and vegetation decompose when land is converted from one use to another. The initial or reference land use is an important assumption when estimating land use change emissions from cropland conversion, as some land uses have larger initial carbon stocks than others. There is high uncertainty associated with land use change estimates, as most accounting methodology relies on general assumptions that do not take into account soil carbon dynamics, soil texture, or crop type (Goglio et al., 2015). Five studies included an estimate of land use change emissions (Miller and Kumar, 2013; Stratton et al., 2010; Ukaew et al., 2016; Uusitalo et al., 2014; Wong, 2008).

Co-product allocation

Co-product allocation is the method by which emissions are allocated between multiple outputs of a product system. The allocation of emissions between the main product and any co-products can occur based on an allocation ratio or a displacement credit. An allocation ratio is based on properties of the products, such as mass, energy content, or market value. A displacement credit subtracts emissions from the life cycle of the main product, assuming that a co-product displaces a similar product in the global market (Figure 5.1). The two HEFA life cycle stages where co-products are produced are oil extraction and reﬁning. The co-product from oil extraction is oilseed meal, which is typically sold as animal feed. HEFA reﬁning co-produces renewable fuels, the amounts of which

54 3.2. REVIEW METHODOLOGY

vary depending on the refining technology and on whether the main fuel product is diesel or jet fuel. Although the feedstock production stage also co-produces straw, this is rarely treated as a co-product due to uncertainty surrounding its use (Uusitalo et al., 2014). Oilseed straw is typically generated in low volumes and is usually left on the field to decompose and replenish soil nutrients. Field studies have linked the removal of straw to direct land use change emissions through the loss of soil carbon (Janzen et al., 1998). There are advantages and disadvantages to all co-product allocation methods, and different allocation methods can generate very different estimates of life cycle emissions. To increase transparency, many LCAs perform a sensitivity analysis on co-product allocation method. Ten studies reviewed herein compared the emissions intensity of HEFA fuels under different allocation methods (Han et al., 2013; Huo et al., 2009; Miller and Kumar, 2013; Nikander, 2008; Shonnard et al., 2010; Uusitalo et al., 2014).

Reﬁning technology

Four companies have patented HEFA reﬁning technologies that have been licensed by commercial fuel producers: UOP Honeywell, Neste, Haldor Topsoe, and Axens. Nat- ural Resources Canada (NRC) also developed a process for HEFA diesel production called Super Cetane, but this process is not commercialized. Super Cetane technology was modelled by one study in this review (Huo et al., 2009). Three studies presented Neste process data (Arvidsson et al., 2011; Edwards et al., 2014; Nikander, 2008; Uusitalo et al., 2014) and all other studies used UOP process data. UOP is the most widely licensed HEFA reﬁning technology (see Chapter 2). Only two studies

55 3.3. RESULTS AND DISCUSSION

compared the emissions intensity of HEFA fuels from diﬀerent reﬁning technologies (Edwards et al., 2014; Huo et al., 2009).

3.3 Results and Discussion

3.3.1 Distribution of results across all scenarios

The median emissions intensity of HEFA fuel in the literature was 40.8 g CO2eq/MJ from 69 scenarios for diesel, and 37.5 g CO2eq/MJ from 55 scenarios for biojet (Figure 3.3). The range of reported results for both fuels was large, between -28.4 and 433.2 g CO2eq/MJ for diesel and between -18.3 and 564.2 g CO2eq/MJ for biojet. No statistical analysis was performed on the diﬀerence in emissions intensity between each fuel product, as the purpose of this analysis was to visually display variation in the data. The spread of results for jet fuel is slightly smaller than that of diesel, but results from both fuels are overlapping.

3.3.2 Feedstock

Five studies compared the emissions intensity of HEFA fuel across diﬀerent oilseed feedstocks. Results from inter-study feedstock analyses are compared in Figure 3.4. Separate results from Miller and Kumar (2013) are presented each for market-based allocation ($/$) and mass-based allocation (M/M), to separate the eﬀect of a co- product allocation analysis from the feedstock analysis. HEFA fuel from canola was included in all intra-study comparisons. In all cases, HEFA fuel from canola had a higher emissions intensity than fuel from other feedstocks. No conclusive ranking of fuel from other feedstocks could be drawn, as no non-canola feedstock pairs were analyzed in more than one study.

56 3.3. RESULTS AND DISCUSSION

200

150 eq/MJ

2 100 g CO

diesel jet HEFA fuel product

Figure 3.3: The distribution of reported emissions intensities for HEFA diesel (n=69) and HEFA biojet (n=55) from scenarios that were reviewed within the literature. A data limit at 200 g CO2eq/MJ, beyond which any outliers (indicated by red crosses) appear evenly distributed, was set to increase readability of the plot.

The production and application of nitrogen fertilizer and the associated N2O emissions are major contributors to the life cycle emissions of fuel from oilseeds (Han et al., 2013), and may contribute to the higher emissions intensity reported for fuels from canola. N2O is emitted naturally from soil through the microbial and abiotic oxida- tion of organic nitrogen compounds, but the application of nitrogen fertilizer results in increased N2O emissions. The highest nitrogen application rates of any feedstock 57 3.3. RESULTS AND DISCUSSION

75 canola (6) 70 soy (3) sunflower (2) 65 camelina (4) carinata (1) 60

eq/MJ HEFA 50 2 45 g CO 40

Han et al. 2013

Edwards et al. 2014 Miller et al. 2013, $/$ Stratton et al. 2010a Miller et al. 2013, E/E Sieverding et al. 2016 Figure 3.4: Studies comparing the emissions intensity of HEFA fuels from diﬀerent feedstocks were reported for canola. Soybean requires little added nitrogen as legumes can convert atmospheric nitrogen (N2) into plant-usable forms. Camelina, pennycress, and carinata all reportedly require less added nitrogen than canola, although the low production levels of alternative oilseeds in North America means that the nitrogen application rates are subject to uncertainty (Shonnard et al., 2010). Sensitivity analyses have shown that nitrogen application rate and crop yield are signiﬁcant sources of variability in reported

58 3.3. RESULTS AND DISCUSSION

Table 3.2: Co-product allocation methods applied in GHG-LCAs of HEFA fuels

Method Method applied to each co-product Straw Oilseed meal Other fuels $/$ None Market Market $/E None Market Energy D/D None Displacement Displacement D/E None Displacement Energy E/$ None Energy Market E/D None Energy Displacement E/E None Energy Energy E/M None Energy Mass M/E None Mass Energy M/M None Mass Mass no/$ None None Market no/E None None Energy no/M None None Mass M/M/M Mass Mass Mass emissions intensity (Li and Mupondwa, 2014; Shonnard et al., 2010).

3.3.3 Co-product allocation method

Fourteen diﬀerent combinations of co-production allocation methods were applied in the studies reviewed herein. The co-products to which emissions were allocated are oilseed meal and other renewable fuels, although oilseed straw was also treated as a co-product by one study (Miller and Kumar, 2013). All studies applied one of the following co-product allocation methods to each co-product: energy-based, mass-based, market-based, displacement, or no allocation (Table 3.2). Ten studies compared two or more co-product allocation methods. A total of ﬁfteen intra-study co-product allocation comparisons are shown in Figure 3.5. Land use change scenarios from Wong (2008) and Uusitalo et al. (2014) were excluded from

59 3.3. RESULTS AND DISCUSSION

100 $/$ (6) 80 $/E (3) D/D (10) 60 D/E (5) E/$ (3) 40 E/D (1) E/E (12) E/M (1)

eq/MJ HEFA 20 2 M/E (3) 0 M/M (7)

g CO no/$ (1) -20 no/E (1) no/M (1) -40 M/M/M (1)

Wong 2008

Han et al. 2013 Nikander 2008 Uuitalo et al. 2014 Fan et al. 2013, jet de Jong et al. 2017 Fan et Huoal. 2013, etHuo al. diesel 2009,et al. 2009,UOP NRC Shonnard et al. 2010,Miller jet et al. 2013, canola Miller et al. 2013, camelina Shonnard et al. 2010, diesel Ukaew et al. 2016, low prices Ukaew et al. 2016, high prices Figure 3.5: Studies comparing the emissions intensity of HEFA fuels when diﬀerent co-product allocation methods are applied this analysis to enable readability, but are included in Figure 3.6 on land use change. In most scenarios, HEFA fuels had a higher emissions intensity when market- based ($/$) allocation was applied, compared to when displacement (D/D), mass- based (M/M), or energy-based (E/E) allocation was applied. Scenarios applying M/M or M/E allocation generated slightly a slightly lower emissions intensity than E/E scenarios. In all but one study, D/D allocation resulted in the lowest emissions intensity for HEFA fuel when compared to other allocation methods. Across the

60 3.3. RESULTS AND DISCUSSION

literature reviewed, scenarios that produced a negative or very low emissions intensity (<10 g CO2eq/MJ) applied displacement allocation to all co-products generated (Fan et al., 2013; Huo et al., 2009; Li and Mupondwa, 2014; Shonnard et al., 2010; Ukaew et al., 2016). The E/D and D/D allocation methods credit the HEFA fuel product with the production and combustion of displaced petroleum-derived fuels that have a high and well-known emissions intensity. If the yield of renewable fuel co-products is relatively high, as is the case when jet fuel production is maximized (see Chapter 2), the displacement credit can be very large. HEFA biojet fuel scenarios generated a much lower emissions intensity than HEFA diesel scenarios when the displacement method was applied to the refining stage, as the co-product yield was much higher (Fan et al., 2013; Shonnard et al., 2010). There was little difference observed in the reported emissions intensity of HEFA fuels for other refining allocation methods, as the mass, energy content, and market value relationships between products are similar (Han et al., 2013). In the oil extraction stage, the choice of allocation method was more important. For example, canola seed is 44% oil by mass but 69% oil by energy content, so energy-based methods allocate more emissions to canola oil. The displacement credit applied to the meal co-product is also subject to more variation. Stratton (2010) found that the emissions intensity of HEFA fuels was very sensitive to the choice of corn, barley, or soy as the animal feed displaced by oilseed meal. Soybean meal is the most common animal feed because of its high protein content, and it is often assumed to be the product displaced by oilseed meal production (Weidema, 1999). However, the emissions intensity of the displaced soybean meal is subject to variation from user

61 3.3. RESULTS AND DISCUSSION

assumptions.

3.3.4 Land use change

Five studies included estimates of land use change emissions. All emissions considered were explicitly from direct land use change (see Chapter 4), except in the case of Wong (2008). In Wong (2008), existing literature values for global indirect land use change emissions from an increase in corn ethanol cultivation in the United States (Searchinger et al., 2008) were adopted for soybean cultivation. All studies except Ukaew et al. (2016) compared the emissions intensity of HEFA fuels with land use change to a baseline scenario without land use change. In all scenarios, the inclusion of land use change increased the emissions intensity of the final product compared to the baseline (Figure 3.6). Where co-product allocation analyses were also performed, results are presented separately (Ukaew et al., 2016; Uusitalo et al., 2014; Wong, 2008). There were five conversion scenarios: forest to cropland, grassland to cropland, a global average (indirect land use change (iLUC)), and cropland to canola cropland under high and low canola prices. The conversion of set-aside land to cropland was grouped under grassland conversion scenarios due to the two land uses having similar initial carbon stocks. Set-aside land is land that was previously cultivated, where some or all of the SOC has been restored. Set-aside programs were put in place in Europe in the 1980s to cope with agricultural surpluses and restore the ecosystem benefits of grasslands (European Economic Community, 1988). The emissions intensity of HEFA fuel from LCAs assuming forest to cropland conversion was much higher than other land conversion scenarios. Both temperate

62 3.3. RESULTS AND DISCUSSION

no LUC (10) 400 grassland (6) forest (4) global avg (4) 300 cropland (high P) cropland (low P) 200 eq/MJ HEFA 2 100 g CO

Miller Wonget al. 2013Wong 2008, 2008,WongD/E Wong M/E2008, 2008,$/E E/E

UuitaloUuitalo et al. 2014,et al. 2014,E/E D/D UkaewUkaew et al. 2016, et al. 2016,D/D E/E Uuitalo et al. 2014, no/E Ukaew et al. 2016, M/M

Stratton et al. 2010a, canola Stratton et al. 2010a, soybean Figure 3.6: Studies comparing the emissions intensity of HEFA fuels under diﬀerent land use change assumptions and tropical forests tend to have higher carbon stocks than grasslands due to the larger stocks of above-ground biomass (Aalde et al., 2006). Global average conversion scenarios also generated high emissions estimates, as the conversion of higher carbon stock land uses, including temperate and tropical forests, are included (Searchinger et al., 2008). Cropland to canola cropland conversion modeled by Ukaew et al. (2016) generated the lowest emissions intensity for HEFA fuel including land use change reported in

63 3.3. RESULTS AND DISCUSSION

the literature, although there was no baseline provided for comparison. An increase in canola prices resulted in an increase in emissions intensity as canola displaced crops with higher soil carbon inputs, such as wheat and sunflower (Ukaew et al., 2016). Cropland has a low initial carbon stock which limits the losses that can occur from land use change. Grassland conversion scenarios also generated low life cycle emissions, as the initial carbon stocks are generally lower than those of other ecosystem types (Aalde et al., 2006). Two studies compared land use change scenarios across different co-product allocation methods. The emissions intensity of HEFA fuel under displacement allocation was very different in each study. Uusitalo et al. (2014) calculated the highest emissions intensity when displacement was applied, whereas Wong (2008) calculated the lowest emissions intensity. Oilseed meal was assumed to displace soybeans in both studies, but unlike Wong (2008), Uusitalo et al. (2014) did not include land use change in the emissions credit for the displaced soybeans. Uusitalo et al. (2014) also applied displacement to the renewable fuel co-product (D/D) whereas Wong (2008) applied energy-based allocation (D/E). Emissions allocated to the renewable fuel co-product increase proportionally when land use change is included under an allocation ratio like energy-based allocation, but not under displacement, which could contribute to the lower emissions intensity reported by Wong (2008).

3.3.5 HEFA technology

Different refining technologies were compared in two studies. There was almost no difference in the emissions intensity of HEFA fuel from Neste and UOP when compared in Edwards et al. (2014), where displacement was applied to all co-products

64 3.4. CONCLUSION AND RECOMMENDATIONS

(Figure 3.7). However, when UOP and NRC technologies were compared across a range of co-product allocation methods, there was a bigger diﬀerence in emissions intensity (Huo et al., 2009). The largest diﬀerence in emissions intensity between technologies was observed when displacement was applied to the renewable fuel co- product (D/D). Displacement generated a negative emissions intensity for fuel from NRC, but not from UOP. Process inputs were similar between the two technologies, but the co-product yield was much higher for the NRC process. This resulted in a lower emissions intensity for fuel from the NRC process under displacement allocation, but a higher emissions intensity when market allocation was applied, as the main fuel product had a higher market value than the co-products.

3.4 Conclusion and recommendations

The wide range in reported emissions intensity for HEFA fuels is a barrier to the development of a HEFA drop-in fuel supply chain in Canada. The aim of this review was to analyze feedstock, co-product allocation method, land use change inclusion, and reﬁning technology as potential sources of variability in the GHG-LCA of HEFA fuels so that important LCA parameters could be better understood. Scenarios from twenty GHG-LCAs were compared using intra-parameter analyses that allowed both intra-study and inter-study variability to be reported for each parameter. Previous reviews have included very few of the GHG-LCAs on HEFA fuels reviewed herein and have not quantitatively analyzed methodological variability from variation in key parameters. When compared to fuel from other feedstocks, life cycle emissions were highest

65 3.4. CONCLUSION AND RECOMMENDATIONS

60 UOP (6) 50 Neste (1) NRC (5) 40

eq/MJ HEFA 10 2 0 g CO -10

-20

-30

Edwards et al. Huo2014 et al. 2009,Huo $/$ et al. 2009,Huo D/D et al. 2009,Huo D/E et al. 2009,Huo E/$ et al. 2009, E/E Figure 3.7: Studies comparing the emissions intensity of HEFA fuels from different refining technologies for HEFA fuels from canola, likely due to high nitrogen fertilization. Forest to cropland land use change scenarios were associated with the highest emissions intensity reported in the literature. Across a wide range of co-product allocation methods, emissions intensity was highest when displacement allocation (D/D) was applied, and the lowest when mass-based allocation (M/M) was applied. The influence of refining technology was not widely compared in the literature and depended on the allocation method applied.

66 BIBLIOGRAPHY

Feedstock, co-product allocation method and land use change inclusion were iden- tiﬁed as important sources of variability. Because of high uncertainty and to commu- nicate the inherent variability in LCA, land use change estimates should be paired with a baseline scenario and more than one co-product allocation method should be applied in an LCA. Robust sensitivity analyses on parameters such as co-product allocation method will help inform future LCA of biofuels. Despite variability in reported emissions intensity, oilseeds are likely to be used for drop-in fuel production in Canada and elsewhere. The GHG-LCAs reviewed here indicate both the need to mitigate land use change concerns and the opportunity for oilseeds that require fewer agricultural inputs, such as camelina and carinata, to supplement canola as feedstocks for HEFA production.

Bibliography

Aalde, H., Gonzalez, P., Gytarsky, M., Krug, T., Kurz, W. A., Lasco, R. D., Martino, D. L., McConkey, B. G., Ogle, S., Paustian, K., Raison, J., Ravindranath, N. H., Schoene, D., Smith, P., Somogyi, Z., van Amstel, A., and Verchot, L. Generic Methodologies Applicable to Multiple Land-Use Categories. In Eggleston, H., Buendia, L., Miwa, K., Ngara, T., and Tanabe, K., editors, 2006 IPCC Guide- lines for National Greenhouse Gas Inventories, Volume 4: Agriculture, Forestry and Other Land Use, chapter Chapter 2, pages 2.1–2.59. Global Environmental Strategies (IGES) for the IPCC, 2006.

Arvidsson, R., Persson, S., Fr¨oling,M., and Svanstr¨om,M. Life cycle assessment of hydrotreated vegetable oil from rape, oil palm and Jatropha. Journal of Cleaner

67 BIBLIOGRAPHY

Production, 19(2-3):129–137, 2011. ISSN 09596526. doi: 10.1016/j.jclepro.2010.02. 008.

Braun-Unkhoﬀ, M., Riedel, U., and Wahl, C. About the emissions of alternative jet fuels. CEAS Aeronautical Journal, 8(1):167–180, 2017. ISSN 18695590. doi: 10.1007/s13272-016-0230-3.

Edwards, R., Hass, H., Lariv´e,J.-F., Lonza, L., Mass, H., and Rickeard, D. Well-to wheels analysis of future automotive fuels and powertrains in the European context v 4.a. Technical report, European Commission Joint Research Centre, 2014.

Elgowainy, A., Han, J., Wang, M., Carter, N., Stratton, R., and Hileman, J. Life Cycle Analysis of Alternative Aviation Fuels in GREET. Technical report, En- ergy Systems Division, Argonne National Laboratory, U.S. Department of Energy, Argonne, Illinois, USA, 2012.

European Economic Community. Commission Regulation (EEC) No 1272/88 of 29 April: Laying down detailed rules for applying the set-aside incentive scheme for arable land. Technical report, European Economic Community, Brussels, apr 1988.

Fan, J., Handler, R. M., Shonnard, D. R., and Kalnes, T. N. A review of life cycle greenhouse gas emissions of hydroprocessed jet fuels from renewable oils and fats. International Journal of Environmental Science and Engineering Research, 3(3): 114–138, 2012. doi: 10.1088/0256-307X/27/11/116401.

Fan, J., Shonnard, D. R., Kalnes, T. N., Johnsen, P. B., and Rao, S. A life cycle assessment of pennycress (Thlaspi arvense L.) -derived jet fuel and diesel. Biomass

68 BIBLIOGRAPHY

and Bioenergy, 55:87–100, 2013. ISSN 09619534. doi: 10.1016/j.biombioe.2012.12. 040.

Fargione, J., Hill, J., Tilman, D., Polasky, S., and Hawthorne, P. Land Clearing and the Biofuel Carbon Debt. Science, 319(1235):1235–1237, 2008. ISSN 15829596. doi: 10.1126/science.1152747.

Han, J., Elgowainy, A., Cai, H., and Wang, M. Q. Life-cycle analysis of bio-based aviation fuels. Bioresource Technology, 150:447–456, 2013. ISSN 09608524. doi: 10.1016/j.biortech.2013.07.153.

Huo, H., Wang, M., Bloyd, C., and Putsche, V. Life-cycle assessment of energy use and greenhouse gas emissions of soybean-derived biodiesel and renewable fuels. Environmental Science and Technology, 43(3):750–756, 2009. ISSN 0013-936X. doi: 10.1021/es8011436.

Janzen, H., Campbell, C., Izaurralde, R., Ellert, B., Juma, N., McGill, W., and Zentner, R. Management eﬀects on soil C storage on the Canadian prairies. Soil and Tillage Research, 47(3):181–195, 1998. ISSN 01671987. doi: 10.1016/S0167- 1987(98)00105-6.

Li, X. and Mupondwa, E. Life cycle assessment of camelina oil derived biodiesel and

69 BIBLIOGRAPHY

jet fuel in the Canadian Prairies. Science of the Total Environment, 481(1):17–26, 2014. ISSN 00489697. doi: 10.1016/j.scitotenv.2014.02.003.

Lokesh, K., Sethi, V., Nikolaidis, T., Goodger, E., and Nalianda, D. Life cycle greenhouse gas analysis of biojet fuels with a technical investigation into their impact onjet engine performance. Biomass and Bioenergy, 77:26–44, 2015. ISSN 1873-2909. doi: 10.1016/j.biombioe.2015.03.005.

Nikander, S. Greenhouse gas and energy intensity of product chain: case transport biofuel. Masters thesis, Helsinki University of Technology, 2008.

Shonnard, D. R., Williams, L., and Kalnes, T. N. Camelina-Derived Jet Fuel and Diesel: Sustainable Advanced Biofuels. Environmental Progress & Sustainable En- ergy, 29(382-392), 2010. ISSN 19447450. doi: 10.1002/ep.

Shonnard, D. R., Klemetsrud, B., Sacramento-Rivero, J., Navarro-Pineda, F., Hilbert, J., Handler, R., Suppen, N., and Donovan, R. P. A Review of Environmental Life Cycle Assessments of Liquid Transportation Biofuels in the Pan American

70 BIBLIOGRAPHY

Region. Environmental Management, 56(6):1356–1376, 2015. ISSN 14321009. doi: 10.1007/s00267-015-0543-8.

Sieverding, H. L., Zhao, X., Wei, L., and Stone, J. J. Life-Cycle Assessment of Oilseeds for Biojet Production Using Localized Cold-Press Extraction. Journal of Environment Quality, 45(3):967, 2016. ISSN 0047-2425. doi: 10.2134/jeq2015.06. 0313.

(S&T)2 Consultants Inc. GHGenius Model 4.03 Volume 2: Data and Data Sources. Technical report, Natural Resources Canada, Ottawa, ON, 2013.

Stratton, R. W. Life cycle assessment of greenhouse gas emissions and non-CO combustion eﬀects from alternative jet fuels. Masters thesis, Massachusetts Institute of Technology, 2010.

Ukaew, S., Shi, R., Lee, J. H., Archer, D. W., Pearlson, M., Lewis, K. C., Bregni, L., and Shonnard, D. R. Full Chain Life Cycle Assessment of Greenhouse Gases and Energy Demand for Canola-Derived Jet Fuel in North Dakota, United States. ACS Sustainable Chemistry and Engineering, 4(5):2771–2779, 2016. ISSN 21680485. doi: 10.1021/acssuschemeng.6b00276.

71 BIBLIOGRAPHY

Uusitalo, V., Vaisanen, S., Havukainen, J., Havukainen, M., Soukka, R., and Lu- oranen, M. Carbon footprint of renewable diesel from palm oil, jatropha oil and rapeseed oil. Renewable Energy, 69:103–113, 2014. ISSN 09601481. doi: 10.1016/j.renene.2014.03.020.

Weidema, B. P. System expansions to handle co-products of renewable materials. 7th LCA Case Studies Symposium SETAC-Europe, pages 45–48, 1999.

Wong, H. M. Life-cycle assessment of greenhouse gas emissions from alternative jet fuels. Masters thesis, Massachusetts Institute of Technology, 2008.

72 Chapter 4

Land use change

4.1 Introduction

Although they are an important factor to consider when considering the sustainable production of biofuels, land use change emissions are not usually included in the LCA of drop-in aviation fuels. Land use change emissions occur from the decay of vegetation and/or soil carbon when land is converted from one use to another. For biofuel LCAs, land use change emissions are usually assumed to result from the conversion of a natural ecosystem to establish the additional cropland required to grow biofuel feedstocks. Depending on the location of land conversion, land use change emissions can be classiﬁed as either direct or indirect land use change (dLUC or iLUC)(ISO, 2013). Direct land use change emissions occur when land conversion takes place where the production of a material within the product system is located. Indirect land use change emissions occur when the production of a material within the product system results in land conversion elsewhere. iLUC emissions are infrequently included in agricultural LCAs, as there is no standard method, and approaches often require the use of data-intensive global economic models (Goglio et al., 2015).

73 4.1. INTRODUCTION

Direct land use change is more commonly accounted for in biofuel LCAs than indirect land use change, but estimates vary widely, depending largely on the reference land use, or type of land conversion, that is assumed to take place. If only the direct land use change emissions from the conversion of existing cropland to grow biofuel feedstock are accounted for, estimates for direct land use change can be low or even negative (Ukaew et al., 2016), with a calculated emissions intensity that is equal to or lower than scenarios that do not include land use change (Figure 3.6). Estimates of direct land use change can be much higher if the conversion of natural ecosystems, as opposed to cropland, is assumed to occur (Fargione et al., 2008). The assumption that biofuel production results in the conversion of natural ecosystems is not dissimilar to indirect land use change assumptions, except that conversion is assumed to take place locally, and not globally. In Canada, the majority of canola production takes place on the Canadian Prairies in the semi-arid (dry cool temperate) and sub-humid (moist cool temperate) climatic zones (Figure 4.1). Areas with heavier production produce more tonnes per agricultural regional annually. Most area on the Canadian Prairies outside of farmland is classiﬁed as grassland (Ahern et al., 2013). Total farmland in the Canadian Prairies has remained steady over the past decade (Statistics Canada, 2011), but it is possible that increased canola demand for drop-in fuel production could lead to the conversion of grassland to cropland. Although land conversion as a result of biofuel demand has been modeled for other regions and feedstocks (i.e. Hertel et al. (2010); Searchinger et al. (2008)), there is little empirical evidence to suggest that this has actually occurred, in Canada or elsewhere (Kim and Dale, 2011). Before the Cana- dian Renewable Fuel Regulations (Government of Canada, 2013) came into eﬀect in

74 4.1. INTRODUCTION

2013, it was suggested that meeting the 2% biofuel blend mandate with Canadian canola would require approximately 1 million tonnes of seed, or 13% of average annual production (Saville, 2006). Demand could be met by either reducing canola exports, increasing harvested area, or increasing canola yields. Statistics show that although both production and exports of canola have been steadily increasing, harvested area has leveled off since 2010 (Statistics Canada, 2017a,b). This suggests that growing demand for canola has been met with increased yields, and not land conversion. In fact, there is no evidence of land use change occuring in Canada, as the total area of farmland remained constant, and in fact declined slightly, between 1976 and 2011 (Statistics Canada, 2011). Regardless, land use change poses a risk to canola-based biojet fuel development in Canada, especially at higher levels of feedstock demand. There is a limit to the yield increases that can be achieved through advances in crop breeding and enhanced fertilization. Additionally, there are environmental costs associated with high levels of agricultural inputs. Water quality can be severely affected by nutrient and pesticide run-off from agricultural areas (Carpenter et al., 1998), however, these environmental impacts are outside the scope of this study. Although direct land use change (dLUC) emissions from grassland to cropland conversion has previously been accounted for in the life cycle of canola drop-in fuels (Miller and Kumar, 2013; Stratton et al., 2010; Uusitalo et al., 2014), the accounting methodology applied thus far has been limited. Most previous LCAs calculated land use change using default IPCC emission factors for the conversion of a reference land use to cropland (Aalde et al., 2006). However, it is recommended that wherever possible, soil carbon models should be used instead of emission factors to estimate land

75 4.1. INTRODUCTION

Figure 4.1: Canola growing regions of Canada (Canola Council of Canada, 2015). Most production occurs in the semi-arid (dry cool temperate) and sub- humid (moist cool temperate) climatic zones of the Canadian Prairies. use change emissions, as well as site-speciﬁc soil parameters (Goglio et al., 2015). Land use change for grassland to canola cropland on the Canadian Prairies is calculated here using the well-tested Rothamstead soil carbon model (RothC) and soil and climate parameters speciﬁc to the Canadian Prairies. This land use change estimate will inform the GHG-LCA of canola-derived HEFA biojet that is performed in Chapter 5. 76 4.2. MODEL DESCRIPTION

4.2 Model description dLUC emissions from the conversion of an undisturbed native grassland to canola cropland were calculated using a model of soil carbon fluxes for the semi-arid climate zone of the Canadian Prairies. Grassland was selected as the initial land use because 85% of the uncultivated land in the Canadian Prairies is classified as grassland (Ahern et al., 2013). Although the conversion of forest to cropland has been modeled in other LCAs, only 1% of undisturbed land on the Canadian Prairies is forested (Ahern et al., 2013), so forest conversion was not considered here. Mathematical relationships between model components, including soil decay constants, were taken from the RothC model user manual (Coleman and Jenkinson, 2014). Athough the RothC model has been previously applied in agricultural LCAs, (Goglio et al., 2015; Yao et al., 2017), no literature was found that applied it to the LCA of biofuels. The RothC model was established by soil scientists using long- term data from an experimental site in the United Kingston and is one of the most well-known soil carbon models (Davidson and Janssens, 2006; Jenkinson et al., 1990). The RothC model is similar to CENTURY (Parton et al., 1987), another of the most well-known soil carbon models (Davidson and Janssens, 2006). Both models fit experimental data fairly well, but RothC requires fewer data inputs (Falloon and Smith, 2006). A shortcoming of the RothC model is that it is a ‘black box’ with respect to the derivation of output data. To overcome this shortcoming and provide more transparency, the model was re-constructed in Matlab. In RothC, annual carbon inputs can be back-calculated from initial soil organic carbon (SOC) level (Coleman and Jenkinson, 2014). Here, annual carbon inputs were obtained from the literature, and

77 4.2. MODEL DESCRIPTION

were selected to correspond to the selected cropping system and geographic region as closely as possible.

4.2.1 Data sources

Initial SOC in the top 30 cm soil layer was derived from published ﬁeld measurements and established depth conversion factors (Janzen et al., 1998; Qin and Huang, 2010; Shrestha et al., 2014). Shrestha et al. (2014) measured SOC content in a long-term cultivated plot in Swift Current, Saskatchewan (SK) as 30.8 metric tonnes per hectare (t C/ha) in the top 0-15 cm soil layer. Based on estimates that approximately one- third of SOC has been lost from the Canadian Prairies since cultivation began in the early 1900s, initial SOC was determined to be 46.2 t/ha (Janzen et al., 1998; Mcconkey et al., 2000). A depth conversion factor of 1.665 (Qin and Huang, 2010) was applied to convert this value to the initial SOC in the 0-30 cm soil layer. Although SOC changes decline with depth (Phillips et al., 2015), the Intergovernmental Panel on Climate Change (IPCC) recommends that estimates of land use change should account for at least the top 30 cm of soil (Penman et al., 2003). Annual carbon inputs to the ecosystem were based on a wheat-oilseed-pulse (WOP) crop rotation, since canola is nearly always grown in rotation with non-oilseed crops (Canola Council of Canada, 2012). An annual organic carbon input of 2.007 t/ha was derived from Shrestha et al. (2014) using an average of the crop residue and below- ground biomass (0-30 cm) for the three crops in rotation and a dry-matter carbon content of 42.3%. An increasing percentage of crops on the Canadian Prairies, especially oilseeds and pulses, are grown using zero-tillage (Boame, 2005). A zero-tillage system was modeled for the selected cropping system by using a positive value for a

78 4.2. MODEL DESCRIPTION

Table 4.1: Inputs into soil carbon model

Input Quantity Units Reference Location Comments Precipitation monthly avg mm/month Environ- Swift Current climate nor- ment Canada SK mals 1981- (2017) 2010 Mean air monthly avg deg Celcius Environ- Swift Current climate nor- temperature ment Canada SK mals 1981- (2017) 2010 Potential ET monthly avg mm/month Alberta Gov- Medicine Hat 2000-2009 ernment AB (2013) Clay content 18 % dry weight Shrestha Swift Current Swinton silt et al. (2014) SK loam, Orthic Brown Cher- nozem Initial SOC 76.42 t C/ha Shrestha Swift Current 0-30 cm depth et al. (2014) SK Annual C 2.007 t C/ha Shrestha Swift Current 0-30 cm, input et al. (2013) SK WOP, distributed evenly between May- Oct monthly soil cover constant. A positive soil cover constant slows the decay of all SOC pools in the model. This assumption may be optimistic considering that only 40% of canola producers used a zero-tillage system in 2011, although the greatest adoption of zero-tillage was in southern Saskatchewan, where Swift Current is located (Smith and Barbieri, 2012) The monthly soil cover constants and the monthly per hectare soil carbon inputs are located in Table B.1 in Appendix B. Other model inputs are listed in Table 4.1. Monthly mean air temperature, potential evapotranspiration (PET), and precipitation were sourced from governmental datasets for the Canadian Prairies region. Clay content was assumed to be 18% of soil by dry weight (Shrestha et al., 2013). In the RothC model, a higher clay content causes soil to dry out more quickly, which slows decay.

79 4.2. MODEL DESCRIPTION

4.2.2 Carbon ﬂows

Soil carbon was split according to the RothC model into four active pools and one inert pool (Coleman and Jenkinson, 2014) (Table 4.2). Monthly changes to each active pool depended on the carbon inputs to the pool and the first-order decay rate of each pool. Partitioning constants determined the proportion of carbon inputs that flowed to each pool. The DPM/RPM ratio refers to how incoming carbon is split into the two pools that represent vegetation: decomposable plant matter (DPM) and resistant plant matter (RPM). The DPM/RPM ratio was set to 1.44 as a default for agricultural conditions, but can be modified to reflect grassland or forest conditions (Coleman and Jenkinson, 2014). Under agricultural conditions, more incoming carbon flows to the DPM pool. Decay rates for each pool were obtained from the RothC documentation and were modified by environmental and soil cover factors that varied monthly (Coleman and Jenkinson, 2014). More information on the calculation of rate modifying factors can be found in Appendix B. Initial pool sizes were derived from initial SOC level and pool allocation percentages. The pool allocation percentages were generated from a model initialization run over 1000 years for grassland conditions (plant inputs 4.7 t/ha/yr, initial soil carbon=0, DPM/RPM= 0.67). The size of the inert organic matter (IOM) pool, which does not change over time, was calculated from initial SOC using Equation 4.1 (Falloon et al., 1998).

IOM = 0.049 ∗ SOC1.139 (4.1)

Annual land use change emissions are calculated as the sum of monthly SOC

80 4.2. MODEL DESCRIPTION

Table 4.2: Initial size and decay constants of SOC pools

SOC pool Abbrev. Initial t C/ha % active SOC Decay constant (0-30 cm)

Inert Organic Matter IOM 6.84 - - Decomposable Plant Material DPM 1.927 2.77% 10 Resistant Plant Material RPM 11.786 16.94% 0.3 Humiﬁed Organic Matter HUM 54.434 78.24% 0.02 Microbial Biomass BIO 1.426 2.05% 0.66 Total SOC 76.41 changes across all active pools, annualized over a designated cropland lifetime (Equa- tions 4.2 to 4.5). Emissions are normalized using yield parameters and included in the canola biojet product system where appropriate. A cropland lifetime or amortization period of 20 years is recommended by the European Renewable Energy Directive (RED), the international Roundtable on Sustainable Biomaterials (RSB), and the IPCC (European Parliament, 2009; Penman et al., 2003). A 30 year amortization period has been used in some studies (Fargione et al., 2008; Searchinger et al., 2008). Results from 20 and 30 year periods are reported here, although only the land use change emissions from the 20 year period are used in Chapter 5. Long term results (100 and 500 years) were also reported to explore model behavior and determine total loss of SOC. Equations governing the calculation of land use change emissions from the model results are as follows:

SOC(t) = DPM(t) + RPM(t) + HUM(t) + BIO(t) + IOM(t) (4.2)

81 4.2. MODEL DESCRIPTION

∆SOC = SOC(i) − SOC(t) (4.3)

landusechange(tCO2/ha/yr) = ∆SOC/(t/12) ∗ 3.67 (4.4)

landusechange(tCO2/MJ) = landusechange(tCO2/ha/yr)/k (4.5)

Where DPM, RPM, HUM, BIO, and IOM are the ﬁve SOC pools deﬁned in

Table 4.2, 3.67 is the conversion factor for carbon to CO2 (Smith et al., 1997), and t/12 is the amortization period in years where t is time in months. k is a constant representing the hectares of cropland required per megajoule (MJ) of biojet, using yields and energy density from Table 5.1.

4.2.3 Model evaluation

The ﬁt of the model to conditions on the Canadian Prairies was evaluated by comparing the output of the model over two time periods, 100 and 30 years, to experimental observations of soil loss. Based on an overview of long-term research sites on the Canadian Prairies with an average experimental duration of 30 years, Janzen et al. (1998) concluded that cropped areas have between 20-30% less soil carbon than ad- jacent uncultivated land. This estimate was used to evaluate the ﬁt of the model. The evaluation run simulated a cropping system of wheat-corn-fallow (1.8 t C/ha/yr (Shrestha et al., 2013)), using the climatic and soil variables from Table 4.1 to represent environmental conditions on the Canadian Prairies. The fallow-containing

82 4.2. MODEL DESCRIPTION

rotation was modeled using a negative soil cover parameter from Nov-Feb. This cropping system was chosen because fallow was a common agronomic practice in the early 20th century (Carlyle, 1997). The model produced a 26% loss of soil carbon over a 30 year period, which fits within the range of 20-30% soil loss estimated by Janzen et al. (1998) based on experimental observations. A 40% loss in soil carbon was modeled over the 100 year period, which may overestimate the potential loss of soil carbon. Janzen et al. (1998) found that soil carbon loss on the Canadian Prairies mostly abated after the first few decades, and that SOC may actually be increasing now due to the adoption of management practices like zero-tillage and decreasing fallow. During the development of the model, it was observed that there was a maximum SOC level associated with each level of annual carbon inputs. An underestimation of average annual carbon inputs could explain the over-prediction of soil carbon loss over the 100 year period compared to findings by Janzen et al. (1998). However, the lack of field experiments older than 50 years make it difficult to compare the 100 year model output to experimental data. Based on the results of the evaluation, the model was deemed satisfactory for estimating land use change emissions on the Canadian Prairies over the short term (20-30 years). It is acknowledged that there are certain limitations in the accuracy of losses estimated due to the simplistic nature of the soil carbon dynamics modeled. For more in-depth information on soil carbon modeling in the Canadian Prairies, the reader is referred to Campbell et al. (2007); Falloon and Smith (2006); Liang et al. (2005).

83 4.2. MODEL DESCRIPTION

Table 4.3: Results of the soil carbon model across ﬁve amortization periods

Amortization period (yrs) ∆ SOC (t C/ha) t CO2/ha/yr t CO2/t/yr g CO2/MJ 20 14.8 2.72 1.26 107.6 30 17.1 2.09 0.97 82.9 100 26.6 0.98 0.45 38.7 500 36.9 0.27 0.13 10.7

4.2.4 Results and Discussion

The soil carbon model was run for 500 years, with SOC loss recorded at 20, 30, 100, and 500 years. The total change in SOC, percentage of SOC lost, and annual land use change emissions calculated per hectare and per megajoule of biojet fuel is shown for each time period in Table 4.3. Figure 4.2 shows the decay of total SOC and change in the SOC in each of the four active pools over the first 100 years after grassland conversion was modeled to occur. SOC was lost exponentially over time, with 19% of the initial SOC stock lost during the first 20 years and 22% lost during the first 30 years (Table 4.3). Annual land use change emissions were lower for the 30 year scenario than the 20 year scenario because of the longer period over which emissions are annualized. This makes the amortization period a very important factor in estimates of annual land use change. Life cycle yields also strongly dictated estimates of land use change emissions.

Once annual emissions were calculated in tonnes of CO2 per hectare, canola seed yield, canola oil extraction yield, and biojet fuel yields were used along with a carbon to CO2 conversion factor of 3.67 to convert per-hectare emissions to g CO2eq/MJ. A sensitivity analysis conﬁrmed that variation in any yield had a linearly proportional impact on land use change emissions per megajoule. Soil loss leveled oﬀ at approximately 400 years for a total loss of 39.5 t/C/ha, or

84 4.2. MODEL DESCRIPTION

80 SOC (20 yrs) SOC Pools 70 = -14.8 t C/ha SOC (30 yrs) Resistant plant matter = -17.1 t C/ha Decomposable plant matter 60 Humus Microbial biomass 50 Total SOC

Soil organic carbon (SOC) [t C/ha] 10

0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Years since land conversion Figure 4.2: Change in total soil organic carbon (SOC) and change in SOC in each of four active pools over 100 years as a result of grassland to cropland conversion

51% of the initial SOC level. By 20 and 30 years, 38% and 43% of total SOC loss had occurred, which considering experimental observations that most SOC loss abates after 30 years (Janzen et al., 1998), could indicate that the model underestimated the decay of SOC. However, the model represents a crop rotation that contains no fallow (positive soil cover parameter every month), which does not represent the average cropping conditions of the 20th century (Janzen et al., 1998), but does reﬂect current best practices in the canola farming industry Smith and Barbieri (2012).

85 4.2. MODEL DESCRIPTION

Comparison to literature

Land use change emissions over 20 and 30 years fell within ±60% of other estimates of emissions from grassland to oilseed cropland conversion reported in the literature (Miller and Kumar, 2013; Stratton, 2010; Uusitalo et al., 2014; Wong, 2008). Emis- sions calculated here were lower that the literature in all cases except one, an estimate from Uusitalo et al. (2014) of only 1.101 t CO2/ha/yr. Uusitalo et al. (2014) assumed that set-aside land was converted to cropland. Set-aside land is land that was once cropland, but has since been entirely or partially restored to natural cover. The initial SOC level of set-aside land estimated by Uusitalo et al. (2014) was much smaller than other estimates for set-aside land, i.e. Stratton et al. (2010). Land use change emissions from the 20 year scenario were only 17% lower than similar estimates from Miller and Kumar (2013), who also used Canadian Prairie grassland as the reference land use, although land use change emissions were calculated using IPCC emission factors. Emissions calculated here could be lower on average than other estimates in the literature because of the dynamic nature of the soil carbon model. Other estimates using IPCC Tier I/II emission factors assume a loss of SOC based on default values for cropland and grassland (Penman et al., 2003). However, as described above, less than half of the total SOC loss from grassland to cropland conversion had occurred after 30 years. Therefore, the model may overestimate the SOC stocks in cropland at 20 and 30 years, compared to estimates that may treat cropland SOC as being at equilibrium. A more rapid rate of decay would have generated lower SOC levels by the 20 year mark.

86 4.2. MODEL DESCRIPTION

Model advantages

There are several advantages to using soil carbon modeling to estimate land use change emissions. Because final SOC levels are not only a function of initial SOC stocks, results are less sensitive to variation in initial SOC estimates. A 10% increase in initial SOC level increases annual per-hectare land use change emissions by 50% when land use change is calculated simply by subtracting 20 year final SOC level from inital SOC level, as is done in IPCC Tier I/II methodology. The same 10% change within the model resulted in only a 16% change in annual emissions. Because of regional variability in soil carbon stocks, it is an advantage to LCA practionners to use land use change estimation methods that are robust to this kind of uncertainty. Another advantage offered by soil carbon modeling is the flexibility to calculate emissions for different time periods. Initial and final SOC levels for a range of climatic zones, land uses, and soil types are available in the literature. However, these values represent only a snapshot in time. Without information on the dynamics of soil carbon loss, it is difficult to compare results with literature values for different amortization periods and climatic zones. Soil carbon models that produce step-wise outputs, like the monthly model applied here, allow for a comparison with a wider range of literature values for evaluation and review purposes.

4.2.5 Conclusion

Annual land use change emissions were estimated for grassland to canola cropland conversion in Western Canada using a well-known soil carbon model. Results over 20 and 30 years for grassland to cropland conversion were lower on average to values reported in the literature. However, soil carbon modeling allowed more temporal and

87 BIBLIOGRAPHY

regional ﬂexibility. Although land use change estimates using IPCC emission factors may be larger, and thus more conservative, the goal of this study was to estimate annual land use change emissions using agricultural and environmental parameters that matched the canola farming conditions in Western Canada as closely as possible. The 20 year estimate of land use change emissions will be applied to the GHG-LCA of canola-derived HEFA biojet in Chapter 5.

Bibliography

Ahern, F., Frisk, J., Latifovic, R., and Pouliot, D. Monitoring ecosystems remotely : a selection of trends measured from satellite observations of Canada. Technical report, Canadian Councils of Resource Ministers, 2013.

Alberta Government. Evaporation and Evapotranspiration in Alberta. Technical report, Alberta Environment and Sustainable Resource Development, 2013.

Boame, A. Zero tillage : a greener way for Canadian farms. Technical report, Statistics Canada, Agriculture Division, Ottawa, nov 2005.

88 BIBLIOGRAPHY

Campbell, C. A., VandenBygaart, A. J., Zentner, R. P., McConkey, B. G., Smith, W., Lemke, R., Grant, B., and Jeﬀerson, P. G. Quantifying carbon sequestration in a minimum tillage crop rotation study in semiarid southwestern Saskatchewan. Canadian Journal of Soil Science, 87(3):235–250, 2007. ISSN 0008-4271. doi: 10.4141/S06-018. URL http://pubs.aic.ca/doi/abs/10.4141/S06-018.

Canola Council of Canada. Canola: Production Tips (Summary of the Canola Grow- ers Manual). Technical report, Canadian Canola Council, 2012.

Canola Council of Canada. Photo- Canola Growing Regions of Canada, 2015. URL https://www.canolacouncil.org/media/image-gallery/canola-growing-region- map/.

Carlyle, W. J. The decline of summerfallow on the Canadian prairies. Canadian Geographer, 41(3):267–280, 1997. ISSN 00083658. doi: 10.1111/j.1541-0064.1997. tb01313.x.

Carpenter, S. R., Caraco, N. F., Correll, D. L., W.Howarth, R., Sharpley, A. N., and Smith, V. H. Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecological Applications, 8(1998):559–568, 1998. ISSN 1051-0761. doi: 10.1890/ 1051-0761(1998)008[0559:NPOSWW]2.0.CO;2.

Coleman, K. and Jenkinson, D. S. RothC - A model for the turnover of carbon in soil- model description and users guide (Windows version). Technical report, Rothamsted Research, 2014.

89 BIBLIOGRAPHY

Davidson, E. A. and Janssens, I. A. Temperature sensitivity of soil carbon decom- position and feedbacks to climate change. Nature, 440(7081):165–173, 2006. ISSN 00280836. doi: 10.1038/nature04514.

Environment Canada. Climate Normals 1981-2010 Station Data- Swift Current A Saskatchewan, 2017. URL http://climate.weather.gc.ca/climate{\ }normals/.

European Parliament. Directive 2009/28/EC of the European Parliament and of the Council of 23 April 2009. Oﬃcial Journal of the European Union, 140(16):16–62, 2009. ISSN 02870827. doi: 10.3000/17252555.L 2009.140.eng.

Falloon, P. and Smith, P. Simulating SOC changes in long-term experiments with RothC and CENTURY: model evaluation for a regional scale application. Soil Use and Management, 18:101–111, 2006. ISSN 1475-2743. doi: 10.1079/SUM2001108.

Falloon, P., Smith, P., Coleman, K., and Marshall, S. Estimating the size of the inert organic matter pool from total soil organic carbon content for use in the Rotham- sted carbon model. Soil Biology and Biochemistry, 30(8-9):1207–1211, 1998. ISSN 00380717. doi: 10.1016/S0038-0717(97)00256-3.

Fargione, J., Hill, J., Tilman, D., Polasky, S., and Hawthorne, P. Land Clearing and the Biofuel Carbon Debt. Science, 319(1235):1235–1237, 2008. ISSN 15829596. doi: 10.1126/science.1152747.

90 BIBLIOGRAPHY

Government of Canada. Renewable Fuel Regulations, 2013.

ISO. Greenhouse gases – Carbon footprint of products – Requirements and guidelines for quantiﬁcation and communication (ISO/TS 14067:2013). Technical report, International Standards Organization, Geneva, 2013.

Jenkinson, D., Andrew, S., Lynch, J., Goss, M., and Tinker, P. The Turnover of Organic Carbon and Nitrogen in Soil [and Discussion]. Philosophical Transactions: Biological Sciences, 329(1255):361–368, 1990. doi: www.jstor.org/stable/76840.

Kim, S. and Dale, B. E. Indirect land use change for biofuels: Testing predictions and improving analytical methodologies. Biomass and Bioenergy, 35(7):3235–3240, 2011. ISSN 09619534. doi: 10.1016/j.biombioe.2011.04.039.

Liang, B. C., Campbell, C. A., McConkey, B. G., Padbury, G., and Collas, P. An empirical model for-estimating carbon sequestration on the Canadian prairies. Cana- dian Journal of Soil Science, 85(4):549–556, 2005. ISSN 00084271.

91 BIBLIOGRAPHY

Mcconkey, B., Liang, C., Padbury, G., Pennock, D., and Lindwall, W. Measuring Soil Carbon Change on Cropland : The Prairie Soil Carbon Balance Project. Pro- ceedings for the Climate Change Workshop, 2000.

Parton, W. J., Schimel, D. S., Cole, C. V., and Ojima, D. S. Analysis of Fac- tors Controlling Soil Organic Matter Levels in Great Plains Grasslands. Soil Sci- ence Society of America Journal, 51:1173–1179, 1987. doi: doi:10.2136/sssaj1987. 03615995005100050015x.

Penman, J., Gytarsky, M., Hiraishi, T., Krug, T., Kruger, D., Pipatti, R., Buendia, L., Miwa, K., Ngara, T., Tanabe, K., and Wagner, F., editors. Good practice guidance for land use, land use change and forestry. Institute for Global Environmental Strategies (IGES) for the IPCC, Hayama, Kanagawa, Japan, 2003. ISBN 4-88788- 003-0. doi: citeulike-article-id:1260638.

Phillips, R. L., Eken, M. R., and West, M. S. Soil Organic Carbon Beneath Crop- lands and Re-established Grasslands in the North Dakota Prairie Pothole Re- gion. Environmental Management, 55(5):1191–1199, 2015. ISSN 14321009. doi: 10.1007/s00267-015-0459-3.

Qin, Z. and Huang, Y. Quantiﬁcation of soil organic carbon sequestration potential in cropland: A model approach. Science China Life Sciences, 53(7):868–884, 2010. ISSN 1674-7305. doi: 10.1007/s11427-010-4023-3.

92 BIBLIOGRAPHY

Saville, B. A. Economic Impact Study for a Canola-Based Biodiesel Industry in Canada. Technical report, Canola Council of Canada, BBI Biofuels Canada, Win- nipeg, 2006.

Shrestha, B. M., Mcconkey, B. G., Smith, W. N., Desjardins, R. L., Campbell, C. A., Grant, B. B., and Miller, P. R. Eﬀects of crop rotation , crop type and tillage on soil organic carbon in a semiarid climate. Canadian Journal of Soil Science, 93: 137–146, 2013. doi: 10.4141/CJSS2012-078.

Shrestha, B. M., Desjardins, R. L., McConkey, B. G., Worth, D. E., Dyer, J. A., and Cerkowniak, D. D. Change in carbon footprint of canola production in the Canadian Prairies from 1986 to 2006. Renewable Energy, 63:634–641, 2014. ISSN 09601481. doi: 10.1016/j.renene.2013.10.022.

Smith, E. G. and Barbieri, J. Summary of inputs and production practices used by canola growers in 2011. Technical report, Lethbridge Research Centre, Lethbridge, 2012.

Smith, W. N., Rochette, P., Monreal, C., Desjardins, R. L., Pattey, E., and Jaques, A. The rate of carbon change in agricultural soils in Canada at the landscape level. Canadian Journal of Soil Science, 77(2):219–229, 1997. ISSN 0008-4271. doi: 10.4141/S96-113.

93 BIBLIOGRAPHY

Statistics Canada. Table 004-0002 Census of Agriculture, total area of farms and use of farm land, Canada and provinces, every 5 years, 2011.

Statistics Canada. Table 001-0015 Exports of grains, by ﬁnal destination, monthly (tonnes)(1,2), 2017a.

Statistics Canada. Table 32-10-0359-01 Estimated areas, yield, production, average farm price and total farm value of principal ﬁeld crops, 2017b.

Stratton, R. W. Life cycle assessment of greenhouse gas emissions and non-CO combustion eﬀects from alternative jet fuels. Masters thesis, Massachusetts Institute of Technology, 2010.

94 BIBLIOGRAPHY

Wong, H. M. Life-cycle assessment of greenhouse gas emissions from alternative jet fuels. Masters thesis, Massachusetts Institute of Technology, 2008.

Yao, Z., Zhang, D., Yao, P., Zhao, N., Liu, N., Zhai, B., Zhang, S., Li, Y., Huang, D., Cao, W., and Gao, Y. Coupling life-cycle assessment and the RothC model to estimate the carbon footprint of green manure-based wheat production in China. Science of the Total Environment, 607-608:433–442, 2017. ISSN 18791026. doi: 10.1016/j.scitotenv.2017.07.028.

95 Chapter 5

Life cycle assessment

5.1 Introduction

Renewable jet fuel presents a growing opportunity to reduce the impact of air travel on the environment. To limit climate change impacts under a 4% annual emissions growth trend, the aviation industry has targeted a 50% reduction of 2005 emissions levels by 2050 (ATAG, 2017). Batteries and fuel cells are still too heavy for aircraft, leaving renewable drop-in aviation fuel, or biojet fuel, as the only short term option. Since 2009, six pathways for biojet fuel production have been approved, with more in the evaluation phase (ATAG, 2017). The HEFA biojet production pathway is the most promising pathway to date. The HEFA pathway produces HEFA biojet via the catalytic reﬁning of lipid feedstocks in a process similar to petroleum reﬁning. As of July 2018, HEFA biojet was the only biojet fuel being produced on a commerical scale (IATA, 2015). A recent techno- economic analysis found that out of six potential biojet production pathways, HEFA had one of the lowest production costs and was deemed to be the best option in the short term (de Jong et al., 2015). Although four companies currently produce

96 5.1. INTRODUCTION

renewable diesel from the HEFA pathway and only one company produces biojet, all of the producers who license or plan to license HEFA technology for future production have announced the intention to produce biojet fuel (Table 2.2). Recent years have seen Canada take an interest in developing a HEFA supply chain specifically to produce biojet fuel (Transport Canada, 2015). Most global HEFA fuel production uses inexpensive by-products of the animal and vegetable oil processing industry, such as tallow and inedible vegetable oils, but reliability of supply is an issue. Although tallow has been identified as a promising feedstock for Canadian HEFA production, the average annual supply is only about half of that required for a small production facility (Transport Canada, 2015, Table 7) (Table 2.2). The availability of oilseeds like canola as a feedstock for HEFA production is much greater than that of by-product lipids, but life cycle assessment (LCA) has suggested that crop-based biofuels may actually be worse for the environment than petroleum- derived fuels (Searchinger et al., 2008; Stratton, 2010). LCA is the most widely-used tool to assess the environmental impact of a product or process and is commonly used to compare alternative fuels with petroleum fuels. Although multiple environmental impacts can be assessed with LCA, the life cycle impact of biofuels is usually expressed as an emissions intensity, in grams of carbon dioxide equivalents (CO2eq) per megajoule. There were at least twelve LCAs of HEFA biojet fuel published between 2008 and 2018 (Figure 3.1). To be included under national or international carbon credit schemes that will make biojet more affordable for airlines, LCA must demonstrate that biojet fuel offers substantial emissions reductions compared to petroleum jet fuel. Inconsistencies in LCA methodology have resulted in a wide range of reported

97 5.1. INTRODUCTION

emissions reductions for renewable transportation fuels, including biojet (Shonnard et al., 2010; Stratton et al., 2011). One of the biggest sources of variability is the inconsistent inclusion of land use change emissions. Land use change emissions are released from devegetation and soil carbon loss when land is converted to cropland to satisfy biofuel demand without impacting global food supply. Of the limited number of LCAs on canola-derived HEFA biojet, there are even fewer that include land use change. Among the studies that do include land use change, there is a large amount of variability in reported emissions intensity. Reviews have suggested that variability in reported emissions intensity stems mainly differences in the reference land use, and how emissions are allocated between the main product and co-products (Miller and Kumar, 2013; Stratton, 2010; Uusitalo et al., 2014) (see Chapter 3). Different sources of input data causes additional variation between studies. Most LCAs use the IPCC tier I/II methodology to estimate land use change, where default emission factors for land conversion are applied across global climatic zones (Aalde et al., 2006). A recent review of land use change accounting methods recommended that soil carbon models be used instead of IPCC methodology due to the importance of regional applicability (Goglio et al., 2015). Despite this, soil carbon modeling is almost always outside of the scope of biofuel LCA. In addition to a lack of model-derived land use change estimates, there have been no HEFA biojet LCAs identified that use recent canola farming data from Canada. Because farming emissions make up a large portion of the life cycle emissions of oilseed biojet (Shonnard et al., 2010), and farming practices vary regionally, farming data should be as geographically specific as possible. Most LCAs of canola-based

98 5.2. METHODS

fuels use European farming data compiled by Stratton (2010). Miller and Kumar (2013) use Canadian data, but it is based on canola grower recommendations from 2003. Although a 2011 survey by Agriculture and Agrifood Canada collected data on farming inputs and practices from nearly one thousand Canadian canola growers (Smith and Barbieri, 2012), no LCA of canola-based fuel was found that used this data. This study will be the first LCA of canola HEFA biojet to use model-derived land use change estimates and recent Canadian production data. To address variability, the resulting LCA of the canola HEFA biojet fuel product system will be analyzed with and without land use change across four different co-product allocation methods. A sensitivity analysis will be applied to life cycle inputs and co-product allocation parameters to identify the most important LCA inputs, and to provide recommendations to reduce life cycle emissions. This work will help support the adoption of biojet fuel in Canada by reporting a range of possible GHG reductions for canola biojet fuel compared to petroleum- derived jet fuel. It will also contribute to an understanding of how land use change emissions and co-product allocation methods affect emissions intensity.

5.2 Methods

5.2.1 Life cycle assessment (LCA) in SimaPro

The LCA of canola biojet was performed using SimaPro v8.0.2.13 LCA software. In SimaPro, life cycle stages are represented by objects called assemblies. Each assembly can contain multiple processes that represent economic activities, such as the production of a raw material. The environmental impact of processes are expressed in terms

99 5.2. METHODS

of emissions to the natural environment. The life cycle of canola biojet was modeled by combining the appropriate quantity of relevant assemblies and their processes and applying an impact assessment method to characterize the emissions from each process. Most processes were obtained from the SimaPro Ecoinvent 3.0 database. Cana- dian or North American speciﬁc data was not available for truck transportation, rail transportation, and fertilizer and chemical manufacturing, so processes representing the global average were used. Processes for the production of petroleum fuel inputs, such as natural gas, were constructed in SimaPro using an emissions intensity calculated in GHGenius v.4.03 LCA software for Western Canada (see Appendix C). Electricity processes were available in SimaPro for all Canadian provinces, so electricity production in Western Canada was modelled as the 2012 Alberta average mix.

Impact assessment

The emissions intensity of canola biojet was calculated using the IPCC 2007 GWP 100a impact assessment method. This method assesses the global warming potential

(GWP) of GHGs relative to CO2, over a time horizon of 100 years. GHGs considered by the IPCC method include CO2,N2O, and CH4, with a respective GWP of 1, 298, and 25. Although some non-greenhouse gas environmental impacts such as water quality degradation and natural resource depletion have been previously considered in biojet LCAs, and may be substantial (Li and Mupondwa, 2014), the scope of this study is restricted to GHG emissions because they are easier to quantify in LCA, and this study is only concerned with the range of variability in emissions intensity of canola biojet, not all potential impacts. Emissions intensity was calculated in grams

100 5.2. METHODS

of CO2 equivalents per megajoule of biojet (g CO2eq/MJ) across eight scenarios: four allocation methods, each with and without land use change. Statistics were not used to compare the eight scenarios because the calculated emissions intensity is not a sample from a population, and therefore represents the true value of the data. If there was a measure of uncertainty inherent to the model, i.e. through a reported range in on or more inputs, a Monte-Carlo analysis could have been performed to produce a mean and a conﬁdence interval for the true value of the emissions intensity for each scenario. Statistics could have then been used to compare the mean of scenarios. A Monte-Carlo analysis was performed preliminarily, but it was determined that the range of variability from methodological choices (i.e. allocation method and land use change inclusion) was the primary focus of the study, so results focus on these sources of uncertainty instead. Ultra-low sulfur petroleum-derived jet fuel was used as the reference fuel from which potential GHG reductions from biojet fuel were calculated. The emissions intensity, was calculated as 101.5 g CO2eq/MJ using GHGenius for the year 2017, using Western Canadian production assumptions. The emissions intensity of conventional (0.06% sulfur) jet fuel is 93.3 g CO2eq/MJ, but ultra-low sulfur jet (ULSJ) fuel was used because biojet fuel contains no sulfur and may be more likely to replace ULSJ fuel than conventional jet fuel. More details on the calculation of the emissions intensity of petroleum fuels can be found in Appendix C.

5.2.2 Scope

The typical scope of renewable fuel LCAs is well-to-wake (WTWa), in reference to the life cycle of petroleum products. WTWa includes all inputs and outputs in the HEFA

101 5.2. METHODS

biojet fuel product system, from the production of raw feedstock to the combustion of the fuel in a jet engine (Figure 3.2). The stages in the life cycle of canola HEFA biojet were defined as feedstock production, oil extraction and upgrading, refining, fuel distribution, and combustion. Land use change was included as an additional stage in scenarios that included it. The emissions from water consumption and waste treatment were not included in the LCA. Canadian canola is almost entirely rain-fed Ward (2016)), and as such water consumption is not typically accounted for (Rustandi and Wu, 2010). Waste treatment typically make up less than 1% of biofuel life cycle emissions (Shonnard et al., 2010). Emissions from infrastructure, such as equipment manufacturing and production facility construction, were also not included as is. Transportation of seed is included in the oil extraction stage and transportation of oil is included in the refining stage.

Life cycle inputs

Data on the inputs and outputs of each life cycle stage was sourced from academic literature, LCA databases, and industry publications. Table 5.1 below shows the quantity of all inputs in the LCA. See Appendix C for a detailed explanation of all data sources and for assumptions made about displaced products.

Table 5.1: Inputs and yields of the canola HEFA biojet life cycle

Stage Input Quantity Units Reference Comments Land use change

Soil carbon 2.72 t CO2/ha/yr Ch. 4 20 years loss Continued on next page 102 5.2. METHODS

Table 5.1: Continued from last page

Stage Input Quantity Units Reference Comments

Feedstock production Yield 2.15 t1 seed/ha Smith and Range given Barbieri 1.84-2.59 (2012) Nitrogen 50.92 kg N /t Smith and 48.72% urea, canola Barbieri 5.42% am- (2012) monium sulfate, 10.83% ammonium nitrate, 27.73% liquid ammonia, 7.29% manure Phosphorus 15.36 kg P2O5/t Smith and Phosphate canola Barbieri (2012) Potassium 6.41 kg K/t Smith and Potassium canola Barbieri chloride (2012) Sulfur 10.64 kg S/t Smith and 49% elemen- canola Barbieri tal S, 51% (2012) ammonium sulfate Herbicides 0.273 kg/t canola Smith and Glyphos- Barbieri phate (2012); (S&T)2 Con- sultants Inc. (2013) Continued on next page

1all units are metric: 1 t= 1 Mg= 1000 kg 103 5.2. METHODS

Table 5.1: Continued from last page

Stage Input Quantity Units Reference Comments Insecticides 0.039 kg/t canola Smith and Organophos- Barbieri phorus (2012); (S&T)2 Con- sultants Inc. (2013) Diesel 523.71 MJ/t canola Smith and Combusted Barbieri in tractor, (2012); LHV (S&T)2 Con- sultants Inc. (2013) Natural gas 0.93 MJ/t canola Smith and Storage and Barbieri drying, com- (2012); busted in (S&T)2 Con- boiler, LHV sultants Inc. (2013) Electricity 3.4 kWh/t Smith and Medium volt- canola Barbieri age Alberta (2012); mix (S&T)2 Con- sultants Inc. (2013) Oil extraction and upgrading Yield 2.25 t seed/t oil (Canadian Extrapolated Oilseed Pro- for 2017; cessors As- Canadian sociation, processing 2016) average N-hexane 3.23 kg/t oil Rustandi and Wu (2010) Continued on next page

104 5.2. METHODS

Table 5.1: Continued from last page

Stage Input Quantity Units Reference Comments Phosphoric 3.77 kg/t oil Rustandi 85% solution acid and Wu (2010) Sodium hy- 0.93 MJ/t oil Rustandi 50% solution droxide and Wu (2010) Natural gas 2340 kg/t oil Rustandi Process heat, and Wu combusted in (2010) boiler, LHV Electricity 114.5 kWh/t oil Rustandi High voltage, and Wu Alberta mix (2010) Seed trans- 225 tkm/t oil (S&T)2 Con- 100 km avg port sultants Inc. (2013) Reﬁning Yield 2.02 t oil/t jet UOP (2013) Range given: 1.72- 2.41 Hydrogen 63.24 kg/t jet UOP (2013) LHV, steam methane reforming Natural gas 14,818 MJ/t jet Li and Process heat, Mupondwa combusted in (2014) boiler Electricity 97.72 kWh/t jet Li and High voltage, Mupondwa Alberta mix (2014) Oil transport 100.5 tkm/t oil (S&T)2 Con- 50 km aver- sultants Inc. age distance (2013) Distribution Continued on next page

105 5.2. METHODS

Table 5.1: Continued from last page

Stage Input Quantity Units Reference Comments

Biojet to air- 225 tkm/t jet (S&T)2 Con- 225 km aver- port (truck) sultants Inc. age distance (2013) Biojet to 225.26 tkm/t jet (S&T)2 Con- 225 km aver- airport (rail) sultants Inc. age distance (2013) Combustion HEFA bio- 44100 MJ/t (S&T)2 Con- LHV jet energy sultants Inc. density (2013) Biojet com- 0.654 g CO2eq/MJ Elgowainy Non-CO2 bustion biojet et al. (2012) emissions

5.2.3 Co-product allocation

Co-products are any outputs generated in the life cycle of a product that are not the final product or an intermediate product. Where the emissions from a process cannot be subdivided, ISO standard 14040 governing the principles of LCA specifies that consideration must be made to the allocation of process emissions between multiple outputs. In the canola biojet life cycle, two co-products were considered: canola meal, produced in the oil extraction stage, and renewable diesel, naphtha, and propane, produced in the refining stage (see Chapter 2). Canola straw is not treated as a co-product of the feedstock production stage because it decomposes rapidly and is typically left on the field to reduce soil carbon loss (Canola Council of Canada, 2012). Co-product allocation methods for a product system are based either on ratio allocation, a displacement credit, or a mixture of both. Figure 5.1 illustrates the basic differences between ratio allocation and displacement allocation. Following

106 5.2. METHODS

Figure 5.1: An example of the application of ratio allocation and displacement allocation to the ﬂow of process emissions between multiple outputs of a life cycle. ratio allocation, emissions produced in a life cycle stage (Process 1 in Figure 5.1 are split between the main product (Product A) and any co-products (Product B) based on properties of the products. The common properties used to allocate emissions are energy content, mass, or market value. The displacement credit method is quite diﬀerent. Following displacement, emissions produced in a life cycle stage (Process 1 in Figure 5.1 are entirely allocated to the main product, Product A. However, an emissions credit is calculated and subtracted from the life cycle of the product being evaluated. The emission credit is equivalent to the emissions intensity of a product in the market (Product C) that the co-product, Product B, displaces. The appropriate allocation method depends on the nature of the co-products and the goal and scope of the assessment. Ratio allocation is easier to standardize and as such is recommended by the European Renewable Energy Directive (RED) (European Parliament, 2009), but displacement is in closer compliance with ISO recommendations for LCA, as it avoids narrowing system boundaries (ISO, 2006). Accordingly, LCAs often compare results using two or more allocation methods.

107 5.2. METHODS

Comparison of co-product allocation methods

In this study, four co-product allocation methods were compared. Each method is comprised of two parts: the allocation principle applying to canola oil/canola meal, and the allocation principle applying to biojet/renewable fuel. The four co-product allocation methods that were compared are summarized in Table 5.2. Three methods apply energy-based allocation to biojet/renewable fuel, and one method applies displacement. In the LCA of biofuels, energy-based allocation is usually applied to fuels, as the energy ratio of products can easily be calculated. Mass and market-based allocation were not among the methods compared, because the resulting allocation ratios are similar to the energy ratio (Han et al., 2013). Neither canola oil or canola meal are valued strictly on mass or energy content, which makes selecting an allocation principle for canola oil/canola meal diﬃcult. Energy-based, mass-based and displacement allocation were compared (Table 5.2). Market value was not included here due to its ﬂuctuation over time. Market-based allocation was compared to mass-based allocation in Miller and Kumar (2013) for the canola biojet life cycle and resulted in a lower emissions intensity. For methods applying displacement to the canola oil/canola meal product pair (D/E and D/D), it was assumed that canola meal displaced soybean meal on a protein- equivalent basis. For methods applying displacement to the biojet/renewable fuels product pair (D/D), it was assumed that renewable fuels displaced petroleum fuels on an energy-equivalent basis. Processes pertaining to the product system of the displaced products were constructed in SimaPro using production assumptions for Western Canada. See Appendix C for more information on the calculation of the emissions intensity of displaced products and the application of co-product allocation

108 5.2. METHODS

Table 5.2: Co-product allocation methods compared in this study

Allocation Main product/co-product pair method Canola oil/canola meal Biojet/renewable fuels Emissions allocation ratio (%) or product displaced E/E energy-based energy-based 69/31 59/41 M/E mass-based energy-based 44/54 59/41 D/E displacement energy-based soy meal 59/41 D/D displacement displacement soy meal petroleum fuels

to the calculation of life cycle emissions.

5.2.4 Sensitivity analysis on inputs

The sensitivity of the emissions intensity to parameters such as crop yield or fertilization rate have been assessed in a number of HEFA LCAs (Bailis and Baka, 2010; Miller and Kumar, 2013; Shonnard et al., 2010; Sieverding et al., 2016), but the impact of variation in input parameters has rarely been analyzed systematically. A sensitivity analysis was performed on the quantity of the inputs in the canola biojet life cycle. The eﬀect of a ±10% variation in any inputs with a GHG contribution to the life cycle greater than 1% was analyzed. The GHG contribution of an input to the life cycle is linearly proportional to both the quantity of that input in the life cycle and to its emissions intensity per unit. The equation below explains the relationship between life cycle emissions and the quantity and emissions intensity of each input:

109 5.2. METHODS

n X L = ein ∗ qn (5.1) i=1

Where L is the emissions intensity of canola HEFA biojet, ein is the emissions intensity of one unit of input n, and qn is the number of units of input n.

Because of the linearity of this relationship, manipulating either qn or ein in the sensitivity analysis would generate the same result. The number of units of input n in the life cycle reﬂect the allocation method applied. If a ratio allocation method is applied, the total number of units is split between the main product and the co-product(s) accordingly. If a displacement credit is applied, units of inputs associated with the displaced product are subtracted from the life cycle (Figure 5.1).

5.2.5 Sensitivity analysis on co-product allocation parameters

Co-product allocation parameters are the allocation ratios and displacement credits that are applied to the canola biojet life cycle. Although many studies have compared multiple co-product allocation methods, no HEFA LCA was found that systematically analyzed the impact of variation in co-product allocation parameters on life cycle emissions. Bailis and Baka (2010) compared three diﬀerent displacement assumptions for jatropha husks, but the allocation ratios for other co-product allocation methods were not varied. Here, the percentage change in the emissions intensity of canola biojet from a ±10% change in the size of the co-product allocation parameters was calculated across all four allocation methods both with and without land use change. The ranges

110 5.3. RESULTS AND DISCUSSION

of variation in the ratios were 53.1-64.9% for the biojet allocation ratio (energy- based), 39.6-48.4% for the oil allocation ratio (mass-based) and 62.1-75.9% for the oil allocation ratio (energy-based). The size of displacement credits was manipulated by multiplying the size of the credit by 0.9 or by 1.1. Parameters were manipulated one at a time so that the relative importance of parameters applied to each product pair.

5.3 Results and Discussion

Across co-product allocation methods, the emissions intensity of canola biojet fuel

ranged from 46.4-66.3 g CO2eq/MJ without land use change and 80.5-153.1 g CO2eq/MJ with land use change (Figure 5.2). The potential GHG emissions reductions from canola biojet fuel compared to the reference petroleum-derived jet fuel were between 35% to 54% without land use change and between -51% to 10% with land use change. When land use change was included, canola biojet had a higher emissions intensity than petroleum-derived jet fuel under all allocation methods except mass-based allocation (M/E). The relationship between the co-product allocation method that was applied and the emissions intensity of canola biojet was aﬀected by the inclusion of land use change. The quantity of life cycle emissions allocated to co-products under displacement allocation do not increase proportionally to life cycle emissions, so displacement scenarios were more aﬀected by the inclusion of land use change than were ratio allocation methods. Canola biojet consistently had the lowest emissions intensity under M/E allocation, both with and without land use change (46.4 g CO2eq/MJ and 80.5 CO2eq/MJ). The M/E method allocates a higher percentage of emissions to co-products than the energy-based (E/E) method: 69% compared to 54% (Table

111 5.3. RESULTS AND DISCUSSION

Figure 5.2: Emissions intensity of canola biojet fuel across four co-product allocation methods with and without land use change (LUC), compared to petroleum-derived jet fuel (P. Jet). Negative emissions contributions (displacement credits) are shown in light red and positive emissions contributions are shown in light blue.

5.2). Without land use change, the emissions intensity of canola biojet was the highest under the displacement-energy (D/E) method; with land use change, it was the highest under the displacement method (D/D). The emissions intensity of canola biojet was the most sensitive to land use change inclusion under the D/D method. When displacement was applied to both co- products, the emissions intensity of canola biojet increased by 185%, double that of any other method. Under D/D, the emissions allocated to the renewable fuels co-product are ﬁxed, so additional emissions from land use change inclusion were entirely allocated to the main product. D/E was much less sensitive to land use change inclusion (69% increase) because the emissions allocated to the canola meal

112 5.3. RESULTS AND DISCUSSION

co-product were not ﬁxed. The emissions allocated to canola meal increased when land use change was included because land use change emissions from soybean meal were included in the displacement credit. Other studies have found displacement to be more than twice as sensitive to land use change inclusion as other allocation methods, depending on the land use change assumptions for displaced products (Bailis and Baka, 2010; Uusitalo et al., 2014). There was also a greater sensitivity to land use change inclusion under the E/E method (93% increase) than under the M/E method (73% increase) because a higher percentage of land use change emissions were allocated to the main product. This has been supported by similar ﬁndings in the literature (Bailis and Baka, 2010; Wong, 2008).

5.3.1 Contribution analysis

Figure 5.3 shows the contribution of each life cycle stage to total life cycle emissions across all four co-product allocation methods. The emissions intensity for each allocation scenario, including the negative contribution from displacement credits, is depicted with a labeled transparent bar. Land use change emissions made up 42-52% of the life cycle emissions of canola biojet fuel. Land use change was the most emissions-intensive life cycle stage, followed by feedstock production and refining. Before any allocation method is applied, emissions from feedstock production and refining constituted 50% and 36% of life cycle emissions respectively, without land use change. Both feedstock production and refining were associated with processes or material inputs that had a high emissions intensity. The impact of variation in the quantity of life cycle inputs was analyzed

113 5.3. RESULTS AND DISCUSSION

Figure 5.3: The contribution of each life cycle stage and displacement credit to total life cycle emissions, across four co-product allocation methods with and without land use change (LUC) in Section 5.3.2. The oil extraction stage contributed to 12% of unallocated life cycle emissions. The contribution of emissions from other stages was very small; less than 1% for fuel distribution and 0.05% for fuel combustion.

5.3.2 Sensitivity analysis

Inputs

The eﬀect of a ±10% variation in the quantity of any input contributing more than 1% to overall life cycle emissions was reported in Figure 5.4. Following a tornado chart format, the inputs that result in the greatest change in the emissions intensity are listed in descending order. The width of the horizontal bars depicts the magnitude of the change in emissions intensity, in g CO2eq/MJ, for both positive (blue) and

114 5.3. RESULTS AND DISCUSSION

Sensitivities

NG (ref.) Hydrogen N O (fert.) 2 N O (resid.) 2 Urea (N fert.) NG (oil ex.) Farm diesel Ammonia (N fert.) Elec. (oil ex.) Phosphate (P fert.) Truck to ref. Elec. (ref.) Truck to airport Truck to oil ex.

54.5 55 55.5 56 56.5 57 57.5 Emissions intensity (g CO eq/MJ) 2 Figure 5.4: Sensitivity analysis showing the effect on the emissions intensity of canola biojet fuel from a ±10% change in the quantity of any input making ≥ 1% contribution to total emissions. The effect of a -10% change is shown in red and the effect of a +10% is shown in blue.

negative (red) changes in the quantity of an input. The baseline emissions intensity is for E/E allocation without land use change. The relative importance of inputs was no different when land use change was included, but a ±10% change in land use change did have more of an effect on the emissions intensity than did variation in any other input. Considering the size of the land use change life cycle stage, this was to be expected (Figure 5.3). Life cycle emissions were the most sensitive to variation in the quantity of natural gas consumed in the refining stage (NG ref.), and hydrogen production. Hydrogen production was modeled in SimaPro as steam methane reforming (SMR) of Western Canadian natural gas. Some studies have suggested that producing hydrogen from renewable co-product fuels instead of natural gas could reduce life cycle emissions by

115 5.3. RESULTS AND DISCUSSION

10-20% (Fan et al., 2013; Ukaew et al., 2016). Renewable co-product fuels are used by commercial HEFA producer Neste to reduce natural gas demand at their facilities (Neste, 2016; Nikander, 2008). However, if energy-based allocation is applied, renewable fuels have the same emissions intensity as canola HEFA biojet on a per-MJ basis. Considering natural gas has an emissions intensity of only 67.9 g CO2eq/MJ ((S&T)2 Consultants Inc., 2013), there could be little environmental beneﬁt to producing hydrogen from renewable fuel co-products instead of natural gas.

The next most important inputs were N2O emissions from nitrogen fertilizer (N2O

N fert.) and N2O emissions from the natural decay of crop residues (N2O N resid.), followed by the production of urea (nitrogen fertilizer), natural gas consumed in the oil extraction stage (NG oil ex.), and diesel use in feedstock production. 40% of the nitrogen applied to canola crops in Canada is in the form of urea, which is synthesized from anhydrous ammonia (NH3) (Smith and Barbieri, 2012). Replacing urea with liquid NH3 has been suggested to reduce the emissions from of camelina production by 10% due to lower manufacturing emissions (Shonnard et al., 2010). The relative importance of the inputs analyzed in this sensitivity analysis was the same for M/E allocation as it was for E/E allocation (Figure 5.4). For D/D and D/E allocation, the relative importance of inputs was slightly different, as there were also inputs associated with displaced products. N2O emissions from synthetic fertilizer were more important than natural gas demand under D/D and D/E allocation than E/E or M/E allocation. Allocation ratio methods allocate a larger percentage of emissions to co-products from feedstock production than from refining, because the allocation ratio for canola oil/canola meal is not applied to the refining stage. Therefore, inputs to the feedstock production stage have less of an impact than they

116 5.3. RESULTS AND DISCUSSION

Figure 5.5: Change (%) in the emissions intensity of canola biojet fuel from a ±10% change in the size of allocation parameters applied to canola oil (oil) or biojet (jet), with and without land use change (LUC) across four allocation methods would under displacement allocation, where all emissions are allocated to the main product before displacement credits are applied.

Co-product allocation parameters

Across all co-product allocation methods, there was a greater change in the emissions intensity of canola biojet fuel when the parameter affecting the allocation of emissions to biojet fuel was varied by 10% than when the parameter affecting emissions allocation to canola oil was varied by 10% (Figure 5.5). For the E/E and M/E methods, there was greater sensitivity to variation in the biojet parameter because it is applied to the refining life cycle stage and to all stages downstream. The canola oil parameter does not apply to the refining stage, which is 117 5.3. RESULTS AND DISCUSSION

the most emissions intensive stage after land use change and feedstock production. For the D/D and D/D methods, there was a much greater change (11% and 13%) in the emissions intensity from variation in the biojet parameter. The displacement credit for renewable fuels was greater than the credit for canola meal, and large relative to total life cycle emissions. The sensitivity of the calculated emissions intensity to variation in the biojet parameter decreased when land use change was included under the D/D method, but not under the D/E method. Under the D/D method, the displacement credit for renewable fuel made up a smaller proportion of total life cycle emissions when land use change was included, making results less sensitive to variation in that parameter. When land use change is included under the D/E method, the displaced land use change emissions in the canola meal credit decreases total life cycle emissions, and thus increases the sensitivity of results to variation in the biojet parameter. The eﬀect on the calculated emissions intensity from variation in the size of the canola meal displacement credit is very small without land use change (< 1%). The increase in sensitivity when land use change was included was noticeable only for the oil parameter under the E/E and M/E methods. Adding land use change emissions to the canola biojet life cycle increases the quantity of emissions to which the oil allocation ratio is applicable by a larger percentage than it increases the quantity of emissions to which the biojet allocation ratio is applicable. This is because the biojet allocation ratio already applies to the reﬁning stage.

118 5.4. CONCLUSION

5.4 Conclusion

Canola-derived HEFA biojet produced in Western Canada has the potential to reduce GHG emissions compared to petroleum-derived jet fuel by between 35% and 54% if land use change emissions from potential natural land conversion are not accounted for. The inclusion of land use change reduced the possibility that canola biojet could provide any emissions reductions under all allocation methods except the mass-based (M/E) method. Land use change inclusion increased the emissions intensity of canola biojet by 185% when displacement was applied to both co-products, by twice as much as under any other allocation method. After land use change, feedstock production and refining are the most emissions- intensive stages in the canola biojet life cycle. A sensitivity analysis on the quantity of inputs showed that natural gas and hydrogen demand in the refining stage are the most important inputs, followed by N2O emissions from the application of nitrogen fertilizers and the natural decay of canola residues. Variation in the biojet allocation parameter had more of an effect on the calculated emissions intensity than the canola oil parameter for two reasons. First, for allocation ratio methods, the biojet parameter applies to all life cycle stages downstream, and second, for displacement allocation, the renewable fuels credit is large relative to the canola meal credit. This highlights the sensitivity of results to biojet fuel yield. Biojet fuel yield can vary from 40% to 60% of refining product by weight (Pearlson et al., 2013; Stratton et al., 2010; UOP, 2013). Methodological variability results in a large range of potential GHG emissions reductions that could be achieved from the adoption of canola-derived HEFA biojet fuel. Co-product allocation method is an unavoidable source of variability in LCA that

119 BIBLIOGRAPHY

can also increase the importance of land use change assumptions. To overcome these challenges, a standardized allocation method should be adopted by any agencies that certify renewable transportation fuels. Out of the methods reviewed here, energy- based allocation is recommended because it is more conservative than mass-based allocation but less sensitive to land use change inclusion than displacement. Energy- based allocation is applied by the European RED in their LCA methodology. Land use change has an overwhelming inﬂuence on the calculated emissions intensity of biojet fuel. LCA may not be an appropriate tool for quantifying the risks of land use change, as it is largely dependent on economics, and the scale of uncertainty in land use change estimates is far greater than that of any other life cycle input (Chapter 3). However, including land use change in LCA may help steer policymakers away from biofuel pathways that are at an increased risk for land use change. Here, an estimate of land use change is included under the pessimistic assumption that a greater demand for canola as biojet fuel feedstock could lead to cropland expansion in the Canadian Prairies. This research will help facilitate inform the ongoing assessment of canola-derived HEFA biojet fuel as a tool to reduce emissions from aviation.

Bibliography

120 BIBLIOGRAPHY

Buendia, L., Miwa, K., Ngara, T., and Tanabe, K., editors, 2006 IPCC Guide- lines for National Greenhouse Gas Inventories, Volume 4: Agriculture, Forestry and Other Land Use, chapter Chapter 2, pages 2.1–2.59. Global Environmental Strategies (IGES) for the IPCC, 2006.

ATAG. Beginner’s Guide to Sustainable Aviation Fuel. Technical Report Edition 3, Air Transport Action Group, 2017.

Bailis, R. E. and Baka, J. E. Greenhouse gas emissions and land use change from Jatropha curcas-based jet fuel in Brazil. Environmental science & technology, 44 (22):8684–8691, 2010. ISSN 1520-5851. doi: 10.1021/es1019178.

Canadian Oilseed Processors Association. Overview of Canada’s Oilseed Processing Sector. Technical report, Canadian Oilseed Processors Association, 2016.

Canola Council of Canada. Canola: Production Tips (Summary of the Canola Grow- ers Manual). Technical report, Canadian Canola Council, 2012. de Jong, S., Faaij, A., Slade, R., Mawhood, R., and Junginger, M. The feasibility of short term production strategies for renewable jet fuels- a comprehensive techno- economic comparison. Biofuels, Bioproducts and Bioreﬁning, 6(3):246–256, 2015. ISSN 1932104X. doi: 10.1002/bbb.

121 BIBLIOGRAPHY

Fan, J., Shonnard, D. R., Kalnes, T. N., Johnsen, P. B., and Rao, S. A life cycle assessment of pennycress (Thlaspi arvense L.) -derived jet fuel and diesel. Biomass and Bioenergy, 55:87–100, 2013. ISSN 09619534. doi: 10.1016/j.biombioe.2012.12. 040.

Han, J., Elgowainy, A., Cai, H., and Wang, M. Q. Life-cycle analysis of bio-based aviation fuels. Bioresource Technology, 150:447–456, 2013. ISSN 09608524. doi: 10.1016/j.biortech.2013.07.153.

IATA. IATA 2015 Report on Alternative Fuels- 10th edition. Technical report, Inter- national Air Transport Association, Montreal-Geneva, 2015.

ISO. Environmental management Life cycle assessment Requirements and guidelines (ISO 14044). Technical report, International Standards Organisation, 2006. URL http://www.springerlink.com/index/10.1007/s11367-011-0297-3.

Li, X. and Mupondwa, E. Life cycle assessment of camelina oil derived biodiesel and jet fuel in the Canadian Prairies. Science of the Total Environment, 481(1):17–26, 2014. ISSN 00489697. doi: 10.1016/j.scitotenv.2014.02.003.

122 BIBLIOGRAPHY

Neste. Neste Renewable Diesel Handbook. Technical report, Neste, 2016.

Nikander, S. Greenhouse gas and energy intensity of product chain: case transport biofuel. Masters thesis, Helsinki University of Technology, 2008.

Rustandi, F. and Wu, H. Biodiesel production from canola in Western Australia: energy and carbon footprints and land, water, and labour requirements. Industrial and Engineering Chemistry Research, 49(22):11785–11796, 2010. ISSN 08885885. doi: 10.1021/ie1013162.

Sieverding, H. L., Zhao, X., Wei, L., and Stone, J. J. Life-Cycle Assessment of Oilseeds for Biojet Production Using Localized Cold-Press Extraction. Journal of 123 BIBLIOGRAPHY

Environment Quality, 45(3):967, 2016. ISSN 0047-2425. doi: 10.2134/jeq2015.06. 0313.

Smith, E. G. and Barbieri, J. Summary of inputs and production practices used by canola growers in 2011. Technical report, Lethbridge Research Centre, Lethbridge, 2012.

(S&T)2 Consultants Inc. GHGenius Model 4.03 Volume 2: Data and Data Sources. Technical report, Natural Resources Canada, Ottawa, ON, 2013.

Stratton, R. W. Life cycle assessment of greenhouse gas emissions and non-CO combustion eﬀects from alternative jet fuels. Masters thesis, Massachusetts Institute of Technology, 2010.

Sustainable Chemistry and Engineering, 4(5):2771–2779, 2016. ISSN 21680485. doi: 10.1021/acssuschemeng.6b00276.

UOP. UOP Renewable Jet Fuel Process. Technical report, received by email February 2017, 2013.

Ward, W. Personal communication, 2016.

Wong, H. M. Life-cycle assessment of greenhouse gas emissions from alternative jet fuels. Masters thesis, Massachusetts Institute of Technology, 2008.

125 Chapter 6

Conclusion

In Canada, growing emissions from air travel and the environmental commitments of the aviation industry have prompted discussion on the domestic production of canola- based biojet fuel via the rapidly developing HEFA pathway. LCAs have previously demonstrated that replacing petroleum-derived jet fuel with biojet fuel derived from renewable lipid feedstocks may allow the aviation industry to reduce their emissions. However, there is uncertainty as to whether or not biojet fuel from crop-based feedstocks will help or hinder the progress of the industry. Although oilseed crops like canola could help global biojet fuel production attain the level required to achieve substantial emissions reductions, land use change impacts are a risk to sustainability that research shows must be considered. Here, emissions reductions for canola biojet were calculated to be between 35% and 54% without land use change and between -51% and 10% with land use change, depending on co-product allocation method. Although determining the nature of the land use change that could occur as a result of biofuel production was outside the scope of this work, this research showed that when potential impacts are modeled and included in an LCA, the emissions intensity of canola biojet is greater than

126 that of petroleum-derived jet fuel for all allocation methods except mass-based allocation (M/E). An important take-away is that land use change limits the emissions reductions that can achieved by adopting biojet fuel. It is also important to emphasize that regardless of the life cycle emissions intensity, given the availability of land worldwide, biojet fuel from oilseeds could only replace a small portion of petroleum-derivied jet fuel. In Canada, the possible production level of biojet fuel from available canola, soybean and flaxseed was estimated to be 168-392 million litres (ML) (Transport Canada, 2015). That is based on an assumption that there is annual carry-over supplies of each crop available to buffer supply shortfalls and meet market demand (Transport Canada, 2015), although it was beyond the scope of this thesis to examine this assumption in detail. The use of 25-50% of domestically produced animal fat and UCO could add another 68-149 ML of biojet production capacity annually. Canada’s domestic and international jet fuel consumption was over 7,000 ML in 2015 (Transport Canada, 2016), so even at optimistic production levels, biojet fuel could only replace 8% of petroleum-derived jet fuel. This points to the need to plan for longer-term solutions to the problem of emissions from aviation, including the production of drop-in fuels from more widely available feedstocks. The annual supply of lignocellulosic biomass from roundwood, pel- lets, and forest and agricultural residues in Canada is over 100 million oven-dry tonnes (otd), which based on the fast-pyrolysis conversion efficiency for forest residues, could produce 20,100 to 24,000 ML of biojet fuel annually, well exceeding domestic demand (Transport Canada, 2015). Canadian-produced biojet fuel may be just one part of the solution to emissions

127 6.1. THESIS SUMMARY

from aviation that includes:

• Short-term deployment of HEFA biojet fuel from by-product lipids and oilseeds, with precautionary measures put in place to avoid incentivizing natural land conversion both domestically and globally (i.e. international certiﬁcation schemes like RSB)

• Medium-term deployment of second and third generation conversion pathways to produce biojet fuel from lignocellulosic and residential waste that does not compete for arable land

• Lon- term deployment of alternative technologies for aircraft propulsion, including hybrid ﬂight and full-electric ﬂight

6.1 Thesis summary

Here, the emissions intensity of canola-based biojet fuel produced in Western Canada was calculated across a range of assumptions, to address three important sources of variability in LCA: land use change inclusion, co-product allocation method, and input data. A range of possible GHG reductions that could result from the adoption of canola-based biojet fuel were obtained. This thesis informed the development of a domestic biojet supply chain based on canola by:

• Reporting the commercial status of HEFA production worldwide

• Reviewing the existing GHG-LCA literature on biojet production from oilseeds

• Applying a soil carbon model to estimate land use change emissions from canola production in Western Canada 128 6.1. THESIS SUMMARY

• Assessing the emissions intensity of HEFA biojet from canola in a Canadian context across diﬀerent methodological assumptions

Chapter 2 provided background information on the operational status, feedstock concerns, and chemistry of the HEFA biojet fuel production pathway. The HEFA pathway has been operational at the commercial scale for renewable diesel production since 2007 and since 2015 for biojet fuel production. There is increasing interest in HEFA biojet fuel production worldwide, as demonstrated by the announcements for intended production capacity in recent years. Although there are many different biojet production pathways under development, the HEFA pathway is likely the most ready for deployment in Canada. Chapter 3 reviewed twenty GHG-LCAs of HEFA fuels from oilseeds. Variability from feedstock, co-product allocation method, land use change inclusion, and refining technology assumptions was analyzed. Fuel from canola had a higher emissions intensity than fuel from other feedstocks when compared on an intra-study level. There was a wide range of co-product allocation methods applied in the literature. Displacement allocation was associated with lowest 5% of emissions intensities. Land use change assumptions were the greatest source of variability in the literature, with a large difference in calculated fuel emissions intensity between forest conversion and other conversion scenarios. Refining technology was not widely compared in the literature. In Chapter 4, land use change was estimated using a soil carbon model for grassland to canola cropland conversion on the Canadian Prairies over 500 years. The model was evaluated based on long-term field experiments on the Canadian Prairies.

129 6.1. THESIS SUMMARY

After a twenty-year period, 14.8 Mg C/ha were released from the decay of soil carbon. Results for grassland to cropland conversion were slightly lower than estimates from the literature, likely due to different assumptions about initial SOC and land management practices. The results of Chapter 4 were integrated with a GHG-LCA in Chapter 5. It was shown that the emissions intensity of canola-derived HEFA biojet was 46.4-66.3 g CO2eq/MJ without land use change and 80.5-153.1 g CO2eq/MJ with land use change, depending on the co-product allocation method applied. Emissions intensity was twice as sensitive to land use change inclusion under displacement allocation than under any other allocation method. The most important life cycle inputs in the canola biojet life cycle were natural gas and hydrogen demand from the refining stage and N2O emissions from feedstock production. Biojet fuel yield was shown to be one of the most significant life cycle parameters, given the sensitivity of results to variation in the biojet co-product allocation parameter.

6.1.1 Recommendations and future work

In combination with lipid by-product feedstocks, a HEFA production facility based on canola may be able to produce biojet fuel that oﬀers acceptable GHG reductions compared to petroleum-derived jet fuel. Potential GHG reductions could be increased by reducing energy consumption throughout the life cycle. Soil testing and other precision agriculture techniques may help reduce the emissions intensity of canola production by reducing fertilizer demand. Approximately 30% of canola farmers in Western Canada report using soil testing to determine fertilization rate (Smith, 2013). There is little uncultivated land left in the Canadian Prairies, which suggests

130 6.1. THESIS SUMMARY

that using canola as a feedstock for renewable fuels while maintaining canola exports will require agricultural intensiﬁcation through increased yields or increased cropping frequency. Future work should analyze a range of land use change scenarios, including the conversion of degraded or marginal land to produce biofuel feedstock. Co-product allocation method is one of the most important methodological de- cisions in LCA. Applying multiple co-product allocation methods in an LCA, as was done here, facilitates comparison between LCAs. However, when comparing alternative fuels for decision making purposes or for the granting of carbon credits, an allocation method should be speciﬁed as part of a standard LCA methodology. Energy-based allocation is recommended by the European RED and was shown here to be a more conservative allocation method than mass-based allocation. Neither method was compared to economic-based allocation in this study, which is used by the Roundtable on Sustainable Biomaterials (RSB) in their LCA methodology. Economic-based allocation has the advantage of being applicable to pathways that produce small quantities of highly valuable co-products, such as specialty chemicals. The United States Renewable Fuel Standards 2.0 (RFS2) use displacement allocation to compare fuel pathways, however displacement is not recommended here because it is highly sensitive to user assumptions. Because much of the value of LCAs comes from intra-study comparisons, this analyses should be expanded to include other feedstocks in its scope. Non-food oilseeds like camelina and carinata have been of particular interest to Canadian producers due to high seed oil content, lower water requirements and the possibility to be grown as a winter cover crop, thus reducing the risk of land use change (Blackshaw et al., 2011; Moser, 2010; Rakow and Getinet, 1998).

131 6.1. THESIS SUMMARY

Agusdinata et al. (2011) found that the possible emissions reductions from replacing petroleum-derived jet fuel with biojet fuel in the United States was in the range of 55% to 92% of 2005 levels by 2050. Possible emissions reductions were shown to depend on land availability and feedstock prices. A similar feedstock supply analysis, including lignocellulosic feedstocks and agent-based modeling, could help determine the potential emissions reductions that could be achieved in Canada’s aviation sector by 2050 (Agusdinata et al., 2011). Multi-criteria supply-chain optimization could be further applied to design and locate fuel production facilities based on the contraints of feedstock supply, transportation networks, and conversion pathway inputs, to simultaneously achieve economic and environmental targets (Cambero et al., 2016; You et al., 2012). In the future, a Monte-Carlo uncertainty analysis using available ranges for canola seed and biojet fuel yields and potential land use change emissions could be performed to statistically express variation in the potential emissions intensity of canola biojet fuel. Instead, variability was addressed by calculating emissions intensity across a range of methodological assumptions.

6.1.2 Contribution to engineering

LCA is a powerful quantitative tool for environmental decision making, especially when the life cycle impact of two or more products is being compared. This thesis assesses the emissions intensity of a canola-derived HEFA biojet fuel using novel data sources and methodology, including the application of a soil carbon model to estimate land use change emissions from grassland to cropland conversion on the Canadian Prairies. By analyzing the range of emissions intensities across eight land

132 BIBLIOGRAPHY

use change inclusion and co-production allocation scenarios, as well as the sensitivity of life cycle emissions to variation in the quantities of life cycle inputs and the size of co-production allocation parameters, a range of possible GHG reductions from canola-derived biojet fuel was determined. This research contributed to a report for the federally-funded CBSCI project that demonstrated the integration and use of biojet fuel use at a Canadian airport, and to two manuscripts that are being prepared for submission to the journals Biofuels, Bio- products, and Biorefining and Environmental Science and Technology (see Statement of Co-Authorship). Chapters 4 and 5 informed a methodologically novel LCA that is the first LCA of canola-derived biojet fuel in a Western Canadian context. Chapter 3 forms the basis for the only recent and comprehensive review of data sources, methodology, and results of LCA literature in the HEFA field. The transparent reporting of data sources used in this LCA and reported in Appendix A will facilitate further research in this field.

Bibliography

Agusdinata, D. B., Zhao, F., Ileleji, K., and DeLaurentis, D. Life Cycle Assessment of Potential Biojet Fuel Production in the United States. Environmental Science & Technology, 45(21):9133–9143, 2011. ISSN 0013-936X. doi: 10.1021/es202148g.

133 BIBLIOGRAPHY

Cambero, C., Sowlati, T., and Pavel, M. Economic and life cycle environmental optimization of forest-based bioreﬁnery supply chains for bioenergy and biofuel production. Chemical Engineering Research and Design, 107:218–235, 2016. ISSN 02638762. doi: 10.1016/j.cherd.2015.10.040. URL http://dx.doi.org/10.1016/j. cherd.2015.10.040.

Moser, B. R. Camelina (Camelina sativa L.) oil as a biofuels feedstock: Golden opportunity or false hope? Lipid Technology, 22(12):270–273, 2010. ISSN 0956666X. doi: 10.1002/lite.201000068.

Rakow, G. and Getinet, A. Brassica Carinata an oilseed crop for Canada. ISHS Acta Horticulturae, 459:419–428, 1998. doi: 10.17660/ActaHortic.1998.459.50.

Smith, E. G. Presentation- Survey of Canola Growers: Practices, Inputs and Pro- duction. In Soils and Crops 2013, Saskatoon, SK, 2013. Agriculture and Agri-food Canada. URL http://www.usask.ca/soilsncrops/conference-proceedings/2013pdf/ Day{\ }2{\ }Presentations/001-Elwin{\ }Smith.pdf.

Transport Canada. Canada’s Action Plan to Reduce Greenhouse Gas Emissions from Aviation: 2015 Annual Report. Technical report, Government of Canada, Ottawa, ON, 2016.

You, F., Graziano, D. J., and Snyder, S. W. Optimal Design of Sustainable Cellu- losic Biofuel Supply Chains : Multiobjective Optimization Coupled with Life Cycle

134 BIBLIOGRAPHY

Assessment and Input Output Analysis. AIChE Journal, 58(4):1157–1180, 2012. ISSN 14746670. doi: 10.1002/aic.

135 Appendix A

Table of literature reviewed

136 Table A.1: GHG-LCAs of HEFA fuels from oilseeds

1 Author Feedstock Location Technology LUC Allocation N2O EF Results(g Comments 2 3 (%) CO2eq/MJ) HEFA jet

Agusdinata camelina MT USA UOP no D/D 1.325,0 60 et al. (2011) de Jong et al. camelina MT USA UOP no D/E 1.325,1.225 44 (2017) camelina MT USA UOP no E/E 1.325,1.225 47 camelina MT USA UOP no M/E 1.325,1.225 37 camelina MT USA UOP no $/E 1.325,1.225 47

137 Fan et al. pennycress USA UOP no $/$ 1.325,1.225 44.9 (2013) pennycress USA UOP no D/D 1.325,1.225 -18.3 pennycress USA UOP no E/E 1.325,1.225 32.7 Han et al. rapeseed EUR UOP no E/E 1.325,1.225 54 (2013) camelina MT USA UOP no E/E 1.325,1.225 47 soybean EUR UOP no M/E 1.325,1.225 30 soybean EUR UOP no E/D 1.325,1.225 31 soybean EUR UOP no E/E 1.325,1.225 39.5 soybean EUR UOP no E/M 1.325,1.225 39.5 Continued on next page

1Land use change inclusion 2Oil extraction allocation method/reﬁning allocation method 3Emission factors for nitrogen from fertilizer, nitrogen from crop residues; IPCC tier 1 default 1.325/1.225 Table A.1: Continued from last page

Author Feedstock Location Technology LUC Allocation N2O EF Results(g Comments CO2eq/MJ) soybean EUR UOP no $/E 1.325,1.225 41 soybean EUR UOP no E/$ 1.325,1.225 41.5 soybean EUR UOP no D/E 1.325,1.225 43 Li and camelina SK CAN UOP no D/D 0.91,0.91 3 high yield, Mupondwa low N, UAN4, (2014) refining 1 camelina SK CAN UOP no D/D 0.91,0.91 7 high yield, low N, UAN, refining 2 camelina SK CAN UOP no D/D 0.91,0.91 7.5 high yield, 138 low N, urea, refining 1 camelina SK CAN UOP no D/D 0.91,0.91 8.5 high yield, low N, urea, refining 2 camelina SK CAN UOP no D/D 0.91,0.91 27 low yield, high N, UAN, refining 1 camelina SK CAN UOP no D/D 0.91,0.91 28.5 low yield, high N, UAN, refining 2 camelina SK CAN UOP no D/D 0.91,0.91 30.5 low yield, high N, urea, refining 1 Continued on next page

4urea ammonium nitrate Table A.1: Continued from last page

Author Feedstock Location Technology LUC Allocation N2O EF Results(g Comments CO2eq/MJ) camelina SK CAN UOP no D/D 0.91,0.91 31 low yield, high N, urea, reﬁn- ing 2 Lokesh et al. camelina MT USA UOP no E/E 1.325,0 31.4 (2015) Shonnard camelina MT USA UOP no E/E 1.325,1.325 22.36 et al. (2010) camelina MT USA UOP no M/M 20.18 camelina MT USA UOP no D/D -17.03 Sieverding canola ND/SD UOP no E/E 1.325,1.225 44 ±7.7 Monte Carlo 139 et al. (2016) USA analysis using triangular input distribu- tions camelina ND/SD UOP no E/E 1.325,1.225 30.8 ±4.5 USA carinata ND/SD UOP no E/E 1.325,1.225 34.5 ±4.6 USA sunﬂower ND/SD UOP no E/E 1.325,1.225 37.9 ±4.8 USA Stratton rapeseed EUR UOP no $/E 1.325,1.225 54.9 et al. (2010) set aside $/E 1.325,1.225 97.9 LUC 30 years grassland (Fargione et al., 2008) Continued on next page Table A.1: Continued from last page

Author Feedstock Location Technology LUC Allocation N2O EF Results(g Comments CO2eq/MJ) soybean USA UOP no $/E 1.325,1.225 37 soybean USA UOP grassland $/E 1.325,1.225 97.8 LUC 30 years (Fargione et al., 2008) soybean USA UOP rainforest $/E 1.325,1.225 564.2 LUC 30 years (Fargione et al., 2008) Ukaew et al. canola ND USA UOP no D/D 1.74,1.74 43 Highest re- (2014) gional N2O EF, RSB

140 methods canola ND USA UOP no D/D 1.71/1.71 42.7 Lowest regional N2O EF, RSB methods canola ND USA UOP no D/D 45.9 1.325/1.225 Ukaew et al. canola ND USA UOP cropland D/D 1.2/1.2 -12 Low seed (2016) price, LUC 30 years canola ND USA UOP cropland D/D 1.2/1.2 21 High seed price canola ND USA UOP cropland $/$ 1.2/1.2 33 Low seed price Continued on next page Table A.1: Continued from last page

Author Feedstock Location Technology LUC Allocation N2O EF Results(g Comments CO2eq/MJ) canola ND USA UOP cropland $/$ 1.2/1.2 44 High seed price canola ND USA UOP cropland E/E 1.2/1.2 36 Low seed price canola ND USA UOP cropland E/E 1.2/1.2 49 High seed price Wong (2008) soybean USA UOP no M/E 1.325/- 21 1.325 soybean USA UOP no E/E 37.5 soybean USA UOP no $/E 35.2 141 soybean USA UOP no D/E 37.3 soybean USA UOP global avg. M/E 1.325/0 131.5 LUC 30 years (Searchinger et al., 2008) soybean USA UOP global avg. E/E 314.4 soybean USA UOP global avg. $/E 289.0 soybean USA UOP global avg. D/E 38.5 HEFA diesel Arvidsson rapeseed EUR Neste no D/D 1.25,1.25 67.3 Baseline yield, et al. (2011) 1.045 t/ha rapeseed EUR Neste no D/D 1.25,1.25 29.6 High yield, 3.24 t/ha rapeseed EUR Neste no D/D 3.75,3.75 87.5 3x N2O EF Continued on next page Table A.1: Continued from last page

Author Feedstock Location Technology LUC Allocation N2O EF Results(g Comments CO2eq/MJ) Edwards rapeseed EUR Neste no D/D modelled5 57.1 et al. (2014) rapeseed EUR UOP no D/D modelled 57.7 soybean EUR Neste no D/D modelled 55.8 sunﬂower EUR Neste no D/D modelled 45.4 Fan et al. pennycress USA UOP no $/$ 1.325,1.225 40.7 (2013) pennycress USA UOP no D/D 1.325,1.225 13.1 pennycress USA UOP no E/E 1.325,1.225 30.1 Huo et al. soybean USA UOP no D/D 1.325,1.325 34.12 142 (2009) soybean USA UOP no E/E 1.325,1.325 24.65 soybean USA UOP no $/$ 1.325,1.325 24.65 soybean USA UOP no D/E 1.325,1.325 40.78 soybean USA UOP no E/$ 1.325,1.325 34 soybean USA NRC no D/D 1.325,1.325 -28.44 soybean USA NRC no E/E 1.325,1.325 32.23 soybean USA NRC no $/$ 1.325,1.325 36.02 soybean USA NRC no D/E 1.325,1.325 33.2 soybean USA NRC no E/$ 1.325,1.325 30 Continued on next page

5GNOC calculator Table A.1: Continued from last page

Author Feedstock Location Technology LUC Allocation N2O EF Results(g Comments CO2eq/MJ) Kalnes et al. soybean USA UOP no M/M 0,0 12.7 Production (2007) assumptions (Sheehan et al., 1998), H2 from renewable fuels soybean USA UOP no M/M 0,0 14 Production assumptions (Sheehan et al., 1998),

143 H2 from petroleum Kalnes et al. soybean USA UOP no D/D included 41 Production (2009) assumptions (Edwards, 2004) soybean USA UOP no M/M 0,0 17.3 Production assumptions (Sheehan et al., 1998) soybean USA UOP no M/M included 48.8 Production assumptions (Hill et al., 2006) Continued on next page Table A.1: Continued from last page

Author Feedstock Location Technology LUC Allocation N2O EF Results(g Comments CO2eq/MJ) Miller and canola AB CAN UOP no M/M 0.76,0.76 48 Baseline Kumar (2013) canola AB CAN UOP no M/M 0.76,0.76 44 Maximum yield canola AB CAN UOP no M/M 0.76,0.76 55 Minimum yield canola AB CAN UOP no M/M/M 0.76,0.76 33 Also straw co-product canola AB CAN UOP no M/M 1,1 50 144 canola AB CAN UOP no M/M 2,2 58 canola AB CAN UOP no M/M 3,3 67 canola AB CAN UOP no M/M 4,4 75 canola AB CAN UOP no M/M 5,5 84 canola AB CAN UOP no $/$ 0.76,0.76 71 canola AB CAN UOP grassland M/M 0.76,0.76 94 LUC 20 years camelina AB CAN UOP no M/M 0.76,0.76 38 Baseline, no LUC for camelina camelina AB CAN UOP no M/M 0.76,0.76 33 Maximum yield camelina AB CAN UOP no M/M 0.76,0.76 43 Minimum yield camelina AB CAN UOP no M/M 0.76,0.76 30 Also straw co-product Continued on next page Table A.1: Continued from last page

Author Feedstock Location Technology LUC Allocation N2O EF Results(g Comments CO2eq/MJ) camelina AB CAN UOP no M/M 1,1 39 camelina AB CAN UOP no M/M 2,2 44 camelina AB CAN UOP no M/M 3,3 49 camelina AB CAN UOP no M/M 4,4 53 camelina AB CAN UOP no M/M 5,5 58 camelina AB CAN UOP no $/$ 0.76,0.76 57 Nikander rapeseed EUR Neste no D/D 1.325,1.325 59.3 (2008) rapeseed EUR Neste no M/M 1.325,1.325 33.4 Shonnard camelina MT USA UOP no E/E 1.325,1.325 18.04 Baseline, 40% 145 et al. (2010) H2 from SMR, N from urea6 camelina MT USA UOP no M/M 1.325,1.325 15.8 Baseline camelina MT USA UOP no D/D 1.325,1.325 4.2 Baseline camelina MT USA UOP no E/E 1.325,1.325 19.0 2x farm diesel camelina MT USA UOP no E/E 2.65,2.65 23.1 2x N2O EF camelina MT USA UOP no E/E 1.325,1.325 23.4 100% hydrogen from SMR camelina MT USA UOP no E/E 1.325,1.325 16.2 N from liquid ammonia camelina MT USA UOP no E/E 1.325,1.325 21.0 N from ammonium nitrate Continued on next page

6steam methane reforming Table A.1: Continued from last page

Author Feedstock Location Technology LUC Allocation N2O EF Results(g Comments CO2eq/MJ) camelina MT USA UOP no E/E 1.325,1.325 27.0 2x N fertilizer Uusitalo rapeseed EUR Neste no D/D included 36.9 et al. (2014) rapeseed EUR Neste no no/E included 52.8 rapeseed EUR Neste no D/D included 30.2 rapeseed EUR Neste grassland E/E included 46.9 LUC 20 years rapeseed EUR Neste grassland no/E included 49.4 rapeseed EUR Neste grassland D/D included 70.4 rapeseed EUR Neste forest E/E included 257.3 LUC 20 years rapeseed EUR Neste forest no/E included 419.8 146 rapeseed EUR Neste forest D/D included 433.2 BIBLIOGRAPHY

Bibliography

Arvidsson, R., Persson, S., Fr¨oling,M., and Svanstr¨om,M. Life cycle assessment of hydrotreated vegetable oil from rape, oil palm and Jatropha. Journal of Cleaner Production, 19(2-3):129–137, 2011. ISSN 09596526. doi: 10.1016/j.jclepro.2010.02. 008.

de Jong, S. D., Antonissen, K., Hoefnagels, R., Lonza, L., Wang, M., Faaij, A., and Junginger, M. Lifecycle analysis of greenhouse gas emissions from renewable jet fuel production. Biotechnology for Biofuels, pages 1–18, 2017. ISSN 1754-6834. doi: 10.1186/s13068-017-0739-7.

Edwards, R. Well-to-wheels analysis of alternative fuels and powertrains in the Euro- pean context v 1.0. Technical report, European Commission Joint Research Centre, 2004.

147 BIBLIOGRAPHY

Fargione, J., Hill, J., Tilman, D., Polasky, S., and Hawthorne, P. Land Clearing and the Biofuel Carbon Debt. Science, 319(1235):1235–1237, 2008. ISSN 15829596. doi: 10.1126/science.1152747.

Han, J., Elgowainy, A., Cai, H., and Wang, M. Q. Life-cycle analysis of bio-based aviation fuels. Bioresource Technology, 150:447–456, 2013. ISSN 09608524. doi: 10.1016/j.biortech.2013.07.153.

Hill, J., Nelson, E., Tilman, D., Polasky, S., Tiffany, D., Hill, J., Nelsont, E., Tilman, D., Polasky, S., and Tiffanyt, D. Environmental , economic , and energetic costs and benefits of biodiesel and ethanol biofuels. Proceedings of the National Academy of Sciences of the United States of America, 103(30):11206–11210, 2006. doi: 10. 1073/pnas.0604600103.

Kalnes, T., Marker, T., and Shonnard, D. R. Green Diesel : A Second Generation Biofuel. International Journal of Chemical Reactor Engineering, 5, 2007.

Li, X. and Mupondwa, E. Life cycle assessment of camelina oil derived biodiesel and

148 BIBLIOGRAPHY

jet fuel in the Canadian Prairies. Science of the Total Environment, 481(1):17–26, 2014. ISSN 00489697. doi: 10.1016/j.scitotenv.2014.02.003.

Nikander, S. Greenhouse gas and energy intensity of product chain: case transport biofuel. Masters thesis, Helsinki University of Technology, 2008.

Sheehan, J., Camobreco, V., Duﬃeld, J., Graboski, M., and Shapouri, H. Life Cycle Inventory of Biodiesel and Petroleum Diesel for Use in an Urban Bus A Joint Study Sponsored by. Technical report, United States Department of Energy and United States Department of Agriculture, 1998.

149 BIBLIOGRAPHY

Ukaew, S., Beck, E., Meki, M. N., and Shonnard, D. R. Application of the roundtable on sustainable biofuels method to regional diﬀerences in nitrous oxide emissions for the rapeseed hydrotreated renewable jet life cycle. Journal of Cleaner Production, 83(2014):220–227, 2014. ISSN 09596526. doi: 10.1016/j.jclepro.2014.06.004.

Wong, H. M. Life-cycle assessment of greenhouse gas emissions from alternative jet fuels. Masters thesis, Massachusetts Institute of Technology, 2008.

150 Appendix B

Soil carbon model

B.1 Decay of SOC pools

This section describes the soil carbon model in greater detail. All equations are taken from the RothC user manual (Coleman and Jenkinson, 2014). The model calculates the size of each SOC pool at monthly time steps. The size of a pool at month t is calculated as follows, where a, b, and c are rate-modifying factors (described below) and k is a pool-speciﬁc decay constant (Table 4.2).

SOC(t) = SOC(t − 1)e(−a∗b∗c∗k)/12 (B.1)

Incoming carbon is added to the plant matter pools (DPM and RPM) according to the DPM/RPM ratio, which is 0.67 for unimproved grassland, and 1.44 for cropland.

The carbon that decays from each pool every month is split between CO2 and the microbial (BIO) and humus (HUM) pools according to a proportion, biohum. biohum calculated using a parameter, x, that changes with soil clay content (clay).

x = 1.67 ∗ (1.85 + 1.60−0.0786∗clay) (B.2)

151 B.1. DECAY OF SOC POOLS

Table B.1: Monthly parameters in SOC model

Month Mean air Precipita- Potential Ac- Soil cover Plant in- temper- tion (mm) evapotran- cumTSMD (V) put (Mg ature, T spiration, C) (C) PET Jan -10.1 16.6 0 0 0.6 0 Feb -8.7 11.8 5 6.80 0.6 0 Mar -2.5 19.4 39 -12.80 0.6 0 Apr 5.3 19.1 110 -60.83 0.6 0 May 10.9 51.2 173 -60.83 0.6 0.3345 Jun 15.3 77.1 194 -60.83 0.6 0.3345 Jul 18.2 60.1 238 -60.83 0.6 0.3345 Aug 17.6 47.4 206 -60.83 0.6 0.3345 Sep 11.5 36.0 118 -60.83 0.6 0.3345 Oct 4.9 18.9 50 -60.83 0.6 0.3345 Nov -3.9 15.8 9 -54.03 0.6 0 Dec -9.3 19.1 0 -54.03 0.6 0

biohum = 1/(x + 1) (B.3)

The biohum proportion of carbon is split between BIO and HUM according to a ﬁxed BIO/HUM ratio of 0.85.

B.1.1 Rate-modifying factors

Monthly model parameters are listed in Table B.1 and are used at each monthly time step to calculate the decay rate-modifying factors a, b, and c (Equations B.1.1,B.1.1,B.1.1). The data sources for these parameters are detailed in Chapter 4. Accumlated top soil moisture deﬁcit (AccumTSMD) is calculated using PET, precipitation, and MaxTSMD. AccumTSMD is zero unless PET exceeds precipitation for a given month. Then, AccumTSMD is the diﬀerence between PET and precipitation. AccumTSMD cannot exceed MaxTSMD, which is calculated as follows for each

152 B.1. DECAY OF SOC POOLS

month:

20.0 + 1.3(clay) − 0.01(clay2) − ∗ d = MaxT SMD (B.4) 23

Where MaxTSMD is the maximum top soil moisture deﬁcit for each month, calculated based on the percentage by weight of clay in the soil (clay), which is 18% for Swift Current, Saskatchewan (Chapter 4). d is the soil depth in cm, which is 30.z The rate modifying constants a, b, and c are calculated each month according to:

Calculation of a

a(t) = 47.91/(1 + e106.06/(T (t)+18.27)) (B.5)

Calculation of b

if AccumTSMD(t)<0.444*MaxTSMD b(t) = 1; else MaxT SMD − AccumT SMD(t) b(t) = 0.2+(1-0.2)* (B.6) MaxT SMD − 0.444 ∗ MaxT SMD

Calculation of c

c = V (B.7)

153 BIBLIOGRAPHY

Where V is a monthly soil cover constant that is 0.6 if the soil is vegetated and 1 if bare. Monthly soil cover can be seen in Table B.1.

Bibliography

Coleman, K. and Jenkinson, D. S. RothC - A model for the turnover of carbon in soil- model description and users guide (Windows version). Technical report, Rothamsted Research, 2014.

154 Appendix C

Data sources and allocation methodology

C.1 Data sources for life cycle inputs

Input data for every life cycle stage was collected and compiled in an Excel spread- sheet, from which the most representative values were used to built processes in SimaPro v8.0.2.13 LCA software. To avoid data-driven errors, a range of values for each input were collected. By amassing values from many sources, the risk of rely- ing on any one potentially ﬂawed dataset was avoided. Values were selected for the emissions model that represented the chosen production system as accurately as possible based on a combination of technical, temporal, and geographic relevance. These values came from primary data sources, such as industry experts, and secondary data sources, such as publications and LCA databases. Where necessary, inputs were normalized for comparison using reported yields or conversion factors. Where Canadian data was not speciﬁcally used, Ecoinvent processes were used to represent inputs (see Chapter 5).

155 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

C.1.1 Canola farming

Data sources

Data on farming inputs were sourced from a 2011 survey of nearly one thousand canola growers in Western Canada. The survey was performed by Agriculture and Agri-food Canada (AAFC) in collaboration with the Canola Council of Canada (CCC) (Smith and Barbieri, 2012). The 2011 AAFC survey was used in a 2015 update to the canola biofuel pathway in the Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation (GREET) Excel-based LCA model (Cai et al., 2015). Previous versions of GREET used European canola farming data from Stratton et al. (2010). Canadian data was used by GREET because the United States imports much more canola oil from Canada than rapeseed from Europe (Cai et al., 2015). The use of this data in the GREET model underlines its reliability, as the GREET model is widely used in biofuel LCA. Six out of twenty studies reviewed in Chapter 3 used the GREET model for LCA or default input values from the model. A summary of the AAFC survey was obtained directly from the authors. The CCC report cited by Cai et al. (2015) was not available online. Cai et al. (2015) was also referenced for some data which was not available in the survey summary, such as production shares across agricultural region (RUs), and manure and pesticide application rates, which were calculated using average equipment passes.

Fertilizer

The average application rates of the four most widely applied fertilizers, nitrogen, potassium, phosphorus, and sulfur, were reported for each RU in pounds per acre

156 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

(Smith and Barbieri, 2012). A weighted average for each nutrient was calculated, taking into account the proportion of farms that did not apply any. Average fertilizer application rate in kilograms per tonne canola seed was calculated using the average yields for each RU. Using the production share of each RU to total canola production, the average application rate for Western Canada was calculated for each fertilizer. Manure was applied by less than 5% of farms (Smith and Barbieri, 2012) and was included in the average nitrogen fertilizer application rate. Urea and anhydrous ammonia (liquid ammonia in SimaPro) were the most commonly applied forms of nitrogen fertilizer, followed by ammonium nitrate, ammonium sulfuate, and manure (Cai et al., 2015). The average Western Canadian nitrogen fertilizer application mix was calculated using production-weighted reporting for each

RU, as above. Phosphorus was assumed to be from phosphate (P2O5) and potassium was assumed to be from potassium chloride (KCl, from potash). Sulfur was reported as roughly 50:50 elemental sulfur and ammonium sulfate. The amount of ammonium sulfate applied was calculated as the sulfur required after ammonium sulfate for nitro-

gen was applied. N2O emissions were associated with nitrogen fertilizer application as well as with ammonium sulfate application.

Pesticides

Pesticides applied by canola producers include herbicides, fungicides, and some insecticide. Pesticide types were obtained from Smith and Barbieri (2012) and average application rates were obtained from Cai et al. (2015). Herbicides applied to canola crops in Western Canada are mainly glyphosphate, Liberty R (13.5% glufosinate ammonium), Select R /Centrion R (26% clethodim), and Odyssey R (35% imazamox/35%

157 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

imazethapyr). Fungicide, usually Proline R (prothioconazole) was applied by 40% of producers. Insecticide was applied by few producers but was either the organophos- phate Lorsban R /Pyrinex R (chlorpyritos) or the pyrethroids Decis R (deltamethrin) or Matador R /Silencer R (lambda-cyholothrin). For simplicity, and because eco-toxicity was not within the scope of this study, herbicides were assumed to be glyphosphate and insecticide was assumed to be organophosphorus. No fungicide application was modeled. It should be noted that although many pesticides contain a petroleum- based surfactant, only the manufacturing emissions for the active ingredient were accounted for. Herbicide and insecticide application made up less than 1% of total life cycle emissions.

Fuel use

Diesel, natural gas, and electricity usage was obtained from Cai et al. (2015). Farming equipment use but not fuel consumption was reported by Smith and Barbieri (2012). The typical fuel usage of farming equipment was used to calculate fuel usage by Cai et al. (2015). Field operations that consumed fuel (diesel) included seeding, pesticide and fertilizer application, and harvesting (Smith and Barbieri, 2012). Little canola is dried post-harvest in the Prairies (Ward, 2016), so additional energy usage reported by canola producers was likely for storage. Electricity use for canola drying and storage has been reported in other LCAs (Edwards et al., 2014; Miller and Kumar, 2013; Rustandi and Wu, 2010; Uusitalo et al., 2014). Processes were created in SimaPro for Canadian fuels, as there were no Canadian fuels available in the Ecoinvent database. GHGenius 4.03 was used to calculate the emissions intensity of natural gas and diesel extracted and reﬁned in Western Canada.

158 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

Table C.1: Properties of Western Canadian petroleum fuel inputs and jet fuel baseline ((S&T)2 Consultants Inc., 2013)

Fuel EI HHV HHV LHV LHV Density g CO2eq/MJ MJ/L MJ/kg MJ/L MJ/kg kg/L Jet, 15 ppm S, WTWa 101.5 37.40 46.27 34.79 43.04 0.81 Jet, 600 ppm S, WTWa 93.8 37.40 46.27 34.79 43.04 0.81 Diesel, 20 ppm S, WTW (tractor) 105.6 41.36 42.83 38.65 40.03 0.84 Natural gas, WTW (boiler) 67.9 0.04 52.48 0.03 46.53 0.000722 Hydrogen, from natural gas 105.1 12.57 141.24 10.68 120.04 0.09

GHGenius is a Excel-based LCA model developed by the Canadian company (S&T)2 Consultants that calculates the emissions intensity of transportation fuels. Much like GREET, calculations can be customized for region and year. The base year selected was 2017 and the region selected was Western Canada. The emissions intensity for diesel combusted in a tractor and natural gas combusted in an industrial boiler were taken from the Equipment Emission Factors sheet and converted from high heating value (HHV) to low heating value (LHV). Land use change emissions were not included for any fuels. The properties and emissions intensities (EI) of all the fuel inputs used in the canola biojet life cycle, as well as petroleum-dervied jet fuel for comparison purposes, are shown in Table C.1. The HHVs and LHVs of the fuels used are from the Fuel Char sheet of GHGenius v.4.03.

N2O emissions

Nitrous oxide (N2O) is released as microorganisms convert nitrogen in the soil into

- usable forms like nitrate (NO3 ) through the processes of nitriﬁcation and denitriﬁca- tion. These processes release nitrous oxide, a greenhouse gas 298 times as potent as

CO2. Here, N2O emissions are estimated from nitrogen application rates using EFs

159 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

that represent the percentage of applied nitrogen that is emitted as N2O. The EF of 1.06% is used for nitrogen in synthetic fertilizers and the EF of 0.96% is used for nitrogen in manure or crop residues (Cai et al., 2015). Crop residues from one tonne of canola seed were assumed to contribute 22.1 kilograms of nitrogen to the soil (Cai et al., 2015). Canola crop residue is produced in a 4:1 ratio with seed under rainfed conditions (Gan et al., 2009), which represents most of the canola growing conditions on the Prairies (Goodwin, 2006). These EFs were obtained from Cai et al. (2015), who cited the CCC report. The EFs were calculated by the CCC using IPCC Tier 2 methodology (De Klein et al., 2006) and are speciﬁc to the canola growing region of the Canadian Prairies (Cai et al., 2015). They are lower than the default IPCC Tier 1 values of 1.325 and 1.225 which are used in most biojet life cycle assessments (see Appendix A). Other

LCAs have calculated site-speciﬁc N2O EFs that are also lower than the default IPCC values (Miller and Kumar, 2013; (S&T)2 Consultants Inc., 2010; Ukaew et al., 2014).

Although there is still high uncertainty associated with estimating N2O emissions, site-speciﬁc EFs take into account crop management and soil parameters, which are important in explaining ﬁeld N2O measurements (Stehfest and Bouwman, 2006).

C.1.2 Oil extraction and upgrading

Energy use in canola oil extraction was obtained from a 2010 LCA of canola biodiesel prepared for the CCC ((S&T)2 Consultants Inc., 2010), that referenced a 2010 Cana- dian Oilseed Producers Association (COPA) survey. An oil extraction yield of 1 tonne oil per 2.25 tonnes seed was obtained from a 2016 COPA presentation (extrapolation to 2017) (Canadian Oilseed Processors Association, 2016). Canola oil has a density

160 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

of 0.1955 kg/L (Przybylski, 2011) and an average seed oil content of 44% (Ukaew et al., 2016). Using a mass balance approach and assuming no processing losses, 1.25 tonnes of meal are produced for every tonne of canola oil. Yields reported are for crude canola oil. Crude canola oil can contain up to 2.5% phospholipids, which are removed by degumming when canola oil is upgraded to a quality suitable for hydroprocessing. The oil extraction seed yield was adjusted to one tonne oil per 2.31 tonnes seed to account for upgrading losses. Chemical inputs in canola oil extraction and upgrading included hexane as a solvent and sodium hydroxide and phosphoric acid for degumming. Quantities of chemical inputs in kilograms per tonne canola are taken from Rustandi and Wu for an average canola oil extraction plant (Rustandi and Wu, 2010). The density of hexane is 0.655 kg/L and the density of phosphoric acid is 1.685 kg/L.

C.1.3 Reﬁning

Honeywell UOP was contacted for information on the HEFA biojet fuel production process ‘Ecofining’ (UOP, 2013). No feedstock specific information was attainable, but a fact sheet was provided by email describing typical yields and process inputs given as a range (UOP, 2013). The average of the upper and lower values was used. All data was for a process configuration that maximizes biojet fuel yield, although HEFA diesel, propane, and naphtha are produced as co-products. Where necessary, a feedstock oil density of 0.9155 kg/L Przybylski (2011) was used. Electricity and natural gas consumption per tonne of biojet fuel were obtained from Li and Mupondwa (2014), using the average inputs from two processes that were compared. The original source of the data was Miller and Kumar (2013); Pearlson

161 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

et al. (2013); Stratton (2010), who based consumption on estimates from process models, including work by Huo et al. (2009), and on experimental data from Marker (2005), in combination with expert consultation. The emissions intensity of hydrogen was calculated from the life cycle results for compressed hydrogen gas fuel from natural gas and is shown in Table C.1. Emissions from distribution, dispensing, and storage were subtracted as hydrogen is assumed to be produced on site.

C.1.4 Transportation and distribution

Average transportation distances for canola seed, canola oil, and biojet fuel were taken from GHGenius 4.03 ((S&T)2 Consultants Inc., 2013). Distances are based on the assumption that oil extraction occurs locally and reﬁning is more centralized. The average transportation distances were 100 km from farm to oil extraction plant and 200 km from oil extraction plant to HEFA reﬁnery, both by transport truck. Since there is currently no commercial scale hydroprocessing facility in Canada, the distance for canola oil transport has high uncertainty. The average transportation distance for biojet fuel to an airport was 225 km by truck and 225.26 km by rail. All fuel consumed was assumed to be diesel. The emissions intensity of transporting 1 tonne of material 1 kilometer (1 tkm) with a 16-32 tonne truck was calculated as

138.37 g CO2eq in SimaPro, not including emissions from infrastructure, like road construction. Transportation from farm to oil extraction plant are included in the oil extraction and upgrading stage, and transportation from the oil extraction plant to HEFA reﬁnery are included in the reﬁning stage. The transportation of biojet was analyzed as a separate distribution stage.

162 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

C.1.5 Combustion

Emissions from the combustion of biojet fuel were taken from Elgowainy et al. (2012) for non-CO2 GHG emissions. Non-CO2 GHG emissions are any gases besides CO2 that have a radiative forcing effect as defined by the IPCC 2007 GWP method: CH4 and N2O (Forster et al., 2007). Particulate matter, water vapour, NOx, and SOx are not included due to high uncertainty in their GWP. CO2 emissions are biogenic, so they were treated as carbon-neutral in the canola biojet life cycle. Tests have shown that renewable jet fuel generally releases fewer non-CO2 emissions than conventional jet fuel, with the exception of water vapor (Braun-Unkhoff et al., 2017; Stratton et al., 2011).

C.1.6 Displaced soymeal

Inputs and soybean yield for soybean farming were taken from the GHGenius 4.03 user manual for 2010 ((S&T)2 Consultants Inc., 2013). The fertilizer mix, pesticide mix, nitrogen content of crop residues, and N2O emission factor were assumed to be the same for soybean and canola. The same per-hectare land use change emissions were also assumed, as soybeans are oilseeds with a biomass yield similar to canola (Fan et al., 2017). However, the quantity of nitrogen fertilizer applied is much lower because soybeans are legumes. Legumes convert atmospheric nitrogen to usable nitrate through symbiosis with Rhizobacteria in their root systems. Energy use for soybean oil extraction and upgrading was also taken from the GH- Genius user manual, using data from a 2009 National Oilseed Producers Association (NOPA) survey. Given an 18% oil content ((S&T)2 Consultants Inc., 2013) and an oil density of 0.9193 kg/L (Przybylski, 2011), oil extraction yield was assumed to be

163 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

1 tonne oil per 5.56 tonnes soybean. Using a mass balance approach, the amount of meal produced is 4.56 tonnes per tonne oil, or a 0.81 ratio of meal:soybeans. Like canola, crude soybean oil contains phospholipids which must be removed before further processing. Degummed oil yield was 5.78 tonnes per tonne oil to account for up to 4% phospholipid content (Przybylski, 2011). Chemical inputs for soybean oil extraction were also taken from Rustandi and Wu (Rustandi and Wu, 2010).

Displacement method

The application of the displacement method for co-product allocation in a LCA depends on the co-product. For co-products that are displaced by a product whose production is not linked to that of another product, the emissions intensity of the displaced product can simply be subtracted from the life cycle of the main product. This is the case for the renewable fuels produced in the refining stage of the HEFA biojet life cycle. Because they can be produced from many different processes in a petroleum refinery, the displacement of these fuels by their renewable counterparts does not strictly impact the supply of any another petroleum product. Complexity arises when avoided production of the displaced product affects the supply of another product. This is the case for oilseed crushing, where oil cannnot be produced without meal, and vice-versa. Soy oil extraction and canola oil extraction produce the same products, albeit in different ratios. Displacing soybean meal with the canola meal produced in the oil extraction stage also displaces soybean oil. Following the method of Weidema (Weidema, 1999), a type of displacement allocation called system expansion can be performed to calculate the production balance

164 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

Table C.2: Parameters of soybean and canola oil extraction and their co-product relationships

Parameter Canola Soybean Reference Protein content 36% 48% (S&T)2 Consultants Inc. (2010) Oil content 44% 18% (S&T)2 Consultants Inc. (2013); Ukaew et al. (2016) Density (kg/L) 0.9155 0.9193 Przybylski (2011) Phospholipids 2.5% 4.0% Przybylski (2011) Seed:crude oil 2.25:1 5.56:1 Canadian Oilseed Processors Association (2016); (S&T)2 Consultants Inc. (2013) Seed:degummed oil 2.31:1 5.78:1 Meal: degummed oil 1.28:1 4.74:1 Protein ratio 1.33:1 0.75:1

of processes with a complex co-product relationship, like soybean and canola oil extraction. For simplicity, the market for vegetable oil is assumed to consist of only two products: soybean oil and canola oil, and the density of the two oils is assumed to be the same. Likewise, the market for vegetable protein consists of only soybean meal and canola meal, which are perfect substitutes for each other on a protein-equivalent basis. The important relationships between the four products are summarized in Equation C.1. The relationship between soybean oil and canola oil can be calculated from the ratio of meal to oil for canola, the meal to oil ratio for soybean, and the protein ratio of canola meal to soybean meal (Table C.2).

1.28/1.33 = 0.203 (C.1) 4.74

Including losses to impurities, 1.28 t of canola meal are produced for every one tonne degummed canola oil. 0.962 t soybean meal and 0.203 t of soybean oil are displaced by the canola meal. In this simpliﬁed market, the displaced soybean oil

165 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

is replaced by canola oil. To maintain the supply of vegetable oil in the market, an additional 0.203 t of canola oil is produced. This co-produces an additional 0.260 t of canola meal, which displaces 0.195 t of soybean meal and 0.412 (0.2032) t soybean oil. This process can be continued for inﬁnite displacement cycles, but here an unchanging 3 digit ﬁgure, 0.255, that represents the total tonnes soybean oil avoided in the production of one tonne of canola oil, was reached after four such cycles. The soybean oil avoided at each displacement cycle and the total soybean oil avoided in the production of one tonne of canola oil is shown below:

0.203 + (0.203)2 + (0.203)3 + (0.203)4... = 0.255 (C.2)

The production of canola oil required to produce one tonne of canola oil and balance the supply of vegetable oil in the market is shown below:

1 + 0.203 + (0.203)2 + (0.203)3 + (0.203)4... = 1.255 (C.3)

In SimaPro, the extra production of canola oil (0.255 tonnes) is added to the HEFA biojet life cycle and the avoided production of soybean oil (0.225 tonnes) is subtracted. If the emissions intensity of canola oil and soybean oil were the same, there would be no functional allocation of process emissions at the oil extraction stage, and all emissions from the production of one tonne canola oil would be allocated to the canola oil and not the canola meal. However, the emissions intensity of soybean oil is greater than canola oil, because soybeans have a lower oil content than canola seed (Table C.2). If the emissions intensity of soybean oil were less, the production of canola meal would add emissions to the life cycle of canola biojet, by displacing

166 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

the production of a less emissions-intensive product. It is important to note that if the oil:meal ratio was higher for soybean oil extraction than for canola oil extraction, total soybean oil avoided would reach infinity. Combined with differing emissions intensities, this is another reason why the choice of displaced product is so critical. Given the characteristics of market efficiency, it is conventional to assume that changes in demand for a product, in this case vegetable protein meal, affect the most sensitive source of that product. In the case of protein meal, the most sensitive source is soybeans, since soybean is the only non-cereal crop with protein as the main product (Weidema, 1999). It is unlikely that canola meal would displace a protein meal that is not the determining product of its production system. Most biojet life cycle assessments assume that canola meal displaces soybean meal. For soybean oil extraction in the GREET database, soybean meal is assumed to replace whole soybeans, but without system expansion (Stratton et al., 2010). The Weidema method of system expansion is applied for oilseed processing in GHGenius ((S&T)2 Consultants Inc., 2010), however no published HEFA biojet life cycle assessments have been found that used this method. The Weidema method is employed in a life cycle assessment of canola biodiesel performed for the CCC using GHGenius ((S&T)2 Consultants Inc., 2010). The displaced soybean oil was reported therein as 0.25, as it is in Weidema ((S&T)2 Consultants Inc., 2010; Weidema, 1999), compared to 0.255 found here. A lower canola seed oil content and no accounting for the removal of impurities in the crude vegetable oils could explain the slight difference in results.

167 C.1. DATA SOURCES FOR LIFE CYCLE INPUTS

Table C.3: Properties of petroleum fuels displaced in reﬁning stage ((S&T)2 Consul- tants Inc., 2013)

Fuel EI HHV HHV LHV LHV Density g CO2eq/MJ MJ/L MJ/kg MJ/L MJ/kg kg/L Diesel 83.36 41.36 42.83 38.65 40.03 0.84 Propane 71.35 25.60 43.34 23.51 39.80 0.54 Naphtha 93.84 34.67 46.91 32.53 44.01 0.74

C.1.7 Displaced fuels

The quantity of renewable fuels co-produced in the HEFA reﬁning stage were based on yields of naphtha, propane, and diesel from UOP for their HEFA production process (UOP, 2013). The WTW emissions intensities of the petroleum-derived fuels displaced by the renewable fuels were calculated in GHGenius using the Upstream Results sheet for Western Canada, 2017. The emissions subtracted were only to account for the activities avoided by renewable fuel production. Emissions from distribution, storage, and dispensing were not included, as the renewable fuels and petroleum were assumed to have the same emissions for this life cycle stage. Non-CO2 emissions were likewise not included. The emissions intensity (EI), density, and the high and low heating values (HHV and LLV) of petroleum-derived diesel, propane, and naphtha are listed in Table C.3.

168 C.2. APPLICATION OF CO-PRODUCT ALLOCATION TO CALCULATION OF EMISSIONS INTENSITY

C.2 Application of co-product allocation to calculation of emissions intensity

The allocation of emissions from life cycle stage i was calculated as follows:

ei = (E ∗ oil) ∗ jet (C.4)

Where ei is the emissions from stage i allocated to the main product, biojet fuel, and E is the total emissions from stage i. oil and jet are the allocation ratios from Table 5.2 in decimal form. Depending on the stage, one or both of these ratios could be 1. If both ratios are 1, is is the case for displacement allocation, all emissions are allocated to the main product (Figure 5.1). All stages upstream of reﬁning had both oil and jet allocation ratios applied. The reﬁning stage only had the jet allocation ratio applied as meal was not produced in this stage. Distribution and combustion stages had no allocation ratio applied as no co-products were produced. The displacement credits were subtracted from the life cycle emissions of canola biojet as follows: s k X X L = ei − cj (C.5) i=1 j=1 Ps Where L is the life cycle emissions of canola biojet, i=1 ei is the sum of emissions Pk from stage i allocated to the main product for s number of stages, and j=1 cj is the sum of the displacement credit terms cj for k number of displaced products. Depending on the allocation method chosen, there could be one or no displacement credit terms.

169 BIBLIOGRAPHY

Bibliography

Braun-Unkhoﬀ, M., Riedel, U., and Wahl, C. About the emissions of alternative jet fuels. CEAS Aeronautical Journal, 8(1):167–180, 2017. ISSN 18695590. doi: 10.1007/s13272-016-0230-3.

Cai, H., Han, J., Elgowainy, A., and Wang, M. Parameters of Canola Biofuel Pro- duction Pathways in GREET. Technical report, Argonne National Laboratory, 2015.

Canadian Oilseed Processors Association. Overview of Canada’s Oilseed Processing Sector. Technical report, Canadian Oilseed Processors Association, 2016.

De Klein, C., Novoa, R. S., Ogle, S., Smith, K. A., Rochette, P., Wirth, T. C., Mc- Conkey, B. G., Mosier, A., and Rypdal, K. N2O emissions from managed soils, and CO2 emissions from lime and urea application. In Eggleston, H., Buendia, L., Miwa, K., Ngara, T., and Tanabe, K., editors, 2006 IPCC Guidelines for Na- tional Greenhouse Gas Inventories, Volume 4: Agriculture, Forestry and Other Land Use, chapter 11, pages 11.1–11.54. Global Environmental Strategies (IGES) for the IPCC, 2006.

Elgowainy, A., Han, J., Wang, M., Carter, N., Stratton, R., and Hileman, J. Life

170 BIBLIOGRAPHY

Cycle Analysis of Alternative Aviation Fuels in GREET. Technical report, En- ergy Systems Division, Argonne National Laboratory, U.S. Department of Energy, Argonne, Illinois, USA, 2012.

Fan, J., McConkey, B., Janzen, H., Townley-Smith, L., and Wang, H. Harvest in- dexyield relationship for estimating crop residue in cold continental climates. Field Crops Research, 204:153–157, 2017. ISSN 03784290. doi: 10.1016/j.fcr.2017.01.014.

Forster, P., Ramaswamy, V., Artaxo, P., Berntsen, T., Betts, R., Fahey, D., Hay- wood, J., Lean, J., Lowe, D., Myhre, G., Nganga, J., Prinn, R., Raga, G., Schulz, M., and Van Dorland, R. Changes in Atmospheric Constituents and in Radiative Forcing. In Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K., Tignor, M., and Miller, H., editors, Climate Change 2007: The Physical Sci- ence Basis. Contribution of Working Group 1 to the Fourth Assessment Report of the intergovernemental Panel on Climate Change, chapter 2. Cambridge Univer- sity Press, Cambridge, UK and New York, NY, USA, 2007. ISBN 9780521880. doi: 10.1103/PhysRevB.77.220407.

Gan, Y. T., Campbell, C. A., Janzen, H. H., Lemke, R. L., Basnyat, P., and Mc- Donald, C. L. Carbon input to soil from oilseed and pulse crops on the Canadian prairies. Agriculture, Ecosystems and Environment, 132(3-4):290–297, 2009. ISSN 01678809. doi: 10.1016/j.agee.2009.04.014.

Goodwin, M. Three potential sources for increased canola oil production in the Canadian prairies to meet the needs of biodiesel demand. Technical report, Canola Council of Canada, 2006.

Huo, H., Wang, M., Bloyd, C., and Putsche, V. Life-cycle assessment of energy 171 BIBLIOGRAPHY

use and greenhouse gas emissions of soybean-derived biodiesel and renewable fuels. Environmental Science and Technology, 43(3):750–756, 2009. ISSN 0013-936X. doi: 10.1021/es8011436.

Marker, T. Opportunities for biorenewables in oil reﬁneries. Technical Report 1-4, United States Department of Energy, 2005.

Przybylski, R. Canola Oil: Physical and Chemical Properties. Technical report, 2011.

Smith, E. G. and Barbieri, J. Summary of inputs and production practices used by canola growers in 2011. Technical report, Lethbridge Research Centre, Lethbridge, 2012. 172 BIBLIOGRAPHY

(S&T)2 Consultants Inc. Lifecycle analysis canola biodiesel. Technical report, Canola Council of Canada, Winnipeg, 2010.

(S&T)2 Consultants Inc. GHGenius Model 4.03 Volume 2: Data and Data Sources. Technical report, Natural Resources Canada, Ottawa, ON, 2013.

Stehfest, E. and Bouwman, L. N2O and NO emission from agricultural ﬁelds and soils under natural vegetation: Summarizing available measurement data and modeling of global annual emissions. Nutrient Cycling in Agroecosystems, 74(3):207–228, 2006. ISSN 13851314. doi: 10.1007/s10705-006-9000-7.

Stratton, R. W. Life cycle assessment of greenhouse gas emissions and non-CO combustion eﬀects from alternative jet fuels. Masters thesis, Massachusetts Institute of Technology, 2010.

Stratton, R. W., Wolfe, P. J., and Hileman, J. I. Impact of aviation non-CO2 combustion eﬀects on the environmental feasibility of alternative jet fuels. Environ- mental Science and Technology, 45(24):10736–10743, 2011. ISSN 0013936X. doi: 10.1021/es2017522.

173 BIBLIOGRAPHY

UOP. UOP Renewable Jet Fuel Process. Technical report, received by email February 2017, 2013.

Ward, W. Personal communication, 2016.

Weidema, B. P. System expansions to handle co-products of renewable materials. 7th LCA Case Studies Symposium SETAC-Europe, pages 45–48, 1999.

174