Fuel Economic Analysis of the Potential Value to Petroleum Refiners for Low Carbon Boosted Spark Ignition Bio-blendstocks --Manuscript Draft-Manuscript Number: JFUE-D-22-02495 Article Type: Research Paper Keywords: Refinery modelling, biofuels, boosted spark ignition, economic value, method development, low carbon Abstract: A refinery optimization framework is developed to estimate the benefits of directly blending high-quality bio-blendstocks into a refinery’s gasoline pool. We hypothesized that bio-blendstocks with favorable fuel properties could provide an opportunity for refineries to rebalance operations in potentially profitable ways, and this potential value of the bio-blendstock is characterized by calculating the breakeven value. In this work, biofuels are blended directly with refinery sourced blendstocks such that we produce a finished premium grade gasoline (also termed as Co-Optima Boosted SI gasoline here) along with other refinery products. The proposed modeling framework relied on extensive data for projected product demands (including the share of premium gasoline) over the next few decades, variations in crude oil and refinery product pricing, and fuel quality specifications. The complete refinery models serve as a basis for assessing the impact on profitability on representative petroleum refinery design configurations. Our analysis indicates that breakeven value (BEV) of biofuels varied widely from $20-$120/bbl depending on the fuel blending level and benchmark crude prices. Further, the BEV was correlated with the fuel characteristics such octane numbers, both antiknock index (average of research and motor octane numbers) and sensitivity (difference between the research and motor octane numbers), with a slightly higher correlation with the sensitivity. Implications towards refiners are explored including opportunities to rebalance operations, access to high-value fuel markets, and synchronization with broader transportation industry trends. Furthermore, results indicate the value of Co-Optima boosted spark ignition (BSI) efficiency gains can extend to refiners to incentivize decarbonization and diversified feedstock production. Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation Highlights (for review) Highlights ο· We demonstrate of a new approach to define the value of a biofuel linked to its fuel characteristics, hereby characterized by a metric called the breakeven value. ο· The breakeven value of a biofuel to refiners facing dynamic economic and market conditions, depends on fuel properties such as octane numbers (antiknock index, AKI i.e., (RON+MON)/2 and sensitivity, S, i.e., (RONMON)) in addition to the benchmark crude oil price. ο· This approach can be applied to any biofuel in gasoline, diesel, or jet range to find its economic value and accelerate towards lowering a fuel carbon’s footprint. Graphical Abstract (for review) Click here to access/download;Graphical Abstract (for review);Abstract Image.JPG Manuscript Click here to access/download;Manuscript;Co-Optima BSI Manuscript_clean.docx 1 Economic Analysis of the Potential Value to Petroleum Refiners for Low Carbon Boosted Spark Ignition Bio- 2 blendstocks 3 Nicholas A. Carlson*1, Avantika Singh*1, Michael S. Talmadge1, Yuan Jiang2, George G. Zaimes3, Shuyun Li2, Troy R. 4 Hawkins3, Daniel J. Gaspar2, Robert L. McCormick1, Magdalena M. Ramirez-Corredores4, Lauren Sittler1, Aaron Brooker1 5 1 6 2 7 3 8 4 9 *Equal contributions. Corresponding authors: Nicholas.Carlson@nrel.gov and Avantika.Singh@nrel.gov National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, CO 80401 Pacific Northwest National Laboratory, 902 Battelle Boulevard, Richland, WA 99354 Argonne National Laboratory, 9700 Cass Avenue, Lemont, IL 60439 Idaho National Laboratory, 1955 North Fremont Ave, Idaho Falls, ID 83415 10 11 Abstract 12 13 A refinery optimization framework is developed to estimate the benefits of directly blending high-quality bio-blendstocks into 14 a refinery’s gasoline pool. We hypothesized that bio-blendstocks with favorable fuel properties could provide an opportunity 15 for refineries to rebalance operations in potentially profitable ways, and this potential value of the bio-blendstock is 16 characterized by calculating the breakeven value. In this work, biofuels are blended directly with refinery sourced blendstocks 17 such that we produce a finished premium grade gasoline (also termed as Co-Optima Boosted SI gasoline here) along with other 18 refinery products. The proposed modeling framework relied on extensive data for projected product demands (including the 19 share of premium gasoline) over the next few decades, variations in crude oil and refinery product pricing, and fuel quality 20 specifications. The complete refinery models serve as a basis for assessing the impact on profitability on representative 21 petroleum refinery design configurations. Our analysis indicates that breakeven value (BEV) of biofuels varied widely from 22 $20-$120/bbl depending on the fuel blending level and benchmark crude prices. Further, the BEV was correlated with the fuel 23 characteristics such octane numbers, both antiknock index (average of research and motor octane numbers) and sensitivity 24 (difference between the research and motor octane numbers), with a slightly higher correlation with the sensitivity. 25 Implications towards refiners are explored including opportunities to rebalance operations, access to high-value fuel markets, 26 and synchronization with broader transportation industry trends. Furthermore, results indicate the value of Co-Optima boosted 27 spark ignition (BSI) efficiency gains can extend to refiners to incentivize decarbonization and diversified feedstock production. 28 Keywords 29 Refinery modelling, biofuels, boosted spark ignition, economic value, method development, low carbon 30 1. 31 The transportation sector is one of the largest contributors to global anthropogenic greenhouse gas (GHG) emissions, and in the 32 United States (US) alone, accounted for over 28% of total GHG emissions in 2018 [1], [2]. There is a mounting concern to 33 reduce the GHG emissions associated with gasoline and diesel fuels as well as improve the efficiency of combustion engines. 34 The research and development of sustainable gasoline blendstocks which demonstrate a synergistic performance improvement 35 when used in boosted spark ignition (BSI) engines was conducted through the Co-Optimization of Fuels & Engines (Co- 36 Optima) project[3]. Such biofuels, owing to their superior fuel properties such as higher-octane ratings, can be designed for use 37 in low-emission vehicle engines with potential efficiency improvements up to 15% [4], [5]. A recent publication elucidates the 38 benefits of using co-optimized fuels and engines that increase engine efficiency, lower sector wide GHG and other air pollutant 39 emissions, reduce water consumption, and support domestic job market growth [6]. Previous research, from Co-Optima and 40 other projects, has also focused on cost-benefit analysis and scalability of producing bio-blendstocks, but none so far have 41 explored the economic potential for market adoption of biofuels from a full refinery operations perspective [6], [7]. The 42 objective of this work is to complement previous analyses by assessing the potential economic value proposition for petroleum 43 refiners with respect to gasoline-range blendstocks identified in Co-Optima research. 44 Commercially available gasoline is a mixture of several blendstocks, produced from a variety of separation and upgrading 45 refinery operations that are ultimately blended into a gasoline pool. These blendstocks range from light hydrocarbons such as 46 butane to intermediate and heavy naphtha sourced from crude distillation (straight run), fluidized catalytic crackers, 47 hydrocrackers, cokers, visbreakers, alkylation, isomerization, and reforming units. Table A1 summarizes the typical volume 48 concentration of these traditional gasoline blendstocks, the ranges of their octane values, and the sources of their production in 49 the refinery. 50 Progressively stringent fuel standards and heightened regulatory requirements are forcing gasoline producers to modify their 51 operations and creates a unique opportunity for the expanded use of biofuels in refinery operations. Advancements in fuel Introduction 52 economy standards, known as Corporate Average Fuel Economy (CAFÉ) standards [8], and the proposal from automobile 53 manufacturers to create Research Octane Number (RON) 95 as the new uniform standard for gasoline will require an increased 54 share of high octane blendstocks, such as reformate, in the fuel mix [9]. The rising price of octane over the last five to six years 55 stemming from an increasing demand for premium gasoline and the increasing consumption of light shale oil [10] gives 56 evidence that refineries are already facing operational constraints [11]. Additionally, the U.S. Environmental Protection 57 Agency (EPA) introduced Tier 3 regulations that require all gasoline sold in the US to meet a 10-ppm annual sulfur average 58 criterion [12]. Tighter sulfur restrictions could render some refineries unable to sell finished gasoline. Integrating bio- 59 blendstocks with superior fuel properties may help less advantaged refineries remain competitive despite changing market 60 demands and increasingly stringent fuel quality specifications. 61 Co-optima researchers have previously evaluated several bio-blendstocks based on their fuel properties, scalability, and 62 compatibility with existing infrastructure [6]. Candidates for the BSI combustion mode were systematically screened in a tiered 63 manner to down select promising candidates which encompass a range of functional groups [13]. An empirical relationship was 64 developed to estimate the theoretical gains in engine efficiency as a function of bio-blendstock properties such as octane 65 number, sensitivity (defined as the difference between research octane number (RON) and motor octane number (MON)), and 66 heating value, among others [5]. Promising bio-blendstocks which met the fuel performance and efficiency improvement 67 criteria were further evaluated to estimate their production costs using techno-economic principles as well as environmental 68 impact metrics [5], [13], [14]. Of the top Co-Optima BSI bio-blendstocks that were identified, seven of those: ethanol (ETC), 69 iso-propanol (IPR), n-propanol (NPR), iso-butanol (IBU), di-iso-butylene (DII), cyclopentanone (CYC), and a furans mixture 70 (FUR) are investigated in this study, primarily because of the availability of extensive blending data of these bio-blendstocks 71 with hydrocarbon blendstocks for oxygenated blending (BOBs). 72 Experimental biofuels property data from Co-Optima research was leveraged to build economic blending models based on 73 optimization tools, consistent with the methodology often deployed at a refinery [15]. Analysis efforts predicated upon these 74 models have two key outcomes of interest. First, identifying the value propositions of blending performance advantaged 75 biofuels into gasoline, which would be useful both to the refiners and the biofuel producers. Understanding the impact of 76 favorable biofuel properties such as octane, sensitivity, or Reid Vapor Pressure (RVP) on refinery economics can help generate 77 market pull for new bio-blendstocks from refiners. Second, coupling refinery models with life cycle assessment (LCA) can 78 identify the environmental benefits and tradeoffs of Co-Optima fuel blending (this will be the subject of an upcoming 79 publication). 80 With the aforementioned interests in mind, we hereby explore the economic opportunity of blending the most promising Co- 81 Optima BSI blendstocks into the gasoline pool. While all refinery and terminal blending operations are unique based the 82 regional availability and fuel specifications, we hereby demonstrate a framework to evaluate new biofuel options. To the best 83 of our knowledge, such analysis is not comprehensively discussed in scientific literature, and is largely limited to refinery 84 operators, particularly for quantifying the economic value of new bio-blendstocks. So, to establish that framework, this paper is 85 divided into three sections: (1) method development for benchmarking and analysis, (2) key results and discussions, and 86 subsequently, (3) future outlook. In the methods section, we outline the assumptions for establishing a model baseline with an 87 emphasis on the development of refinery pricing structures. Other key underpinnings of the analysis included in the modelling 88 approach are gasoline specifications, fuel demand projections, and blending properties of bio-blendstocks as discussed below. 89 Further, in the results section we dive into the specific quantification of the bio-blendstock value to refineries, particularly as a 90 function of its fuel properties at different blending ratios and different crude oil prices, while the future outlook section 91 highlights some of the challenges and shortcomings of this analysis approach. Overall, the objective of this analysis is to 92 evaluate if there are opportunities to improve the economics of a traditional petroleum refinery by utilizing the unique and 93 potentially valuable properties of the Co-Optima blendstocks. 94 95 2. 96 AspenTech’ s Process Industry Modeling System (PIMS) software [16] was used to develop and optimize three different 97 refinery configuration models using the “Gulf-Coast” (belonging to the Petroleum Administration for Defense District 3, 98 PADD3) example as a starting point. Differentiations between the three refinery configurations included varying complexities, 99 production rate targets, crude oil types, crude and product pricing, and fuel specifications relevant for Co-Optima analysis. Materials and Methods 100 These are labeled as PADD3-Large, PADD3-Medium, and PADD3-Small intending to represent refineries of decreasing 101 conversion capacity and complexity, all located within the U.S. Gulf-Coast PADD3 region. 102 Information on the crude oils accessible to the refinery and their property characteristics and assays are incorporated in the 103 model based on published information pertaining to the PADD3 region [17]. Primary petroleum processing units and their 104 corresponding capacities shown in Table A2 along with typical yield compositions are also built into the model [18]. Ideas to 105 expand the scope of the model are presented after results in the model customization section. 106 The objective function in PIMS advanced optimization mode aims to maximize the refinery model’s gross margin subject to 107 constraints on crude availability, limitations on process unit capacities and operating conditions, and specifications on finished 108 fuels. The mass and energy balances are consistently maintained in all operations. The refinery models considered here 109 produce two grades of gasoline (a) regular and (b) premium for two blend types - conventional and reformulated - as well as 110 introducing a high-efficiency co-optimized gasoline (similar in specs to reformulated premium) containing one of the several 111 candidate bio-blendstocks. Notably, the co-optimized gasoline blend enacts constraints on two gasoline features to improve 112 efficiency and performance: (i) a novel minimum sensitivity of 10, and (ii) a minimum RON of 98. Table A3 displays the 113 finished fuel specifications applied in the analysis. 114 The following subsections highlight all components incorporated into the refinery models including the following: (2.1) 115 refinery configuration and capacity data, (2.2) product demand projections, (2.3) pricing structure, (2.4) bio-blendstock 116 properties for blending, and (2.5) break-even value calculation. These bases are derived from other Co-Optima projects, 117 published literature, and experimental data provide a comprehensive framework to assess refinery impacts and valuation results 118 for the BSI bio-blendstocks. ` 119 2.1 Refinery Configuration and Capacity Data 120 Error! Reference source not found.A2 summarizes the process unit capacities of the refinery models employed in this study. 121 These configurations were developed based on U.S. Energy Information Administration (EIA) data on the US refining 122 infrastructure by each Petroleum Administration for Defense Districts (PADD) [18]. The goal was to develop sets of refinery 123 models that span a wide range of refinery configurations and complexities as indicated by the calculated Nelson Complexity 124 Index values in Table 2 Error! Reference source not found.[19]. Our analysis focuses on the PADD3-Large refinery 125 configuration and PADD3-Small configuration which are intended to serve as bookends. 126 2.2 Product Demand Projections from EIA, OPIS and ADOPT 127 Demand for gasoline range fuels for light duty vehicles is expected to rise in the next 30 years. To isolate the economic impact 128 of biofuels blending, time-based scenarios are designed between years 2020 to 2050 to represent changing market demands. 129 Fig. 1 presents the basis for the refinery operating scenarios with projections for the major refinery product categories such as 130 liquified petroleum gas (LPG), gasoline, jet, distillate, and residual fuels. EIA projections from the Annual Energy Outlook 131 (AEO) provide a complete basis for all product groupings [20]. In addition to the EIA gasoline BOBs projection, a second 132 gasoline projection is also considered in the refinery scenarios from the Automotive Deployment Options Projection Tool 133 (ADOPT) [21]. ADOPT is a consumer choice vehicle model developed by the National Renewable Energy Laboratory (NREL) 134 taking parameters like vehicle price, fuel cost, and range into account, which in turn predicts the fuel demand. The ADOPT 135 results predict a sharper decline in gasoline demand owing to a higher level of projected electrification in light-duty vehicles. 136 Product demand projections in terms of volumes can be found in Table A6. 45% Propane and LPG 140% 40% 130% 35% 120% 110% 30% 100% 25% 90% 80% 20% 70% 15% 60% 50% 2020 2025 2030 2035 2040 2045 10% 2050 Gasoline Market Share (%) Product Demand (% of 2020 Demand) 150% EIA - Gasoline BOBs ADOPT Gasoline BOBs Jet/Kerosene Distillate Fuel Oils Residual Fuel Oils Premium Gasoline Market Share Year 137 138 Fig 1. Projected product demands for refinery product (left Y axis) categories from EIA [20] and ADOPT model [21] alongside 139 projected market share of premium gasoline fuels as a percentage of total gasoline fuels (shown by the dotted line linked to the 140 right Y axis) [11]. Note: the percentage projections here were multiplied by 400,000 barrels/day (BPD) and 45,000 BPD for 141 the PADD3-L and PADD3-S models, respectively, to set the maximum sale quantities for each product category in the refinery 142 models. Capping the sales of each product depending on the year and demand projection source (EIA or ADOPT) simulated 143 the dynamic economic conditions between 2020 – 2050. 144 145 In addition to the projections for overall market demands, scenarios also consider a projection for the market share of premium 146 gasoline relative to total gasoline products (Second Y axis, Fig. 1). Premium gasoline demand is important within the context 147 of the analysis because Co-Optima gasoline is categorized within the premium gasolines grouping. Fig. 1 shows the premium 148 market share interpolation between 2020 and 2050 based on data reported by OPIS [11]. 149 2.3 Pricing Structure 150 Crude oil and product pricing models were developed to model the impact of dynamic petroleum markets on bio-blendstock 151 values. To simplify the number of economic inputs feeding the analysis, prices entering the model were correlated to a 152 benchmark crude oil price, namely West-Texas Intermediate (WTI). WTI was chosen because of its longstanding position as 153 the economic benchmark for crude oil pricing in U.S., its prevalence in the PADD3 district being the context for the analysis, 154 and the availability of historical pricing data. Examples for each pricing calculation introduced in this section can be found in 155 the Appendix. 156 2.3.1 Crude Oil Correlation: Quality and Price Determination 157 Historically, different crude oil prices are correlated but not equal due to differences in quality, production costs, or 158 transportation costs. Any crude oil historical pricing data that could be found in timestep with WTI was aggregated from 159 multiple data sources and used to linearly regress the crude’s price as a function of WTI price. After regression, a given crude 160 oil’s price could be modelled using equation 1: (1) ππππ’ππ,$ = ππ ππππΌ,$ + ππ 161 162 Where slope (mc) and intercept (bc) were fitted parameters yielded from linear regression. Example of a crude oil pricing 163 calculation is shown in the Example A in the Appendix. Various pricing data sources were utilized including the Energy 164 Information Administration (EIA), Oil Price Information Service (OPIS), Bloomberg Terminal, and Baker & O’Brien’s 165 database to dilute any bias coming from an individual source. Modelling crude oil prices as a linear function of WTI price was 166 the preferred crude oil pricing method because of the high-quality correlations typically observed. 167 When there was an insufficient amount of historical crude oil pricing data, a backup method for obtaining crude oil price as a 168 function of crude qualities was developed. The variables in the pricing model include: the American Petroleum Institute (API) 169 gravity measure, sulfur content (weight %), Total Acid Number (TAN), and WTI benchmark price. API gravity is a commonly 170 reported index of crude oil density and generally reflects the degree of conversion necessary to make high value products such 171 as gasoline and diesel fuel. Sulfur content and TAN were measures intended to reflect the level of impurities in the crude oil 172 and the corresponding level of treatment needed to make finished fuels. The WTI benchmark price was included to give the 173 model a reference point and gave the model identical inputs when compared with the other pricing correlation models being 174 utilized. 175 A multiple linear regression model performed well in capturing the crude oil pricing variability as a function of the four 176 independent variables described above. Consequentially, crude oil prices determined by equation 2 supplanted pricing data 177 determined by equation 1 when historical data limitations were met: 178 (2) ππππ’ππ,$ = π1 (π΄ππΌ) + π2 (ππ’πππ’ππ€π‘% ) + π3 (ππ΄π) + π4 (ππππΌ,$ ) + π 179 An example calculation using the crude quality model can be found in Example B in the Appendix. 180 2.3.2 Product Price Determination 181 Crude oil purchases are proportionally the largest cost associated with refinery operations leading to a high degree of 182 correlation between the prices of crude oils and refinery products. Consequentially, product pricing model development with 183 simple linear correlations was similarly enabled, when compared to the crude oil pricing models, as depicted in equation 3. (3) ππππππ’ππ‘,$ = ππ ππππΌ,$ + ππ 184 185 Gaseous products such as natural gas or propane are less correlated to crude oil because they aren’t exclusively sourced from 186 crude oil production. However, the lower value of these products ultimately served to dampen the effects of pricing uncertainty 187 on the model’s profitability. Refinery intermediates were also priced to allow the PIMS-model to sell all molecules in case 188 certain finished product specifications could not be satisfied. The same set of diverse sources used for crude oil pricing were 189 again used to lessen the effects of bias from any one data source. An example calculation of gasoline price given a benchmark 190 WTI crude price can be found in Example C in the Appendix. 191 2.3.3 Octane Pricing Correlation 192 The proposed performance-enhancing qualities of co-optimized fuels for BSI engines are two-fold including 1) increased 193 octane which is historically priced by the market through a spread between premium and regular gasolines and 2) increased 194 sensitivity which has not been traditionally priced by the market. In an attempt to reliably and conservatively price co- 195 optimized fuels with otherwise unknown values, an octane pricing model was developed with reference to regular gasoline. 196 The pricing model is conservative due to its lack of accountability for increased sensitivity in the value of the co-optimized 197 fuel. Increased engine efficiencies from higher sensitivity fuels will certainly increase value to end-users so not accounting for 198 sensitivity effectively underprices co-optimized gasolines and the corresponding value bio-blending brings to refiners. 199 A single layer neural network model was adapted to historical data to predict octane value in terms of $/octane-gallon. 200 Therefore, the model’s output (Y) could be utilized in equation 4 to determine a price for co-optimized gasoline. 201 (4) πππ−ππ,$ = π ( $ πππ‘βπππ ) (π΄πΎπΌππ−ππ − π΄πΎπΌπππ )(πππ‘) + ππππ,$ ( $ πππ ) 202 where octane value was multiplied by the difference in anti-knock index (AKI – calculated as the average of RON and MON) 203 of the co-optimized gasoline and regular, conventional gasoline. 204 205 The neural network model inputs included a baseline regular gasoline price, total U.S. octane demand, and the differentials of 206 benzene, toluene, and xylene commodity prices to regular gasoline price. The baseline regular gasoline price was included as a 207 predictor to reflect the overall supply-and-demand equilibration constantly induced in the U.S. gasoline market. Total U.S. 208 octane demand was calculated as the sum-product of regular and premium gasoline demands, respectively, and their 209 corresponding octane numbers found. This predictor was included to account for any octane-barrel supply constraints the U.S. 210 refining industry may have faced historically that would have induced higher prices. Price differentials between commodity 211 benzene, toluene, and xylene (BTXs) were also included because aromatics are common high-octane blendstocks refiners use 212 to upgrade their regular gasoline pool. The price difference between regular gasoline and the constituents that must be added to 213 form premium blends quantifies the value of octane to refiners to some degree. Predictors were highly correlated, so they were 214 grouped into two principled components using principled component analysis (PCA) prior to fitting the non-linear neural 215 network. The model was trained on 249 samples yielding a testing set R 2 of 0.97 on 62 samples with ten-fold cross-validation. 216 Fig. A1 shows a parity plot of the model’s predictions and historically realized premium spreads. 217 2.4 Bio-Blendstock Properties for Blending 218 All the identified bio-blendstocks considered in this analysis demonstrate a significant boost in the octane values of final blends 219 relative to the impact predicted by linear volumetric blending alone (Fig. 2, numerical data in Table A4). 120 Neat 10 vol% 20 vol% 30 vol% 10 vol% 20 vol% 30 vol% Blending RON 120 80 100 60 40 20 60 40 0 ETC 40 80 20 0 IPR Neat NPR 10 vol% IBU DII 20 vol% CYC FUR 30 vol% 35 ETC 140 IPR Neat NPR 10 vol% IBU DII 20 vol% CYC FUR 30 vol% 120 Blending RVP (kPa) Blending Sensitivity Neat 140 100 Blending MON 160 30 100 25 20 15 10 5 0 80 60 40 20 0 ETC IPR NPR IBU DII CYC FUR ETC IPR NPR IBU DII CYC FUR 220 Fig 2. Graphical summary of blending properties applied in this analysis for Co-Optima BSI bio-blendstocks which show 221 nonlinear effect on octane number, RVP, and other properties when combined with petroleum BOBs. Abbreviations are CYC: 222 cyclopentanone, DII: diisobutylene, ETC: ethanol, FUR: furans mixture, IBU: isobutanol, IPR: isopropanol, NPR: n-propanol. 223 The numerical values are described in Appendix Table A4. 224 225 To more accurately reflect the impacts of non-linear blending, effective blending octane numbers were calculated for each 226 blendstock at each blend level. The concept of blending octane number has been the subject of numerous studies [7] and can be 227 summarized as by the relationship shown in Equation 5 where RON i represents the blending research octane number of a given 228 bio-blendstock. 229 π πππππππ = π£π΅ππ΅ π πππ΅ππ΅ + π£π π πππ (5) π πππ = 230 π πππππππ −π£π΅ππ΅ π πππ΅ππ΅ π£π 231 Here, π£π΅ππ΅ and π£π represent the volume fractions of BOB and biofuel respectively, π πππππππ is the measured RON of the final 232 biofuel/gasoline blend, and π πππ΅ππ΅ is the measured RON of gasoline BOB. The data for 10%, 20%, and 30% blending levels 233 were found in the NREL’s Fuel Properties Database and are tabularized in Table A4 [22]. Blending research/motor octane 234 numbers and their resulting blending sensitivities were used as inputs to the refinery models to capture the effects of blend 235 level on bio-blendstock values. 236 237 2.5 Bio-Blendstock Breakeven Value (BEV) Characterization 238 The break-even value (BEV) for a bio-blendstock, or any other refinery input, is defined as the calculated maximum 239 purchasing cost that would preserve an equivalent gross margin under the same conditions without the bio-blendstock 240 purchase. The use of this calculation for valuation is consistent with industry practices for decisions like crude purchases as 241 well as other bio-blendstock research [7]. The BEV calculation procedure in this analysis was as follows: 1) fix model inputs 242 including benchmark WTI price, year, demand projection source, bio-blendstock candidate, and blending level 2) optimize 243 PIMS refinery model with no bio-blendstock purchasing and Co-Optima gasoline production allowed to form a baseline 3) re- 244 optimize the PIMS refinery model with bio-blendstock prices set to zero and Co-Optima gasoline production allowed 4) divide 245 the difference between observed profits by the volume of bio-blendstock purchased in the Co-Optima scenario as shown in 246 Equation 6. The units for the BEV are US$ per volumetric unit of bio-blendstock (denoted by BBS below), typically calculated 247 in barrels or gallons. 248 (6) π΅πΈππ΅π΅π = ππππππ‘π΅π΅π − ππππππ‘π΅ππ π π΅π΅π π΅ππππππ ππππ’ππ 249 The bio-blendstock BEVs are ultimately a function of product demands, crude and product prices, refinery configurations, 250 refinery process unit capacities and constraints, blendstock properties, and finished fuel specifications. All of these elements 251 are considered by the Aspen PIMS model in optimizing refinery operating scenarios. 252 253 3. Results 254 3.1 Establishing Baseline Refinery Gross Margin 255 A discussion on prospective refining strategies must consider the major market changes refiners are expected to endure over 256 the next 30 years. As projected in EIA’s Annual Energy Outlook and other industry perspectives, the refinery product slate is 257 expected to undergo substantial transformation with the demand for gasoline in the light duty transportation sector declining 258 while long haul transportation fuel demand is poised to increase [20] [23]. Fig. 3 depicts the PADD3-L model’s projected 259 refinery gross margin extending to 2050 under both demand projections given by EIA and ADOPT projections, respectively. 260 Gross margin values are given as percentages of 2020 gross margin. Whiskers represent the range of values observed over WTI 261 benchmark crude prices varied from $20-100/bbl. Gross Margin (% of 2020 Value) 120% 110% 100% 90% 80% EIA 70% ADOPT 60% 2015 262 2020 2025 2030 2035 2040 2045 2050 2055 Year 263 Fig 3. PADD3-Large refinery gross margins normalized to 2020 values given demand projections from both the EIA’s 264 Annual Energy Outlook (2019) and ADOPT consumer choice model. 265 266 U.S. refineries are typically geared towards gasoline production (45%–50%), with diesel production approximately at 25% of 267 an overall refinery’s product slate [22]. Thus, the higher level of electric vehicle market penetration predicted by the ADOPT 268 model has an increasingly large effect on refinery profitability going out to 2050. The assumptions made in the EIA and 269 ADOPT models may be considered bookends with ADOPT being the least favorable outcome for the refining industry. 270 This work leveraged the expected trends in fuel demand to benchmark the change in refinery profitability with the assumption 271 that the refining sector does not reconfigure. While this was an unlikely scenario, in the absence of any verifiable data, this 272 assumption served as a basis to understand the value of bio- blendstock integration into the gasoline market. 273 3.2 Estimating Bio-Blendstock Break Even Values 274 Fig. 4 depicts each bio-blendstocks’ break-even value when blended at 10, 20, and 30 vol.% with a WTI benchmark crude 275 price of $60/bbl in year 2050 under EIA demand projections. This scenario was treated as a baseline as it represents a middle 276 range WTI price during a fully developed Co-Optima gasoline market in 2050. Bio-blendstocks without a bar corresponding to 277 the 10 vol.% blending level do not have a high enough sensitivity to meet the minimum Co-Optima gasoline sensitivity 278 specification of 10. 279 Break-Even Value of BSI BioBlendstocks ($/bbl) 140 120 100 80 10 vol% 60 20 vol% 30 vol% 40 20 0 ETC IPR NPR IBU DII CYC FUR 280 281 Fig. 4. BSI Bio-Blendstock break-even values ($/bbl) for the following scenario: $60/bbl WTI benchmark crude price, year 282 2050, and EIA demand assumptions. 283 284 3.3 Determining Bio-Blendstock Value and the role of Sensitivity parameter 285 Traditionally, refineries provide value to gasoline consumers by increasing octane to enable consumption in higher 286 performance engines. Co-optimized gasolines leverage increased sensitivity alongside octane to further increase engine 287 performance over modern high compression engines [24]. Common gasoline blendstocks typically have low sensitivities as 288 when compared to the bio-blendstocks sensitives (see Tables A4-A5). Gasoline upgrading units (i.e., reforming, isomerization, 289 alkylation, etc.) focus on increasing octane and provide little to no operational levers to adjust blendstock sensitivities. 290 Blending varying levels of BSI blendstocks being considered in Co-Optima bio-blendstocks featuring relatively high 291 sensitivities and low levels of impurities, to not interfere with other gasoline specifications, could provide a control mechanism 292 for refineries to adjust sensitivity. 293 Figs. 5 and 6 show the break-even values (BEVs), averaged over WTI benchmark prices of 20-100 $/bbl, of the seven bio- 294 blendstocks under consideration at blending levels of 10, 20, and 30 vol.%. Adjoining whiskers represent minimums and 295 maximums, respectively, observed in the 20-100 $/bbl WTI benchmark range. Fig. 5 plots BEVs against blending octanes, 296 accounting for variable octane boosting at different blending levels. A positive correlation is observed between the break-even 297 values and increasing octanes. 200 ETC IPR NPR 160 IBU DII CYC 140 FUR Break Even Value ($/bbl) 180 120 100 80 60 40 y = 1.4066x - 85.105 R² = 0.2028 20 0 100 298 105 110 115 120 125 130 135 Blending Octane (RON + MON)/2 299 Fig. 5. Break even values for seven candidate BSI bio-blendstocks, at 10%, 20%, and 30% blending levels plotted 300 against corresponding blending octanes. CYC: cyclopentanone, DII: diisobutylene, ETC: ethanol, FUR: furans 301 mixture, IBU: isobutanol, IPR: isopropanol, NPR: n-propanol. 302 303 Fig. 6 similarly plots break-even values but juxtaposes them against the different bio-blendstock sensitivities taken as the 304 difference between blending RONs and MONs. A stronger correlation between blendstock value and sensitivity is observed 305 when compared to Fig. 5 (determined by the R-squared value). Fig. A2 and A3 plots the same data but shows three individual 306 regression lines for 10, 20, and 30 blending vol.% and shows the bulk of the correlation is attributed to 10 vol.%. 200 Break Even Value ($/bbl) 180 160 ETC IPR NPR IBU DII CYC FUR 140 120 100 80 60 40 y = 2.2572x + 22.01 R² = 0.4697 20 0 7 12 17 22 27 32 37 Blending Sensitivity (RON - MON) 307 308 Fig. 6. Break even values for seven candidate BSI bio-blendstocks, at 10%, 20%, and 30% blending levels plotted 309 against corresponding blending sensitivities determined as the difference between blending RON and MON. CYC: 310 cyclopentanone, DII: diisobutylene, ETC: ethanol, FUR: furans mixture, IBU: isobutanol, IPR: isopropanol, NPR: n- 311 propanol. 312 313 The data indicates that although bringing high octane bio-blendstocks to the refinery is valuable, incremental improvements in 314 sensitivity are more valuable when co-optimized engines create a market demand for high sensitivity fuel. Refineries don’t 315 necessarily need outsourced blendstocks to meet higher octane specifications or to fill excess demand since gasoline demand is 316 projected to remain stable if not decline steadily. The true value proposition bio-blending seems to bring to refineries is the 317 capability of producing specialty fuels with property considerations other than octane. These fuels could become additional 318 value chains for refineries and could supplant lost revenues resulting from decreasing gasoline demands. 319 320 3.4 Determining the Optimal Blending Volumes 321 Fig. 7A also plots break even value averages for PADD3-L, with minimums and maximums observed across the 20-100 ($/bbl) 322 range again represented as whiskers. Each bio-blendstock has a clear optimum blending level, most consistently around the 20 323 vol.% mark except for the furan mixture. The majority of the blendstocks having an optimal value indicates that beyond a 324 certain level, the added sensitivity brought through bio-blending no longer has a marginal benefit to the refinery. As shown in 325 Fig. 2, furans have an octane sensitivity significantly higher than other bio-blendstocks, and therefore a lower blend level is 326 preferred from the economic perspective. Fig. 7B relates each bio-blendstocks’ optimal blending volume to its pure sensitivity and depicts a clear inverse relationship. 328 Optimal blending volumes were determined by finding the maximums of quadratic functions fit to the data presented in Fig. 329 7A. The furans mixture is excluded because additional runs below the 10% blending level would be required to find an optimal 330 blending volume. With increasing sensitivity, less bio-blendstock is required to meet the co-optimized gasoline pools 331 sensitivity specification. 200 180 160 140 120 100 80 60 40 20 0 30 Optimal Blending Level (Vol%) Break Even Value ($/bbl) 327 25 20 15 10 y = -0.7165x + 36.386 R² = 0.7677 5 0 5 10 15 20 25 30 35 15 Bio-blendstock Blending Level (Vol%) (A) 20 25 30 Blending Sensitivity (RON - MON) 35 (B) Fig. 7. (A) Bio-blendstock break even values, averaged over 20-100 $/bbl WTI benchmark prices and EIA/ADOPT projections from 2020-2050, plotted against blending level in co-optimized gasoline. (B) Optimal blending level plotted against pure bio-blendstock sensitivity. CYC: cyclopentanone, DII: diisobutylene, ETC: ethanol, FUR: furans mixture, IBU: isobutanol, IPR: isopropanol, NPR: n-propanol. 332 333 334 3.5 Bio-Blendstock Integration Value Over Time 335 Integrating bio-blendstock can give refiners access to novel value-chains with the potential to replace expected lost revenues 336 stemming from declining gasoline demand. Fig. 8 depicts the average break-even value for PADD3-L across all bio- 337 blendstocks and 20-100 ($/bbl) WTI benchmark price levels with observed minimums and maximums as whiskers. Using the 338 specific methods described in this analysis, the value of bio-blending to refiners does not appear to be agnostic to the gasoline 339 market’s projected contraction. The aggregate break-even values appear to correlate with the decreasing refinery gross-margin 340 depicted in Fig. 3 with a more severe decline under ADOPT gasoline demand projections. 140 EIA ADOPT Break Even Value ($/bbl) 120 100 80 60 40 20 0 2025 2030 2035 2040 2045 2050 2055 Year 341 342 Fig. 8. Break even values averaged over the seven bio-blendstocks, WTI benchmark prices ranging from $20-100/bbl, 343 and 10%, 20%, and 30% blending levels shown from 2030 to 2050. 344 345 More notably, the average break-even values remain steadily high during the anticipated gasoline market contraction from 346 2030-2050 indicating consistent value to refiners through unfavorable economic conditions. Blending co-optimized fuels 347 combats decreasing gasoline margins by adding new value chains to the gasoline market while diversifying feedstock sourcing 348 with the displacement of traditional fossil feedstocks with bio-derived feedstocks. 349 3.6 Bio-Blendstock Value to Low-Conversion Refinery 350 Although sensitivity has been identified as the key value driver for refineries to integrate Co-Optima bio-blendstocks, the 351 supplemental octane boost can be an additional attraction for certain refineries facing conversion or octane constraints. Lower 352 complexity refineries with less conversion capacity have inherently less flexibility in their blending operations and can often be 353 octane constrained. The value proposition of bio-blending to conversion constrained refineries was studied using the PADD3- 354 Small refinery model. The main changes from the PADD3-Large model included reducing the capacity from 400,000 bbl/day 355 to 43,500 bbl/day and removing the gasoil hydrotreating, catalytic cracking, hydrocracking, and delayed-coking units. 356 357 Without bio-blending, the PADD3-Small model could not produce enough high-octane gasoline blendstocks to sell premium 358 gasoline. As shown in Fig. 9, integrating BSI bio-blendstocks induced a step change in profitability for the small refinery 359 because the bio-blendstocks’ high octane compensated for the refineries lack of octane-barrels. A portion of the refineries’ 360 lower-quality gasoline blendstocks could be diluted by the high-quality bio-blendstocks in the co-optimized gasoline pool, 361 featuring a large pricing premium over regular gasoline. At the 10% blending level, there was not enough high-octane bio- 362 blendstock purchasable by the model to compensate for the low-quality gasoline blendstocks, so no co-optimized fuel could be 363 sold. For this reason, Fig. 9 only compares the 20% and 30% blending level scenarios observed for the PADD3-Small and 364 PADD3-Large models. 200 PADD3L Break Even Value ($/BBL) 180 PADD3S 160 140 120 100 80 60 40 20 0 2025 2030 2035 2040 2045 2050 2055 Year 365 366 Fig. 9. Break even values averaged over the six candidate BSI bio-blendstocks, 20-100 $/bbl WTI benchmark prices, 367 and 20% and 30% blend levels with comparison between the PADD3-Large and PADD3-Small refinery models. 368 369 Fig. 9 shows that bio-blending is more valuable to lower complexity refineries because it not only brings in capabilities to 370 produce specialty, high-value fuels, but also introduces high octane-barrels into their otherwise constrained blending 371 operations. 372 373 4. 374 4.1 Model Customization and Improvement 375 This work demonstrates the approach to define the value of a biofuel linked to its fuel characteristics such as octane number, 376 and sensitivity. Though this analysis demonstrates high sensitivity/RON as fuel properties driving value to refiners producing 377 Co-Optimized BSI gasoline, there are certainly many avenues one could take to extend or reframe the research. As seen in the Discussion 378 previous section, the value of a bio-blendstocks to a refiner increases as a function of increasing RON and increasing 379 sensitivity. However, when we tried to rank which of these two factors may have a greater impact for the refiner, we find that 380 both octane number and sensitivity are highly correlated and trying to delineate the two did not give a very clear ranking 381 between RON and sensitivity. Higher resolution blending data could better inform the non-linear blending models featured 382 here. 383 Additionally, the modelling approach as discussed here can be customized and adapted for improved accuracy. For example, 384 we were working with limited blending data from 3 summer BOBs (combination of conventional and reformulated BOBs) to 385 obtain blending in numbers. However, previous research has shown that determination of specific BOB characteristics 386 (PIONA, RVP, etc.) would allow for a more specific blending octane number and a higher degree of accuracy rather than a 387 generic calculation for blending octane numbers [25]–[28]. The same applies for RVP [29]–[31]. More specific determination 388 of blending properties (RON, S, RVP) can enable more accurate calculations and can be tailored for specific refinery outputs 389 and characteristics. 390 391 Another notable aspect is the impact on ease of vaporization or the distillation profile (D86 curve), which determines the 392 drivability experience, such as knocking, etc. [29], [32], [33]. In this analysis, we were only solving for regulatory specs such 393 as RVP and octane. However, inclusion of D86 data could be an additional step that can be takes to make the analysis more 394 comprehensive. 395 396 We also would like to highlight that the analysis presented here for gasoline closely mirrors the reformulated summer gasoline 397 specifications, which is more stringent in terms of RVP, as opposed to winter, which has a higher RVP minimum [34]. We 398 hypothesize that high-octane blendstocks which hit the RVP constraint of summer gasoline, for example, could potentially still 399 be valuable for winter gasoline when cheaper naphtha streams and butane are available for blending. This would allow refiners 400 to still create finished gasoline with higher biofuel blending specifically in situations where the constraining criteria is seasonal. 401 402 Similarly, PADD and fuel specifications vary not only in the US but also in other countries (cite). Also, it is known that every 403 refinery is unique in its location, design, crude slate, operating limits, and refined products. Therefore, the analysis presented 404 here is limited in its applicability to the hypothetical refinery models considered. Refineries outside of the PADD3 district 405 which consume different crudes and meet different specifications could also be studied. Hence, this study which was 406 developed for PADD3 refinery configuration can be customized or adapted for more accuracy. The fuel specifications as well 407 as the characteristics of a specific refinery (crude/product slate) vary across different geographies and different continents so a 408 different more specialized refinery case study might be warranted to realize the benefits of biofuels. Here, we are 409 demonstrating a proof of concept as well as a framework on how to approach blending of biofuels and how to value a biofuel 410 for a specific refinery, so this technique should be applicable even with different sets of data leading, potentially leading to 411 different results in the in terms of value of biofuel. Some of these aspects, such as variability in crude pricing, are covered in 412 the results sections before. 413 A potential drawback with this approach stems from the lack of wide scale adoption, i.e., second generation biofuel, from 414 feedstocks such as herbaceous biomass or woody feedstocks, that are not currently available in sufficient volumes and that lack 415 sufficient infrastructure for production, transportation, and distribution [14], [35]. With the limited availability of bio- 416 blendstocks, their prices might be constrained by the supply and demand which can potentially impact refinery economics. In 417 this analysis we assume that sufficient biofuel is available for all refineries to meet their demand, but it would be dictated by 418 supply and demand factors. Some of these aspects are discussed in an upcoming publication. 419 Finally, in this model we correlated the value of products associated with a single WTI price benchmark and octane price gradient 420 and calculated different crude oil prices as well as product prices as a function of WTI. Consequently, there is uncertainty built 421 in the model, as seen by the whisker plots). This approach works to present results as a function of WTI pricing alone for the 422 purposes of this research work. An alternate approach could entail using an average of market data from the previous years as 423 well as future projections to quantify the value of premium fuel. 424 4.2 Outlook and Next Steps 425 The refinery models detailed here could serve as the basis for even more comprehensive models. Potential modifications could 426 range from PIMS sub-model alterations, the addition of new bio-blendstocks/products, to simple backend calculations using the 427 material and cash flows produced by the PIMS models, which may have much wider implications. More specific ideas to build 428 upon the analysis are presented below: 429 1. Shared value between biofuel producers and refiners: A more complete understanding of each bio-blendstock’s 430 economic feasibility could be obtained be taking the difference between BEV and minimum selling price (MSP), 431 derived from pathway techno-economic analysis (TEA) for producing the bio-blendstock. These differences would 432 represent the potential shared benefits between the bio-blendstock producer and petroleum refiner. Alternatively, 433 calculating the price premiums, over a benchmark gasoline, needed to incentivize the refinery to purchase bio- 434 blendstocks at their TEA derived MSPs, combined with the anticipated efficiency gains expected from Co-Optima 435 engines, would yield an even more comprehensive picture of each bio-blendstock’s economics. 436 437 2. Integration with SAFs/HD/petrochemical plants: The biggest impact of this work lies in the potential for integration of 438 drop-in biofuels not just for the gasoline pool, but adopting the same approach for the entire product pool at a refinery. 439 This methodology can be applied for sustainable aviation fuels as well as heavy duty fuels, which can alternatively 440 route more fossil derived streams to petrochemicals plants such as BTX or other high value chemicals, ultimately 441 leading to maximum profitability and increased sustainability at a refinery. The potential for biofuel blending in the 442 heavy-duty market is the subject of companion a manuscript. 443 3. GHG Emissions Reduction: Future work will consider the life cycle greenhouse gas benefits of blending biofuels into 444 the gasoline and diesel-range fuel pools, by coupling ASPEN PIMS refinery simulation with life cycle assessment 445 (LCA). This novel perspective accounts for energy use and emissions in the foreground refinery complex, such as utility 446 requirements and process energy use and emissions across individual refinery processing trains, as well as background 447 changes in the supply chains of purchased goods consumed at the refinery over time (e.g., changes in the carbon intensity 448 of purchased electricity, crude oil supply, etc.). Life cycle impacts can be aggregated refinery-wide, or shown for 449 individual refinery products, and benchmarked against a ‘reference’ refinery case to contextualize the potential 450 emissions benefits. Further, this approach can track environmental impacts over time and across a wide array of 451 sustainability and climate metrics, and thus provides a broad-based understanding of the impact of key parameters such 452 as changes in demand for refinery products, crude oil price, biofuel blending level, etc. on the environmental 453 sustainability of refinery operations. 454 4. Co-processing as an avenue for more opportunities: A larger refinery-wide optimization with the objective of lowering 455 the carbon footprint as well as maximizing the profitability of the refinery can be accomplished. In this case, we are 456 addressing biofuels blending alone, but coprocessing with pyrolysis oil or algal/terrestrial hydrothermal liquefaction oil 457 can be other potential avenues for bringing bio-derived intermediates into a refinery and upgrading them into finished 458 fuels. This combined system-wide optimization can potentially be done in the near term while the refinery focuses on 459 meeting increasing demands of diesel and jet fields. The California Air Resources Board is actively looking into methods 460 by which they can quantify the biogenic content of the finished fuel from coprocessing, creating markets for new LCFs 461 applications. 462 463 5. Policy Factors for Adoption: Market dynamics such as policy credits whether they are low carbon fuel standard in California or federal RIN credits in the US can foster a market for these biofuels and enable faster biofuel adoption. 464 California alone has been a big draw for renewable diesel similarly policies as they evolve in other countries or other 465 parts of the world especially in Europe can also then able liquid fuels that have a lower carbon footprint than the current 466 100% fossil derived refinery fuels. Other countries also have policies in effect that encourage more drop-in biofuels for 467 blending. 468 469 5 470 This work demonstrates that Co-Optima initiatives to co-design specialty engines and fuels has incentives for the refining 471 industry as well as gasoline consumers. In the case of Co-Optima BSI bio-blendstocks, refineries can leverage bio-blendstocks 472 with superior properties to produce premium grade gasoline that benefits the environment. This value proposition, taken with 473 the chance for refiners to take part in a more circular and diversified feedstock economy, gives multiple reasons for refiners to 474 consider partaking in a low carbon fuel market. Moreover, this analysis is one example of the framework to evaluate the value 475 propositions of integrating biofuels and the refiners facing dynamic economic and market conditions can produce specialty, 476 high-value fuels to stay competitive. Conclusion 477 478 Conflicts of interest 479 No conflict of interest to declare 480 Data Availability 481 For any data requests, please reach out to the corresponding authors: Avantika.Singh@nrel.gov and 482 Nicholas.Carlson@nrel.gov 483 Abbreviations 484 BSI Boosted Spark-Ignition 485 GHG Greenhouse Gas 486 CAFÉ Corporate Average Fuel Economy 487 RON Research Octane Number 488 MON Motor Octane Number 489 AKI Anti-Knock Index 490 S Sensitivity 491 EPA Environmental Protection Agency 492 ETC Ethanol 493 IPR Iso-propanol 494 NPR N-propanol 495 IBU Iso-butanol 496 DII Di-iso-butylene 497 CYC Cyclopentanone 498 FUR Furans Mixture 499 BOB Before Oxygenate Blend 500 RBOB Reformulated Before Oxygenate Blend 501 RVP Reid Vapor Pressure 502 LCA Life Cycle Assessment 503 PIMS Process Industry Modeling System 504 PADD Petroleum Administration for Defense Districts 505 LPG Liquefied Petroleum Gas 506 AEO Annual Energy Outlook 507 ADOPT Automotive Deployment Options Projection Tool 508 NREL National Renewable Energy Laboratory 509 BPD Barrels-per-day 510 WTI West-Texas Intermediate 511 OPIS Oil Price Information Service 512 API American Petroleum Institute 513 TAN Total Acid Number 514 BTX Benzene, Toluene, Xylenes 515 PCA Principled Component Analysis 516 BEV Break-Even Value 517 PIONA N-paraffins, Iso-paraffins, Olefins, Naphthenes, and Aromatics 518 MSP Minimum Selling Price 519 TEA Techno-Economic Analysis 520 BBS Bio-Blendstock 521 DOE Department of Energy 522 LSR Light Straight Run Naphtha 523 FCC Fluid Catalytic Cracking 524 ASTM American Society for Testing and Materials 525 526 Acknowledgements 527 The authors gratefully acknowledge funding support from the Co-Optimization of Fuels & Engines (Co-Optima) project 528 sponsored by the U.S. Department of Energy Office of Energy Efficiency and Renewable Energy, Bioenergy Technologies and 529 Vehicle Technologies Offices. This work was authored in part by the National Renewable Energy Laboratory, operated by 530 Alliance for Sustainable Energy, LLC, for the U.S. Department of Energy (DOE) under Contract No. DE-AC36-08GO28308. 531 The views expressed in the article do not necessarily represent the views of the DOE or the U.S. Government. The U.S. 532 Government retains and the publisher, by accepting the article for publication, acknowledges that the U.S. Government retains 533 a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this work, or allow 534 others to do so, for U.S. Government purposes. 535 Additionally, we would like to acknowledge Amie Sluiter for providing thoughtful comments in her review of this work. 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 Appendix A 552 Table A.1. Primary gasoline blendstocks and their typical RON values, volume share, and sources [36] Abbreviations used- 553 LSR: Light straight run; FCC: Fluid Catalytic Cracking; RVP: Reid Vapor Pressure, RON: Research Octane Number, MON: 554 Motor Octane Number. Description Typical Typical Sources RON Volume Refinery- Range Share (%) Produced β Purchased β Refinery Process Comments Crude Distillation, High RVP- limited Catalytic Cracking during summer 92 2-4 60-70 2-3 LSR Naphtha 60-70 8-10 β Crude Distillation Alkylate 93-96 10-12 β Alkylation Isomerate 82-86 5-7 β Isomerization Potential RVP issues Large RON/MON 23-25 β Catalytic 90-100 Reforming spread Aromatics Higher value in Extraction petrochemical market Butane β Natural Gasoline Reformate (full range) Toluene 114 <1 FCC Naphtha 89-95 33-35 β β Main gasoline Catalytic Cracking component (full range) Light 85-88 <5 117 10 β Hydrocracking Hydrocrackate Ethanol 555 556 β Limited at 10 vol% 557 Table A.2. Refinery configurations applied for Co-Optima refinery impact and bio-blendstock valuation analysis for PADD3 558 representative refineries [18]. Nelson Complexity Index is calculated using the methods shown in [19]. Symbols: * denotes 559 that * not applied in BSI bio-blendstock analysis. Abbreviations: kbbl/day is the volume of thousand barrels crude oil input to 560 the refinery each day. The minimum turndown capacities for each unit operation are set to 50% of the design capacity. Unit Operation PADD3 PADD3 PADD3 Large Medium* Small Atmospheric Distillation 400 160 45 Vacuum Distillation 220 87 9.7 Fluid Catalytic Cracking 142 85 Delayed Coking 105 46 Hydrocracking 49 Gas Oil Hydrotreating 72 72 Diesel Hydrotreating 96 101 Jet/Kero Hydrotreating 65 Gasoline Hydrotreating 38 46 Naphtha Hydrotreating 116 54 13.4 Naphtha Reforming 108 52 9.3 Aromatics/Isomerization 30 Alkylation/Polymerization 39 (Capacities in kbbl / Day) 561 562 563 564 565 566 12.8 6 15 567 Table A.3. Finished fuel property specifications applied in Co-Optima BSI impact analysis from American Society for Testing 568 and Materials (ASTM) [24]. Spec Conventional Conventional Reformulated Reformulated Co-Optima Type Regular Premium Regular Premium Boosted SI Max 9 9 7.8 7.8 7.8 Blend Property RVP (psi) Research Octane Min 98 Number, RON Anti-Knock Index Min 87 93 87 93 Octane Sensitivity, Min 10 RON – MON 50 Vol % Evap, °F (°C) Min 150 (66) 150 (66) 150 (66) 150 (66) 150 (66) Sulfur, wppm Max 10 10 10 10 10 569 570 Table A.4. Blending properties applied for Co-Optima BSI bio-blendstocks from Fuels Properties Database [18]. Bio-Blendstock Blend Vol.% Blending MON Blending RON Blending S Blending RVP (kPa) 10% 106.1 132.8 26.8 116.0 20% 112.7 145.9 33.2 116.7 30% 102.4 128.4 26.0 68.1 Neat BBS 90.0 109.0 19.0 18.6 10% 102.8 117.2 14.4 88.5 20% 109.4 127.7 18.3 85.8 30% 102.5 120.6 18.2 54.8 Neat BBS 96.7 112.5 15.8 14.6 10% 98.3 117.4 19.1 74.0 20% 102.7 127.4 24.7 69.5 30% 94.8 118.7 23.9 48.0 Neat BBS 89.0 104.0 15.0 9.7 10% 95.1 110.3 15.1 53.6 Ethanol Iso-propanol N-propanol Iso-butanol 20% 99.0 118.4 19.5 46.6 30% 93.6 113.0 19.4 38.4 Neat BBS 93.0 107.0 14.0 1.7 10% 99.5 113.4 13.9 33.9 20% 103.1 123.1 19.9 19.8 30% 93.3 117.3 24.1 16.2 Neat BBS 87.0 106.0 19.0 11.0 10% 97.8 108.1 10.3 53.6 20% 101.6 117.4 15.7 43.8 30% 93.4 115.8 22.3 28.3 Neat BBS 89.0 101.0 12.0 1.5 10% 100.7 131.4 30.7 41.6 20% 104.3 141.3 37.0 34.4 30% 93.7 120.5 26.8 36.5 Neat BBS 91.6 109.0 17.4 7.8 Diisobutylene Cyclopentanone Furan Mix 571 572 573 Table A.5. Common refinery sourced gasoline blendstock sensitivity values [37]. Gasoline Blendstock 574 575 576 Research Octane Motor Octane Sensitivity Number (RON) Number (MON) (RON – MON) Butane 93.8 89.6 4.2 Hydrotreated light straight run naphtha 50-65 55-60 -5-5 Full-range reformate 94-104 79-92 12-22 Full-range fluidized catalytic cracking gasoline 90-93 79-81 11-12 Alkylate 95 93 2 Isomerate 84 81 3 577 Table A.6. Refinery product demand projections from Energy Information Administration’s (EIA) Annual Energy Outlook 578 (AEO) and the National Renewable Energy Laboratory’s (NREL) Automotive Deployment Options Projection Tool (ADOPT). Year 579 580 581 582 EIA (Billion Gallons / Year) Gasoline Jet / LPG BOBs Kero Distillates Residuals ADOPT (Billion Gallons / Year) Gasoline Jet / LPG BOBs Kero Distillates Residuals 2020 48.65 133.28 30.30 67.65 4.72 48.65 129.12 30.30 67.65 4.72 2021 49.46 132.20 30.16 66.97 4.62 49.46 128.12 30.16 66.97 4.62 2022 50.53 130.39 29.95 67.15 3.95 50.53 126.97 29.95 67.15 3.95 2023 51.54 127.96 29.98 66.48 4.81 51.54 125.70 29.98 66.48 4.81 2024 52.28 125.41 30.13 66.30 4.88 52.28 124.62 30.13 66.30 4.88 2025 52.80 122.79 30.39 66.03 4.92 52.80 123.52 30.39 66.03 4.92 2026 53.22 120.78 30.63 65.97 4.72 53.22 122.24 30.63 65.97 4.72 2027 53.62 119.05 30.89 65.63 4.65 53.62 121.01 30.89 65.63 4.65 2028 54.10 117.52 31.19 65.48 4.34 54.10 119.75 31.19 65.48 4.34 2029 54.58 116.14 31.47 65.13 4.32 54.58 118.30 31.47 65.13 4.32 2030 55.16 114.97 31.76 64.63 4.65 55.16 116.68 31.76 64.63 4.65 2031 55.65 113.98 32.05 64.36 4.62 55.65 114.94 32.05 64.36 4.62 2032 56.11 112.94 32.35 64.07 4.61 56.11 112.99 32.35 64.07 4.61 2033 56.55 111.99 32.64 63.82 4.57 56.55 110.97 32.64 63.82 4.57 2034 57.02 111.08 32.92 63.69 4.56 57.02 108.82 32.92 63.69 4.56 2035 57.52 110.20 33.19 63.57 4.51 57.52 106.61 33.19 63.57 4.51 2036 57.90 109.51 33.46 63.56 4.20 57.90 104.40 33.46 63.56 4.20 2037 58.29 108.92 33.74 63.39 4.17 58.29 102.20 33.74 63.39 4.17 2038 58.67 108.44 34.00 63.29 4.07 58.67 100.09 34.00 63.29 4.07 2039 59.10 108.07 34.28 63.18 4.02 59.10 98.02 34.28 63.18 4.02 2040 59.56 107.87 34.58 63.25 3.94 59.56 96.00 34.58 63.25 3.94 2041 59.98 107.76 34.88 63.26 3.99 59.98 93.99 34.88 63.26 3.99 2042 60.52 107.76 35.19 63.39 3.81 60.52 92.07 35.19 63.39 3.81 2043 60.96 107.85 35.52 63.53 3.77 60.96 90.19 35.52 63.53 3.77 2044 61.47 108.02 35.86 63.76 3.63 61.47 88.40 35.86 63.76 3.63 2045 61.98 108.22 36.22 63.90 3.71 61.98 86.63 36.22 63.90 3.71 2046 62.58 108.53 36.60 64.17 3.55 62.58 84.91 36.60 64.17 3.55 2047 63.08 108.93 36.99 64.38 3.52 63.08 83.23 36.99 64.38 3.52 2048 63.57 109.34 37.38 64.54 3.50 63.57 81.69 37.38 64.54 3.50 2049 64.18 109.79 37.79 64.73 3.46 64.18 80.35 37.79 64.73 3.46 2050 64.58 110.31 38.20 64.83 3.42 64.58 79.05 38.20 64.83 3.42 583 0.10 Predicted Premium Spread ($/gal) 0.09 0.08 0.07 0.06 0.05 0.04 0.03 R2 = 0.97 0.02 0.01 0.00 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10 Historical Premium Spread ($/gal) 584 585 Fig A.1. Parity plot of predicted versus actual premium gasoline spreads dating between 1994 and 2019. 140 Break Even Value ($/BBL) 120 100 y = -0.3114x + 120.07 R² = 0.1675 80 y = -0.6757x + 146.92 R² = 0.7153 60 40 10% y = 4.0683x - 388.6 R² = 0.5886 20% 20 30% 0 100 586 105 110 115 120 125 130 135 Blending Octane (RON + MON)/2 587 Fig A.2. Break even values for seven candidate BSI bio-blendstocks, at 10%, 20%, and 30% blending levels plotted 588 against corresponding blending octanes. CYC: cyclopentanone, DII: diisobutylene, ETC: ethanol, FUR: furans 589 mixture, IBU: isobutanol, IPR: isopropanol, NPR: n-propanol. 590 140 Break Even Value ($/bbl) 120 100 y = 0.1068x + 81.157 R² = 0.0234 10% 80 60 20% 40 30% y = 0.3549x + 65.989 R² = 0.1118 y = 4.2709x - 23.188 R² = 0.8889 20 0 5 591 10 15 20 25 30 35 40 Blending Sensitivity (RON - MON) 592 Fig A.3. Break even values for seven candidate BSI bio-blendstocks, at 10%, 20%, and 30% blending levels plotted 593 against corresponding blending sensitivities. CYC: cyclopentanone, DII: diisobutylene, ETC: ethanol, FUR: furans 594 mixture, IBU: isobutanol, IPR: isopropanol, NPR: n-propanol. 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 Appendix B 612 Example A: Crude Oil Price Calculation from Benchmark 613 The following data was collected from both the Bloomberg and Baker & Obrien databases. The prices in timestep of both 614 Cabinda and West Texas Intermediate (WTI) crude oils were plotted against one another revealing a high degree of correlation 615 enabling an expression of Cabinda price as a simple linear function of WTI price. 140 y = 1.22x - 6.55 R² = 0.95 Cabinda ($ / BBL) 120 100 80 60 Bloomberg Terminal 40 Baker & Obrien Database 20 0 30 40 50 60 70 80 WTI ($ / BBL) 90 100 110 616 617 Fig B.1. Cabinda crude historical prices plotted with reference to West Texas Intermediate (WTI) crude prices. 618 Linear regression dictated a slope (m) of 1.22 with intercept (b) -6.55. Therefore, given a WTI price of 60 $/bbl, the price of 619 Cabinda could be calculated as: ππΆππππππ = 1.22 β 60 620 $ $ − 6.55 = 67 πππ πππ 621 Example B: Crude Oil Price Calculation from Properties 622 Some crude oils, such as Escalante, had little to no pricing data available in the multiple resources used for this analysis. In 623 order to price these crude oils, a multiple linear regression model of crude oil price was fit as a function of API gravity, sulfur 624 weight percent, and total acid number (TAN). Using the fitted parameters of this model, a price for Escalante given a 625 benchmark West Texas Intermediate price of 60 $/bbl could be calculated using the property data below: 626 API Gravity = 23.2° 627 Sulfur = 0.16 wt% 628 TAN = 0.50 ππΈπ ππππππ‘π = 0.22(23.2π΄ππΌ ) − 1.71(0.16ππ’πππ’π ) − 1.11(0.61 ππ΄π ) − 13.86 = 72.8 629 $ πππ 630 Example C: Refinery Product Price Calculation from Benchmark 631 Refinery product pricing data was also collected in timestep with West Texas Intermediate (WTI) crude oil prices. The 632 following RBOB data was collected from Oil Price Information Service (OPIS), Bloomberg, Energy Information 633 Administration (EIA), and Baker & Obrien datasets. 3.5 RBOB ($ / Gal) 3.0 y = 0.03x + 0.24 R² = 0.91 2.5 2.0 1.5 1.0 OPIS Bloomberg EIA PRISM 0.5 0.0 20 40 60 WTI ($ / BBL) 80 100 634 635 Fig B.2. Reformulated Before Oxygenate Blending (RBOB) gasoline blend historical prices plotted with reference to 636 West Texas Intermediate (WTI) crude prices. 637 Linear regression dictated a slope (m) of 0.03 with intercept (b) 0.24. Therefore, given a WTI price of 60 $/bbl, the price of 638 RBOB could be calculated as: ππ π΅ππ΅ = 0.03 β 60 639 $ $ + 0.24 = 1.82 πππ πππ 640 Example D: Octane Price Calculation 641 A neural network with the format below was developed to predict the incremental price of octane ($/octane-gallon) based on 5 642 independent variables being conventional regular gasoline price ($/gal), benzene to regular gasoline price differential 643 ($/octane-gallon), toluene to regular gasoline price differential ($/octane-gallon), xylenes to regular gasoline price differential 644 ($/octane-gallon), and total U.S. octane demand (octane-gal/day). These five input variables were ultimately determined by the 645 West Texas Intermediate (WTI) benchmark price, refinery product demand model (ADOPT or EIA), and the year in the 646 following way: 647 Assuming: 648 WTI Price = 60 $/bbl 649 Demand Projection = EIA Assumptions 650 Year = 2030 651 1) Conventional Regular Gasoline Price: 652 Given a benchmark WTI price of 60 $/bbl and using the method described in Example C with linear regression 653 yielding slope (m) of 0.03 and intercept (b) of 0.27 the price of conventional regular gasoline octane with AKI of 87 654 can be calculated as: 655 πππππ£−πππ πππ πππππ = 0.03 β 60 656 2) Benzene to Conventional Regular Gasoline Price Differential: $ $ + 0.27 = 1.81 πππ πππ 657 The price of benzene is calculated given a benchmark WTI price of 60 $/bbl and linear regression coefficients of slope 658 (m) of 0.04 and intercept (b) of 0.50 as: πππππ§πππ = 0.04 β 60 659 660 Implying, πππππ§πππ−ππππ = 661 662 $ $ + 0.50 = 2.68 πππ πππ πππππ£−πππ πππ πππππ πππππ§πππ 2.68 1.81 $ − = − = 0.0026 π΄πΎπΌππππ§πππ π΄πΎπΌππππ£−πππ πππ πππππ 115 87 πππ‘ β πππ 3) Toluene to Conventional Regular Gasoline Price Differential: 663 The price of toluene is calculated given a benchmark WTI price of 60 $/bbl and linear regression coefficients of slope 664 (m) of 0.03 and intercept (b) of 0.32 as: πππππ§πππ = 0.03 β 60 665 666 667 668 $ $ + 0.32 = 2.32 πππ πππ Implying, ππ‘πππ’πππ−ππππ = πππππ£−πππ πππ πππππ ππ‘πππ’πππ 2.32 1.81 $ − = − = 0.0016 π΄πΎπΌπ‘πππ’πππ π΄πΎπΌππππ£−πππ πππ πππππ 104 87 πππ‘ β πππ 4) Xylenes to Conventional Regular Gasoline Price Differential: 669 The price of xylenes is calculated given a benchmark WTI price of 60 $/bbl and linear regression coefficients of slope 670 (m) of 0.04 and intercept (b) of 0.35 as: πππππ§πππ = 0.04 β 60 671 672 Implying, ππ‘πππ’πππ−ππππ = 673 674 $ $ + 0.35 = 2.48 πππ πππ πππππ£−πππ πππ πππππ ππ‘πππ’πππ 2.48 1.81 $ − = − = 0.0003 π΄πΎπΌπ‘πππ’πππ π΄πΎπΌππππ£−πππ πππ πππππ 118 87 πππ‘ β πππ 5) Total U.S. Octane Demand 675 Under EIA Assumptions, in year 2030 the U.S. gasoline demand is projected to be 31,500,000 gal/day with a premium 676 market share of 23%. Therefore, the octane demand can be calculated as: 677 π·πππ‘ ππππππ = 31,500,000 β (1 − 0.23) β 87π΄πΎπΌ + 31,500,000 β 0.23 β 92π΄πΎπΌ = 2,780,000,000 πππ‘ β πππ πππ¦ 678 679 Prior to entering the neural network, the input variables were centered by subtracting the mean, scaled by dividing the standard 680 deviation, and grouped into two principled components using principled component analysis (PCA) as show below 681 682 1) Center and scaling: πππππ£−πππ πππ πππππ, π πππππ = 1.81 − 1.52 = 0.36 = π1 0.81 683 684 π·πππ‘ ππππππ, π πππππ = 2,780,000,000 − 4,157,396,000 = −0.98 = π2 1,405,534,000 685 686 πππππ§πππ−ππππ, π πππππ = 0.0026 − 0.0019 = 1.54 = π3 0.0005 ππ‘πππ’πππ−ππππ, π πππππ = 0.0016 − 0.0011 = 1.10 = π4 0.0005 ππ₯π¦πππππ −ππππ, π πππππ = 0.0003 − (−0.0003) = 1.36 = π5 0.0004 687 688 689 690 691 692 2) Principled Component Reduction 693 After principled component analysis, a series of weights are determined that can applied to the original input variables 694 to form a series of principled components that are orthogonal, meaning they have no correlation. Two principled 695 components were calculated for the neural network: 696 697 698 ππΆ1 = 0.55 β π1 + 0.48 β π2 + 0.54 β π3 + 0.35 β π4 − 0.22 β π5 = 0.55 β 0.36 + 0.48 β 1.54 + 0.54 β 1.10 + 0.35 β 1.36 − 0.22 β (−0.98) = 2.23 699 700 701 ππΆ2 = 0.20 β π1 − 0.38 β π2 + 0.25 β π3 − 0.56 β π4 − 0.65 β π5 = 0.20 β 0.36 − 0.38 β 1.54 + 0.25 β 1.10 − 0.56 β 1.36 − 0.65 β (−0.98) = −0.36 702 703 After pre-processing, the principled components were fed to a neural network with five units in one hidden layer using a 704 sigmoidal transfer function. The hidden layer outputs are computed as follows: 705 706 707 β1 = −0.50 β ππΆ1 − 1.19 β ππΆ2 − 0.47 = −0.50 β 2.23 − 1.19 β (−0.36) − 0.47 = −1.15 → π ππ(β1 ) = 1 1+π −β1 = 1 1+π 1.15 = 0.24 708 709 710 β2 = −0.32 β ππΆ1 + 0.15 β ππΆ2 − 1.59 = −0.32 β 2.23 + 0.15 β (−0.36) − 1.59 = −2.36 → π ππ(β2 ) = 1 1 = = 0.09 −β2 1+π 1 + π 2.36 711 712 713 β3 = −0.62 β ππΆ1 − 1.51 β ππΆ2 − 0.13 = −0.62 β 2.23 − 1.51 β (−0.36) − 0.13 = −0.96 → π ππ(β3 ) = 1 1 = = 0.28 −β3 1+π 1 + π 0.96 714 715 716 β4 = −0.44 β ππΆ1 − 0.25 β ππΆ2 − 1.85 = −0.44 β 2.23 − 0.25 β (−0.36) − 1.85 = −2.74 → π ππ(β4 ) = 1 1 = = 0.06 −β4 1+π 1 + π 2.74 717 718 719 720 β5 = −0.50 β ππΆ1 − 0.88 β ππΆ2 − 2.00 = −0.50 β 2.23 − 0.88 β (−0.36) − 2.00 = −2.78 → π ππ(β5 ) = 1 1 = = 0.06 1 + π −β5 1 + π 2.78 721 Ultimately, the neural network output being the incremental octane price under the given economic conditions was computed 722 as: 723 π = −0.84 β π ππ(β1 ) + 0.22 β π ππ(β2 ) + 0.54 β π ππ(β3 ) − 0.45 β π ππ(β4 ) + 0.54 β π ππ(β5 ) + 0.07 724 = −0.84 β (−1.15) + 0.22 β (−2.36) + 0.54 β (−0.96) − 0.45 β (−2.74) + 0.54 β (0.06) + 0.07 725 = 0.04 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 $ πππ‘ β πππ 744 References 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] Sources of Greenhouse Gas Emissions | Greenhouse Gas (GHG) Emissions | US EPA. https://www.epa.gov/ghgemissions/sources-greenhouse-gas-emissions (accessed Nov. 23, 2020). H. Ritchie and M. Roser, “COβ and Greenhouse Gas Emissions,” Our World in Data, May 2020, Accessed: Jan. 13, 2022. [Online]. Available: https://ourworldindata.org/emissions-by-sector CO-OPTIMA RESEARCH, Energy.gov. https://www.energy.gov/eere/bioenergy/co-optima-research (accessed Jan. 13, 2022). J. Farrell, “Co-Optimization of Fuels & Engines (Co-Optima) Initiative: Recent Progress on Light-duty Boosted SparkIgnition Fuels/Engines,” p. 51. J. P. Szybist et al., “What fuel properties enable higher thermal efficiency in spark-ignited engines?,” Progress in Energy and Combustion Science, vol. 82, p. 100876, Jan. 2021, doi: 10.1016/j.pecs.2020.100876. J. B. Dunn et al., “Environmental, Economic, and Scalability Considerations and Trends of Selected Fuel EconomyEnhancing Biomass-Derived Blendstocks,” ACS Sustainable Chem. Eng., vol. 6, no. 1, pp. 561–569, Jan. 2018, doi: 10.1021/acssuschemeng.7b02871. Y. Jiang, S. D. Phillips, A. Singh, S. B. Jones, and D. J. Gaspar, “Potential economic values of low-vapor-pressure gasoline-range bio-blendstocks: Property estimation and blending optimization,” Fuel, vol. 297, p. 120759, Aug. 2021, doi: 10.1016/j.fuel.2021.120759. Ethanol Touts CAFE Proposal as Opening for High-Octane Fuels. https://www.dtnpf.com/agriculture/web/ag/blogs/ethanol-blog/blog-post/2018/08/02/ethanol-touts-cafe-proposal-opening (accessed Nov. 24, 2020). House Bill Would Set High-Octane Standard. https://www.dtnpf.com/agriculture/web/ag/news/article/2020/09/25/housebill-set-high-octane-standard (accessed Nov. 23, 2020). Understanding Octane Value in North America - ADI Analytics. http://adi-analytics.com/2016/06/20/understandingoctane-value-in-north-america/ (accessed Dec. 12, 2021). D. Cinquegrana, “Behind the Battle for Octane.” Oil Price Information Service (OPIS), an IHS Markit Company. [Online]. Available: http://info.opisnet.com/ebook-behind-the-battle-for-octane Control of Air Pollution From Motor Vehicles: Tier 3 Motor Vehicle Emission and Fuel Standards; Final Rule. R. L. McCormick et al., “Selection Criteria and Screening of Potential Biomass-Derived Streams as Fuel Blendstocks for Advanced Spark-Ignition Engines,” SAE Int. J. Fuels Lubr., vol. 10, no. 2, pp. 442–460, Mar. 2017, doi: 10.4271/201701-0868. D. J. Gaspar et al., “Top Ten Blendstocks Derived From Biomass For Turbocharged Spark Ignition Engines: Bioblendstocks With Potential for Highest Engine Efficiency,” PNNL-28713, 1567705, Sep. 2019. doi: 10.2172/1567705. Optimization | McKinsey Energy Insights. https://www.mckinseyenergyinsights.com/resources/refinery-referencedesk/optimization/ (accessed Dec. 17, 2020). Aspen PIMS | Trusted by 400 Over Refineries | AspenTech. https://www.aspentech.com/en/products/msc/aspen-pims (accessed Dec. 01, 2020). International Feedstocks Pricing Report, OPIS. https://www.opisnet.com/product/pricing/spot/international-feedstocksintelligence-report/ (accessed Dec. 12, 2021). Refinery Capacity Report. https://www.eia.gov/petroleum/refinerycapacity/ (accessed Dec. 01, 2020). D. Johnston, “Complexity index indicates refinery capability, value,” Oil and Gas Journal, vol. 94, no. 12, Mar. 1996, Accessed: Dec. 12, 2021. [Online]. Available: https://www.osti.gov/biblio/207885-complexity-index-indicates-refinerycapability-value EIA - Annual Energy Outlook 2019. https://www.eia.gov/outlooks/archive/aeo19/ (accessed Dec. 01, 2020). A. Brooker, J. Gonder, S. Lopp, and J. Ward, “ADOPT: A Historically Validated Light Duty Vehicle Consumer Choice Model,” Apr. 2015, pp. 2015-01–0974. doi: 10.4271/2015-01-0974. Co-Optimization of Fuels and Engines: Fuel Properties Database | NREL. https://www.nrel.gov/transportation/fuelsproperties-database/ (accessed Dec. 07, 2020). Global downstream outlook to 2035 | McKinsey. https://www.mckinsey.com/industries/oil-and-gas/our-insights/globaldownstream-outlook-to-2035 (accessed Dec. 12, 2021). D02 Committee, “Specification for Automotive Spark-Ignition Engine Fuel,” ASTM International. doi: 10.1520/D481420A. T. L. Alleman, A. Singh, E. D. Christensen, E. Simmons, and G. Johnston, “Octane Modeling of Isobutanol Blending into Gasoline,” Energy Fuels, vol. 34, no. 7, pp. 8424–8431, Jul. 2020, doi: 10.1021/acs.energyfuels.0c00911. P. Ghosh, K. J. Hickey, and S. B. Jaffe, “Development of a Detailed Gasoline Composition-Based Octane Model,” Ind. Eng. Chem. Res., vol. 45, no. 1, pp. 337–345, Jan. 2006, doi: 10.1021/ie050811h. E. Christensen, J. Yanowitz, M. Ratcliff, and R. L. McCormick, “Renewable Oxygenate Blending Effects on Gasoline Properties,” Energy Fuels, vol. 25, no. 10, pp. 4723–4733, Oct. 2011, doi: 10.1021/ef2010089. 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 [28] J. E. Anderson, U. Kramer, S. A. Mueller, and T. J. Wallington, “Octane Numbers of Ethanol− and Methanol−Gasoline Blends Estimated from Molar Concentrations,” Energy Fuels, vol. 24, no. 12, pp. 6576–6585, Dec. 2010, doi: 10.1021/ef101125c. [29] D. J. Gaspar et al., “Measuring and predicting the vapor pressure of gasoline containing oxygenates,” Fuel, vol. 243, pp. 630–644, May 2019, doi: 10.1016/j.fuel.2019.01.137. [30] S. Aghahossein Shirazi, B. Abdollahipoor, J. Martinson, B. Windom, T. D. Foust, and K. F. Reardon, “Effects of dualalcohol gasoline blends on physiochemical properties and volatility behavior,” Fuel, vol. 252, pp. 542–552, Sep. 2019, doi: 10.1016/j.fuel.2019.04.105. [31] A. Landera, N. Mac Dowell, and A. George, “Development of robust models for the prediction of Reid vapor pressure (RVP) in fuel blends and their application to oxygenated biofuels using the SAFT-γ approach,” Fuel, vol. 283, p. 118624, Jan. 2021, doi: 10.1016/j.fuel.2020.118624. [32] T. J. Theiss et al., “Summary of High-Octane Mid-Level Ethanol Blends Study,” Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States), ORNL/TM-2016/42, Jul. 2016. doi: 10.2172/1286966. [33] Motor-gas-tech-review. Chevron Corporation, 2009. Accessed: Mar. 01, 2022. [Online]. Available: https://www.chevron.com/-/media/chevron/operations/documents/motor-gas-tech-review.pdf [34] O. US EPA, Gasoline Reid Vapor Pressure, Aug. 07, 2015. https://www.epa.gov/gasoline-standards/gasoline-reid-vaporpressure (accessed Jan. 13, 2022). [35] PEI Journal - 2nd Quarter 2017 - 48. https://digital.peijournal.org/peijournal/Q22017?pg=48&lm=1517078446000 (accessed Jan. 13, 2022). [36] The New Economics of Octane. Oil Price Information Service (OPIS). [Online]. Available: https://www.opisnet.com/product/news/exclusive-study-new-economics-octane/ [37] J. Farrell, R. Wagner, C. Moen, and D. Gaspar, “Transportation Future with Science in the Driver’s Seat: Mapping a Viable Route Forward for Affordable, Efficient, and Clean Fuels and Engines,” p. 12. Click here to access/download;Figure;Figure 1.JPG Click here to access/download;Figure;Figure 2.JPG Click here to access/download;Figure;Figure 3.JPG Click here to access/download;Figure;Figure 4.JPG Click here to access/download;Figure;Figure 5.JPG Click here to access/download;Figure;Figure 6.JPG Click here to access/download;Figure;Figure 7.JPG Click here to access/download;Figure;Figure 8.JPG Click here to access/download;Figure;Figure 9.JPG