REFINING ECONOMICS OF A SINGLE OCTANE. For. The Alliance of Automobile Manufacturers. MathPro Inc. P.O. Box West Bethesda, Maryland

Transcription

1 FINAL REPORT REFINING ECONOMICS OF A SINGLE OCTANE NATIONAL CLEAN GASOLINE STANDARD For The Alliance of Automobile Manufacturers by MathPro Inc. P.O. Box West Bethesda, Maryland October 8, 2010

2 Standard Final Report T A B L E O F C O N T E N T S 1. Introduction and Summary Proposed NCG Standard Objective of the Study Technical Approach Effects of NCG and Single Octane NCG Standards on Refining Operations Summary of Quantitative Results Organization of the Report 8 2. Some Key Terms and Concepts Measures of Refining Economics Used in This Report PADDs Refining Processes Information Used to Develop and Calibrate the Refining Models Projected Refinery Operations for Technical Approach for Regional Refinery Modeling Scope of the Analysis Key Assumptions and Premises Refined Product Volumes Held Constant Energy and CO 2 Accounting Framework in the Refining Models U.S. Refinery Energy Use Conceptual Framework for Estimating Refinery Energy Use Energy Accounting in the Refining Models Comparison with EIA Framework for Reporting U.S. Refinery Energy Use CO 2 Emissions Accounting in the Refining Models Results of the Calibration, Reference, and Study Cases Primary Effects of the NCG and the 94 RON and 95 RON NCG Standards Detailed Modeling Results Calibration of the Regional Models Summary of the Effects of NCG and Single Octane NCG on Refining Operations Additional Comments Interpreting the Estimated Refining Costs Refinery-to-Refinery Differences Affect Responses to New Standards Holding Refined Product Volumes Constant in the Study Cases Over-Optimization The Analysis Focuses on the Current U.S. Refining Sector Regional Average Octane Values in the Reference Cases 37 October 8, 2010 i

3 Standard Final Report 1. INTRODUCTION AND EXECUTIVE SUMMARY This report describes a study conducted by MathPro Inc. to analyze the technical and economic effects on the U.S. refining sector of a national clean gasoline (NCG) standard with a single octane grade. The single octane NCG standard would augment the federal and California standards for reformulated gasoline and would replace the standards for all special gasolines ( boutique fuels ) and for conventional gasoline (CG). 1.1 Proposed NCG Standard A single octane NCG standard would consist of a set of gasoline property standards in combination with a single octane standard of 94 or 95 RON. For purposes of this study, single octane NCG is defined as shown in Table 1.1. Table 1.1: Proposed Standards for Single Octane NCG Property Units Spec. T yp e Comments Octane Octane RON 94 Min With alternative standard of 95 RON Sensitivity RON - MON 8 Min Complex Model Properties RVP (Summer season) psi 7 Sulfur ppm 10 Max Corresponds to 6.8 psi at the refinery gate Max Corresponds to 5 ppm at the refinery gate Benzene vol.% 0.62 Avg The new federal MSAT2 standard Aromatics vol.% 10 Min Olefins vol.% ---- No Olefins limit for NCG E200 vol.% ---- No E200 requirement for NCG E300 vol.% ---- No E300 requirement for NCG Oxygen wt% ---- No oxygen required, but the entire U.S. gasoline pool is assumed to be E10 Ethanol Content vol% 10 Consistent with renewable fuel mandates Driveability Index 1250 Max Represented as 1220 average at the refinery gate Note: Average winter RVP and all other gasoline properties as specified in ASTM D NCG would have lower summer RVP and sulfur content than allowed under current regulatory standards and also would have to meet a Driveability Index standard. 1 NCG still would have to 1 The Driveability Index (DI) standard shown in Table 1.1 applies to the DI definition recently adopted by ASTM (and as specified in a recent report issued by the Alliance of Automobile Manufacturers), namely: DI ( F) = 1.5*T *T *T *(Vol% Ethanol). The last term in the equation adjusts the DI upwards by about 24 numbers when ethanol is blended at 10%, to reflect ethanol s observed adverse effects on driveability. October 8,

4 Standard Final Report meet requirements imposed by the Complex Model in federal RFG areas and by the Predictive Model in California. Single octane NCG would replace the current three gasoline grades regular, mid-grade, and premium, each with its own octane specified in terms of the average of their RON and MON (usually referred to as (R+M)/2)) with one grade having at least 94 RON and with octane sensitivity (the difference between MON and RON) of at least 8 numbers. Table 1.2 shows current national averages for the RON, MON, and (R+M)/2 for the premium and regular grades and for the entire U.S. gasoline pool. A 94 RON NCG standard would increase the U.S. gasoline pool s average RON by about 1 number, and would increase the pool average (R+M)/2 by a maximum of 2 numbers, depending on its effects on octane sensitivity. Table 1.2: Estimated U.S. Average Gasoline Grade Splits and Octane, 2008 Grade Split RON MON (R+M)/2 Pool Premium 10.7% Regular 89.3% Moving from the current U.S. gasoline pool to a single octane NCG would have significant effects on the technical and economic performance of the U.S. refining sector, including increases in (1) refining costs, (2) refinery investment in new process capacity, (3) crude oil use per barrel of gasoline output, (4) refinery energy use, and (5) refinery emissions of CO Objective of the Study The objective of this study was to develop estimates of the effects of the proposed NCG and single octane NCG standards (both at 94 RON and 95 RON) on U.S. refining economics and operations, by region and season. The following sections discuss the technical approach adopted in this study for assessing such effects and the quantitative results of our analysis. 1.3 Technical Approach We conducted the analysis by means of regional refinery LP modeling, using our proprietary refinery modeling system. We developed four refining models, representing aggregate refining October 8,

5 Standard Final Report operations in PADD 1 (the East Coast), PADD 2 (the Mid-west), PADD 3 (the Gulf Coast), and California. 2 The study does not consider PADD 4 (the Mountain States) or PADD 5 ex California (the West Coast), because (1) the total gasoline volume produced and consumed in that region is small and (2) unique issues involving small refineries arise in those areas. The refinery modeling encompassed: (1) Calibration cases for 2007, (2) Reference cases for 2015, and (3) NCG and single octane NCG Study cases for The Calibration cases served to calibrate the regional refining models to reported data for The Reference cases established the baseline for the analysis, in which each regional refining model produces projected 2015 refined product volumes with regulatory business as usual. 3 The Reference cases reflect regional refined product volumes derived from national projections reported by EIA in AEO 2009 (April Revision) and a world crude oil price of about $100/b. 4 The Study cases assessed the progressive effects (relative to the baseline) of imposing the NCG and single octane NCG standards on the refining sector. The modeling was conducted for both the summer and winter seasons. The refinery modeling was designed so that refining investments in new process capacity in the Reference, NCG, and single octane NCG cases all are made independently with respect to baseline 2007 refining capacity. This represents investment planning with perfect foresight regarding future regulatory standards for gasoline. 5 We estimated the regional average effects of the NCG standard and the 94 RON and 95 RON single octane NCG standards by comparing each 2015 Study case with its corresponding 2015 Reference case (not the 2007 Calibration case). 2 PADDs (Petroleum Administration for Defense Districts) are geographical regions of the U.S.; they are defined in Section 2. 3 By business as usual we mean regulations now in place or scheduled to be in place by 2015, but no new ones. 4 EIA refers to the U.S. Department of Energy s Energy Information Administration. AEO refers to the Annual Energy Outlook, an EIA publication that provides long-term projections of U.S. energy production, use, and prices. 5 In preliminary modeling for PADD 3, we also developed a series of cases in which refining investment was sequential; i.e., refining investments made in the Reference case for major refining processes were carried forward to the NCG case; any additional investments made in the NCG case were then carried forward to the single octane NCG case. Because the modeling results with sequential investment were similar to those with foresighted investment, we adopted the assumption of perfect foresight in all subsequent refinery modeling. October 8,

6 Standard Final Report 1.4 Effects of NCG and Single Octane NCG Standards on Refining Operations NCG Standard The primary changes in gasoline properties required to meet the NCG standard would be: (1) a reduction in the RVP of conventional gasoline (CG) in the summer (from 9 psi to 7 psi), and (2) a reduction in the sulfur content of all gasoline (from 30 ppm average to 10 ppm max). Refineries would meet the NCG standard primarily by (1) reducing RVP through additional debutanization and depentanization of gasoline blendstocks (primarily straight run naphtha and FCC naphtha) and (2) further reducing the sulfur content of FCC naphtha, via post-treatment. 6 Both responses, all else constant, would reduce gasoline pool octane (butanes have high octane and more severe desulfurization of FCC naphtha reduces its octane). Removal of butanes and pentanes from the gasoline pool also would reduce gasoline pool volume. To replace the lost volume and octane refineries could (1) increase reformer throughput and severity (which also would increase volume loss); (2) increase FCC conversion to increase FCC naphtha output; (3) produce relatively more gasoline-boiling range material in hydrocracking operations; and (4) increase crude throughput. The NCG standard would have greater effects on refining operations in PADDs 1-3 than in California, because gasoline produced by refineries in PADDs 1-3 in 2015 would have, on average, higher RVP and sulfur content than gasoline produced by California refineries. 94 and 95 RON NCG Standards A 94 RON single octane standard would raise the national gasoline pool s RON by about 1 number and its (R+M)/2 by a maximum of 2 numbers; a 95 RON single octane standard would raise the national gasoline pool s RON by about 2 numbers and its (R+M)/2 by a maximum of 3 numbers. Refineries would meet these additional octane requirements primarily by modifying the operations of their upgrading (octane-generating) processes (1) increasing reformer throughput and severity (as measured by the RON of the produced reformate), with resultant increases in the concentration of reformate in the gasoline; (2) upgrading straight run naphthas via pen/hex isomerization; (3) possibly increasing production of alkylate; and (4) removing from the gasoline pool C5s that have relatively high RVP and low octane. 7 6 FCC naphtha is a key gasoline blendstock. Most of the sulfur in gasoline comes from FCC naphtha. 7 C5s are gasoline blendstocks with relatively high RVP and low octane; hence, removing them from the gasoline pool improves both average gasoline pool RVP and octane (but with volume loss). The octane of straight run naphthas may be improved significantly (with small volume loss) via a refining process known as pen/hex isomerization. Alkylate is a high-octane gasoline blendstock. October 8,

7 Standard Final Report These operational changes (with constant refined product out-turns) would require refineries to increase crude oil throughput to offset the volume losses associated with the removal of lowoctane blendstocks from the gasoline pool and with expanded reforming operations. They also would result in the production of gasoline with: (1) a heavier distillation curve (that is, higher DI); (2) increased aromatics content; (3) higher energy density (which translates into greater fuel economy); and (4) greater octane sensitivity. 8 The effects of the NCG and single octane standards on refining operations would be more pronounced in the summer than in the winter, because refineries are more octane-constrained in the summer than in the winter higher RVP limits in the winter allow blending more highoctane C4s in gasoline. 1.5 Summary of Quantitative Results Table 1.3 summarizes the primary results of our analysis for PADDs 1-3, California, and the total of PADDs 1-3 and California. (We report California separately because the state s unique gasoline standards significantly increase the cost of the single octane standard.) Table 1.3: Estimated Effects of NCG and Single Octane NCG Standards on Refining Cost and on Refinery Crude Oil Use, Energy Use, and CO2 Emissions PADDs 1-3 California Total Single Octane Single Octane Single Octane 94 RON 95 RON 94 RON 95 RON 94 RON 95 RON NCG NCG NCG NCG NCG NCG NCG NCG NCG Effects on Costs Refining Cost Total ($B/y) Capital charge & fixed cost All other Per gallon of gasoline ( /gal) Value of Improved Gasoline Energy Density ( /gal) Refinery Investment ($B) Effects on Operations Change in Crude Oil Input per Barrel of Gasoline (b/b) Percent Increase in Refinery Energy Use (Fuel & Power) 3% 5% 9% 0% 5% 12% 3% 5% 9% CO2 Emissions 2% 4% 6% 0% 3% 9% 2% 4% 6% The top part of the table shows the effects of the NCG and single octane NCG standards on total refining costs, refining costs per gallon of gasoline, the offsetting value of the improved energy 8 Reformate has higher octane sensitivity than almost any other gasoline blendstock. Hence, expanding reformate s share of the gasoline pool increases not only the gasoline pool s RON but also its octane sensitivity. October 8,

8 Standard Final Report density of gasoline, 9 and refinery investments. The bottom part of the table provides estimates of certain physical effects of the changes in refining operations needed to produce NCG and single octane NCG specifically, changes in crude oil inputs per barrel of gasoline output, and increases in refinery energy use and CO 2 emissions. We assessed the effects of the NCG and single octane NCG standards on the refining sector by comparing each of the Study cases to a corresponding 2015 Reference case that represents regulatory business as usual. Thus, the numbers in Table 3 are estimates of the additional refining costs and the changes in refining operations that would arise from meeting each of the gasoline standards NCG, 94 RON NCG, and 95 RON NCG. Effects on Refining Economics Estimated total refining costs (PADDs 1-3 and California) increase progressively from about $3½ billion per year for the NCG standard to about $7½ billion and $12½ billion, respectively, for 94 RON and 95 RON single octane NCG standards. The corresponding increases in average refining costs that is, total refining cost divided by total gasoline production are about 3 /gal, 6 /gal, and 10 /gal, respectively. The increases in refining costs are somewhat offset by the estimated improvements in gasoline s energy density, valued at about ½ /gal, 1¼ /gal, and 2 /gal respectively. 10 The estimated average net costs of the gasoline standards, after adjusting for the value of improved energy density, are about 2½ /gal, 5 /gal, and 8 /gal for NCG, 94 RON NCG, and 95 RON NCG, respectively. 11 The marginal cost of gasoline production is highly sensitive to small changes in the volume of gasoline production, and it depends on the interplay of (i) domestic refining costs, (ii) the prospective supply of imported gasoline, and (iii) the price-sensitivity of domestic demand for gasoline. Analyzing this interplay was beyond the scope of the study. Rather, we held regional gasoline production volumes constant across the Reference and Study cases. In practice, the 9 As explained later in this report, the NCG and single octane NCG standards would increase the average energy density of gasoline, leading to corresponding improvements in fuel economy and reductions in gasoline use. 10 Improvements in gasoline energy density increase vehicle fuel economy, leading to a reduction in aggregate gasoline consumption by motorists at constant vehicle miles traveled. We estimated the value of improvements in gasoline energy density (in cents per gallon) as the percent improvement in energy density times the marginal refining cost of gasoline. Essentially, this is an estimate of the value of the resources (crude oil, power, fuel, etc.) saved in the refining sector from producing less gasoline. We have not included reductions in marketing and distribution costs associated with reduced gasoline consumption, as they are small relative to the refining cost of gasoline. Nor have we included savings obtained by motorists from paying less federal and state gasoline taxes and less marketing and distribution markups, because these are transfer payments (i.e., other parties government and private companies incur offsetting losses). 11 Net costs, as used in this report, may be thought of as social costs the cost of the additional resources society uses to meet the new gasoline standards. October 8,

9 Standard Final Report refining sector s response to the new gasoline standards likely would include reducing refined product out-turns to moderate increases in marginal refining costs. Estimated refining investment to support the NCG standard and 94 RON and 95 RON NCG standards is about $4¼, $5, and $8 billion, respectively. 12 The refining investments shown in Table 1 reflect grassroots investment economics, rather than expansion economics. Typically, the investment for expanding existing refining units is about half to two-thirds the investment for adding the same capacity in grassroots units. To the extent that refineries can expand existing units, rather than add grassroots units, our analysis may somewhat overstate the investment, and corresponding capital charges, associated with the new gasoline standards. Also, capital charges in the regional refinery models and in the cost analysis reflect an assumed after-tax rate of return on investment of 10%. Refiners often use higher rates of return for investment planning, but EPA typically uses lower rates of return (usually about 5%) when estimating the social (national) cost of fuel standards. Use of a 5% after-tax rate of return on investment instead of 10% would reduce estimated capital charges and fixed costs by about 25%. Capital charges and fixed costs represent about 50% of the refining cost of a NCG standard, but only about 20% of the 94 RON and 95 RON NCG standards. The bulk of the estimated refining cost of the 94 RON and 95 RON NCG standards arises from increased refinery energy use. Estimated costs for California have a different profile than those for PADDs 1-3. For PADDs 1-3, the average net cost of the gasoline standards, after adjusting for the value of improved energy density, would be about 2½ /gal, 5 /gal, and 7¼ /gal for the NCG and the 94 RON and 95 RON NCG standards, respectively. The incremental costs of the successive standards (e.g., the cost of moving from the NCG standard to the 94 RON NCG standard) are similar about 2½ /gal (including the value of improved gasoline energy density). For California, the estimated cost of the NCG standard is low, because most gasoline produced by California refineries in 2015 already will have low RVP and very low sulfur content. On the other hand, the estimated incremental cost of moving from NCG to 94 RON NCG is about 5¼ /gal (after adjusting for the value of improved gasoline energy density), significantly higher than the 2½ /gal estimated for PADDs 1-3. And the estimated incremental cost of moving from 94 RON NCG to a 95 RON NCG is even higher -- about 7½ /gal (after adjusting for the value of improved gasoline energy density). 12 The investment cost of the NCG standard could be lower than estimated, depending on how many FCC naphtha desulfurization units outside of California already are capable of controlling sulfur to 5 ppm. (We assume California refineries would have to produce ultra-low sulfur gasoline in the 2015 Reference case to satisfy state standards imposed through the newly-amended Predictive Model.) In this analysis, we assumed that all such units would have to be retrofitted to produce 5 ppm FCC naphtha. If none had to be retrofitted, our estimates of aggregate investments would be reduced by about $3½ billion in PADDs 1-3. This would reduce aggregate refining costs (capital charges and fixed costs) by about $1 billion/year and refining costs per gallon of gasoline by about 1 /gal. October 8,

10 Standard Final Report Refineries in California would incur higher costs to meet the 94 RON and 95 RON NCG standards because California s stringent gasoline standards, implemented through the Predictive Model, tightly constrain the blending space of California-compliant-gasoline; and because California s summer gasoline season is about 8 months long, whereas the summer season in PADDs 1-3 is about 6 months long. California refineries least cost responses to the 94 RON and 95 RON NCG standards would increase VOC, NOx, and toxics emissions, as calculated by the Predictive Model, because they would increase gasoline s aromatics content and raise its T50 and T90 (the temperatures at which 50% and 90% of the gasoline boils off). California refineries would be forced to moderate such property changes or offset their effects by changing other gasoline properties in order to produce California-compliant gasoline. Effects on Refining Operations Estimated crude oil input per barrel of gasoline progressively increases with the successive standards. Crude oil input would increase to offset volume losses associated with (1) removal of C4s, C5s, and naphthas from the gasoline pool; and (2) increased volume and severity of reformer operations. Refinery energy use for the aggregate of the four regions increases by about 3% for the NCG standard and by about 5% and 9%, respectively, for 94 RON and 95 RON NCG standards. Refineries would use more energy to support additional crude oil throughput, expanded upgrading operations, primarily pen/hex isomerization and reforming, and additional fractionation. Refinery CO 2 emissions also increase progressively across the NCG and single octane NCG standards. However, the percentage increases in refinery CO 2 emissions are lower than the percentage increases in energy use because, as NCG octane increases: (1) relatively less refinery energy comes from burning catalyst coke (a high CO 2 energy source) and more comes from natural gas and refinery gases; and (2) the additional hydrogen produced in more extensive reformer operations reduces refinery calls for on-purpose hydrogen, thereby reducing CO 2 coproduced during on-purpose hydrogen production. Estimated increases in energy use and CO 2 emissions by California refineries are significantly greater than those for refineries in PADDs 1-3 for the 94 RON and 95 RON NCG standards. 1.6 Organization of the Report The following sections of the report describe the development of the regional refining models, present the results of the analysis, and discuss analytical issues associated with the refinery modeling. Section 2 contains definitions of some key terms and concepts Section 3 provides information used to develop and calibrate our refinery models (refining capacity, refining inputs & outputs, gasoline properties, prices, etc.). October 8,

11 Standard Final Report Section 4 discusses the projections of prices, refinery inputs, and refined product out-turns developed to establish the baseline (Reference case) for 2015, the target year for the analysis. Section 5 discusses the technical approach for the regional refinery modeling. Section 6 describes the regional refining models accounting framework for refinery energy use and CO 2 emissions Section 7 presents results of the refinery modeling analysis for the Calibration, Reference, and Study Cases. Section 8 summarizes the effects of the NCG and single octane NCG standards on refining operations. Section 9 provides additional comments, bearing on the scope of the analysis and on the interpretation of the results of the analysis. October 8,

12 Standard Final Report 2. SOME KEY TERMS AND CONCEPTS A few terms and concepts that are essential elements of the analysis are likely to be unfamiliar to readers not well-versed in the analysis of refining economics. Accordingly, we define and briefly discuss these terms and concepts here to facilitate understanding of this report. 2.1 Measures of Refining Economics Used in This Report The refining cost of complying with a gasoline specification or standard (such as NCG) is the sum of capital charges and fixed costs associated with investments in refining capacity made in order to produce gasoline complying with the new standard; and reduction in cash flow the algebraic sum of changes in direct operating costs (e.g., the costs of crude oil, other refinery inputs, natural gas, electricity, catalysts and chemicals, etc.) and product revenues resulting from producing gasoline complying to the new standard. The net cost (often called social cost) of complying with a gasoline specification or standard (such as NCG) is the cost of the additional resources society expends to meet the new standard or specification (without regard to which segments of society incur these costs). In this study, the net cost of the various NCG standards is deemed to be the algebraic sum of the associated change in refining cost and the value of the associated increase in gasoline energy density. The marginal cost of gasoline ( /gallon) is the additional refining cost incurred in producing the last volume increment of gasoline, at constant octane. In general, the marginal cost of gasoline increases with increasing production volume. 2.2 PADDs PADDs (Petroleum Administration for Defense Districts) are geographical regions of the U.S.; five in number. They were defined by the federal government during World War II for various administrative purposes, including data collection and reporting related to petroleum supply and consumption. They continue to be widely used for these purposes. The five PADDs are shown on the map below. October 8,

13 Standard Final Report Petroleum Administration for Defense Districts (PADDs) PADD 5 PADD 4 PADD 2 PADD 3 PADD Refining Processes U.S. refineries comprise many specialized refining processes. However, these processes can be thought of in terms of a few broad classes, shown in Table 2.1. Of these processes, a few are particularly relevant to this analysis The catalytic reforming process is used by almost all U.S. refineries that produce gasoline. Reforming is the refinery s primary source of high-octane gasoline blendstocks and incremental octane. Reforming upgrades low-octane crude oil fractions and other refinery streams in the gasoline boiling range to a high-octane gasoline blendstock called reformate, typically with RON > 95 and octane sensitivity > 10. Reformate is the second largest constituent of the U.S. gasoline pool. Reformers also produce, as by-products, hydrogen and still gas. Reformer hydrogen constitutes almost half of U.S. refinery hydrogen consumption; reformer still gas goes to refinery fuel. Reforming units can be operated over a range of severity (defined by the RON of the produced reformate), up to RON or higher. Reforming is the only refinery process that (1) is capable of producing gasoline blendstock with RON > 95 and (2) can be controlled to operate over a wide range of severity. Consequently, the ability of U.S. refineries to produce 94 RON or 95 RON NCG will hinge on the capacities and capabilities of their reforming units. October 8,

14 Standard Final Report Table 2.1: Important Classes of Refining Processes in U.S. Refineries Class Function Examples Crude distillation Separate crude oil charge into boiling range Atmospheric distillation fractions for further processing Vacuum distillation Conversion Break down ("crack") heavy crude fractions into lighter, Fluid cat cracking (FCC) higher-valued streams for further processing Coking, Hydrocracking Upgrading Enhance the blending properties (e.g., octane) and value Catalytic Reforming of gasoline and diesel blendstocks Alkylation, Isomerization Treating Remove hetero-atom impurities from refinery streams Hydrotreating and blendstocks Caustic treating Separation Separate, by physical or chemical means, constituents Fractionation of refinery streams for further processing Extraction Blending Combine blendstocks to produce finished products that meet product specifications and environmental standards Utilities Supply refinery fuel, power, steam, oil movements, Steam generation storage, emissions control, etc. Sulfur recovery The fluid cat cracking (FCC) process is the heart of almost all U.S. conversion refineries the refineries that produce large volumes of gasoline and other transportation fuels. FCC units convert heavy, low-value crude oil fractions into lighter streams that can be blended into the refinery s gasoline, jet fuel, and diesel fuel pools. FCC naphtha, the gasoline fraction produced by FCC units, contains virtually all of the sulfur that must be removed to produce low-sulfur gasolines, such as NCG. (FCC naphtha is the largest constituent of the U.S. gasoline pool.) Alkylation and isomerization produce high-octane gasoline blendstocks from various light refinery streams (C4s for alkylation; C5s and C6s for isomerization). The blendstocks that they produce, alkylate and isomerate, have relatively high octane (though significantly less than reformate) and low sensitivity. These processes are not controllable with respect to octane in the sense that reformers are, and they tend to run at full capacity; hence, they can offer little in the way of additional octane to the gasoline pool, except by expansion of the existing process capacity. October 8,

15 Standard Final Report 3. INFORMATION USED TO DEVELOP AND CALIBRATE THE REFINING MODELS Each regional refining model is defined by a unique set of boundary conditions constraints that represent actual or forecast market and technical conditions to which solutions returned by the refinery model must conform. Boundary conditions in this study include refining capacity; refinery inputs of unfinished oils and blendstocks; refinery out-turns of finished products, such as gasoline and diesel fuel; specifications for the properties of refined products (e.g., octane, RVP, and sulfur for gasoline); and certain regulatory standards that constrain refining operations (e.g., RFG standards implemented through the Complex and Predictive Models). Exhibits A-1 through A-21 provide information that characterizes refined petroleum product markets in the U.S circa We derived this information from public data sources and used this information to establish the boundary conditions defining the regional refining models used in this study. Exhibit A-1 shows petroleum refining capacity, by region for Exhibit A-2 shows the annual supply of refined products by region and source of supply. These data, along with projections in AEO 2009, were used to develop projections of regional refined product supply in Exhibit A-3 shows annual refinery inputs, volumes of feed going to specific refinery processes, average crude oil properties, and net production of refined products, by region for Exhibit A-4 shows the same information as Exhibit A-3, but separately for the summer and winter seasons. Exhibit A-5 shows properties and distillation curves for the composite crude oils used in the regional refinery models. Exhibits A-6a, A-6b, A-7a and A-7b provide information on regional refinery fuel use and fuel use relative to refinery inputs and outputs. Exhibits A-8 through A-13 show average properties of various gasoline classes at different time periods and levels of aggregation. Exhibit A-8 shows PADD-level, seasonal data provided by EPA for 2004 (the most recent year for which data of this type are available); Exhibit A-9 shows national data (ex California), by gasoline type and season, for , Exhibits A-10a and A-10b show estimated properties of RFG for PADDs 1, 2, and 3 for , by season; Exhibit A-11 shows the properties of gasoline produced by California refineries in summer 2006; and Exhibits A-12a and A-12b and A-13 show data on October 8,

16 Standard Final Report gasoline properties and octane for , by producing region, that we derived from the Alliance North American Fuel Surveys. 13 Exhibits A-14 through A-17 provide information that we used to estimate average regional RVP and octane levels for 2007 and to project such levels to Exhibit A-14 shows estimated annual grade splits for in consuming regions by type of gasoline. Exhibit A-15 shows the distribution of gasoline in consuming regions, by type of gasoline, that we estimated for 2007 and projected for 2015 (accounting for the implementation of the 8-Hour Ozone Standard). Exhibit A-16 shows for 2007: the estimated distribution of CG and low-rvp gasoline (i.e., all non-rfg), by consuming region; the derived distribution of regional refinery production of CG and low-rvp gasoline; and the calculated RVPs of the non-rfg gasoline pool produced in each refining region. Exhibits A-17a & A-17b show for 2007 and 2008, respectively, regional estimates of: the share of gasoline production shipped to various consuming regions; grade splits; and gasoline pool octane. (Octane declined slightly in 2008 because of a continued decline across all regions in the market share of premium gasoline.) Exhibits A-18 through A-23 provide information on seasonal crude oil acquisition costs; seasonal refined product and WTI spot prices; seasonal wholesale gasoline prices, by type and grade, and seasonal No. 2 diesel fuel prices; seasonal spot prices for C4s and C5s; annual natural gas prices; and annual electricity prices. 13 Exhibits A-10a&b, A-12, and A-13 show estimated average properties and octane for RFG and CG by PADD of origin, not consumption. We estimated average properties by producing region for both the Alliance surveys and the EPA federal RFG Area Surveys by assigning the surveyed cities to specific PADDs, according to the sourcing of their gasoline supplies. For example, cities assigned to PADD 3 included all of those located in PADD 3, along with the Washington metropolitan area and all cities in the southeast. Cities assigned to PADD 1 included Pittsburgh and all cities in the northeast, except those in New England (e.g., Boston), because the latter areas are supplied largely by imports. All cities in PADD 2 were assigned to PADD 2 October 8,

17 Standard Final Report 4. PROJECTED REFINERY OPERATIONS FOR 2015 Exhibits B-1 through B-7 show our projections of refinery inputs and outputs for These projections establish the boundary conditions for the 2015 Reference cases. We developed the projections primarily using data from the Department of Energy s Petroleum Supply Annuals , in combination with aggregate energy forecasts and regional energy demand forecasts from AEO Exhibits B-1 to B-3 show projections for 2015 of total U.S. petroleum supply and disposition, imports and exports of petroleum products, and regional growth in consumption of refined products. We used these national projections in developing the regional projections shown in the subsequent tables. EIA projects gasoline consumption (including a small volume of E85) to be about 9.4 million b/d in 2015 only about 0.6% higher than consumption in However, projected net imports of finished gasoline and gasoline blending components (adjusted for subsequent ethanol blending) decline from about 1.1 million b/d to about 0.6 million b/d. Hence, domestic production of finished gasoline (net of imported blendstocks and accounting for ethanol blending at terminals) is projected to increase by about 7½ percent. Exhibit B-4 shows the projected annual supply of major refined products, by region. These projections are based on the pattern of supply in 2007 (Exhibit A-2), the aggregate projections shown in Exhibit B-1, and the regional growth rates in product demand shown in Exhibit B-3. Exhibits B-5 and B-6 show, respectively, projected annual and seasonal refinery inputs and refined product out-turns, by region. Exhibits B-7 and B-8 show projected gasoline pool RVP and octane. Projected gasoline pool RVP (for CG and low-rvp gasoline only) declines from 2007 (Exhibit A-15) to 2015 because of implementation of the 8-hour ozone standard and the NCG standard. 14 Projected gasoline pool octanes are similar to those estimated for 2008 (Exhibit A-16b), because we assumed grade splits estimated for 2008 would persist through We conducted the refinery modeling using just two types of gasoline: RFG (federal or California) and all other. The all other category includes CG and low-rvp gasolines. 14 The RVP shown in Exhibit B-6 for NCG incorporates the RVP effect of ethanol blending with the applicability of the 1 psi RVP waiver only to CG. 15 Small differences in octane for PADDs 2 & 3 reflect resumption of shipments of premium grade gasoline from PADD 3 to PADD 2 when E10 becomes ubiquitous. The decline in octane for CG produced by California refineries reflects our assumption that octane for regular grade gasoline would average about 87.5, rather than the three year average of 88.1 for Las Vega calculated using data from the Alliance gasoline surveys. October 8,

18 Standard Final Report Consequently, calculated RVPs for the all other category, shown in Exhibit B-6, reflect the weighted average of finished CG and low-rvp gasolines, incorporating the effects of the 1 psi ethanol RVP waiver. Additionally, we combined premium and regular grade gasoline into a single gasoline pool with octane corresponding to the weighted average of the octanes of the corresponding premium and regular grades (the weighting factors are the projected premium/regular grade splits). This approach simplifies the modeling without significantly affecting the results. Projected shares of RFG, CG, and low-rvp gasoline for 2015 are shown in Exhibit A-15. These projections assume implementation of the 8-hour ozone standard by 2015 and consequent increases in the shares of low-rvp and reformulated gasoline in the gasoline pool. Our estimates of the effects of the 8-hour ozone standard on gasoline markets come from a MathPro study for the American Petroleum Institute (API). 16 Exhibits A-1 (2007 refining process capacity), A-18 (crude oil prices), B-6, B-7, and B-8 establish the primary boundary conditions for the 2015 Reference cases. Nationally, the 2015 projections reflect, relative to refined product out-turns for 2007 shown in Exhibit A-3, an increase in domestic production of finished, ethanol-blended gasoline (net of imported gasoline blending components) of about 9%; 17 a slight decline in jet fuel production; an increase of about 2% in distillate production; and a decline of about 16% in resid plus asphalt production. 16 Potential Effects of 8-Hour Ozone Standard on Gasoline Supply, Demand, and Production Costs; Prepared for the American Petroleum Institute by MathPro Inc.; March The growth in net production of finished gasoline of 9% is higher than the 7½% growth in net production of gasoline estimated from projections in AEO The discrepancy arises because various adjustments incorporated in gasoline data reported in the AEO 2009 and included in Exhibit A-2 are not incorporated in the historical refinery net input and output data provided on EIA s website (and in corresponding tables in the Petroleum Supply Annual) that we used to prepare Exhibit A-3 and to establish the regional base cases for October 8,

19 Standard Final Report 5. TECHNICAL APPROACH FOR REGIONAL REFINERY MODELING 5.1 Scope of the Analysis As described in Section 1.3, we developed and analyzed a series of refinery modeling cases Calibration, Reference, and Study cases to assess the effects of the contemplated NCG and single octane NCG standards in PADDs 1, 2, and 3 and in California. The 2007 Calibration cases served to calibrate the models to historical data for summer and winter The 2015 Reference cases established the 2015 summer and winter baselines to which the Study cases are compared. The Reference cases incorporate refining capacity sufficient to produce projected refined product out-turns, meet the Tier 2 sulfur standards for gasoline and diesel fuel, meet the MSAT 2 standards (benzene control), and supply the volumes of low-rvp gasoline and RFG to meet the (projected implementation of the) 8-Hour Ozone Standard. 18 All gasoline in the Reference cases is E10. The Study cases assessed the progressive effects of first imposing the NCG standard and then imposing the 94 RON and 95 RON standards in combination with the NCG standard. The refinery modeling was designed so that investments in new refining process capacity in the Reference cases and in the NCG and single octane NCG Study cases all are made independently and with respect to baseline 2007 refining capacity. This represents investment planning with perfect foresight regarding future regulatory standards for gasoline. In preliminary modeling for PADD 3, we also developed a series of cases in which refining investment was sequential. Under that approach, refining investments made in the Reference case for major refining processes were carried forward to the NCG case; any additional investments made in the NCG case were then carried forward to the 94 RON NCG case. This represents what might be considered myopic investment with regard to regulatory standards. Because the preliminary modeling results for sequential investment were similar to those of the foresighted investment cases, we adopted the assumption of perfect foresight for all subsequent refinery modeling. 18 In the Reference cases, the regional refining models are endowed with refining process capacity as of The models add refining capacity as needed to produce specified product volumes and to meet refined product standards. We did not incorporate in the regional models any capacity expansions after 2007 reported as planned or under construction, as many such projects have been delayed or placed on hold due to the decline in economic activity. October 8,

20 Standard Final Report 5.2 Key Assumptions and Premises The refinery modeling for the 2015 Reference cases incorporated the following assumptions regarding national and state policies affecting gasoline quality. All gasoline ethanol-blended at 10 vol% (National E10). Implementation of the federal MSAT 2 standard, limiting the average benzene content of gasoline to 0.62 vol%. Implementation of the 8-hour ozone standard. Continuation of the 1 psi RVP waiver for ethanol-blended CG and for low-rvp gasoline in those states (mostly in the Midwest) that now grant the waiver. (The waiver does not apply to low-rvp gasoline during the peak ozone months of July and August.) Continuation of the RVP waiver for NCG and single octane NCG that replace CG, but not for NCG and single octane NCG that replace low-rvp gasoline, even in the Midwest. Continued use of the Complex Model for certifying federal RFG and of the Predictive Model for certifying California RFG. The Reference cases reflect projected regional refined product volumes derived from national projections reported by EIA in AEO 2009 (April Revision) and a world crude oil price of about $100/b. The regional refining models included representations of two types of reforming processes a medium-pressure process (the older type) capable of a maximum severity of 100 RON and a low-pressure process capable of a maximum severity of RON. (These severities refer to the RON of the produced reformate.) The aggregate capacities of the two reforming processes represented in the regional models are consistent with capacities reported for each U.S. refinery. 5.3 Refined Product Volumes Held Constant In the Study cases, the regional refining models are constrained to produce the same volume of major refined products as in the corresponding Reference cases, i.e. refined product volumes are fixed, except for light gases. Inputs of crude oil (the primary input) and light gases are priced and allowed to vary. Volumes of other refinery inputs (natural gasoline, pyrolysis gasoline, naphtha, kerosene, heavy gas oil, and residual oil) are fixed. (For PADDs 1-3, the refining models represent refinery production of finished gasoline containing 10 vol% ethanol. For California, the refining model represents production of a sub-octane California gasoline (CARBOB) for subsequent blending with 10 vol% ethanol to produce finished California gasoline. This approach is necessary in modeling the California refineries, because the California Predictive Model (used for certifying compliance with the October 8,

21 Standard Final Report California standards) operates on the properties of the produced CARBOB, not the finished gasoline.) The regional refining models respond to the imposition of NCG and single octane NCG standards primarily by adjusting refinery crude oil inputs, process capacity use, and process operations, but not by adjusting outputs of gasoline and other major finished refined products. However, we allow the regional refining models to expand sales of C4s, C5s, coker naphtha, and FCC naphtha to comply with NCG and single octane NCG standards. Although refineries, in practice, may alter refined product out-turns in response to the imposition of new gasoline standards, specifying constant product out-turns in the refinery models allows for a straightforward assessment of the costs of the various standards without having to pursue a more complicated equilibrium analysis of the effects of the standards on domestic production, imports of refined products, and refined product supply in general. October 8,

22 Standard Final Report 6. ENERGY AND CO 2 ACCOUNTING FRAMEWORK IN THE REFINING MODELS In the interest of facilitating subsequent life-cycle analysis of single octane NCG, this section provides an overview of the accounting framework in the regional refining models for estimating refinery energy use and the consequent CO 2 emissions. The accounting framework bears exclusively on refining operations. It does not address energy use and CO 2 emissions in the production and transport of crude oil and other refinery inputs (e.g., natural gas, unfinished oils); nor does it address CO 2 emissions from the end-use of refined products (e.g., combustion of transportation fuels). 6.1 U.S. Refinery Energy Use Energy consumption in the U.S. refining sector is about 3 quads per year about 3% of total U.S. energy consumption equivalent to about 0.57 MM BTU/Bbl of refinery charge. U.S. refinery energy consumption, both total and per barrel of crude through-put, tends to increase slowly over time. This trend reflects (1) U.S. refiners gradual shift to a heavier, higher sulfur crude slate and (2) increasingly stringent specifications on refined products, particularly the sulfur standards for gasoline and diesel fuel. Energy use by California refineries is about 0.66 MM BTU/Bbl of refinery charge higher than in the rest of the U.S. refining sector because of the heavy crude slate run by California refineries and because of California s refined product specifications, the most stringent in the U.S. The energy consumed in refining comes from numerous sources, some outside the refinery and some within. However, four sources purchased natural gas and electricity and refineryproduced still gas and catalyst coke account for about 95% of reported U.S. refinery energy consumption. Still gas is a mixture of light gases (methane, ethane, etc.) produced as by-products in various refining processes. These light gas streams are collected, treated, and burned in the refinery fuel system to generate process heat and steam. Catalyst coke coke laid down on FCC catalyst is a by-product of the cracking reactions that occur in the FCC reactor. The coke is burned off the catalyst in the FCC regenerator. The heat of combustion is used to provide process energy for the FCC reactor and to generate refinery steam. 19 Refinery purchases of electricity reported by EIA reflect purchases from the grid and do not include refinery-generated electricity, most of which comes from gas-fired co-generation units. 19 Petroleum coke (or marketable coke) which is not used as a refinery fuel is the primary byproduct of refinery coking units (cokers). Petroleum coke constitutes wt% of coker output and has various non-energy uses outside the refining industry. October 8,

23 Standard Final Report Gross power generation in U.S. refineries averaged about 2.6 gigawatts (63 gigawatt-hours per day) in U.S. refineries sold 30% of their gross power output to the grid, leaving about 1.9 gigawatts of indicated net power generation for internal use. Refinery purchases of natural gas for fuel use that are reported by EIA appear to include natural gas used for refinery power generation, without adjustment for refinery sales of electricity to the grid; but not natural gas used by refineries as feed to hydrogen production units or natural gas used by merchant hydrogen plants as either feed or fuel. 6.2 Conceptual Framework for Estimating Refinery Energy Use In principle, there are several approaches for estimating refinery energy use and CO 2 emissions. The most rigorous approach, from a theoretical standpoint, is to develop complete energy, material, and carbon balances around each refinery. The difference between the energy embodied in all refinery outputs and inputs equals the energy expended in the refinery. Similarly, the difference between the total carbon content of all refinery outputs and inputs equals the refinery s carbon emissions. At first glance, this approach seems appealing because it rests on fundamental engineering principles of heat and material balance. In practice, the approach is unworkable. It requires the development of complete and tight material and energy balances for the refinery (including not only all refinery feed and product streams but also waste streams and losses, such as flue gas, flare gas, fugitive emissions, cooling water, waste water, etc.) and precise estimates of the energy and carbon content of each refinery input and output. Such properties vary with crude type, are subject to day-to-day fluctuation, and in many cases are simply unavailable. Moreover, because the desired results refinery energy use and CO 2 emissions are residuals, the inevitable gaps in refinery material and energy balances and the inaccuracies in energy and carbon content would render such estimates useless. The most useful approach focuses exclusively on energy use within the refinery battery limits. This approach involves estimating energy use in each refining process and then estimating total refinery energy use as the sum of the energy used in each process that is, by summing the direct energy inputs to all refining processes, by energy source (power, steam, and fuel). This is the approach of choice in industry LP models of refining operations and the one followed in the regional refining models used in this study. 6.3 Energy Accounting in the Refining Models Each process representation in the regional refining models includes input/output coefficients representing the process s net per-barrel consumption of fuel (foeb/bbl), steam K lbs/bbl), and power (Kwh/Bbl). For a given process, the values of the fuel, steam, and power coefficients depend on the process operating conditions (e.g., severity, conversion, etc.) and feed streams. In addition, the regional refining models contain representations of three refinery utilities: October 8,