Fuel Economy Labeling of Motor Vehicles: Revisions To Improve Calculation of Fuel Economy Estimates

Federal RegisterFeb 1, 2006

Ask Donna

What actually matters in this document.

Text

ENVIRONMENTAL PROTECTION AGENCY

40 CFR Parts 86 and 600

[EPA-HQ-OAR-2005-0169; FRL-8021-8]

RIN 2060-AN14

Fuel Economy Labeling of Motor Vehicles: Revisions To Improve Calculation of Fuel Economy Estimates

AGENCY:

Environmental Protection Agency (EPA).

ACTION:

Notice of proposed rulemaking.

SUMMARY:

The Environmental Protection Agency (EPA) is proposing changes to the test methods used to calculate the fuel economy estimates that are posted on window stickers of all new cars and light trucks sold in the United States. A fundamental issue with today's fuel economy estimates is that the underlying test procedures do not fully represent real-world driving conditions. Although no single test or set of tests can ever account for the wide variety of conditions experienced by every driver, the new fuel economy estimates would more accurately reflect a number of important factors that drivers are likely to experience on the road. These changes will take effect starting with 2008 model year vehicles. Under the new methods, the City MPG estimates for most vehicles would drop 10 percent to 20 percent from today's labels, depending on the vehicle. The Highway MPG estimates would generally drop 5 percent to 15 percent for most vehicles. Although today's proposed fuel economy test methods would provide more accurate estimates for many consumers, there will always continue to be drivers who get higher or lower fuel economy than the window sticker numbers. Currently the same test procedures are used for both the window sticker estimates and the fuel economy values used to determine a manufacturer's corporate average fuel economy (CAFE). However, this proposal would not alter the test procedures, driving cycles, measurement techniques, or the calculation methods used to determine CAFE.

DATES:

Comments:

Comments must be received on or before April 3, 2006. Under the Paperwork Reduction Act, comments on the information collection provisions must be received by OMB on or before March 3, 2006. See Section VII.A of the

SUPPLEMENTARY INFORMATION

section for more information about written comments.

Hearings:

We will hold a public hearing in Romulus, Michigan, on March 3, 2006. See Section VII.C of the

SUPPLEMENTARY INFORMATION

section for more information about public hearings.

ADDRESSES:

Submit your comments, identified by Docket ID No. EPA-HQ-OAR-2005-0169, by one of the following methods:

•

www.regulations.gov:

Follow the on-line instructions for submitting comments.

• Fax: (202) 566-1741.

• Mail: Environmental Protection Agency, EPA Docket Center (EPA/DC), Air and Radiation Docket, Mail Code 6102T, 1200 Pennsylvania Avenue, NW., Washington, DC 20460, Attention Docket ID No. EPA-HQ-OAR-2005-0169. In addition, please mail a copy of your comments on the information collection provisions to the Office of Information and Regulatory Affairs, Office of Management and Budget (OMB), Attn: Desk Officer for EPA, 725 17th St., NW., Washington, DC 20503.”

• Hand Delivery: Docket Center, (EPA/DC) EPA West, Room B102, 1301 Constitution Ave., NW., Washington, DC, Attention Docket ID No. OAR-2005-0169. Such deliveries are only accepted during the Docket's normal hours of operation, and special arrangements should be made for deliveries of boxed information.

Instructions:

Direct your comments to Docket ID No. EPA-HQ-OAR-2005-0169. EPA's policy is that all comments received will be included in the public docket without change and may be made available online at

www.regulations.gov,

including any personal information provided, unless the comment includes information claimed to be Confidential Business Information (CBI) or other information whose disclosure is restricted by statute. Do not submit information that you consider to be CBI or otherwise protected through

www.regulations.gov

or e-mail. The

www.regulations.gov

Web site is an “anonymous access” system, which means EPA will not know your identity or contact information unless you provide it in the body of your comment. If you send an e-mail comment directly to EPA without going through www.regulations.gov your e-mail address will be automatically captured and included as part of the comment that is placed in the public docket and made available on the Internet. If you submit an electronic comment, EPA recommends that you include your name and other contact information in the body of your comment and with any disk or CD-ROM you submit. If EPA cannot read your comment due to technical difficulties and cannot contact you for clarification, EPA may not be able to consider your comment. Electronic files should avoid the use of special characters, any form of encryption, and be free of any defects or viruses. For additional information about EPA's public docket visit the EPA Docket Center homepage at

http://www.epa.gov/epahome/dockets.htm

. For additional instructions on submitting comments, go to Section VII of the

SUPPLEMENTARY INFORMATION

section of this document.

Public Hearing:

The public hearing will be at the Crowne Plaza Hotel, Detroit—Metro Airport, 8000 Merriman Road, Romulus, Michigan.

Docket:

All documents in the docket are listed in the

www.regulations.gov

index. Although listed in the index, some information is not publicly available,

e.g.

, CBI or other information whose disclosure is restricted by statute. Certain other material, such as copyrighted material, will be publicly available only in hard copy. Publicly available docket materials are available either electronically in

www.regulations.gov

or in hard copy at the EPA Docket Center, EPA/DC, EPA West, Room B102, 1301 Constitution Ave., NW., Washington, DC. This Docket Facility is open from 8:30 a.m. to 4:30 p.m., Monday through Friday, excluding legal holidays. The EPA Docket Center telephone number is (202) 566-1742. The Public Reading Room is open from 8:30 a.m. to 4:30 p.m., Monday through Friday, excluding legal holidays. The telephone number for the Public Reading Room is (202) 566-1744.

FOR FURTHER INFORMATION CONTACT:

Rob French, U.S. EPA, Voice-mail (734) 214-4636; E-mail:

french.roberts@epa.gov

SUPPLEMENTARY INFORMATION:

Regulated Entities

This proposed action would affect companies that manufacture or sell new light-duty motor vehicles. Regulated categories and entities include:

Category

NAICS codes

Examples of potentially regulated entities

Industry

336111, 336112

Motor vehicle manufacturers.

Industry

811112, 811198, 541514

Commercial Importers of Vehicles and Vehicle Components.

North American Industry Classification System (NAICS).

This list is not intended to be exhaustive, but rather provides a guide regarding entities likely to be regulated by this action. To determine whether particular activities may be regulated by this action, you should carefully examine the proposed regulations. You may direct questions regarding the applicability of this action to the person listed in

FOR FURTHER INFORMATION CONTACT.

Table of Contents

I. Introduction

A. History of Federal Fuel Economy Requirements

B. Why is Today's Action Warranted?

C. What New Requirements Are We Proposing?

D. Today's Proposal Does Not Impact or Change CAFE Test Procedures

E. When Will the New Fuel Economy Estimates Take Effect?

F. How Will EPA Communicate to the Public the Transition Between the Old Label Values and New?

G. Statutory Provisions and Legal Authority

II. Description of the Proposed Fuel Economy Label Methodology

A. Proposed Fuel Economy Label Formulae

B. Application of the Formulae To Develop Fuel Economy Labels for Specific Vehicles

C. Derivation of the Proposed 5-Cycle Fuel Economy Formulae

D. Derivation of the MPG-Based Approach

E. Effect of the New Formulae on Fuel Economy Label Values

F. Comparison to Other Onroad Fuel Economy Estimates

III. What Major Alternatives Were Considered?

IV. Revisions to the Fuel Economy Label Format and Content

A. Estimated Annual Fuel Cost

B. Fuel Economy of Comparable Vehicles

C. “Your mileage will vary * * *” Range of Expected Fuel Economy Information

D. Other Format Changes

V. Other Related Proposals

A. Comparable Class Categories

B. Electronic Distribution of Dealer-Supplied Fuel Economy Booklet

C. Testing Provisions

D. Voluntary Fuel Economy Labeling for Vehicles Exceeding 8500 Pounds GVWR

E. Consideration of Fuel Consumption vs. Fuel Economy as a Metric

F. Environmental Information on Fuel Economy Labels

VI. Projected Impacts of the Proposed Requirements

A. Information and Reporting Burden

B. Fees

C. Aggregate Costs

VII. Public Participation

A. How and To Whom Do I Submit Comments?

B. How Should I Submit CBI to the Agency?

C. Will There Be a Public Hearing?

VIII. Statutory and Executive Order Reviews

A. Executive Order 12866: Regulatory Planning and Review

B. Paperwork Reduction Act

C. Regulatory Flexibility Act

D. Unfunded Mandates Reform Act

E. Executive Order 13132: Federalism

F. Executive Order 13175: Consultation and Coordination With Indian Tribal Governments

G. Executive Order 13045: Protection of Children From Environmental Health and Safety Risks

H. Executive Order 13211: Actions That Significantly Affect Energy Supply, Distribution, or Use

I. National Technology Transfer Advancement Act

IX. Statutory Provisions and Legal Authority

I. Introduction

The EPA fuel economy estimates have appeared on the window stickers of all new cars and light trucks since the late 1970's and are well-recognized by consumers. The fuel economy estimates essentially serve two purposes: to provide consumers with a basis on which to compare the fuel economy of different vehicles, and to provide consumers with a reasonable estimate of the range of fuel economy they can expect to achieve. While the estimates historically have been a valuable tool for comparison shopping purposes, attention has been focused recently on how closely the EPA estimates approximate consumers' real-world fuel economy experience.

Today, we are proposing changes to EPA's fuel economy test methods to bring the estimates closer to the fuel economy consumers are achieving in the real-world. We believe these estimates will provide car buyers with useful information when comparing the fuel economy of different vehicles. It is important to emphasize that fuel economy varies from driver to driver for a wide variety of reasons, such as different driving styles, climates, traffic patterns, use of accessories, loads, weather, and vehicle maintenance. Even different drivers of the same vehicle will experience different fuel economy as these and other factors vary. Therefore, it is impossible to design a “perfect” fuel economy test that will provide accurate real-world fuel economy estimates for every consumer. With any estimate, there will always be consumers that get better or worse actual fuel economy. The EPA estimates are meant to be a general guideline for consumers, particularly to compare the relative fuel economy of one vehicle to another. Nevertheless, we do believe that today's new fuel economy test methods will do a better job of giving consumers a more accurate estimate of the fuel economy they can achieve in the real-world.

It is essential that our fuel economy estimates continue to be derived from controlled, repeatable, laboratory tests. However, the inputs to our estimates are based on data from actual real-world driving behavior and conditions. Because the test is controlled and repeatable, an EPA fuel economy test result can be used for comparison of different vehicle models and types. EPA and manufacturers test over 1,250 vehicle models annually and every test is run under identical conditions and under a precise driver's trace, which assures that the result will be the same for an individual vehicle model no matter when and where the laboratory test is performed. Variations in temperature, road grade, driving patterns, and other variables do not impact the result of the test. While such external conditions impact fuel economy on a trip-to-trip basis, they do not change the laboratory test result. Therefore, a repeatable test provides a level playing field for all vehicles, which is essential for comparing the fuel economy of one vehicle to another. Finally, EPA must preserve the ability to confirm the values achieved by the manufacturers' testing, and this can only be achieved with a highly repeatable test or set of tests. No other fuel economy test program provides the level of repeatability as the EPA program.

However, the EPA fuel economy test methods need to reflect real world conditions as well as being a repeatable test. While some organizations have issued their own fuel economy numbers based on on-road driving, this approach introduces a wide number of variables—different drivers, driving patterns, weather conditions, temperatures, etc.—that make repeatability impossible. Our proposed fuel economy test methods are more representative of real-world

conditions than the current fuel economy tests—yet we would retain our practice of relying on controlled, repeatable, laboratory tests.

The methods used today for calculating the city and highway mpg estimates were established in the 1970's, and were adjusted in the mid-1980's. Since these adjustments were made, America's driving behavior has changed. In the past 20 years, speed limits have increased and vehicles have been designed for higher power—as a result, Americans are driving faster and more aggressively than ever before. Vehicle technology has changed markedly, and many more vehicles are equipped with energy-consuming accessories like air conditioning. These and other factors are not accounted for in the current test procedures used to determine the city and highway mpg estimates. Our analyses indicate that if these factors were better accounted for, the city and highway fuel economy label estimates would be generally lower and closer to the average real-world experience of consumers.

A fundamental issue with today's fuel economy estimates is that the underlying test procedures do not fully represent real-world driving conditions. Some of the key limitations are that the highway test has a top speed of only 60 miles per hour, both the city and highway tests are run at mild climatic conditions (75 deg. F), both tests have mild acceleration rates, and neither test is run with the use of accessories, such as air conditioning. However, since the time of the last fuel economy labeling revisions in the mid-1980's, EPA has established several additional test procedures, used for emissions compliance purposes, which capture a much broader range of real-world driving conditions. Specifically, these emissions test cycles capture the effects of higher speeds, more aggressive driving (i.e., higher acceleration rates), the use of air conditioning at higher ambient temperatures, and colder temperature operation. Our analysis indicates that these factors can have a significant impact on fuel economy, and that the impacts can vary widely across different vehicles.

Today, we are proposing that three additional emission tests, already used by manufacturers, could be utilized to derive more accurate fuel economy estimates. These three test procedures encompass a much broader range of real-world driving, as they incorporate the effects of higher speeds, more rapid accelerations, air conditioning use, and cold temperatures. Our proposed approach would utilize these additional emission tests, together with the current two fuel economy tests, so that our fuel economy test methods reflect a much broader range of driving conditions.

In the Energy Policy Act of 2005, Congress required EPA to update or revise adjustment factors to better reflect a variety of real-world factors that affect fuel economy. Section 774 of the Act directs EPA to “* * * update or revise the adjustment factors in [certain sections of the fuel economy labeling regulations] to take into consideration higher speed limits, faster acceleration rates, variations in temperature, use of air conditioning, shorter city test cycle lengths, current reference fuels, and the use of other fuel depleting features.”

Today's proposal does take into account these conditions and would address this statutory requirement.

Pub. L. 109-58, 119 Stat. 835 (2005).

Over the past few years, there have been several independent studies comparing EPA's fuel economy estimates to the real-world experience of consumers. These studies confirm that there is considerable variation in real-world fuel economy, and provide further evidence that EPA's mileage ratings often overestimate real-world fuel economy. Although these studies differ in a number of variables, including their test methods, driving conditions, and fuel economy measurement techniques, they indicate that EPA's approach to estimating fuel economy needs to be improved to better represent some key real-world fuel economy impacts.

Currently the same test procedures are used for both the window sticker estimates and the fuel economy values used to determine a manufacturer's corporate average fuel economy (CAFE), although the label estimates are adjusted downward. This proposal would not alter the test procedures, driving cycles, measurement techniques, or the calculation methods used to determine CAFE. The Energy Policy and Conservation Act of 1975 requires that CAFE values be determined from the EPA test procedures in place as of 1975 (or procedures that give comparable results), meaning that whatever action we take to improve the window sticker estimates must leave in place the existing tests used for CAFE determination. The proposed test methods for determining the new fuel economy label estimates would be incorporated in sections of the regulations that are entirely separate from the CAFE regulations.

This section begins with a history of EPA's involvement in fuel economy programs. Then we discuss why we are taking action, including discussions of the limitations of the current tests, various data sources of real-world fuel economy, the additional real-world driving conditions captured by other emissions tests procedures, and the impact of these factors on fuel economy. We then provide an overview of our proposed new fuel economy test methods (which are discussed in detail in Section II), and conclude with a discussion of the relevant Federal statutes and how they bear on this proposal.

A. History of Federal Fuel Economy Requirements

The Energy Policy and Conservation Act of 1975 (EPCA) established two primary fuel economy requirements: (1) Fuel economy information, designed for public use, in the form of fuel economy labels posted on window stickers of all new motor vehicles, and the publication of an annual booklet of fuel economy information to be made available free to the public by car dealers; and (2) calculation of a manufacturer's average fuel economy and compliance with a standard (later, this compliance program became known as the Corporate Average Fuel Economy (CAFE) program). The responsibilities for these requirements were split between EPA, the Department of Transportation (DOT) and the Department of Energy (DOE). EPA is responsible for establishing the test methods and procedures both for determining the fuel economy estimates to be posted on the window stickers and in the annual booklet, and for the calculation of a manufacturer's corporate average fuel economy. DOT is responsible for administering the CAFE compliance program, including establishing standards for non-passenger automobiles and determining if manufacturers were complying with the applicable CAFE standards, and assessing any penalties as needed. DOE is responsible for publishing and distributing the annual fuel economy information booklet.

EPA published regulations implementing portions of the EPCA statute in 1976.

The provisions in this regulation, effective with the 1977 model year, established procedures to calculate fuel economy values for labeling and CAFE purposes that used the Federal Test Procedure (FTP or “city” test) and the Highway Fuel Economy Test (HFET or “highway” test) data as the basis for the calculations. At that time, the fundamental process for determining fuel economy was the same for labeling as for CAFE, except that the

CAFE calculations combined the city and highway fuel economy into a single number.

See 41 FR 38685, which is promulgated at 40 CFR Part 600.

After a few years of public exposure to the fuel economy estimates on the window stickers of new vehicles, it soon became apparent that drivers were disappointed that they were not often achieving these estimates on the road and that they expected them to be as accurate as possible. In 1978, Congress recognized the concern about differences between EPA estimated fuel economy values and actual consumer experience and mandated a study under section 404 of the National Energy Conservation Policy Act of 1978.

In February, 1980, a set of hearings were conducted by the U.S. House of Representatives Subcommittee on Environment, Energy, and National Resources. One of the recommendations in the subsequent report by the Subcommittee was that “EPA devise a new MPG system for labeling new cars and for the Gas Mileage Guide that provides fuel economy values, or a range of values, that most drivers can reasonably expect to experience.”

Pub. L. 95-619, Title IV, 404 (November 9, 1978).

See House Committee on Government Operations, “Automobile Fuel Economy: EPA's Performance,” Report 96-948, May 13, 1980.

EPA commenced a rulemaking process in 1980 to revise its fuel economy labeling procedures, and analyzed a vast amount of in-use fuel economy data.

In 1984, EPA published new fuel economy labeling procedures that were applicable to 1985 and later model year vehicles.

The decision was made to retain the FTP and highway test procedures, primarily because those procedures were also used for other purposes—emissions certification and CAFE determination. Based on the in-use fuel economy data, however, it was evident that the final fuel economy values put on the labels needed to be adjusted downward in order to more accurately reflect consumers' average fuel economy experience. The final rule, therefore, included downward adjustment factors for both the city and highway label fuel economy estimates. The city values (based on the raw FTP test data) were adjusted downward by 10 percent and the highway values (likewise based on the raw highway test data) were adjusted downward by 22 percent.

See “Passenger Car Fuel Economy: EPA and Road,” U.S. Environmental Protection Agency, Report no. EPA 460/3-80-010, September, 1980, and “Technical Support Report for Rulemaking Action: Light Duty Vehicle Fuel Economy Labeling,” U.S. Environmental Protection Agency, Report no. EPA/AA/CTAB/FE-81-6, October, 1980.

See 49 FR 13845, April 6, 1984, and 49 FR 48149, December 10, 1984.

EPA projected at the time that these adjustments would put the average city and highway MPG values in the middle of the range of fuel economy values experienced by consumers.

During the rulemaking process, the Office of Management and Budget (OMB) expressed concern that fuel economy estimates based on the average experience would result in a significant number of drivers failing to achieve that fuel economy. They requested that EPA provide a range of values on the label that would encompass the expected fuel economy of about 75 percent of the driving population.

To address this concern, in the final rule, EPA required the label to contain the range of city and highway fuel economy that most drivers should expect. Based on our understanding of the frequency distribution of in-use fuel economy data at the time, the range was set at plus or minus 15 percent of the stated city and highway estimates, and appears on fuel economy labels today as small print text. Further in this section, we discuss, in the context of today's proposal, similar issues regarding how best to communicate to the public the level of the city and highway mpg estimates, as well as the range of drivers' fuel economy experience.

See 49 FR 13832, April 16, 1984.

See 49 FR 13835, April 16, 1984.

B. Why Is Today's Action Warranted?

The fundamental problem with the current fuel economy estimates is that the test procedures on which they are based do not reflect a broad enough range of in-use driving conditions. The current test procedures omit several critical factors that are prevalent in the real-world and that can have a significant impact on fuel economy. Key among these are higher speeds, faster accelerations, the use of air conditioning, and colder temperatures. The impact of these factors on fuel economy can vary widely from vehicle to vehicle. However, for emissions compliance, we have already developed additional test procedures to account for these factors, and these test procedures are already being regularly used by the auto companies. Today, we are proposing to use these tests, in conjunction with the existing fuel economy tests, as an input into the calculation of fuel economy estimates. In doing so, the fuel economy test methods would reflect a much broader range of real-world conditions than they do today.

There is broad-based support among automobile manufacturers and other stakeholders proposing changes to current fuel economy estimates. Congress recognized the need for action by including a provision in the Energy Policy Act of 2005 requiring EPA to revise its fuel economy estimates. EPA has worked closely with auto manufacturers, states, and other organizations in developing this proposed rule.

Bluewater Network petitioned EPA to revise the fuel economy labeling test procedures.

EPA published a

Federal Register

notice requesting comments on the petition, and received over 33,000 comments.

Nearly all of these comments support the revision of EPA's fuel economy estimates to better reflect real world driving. Today's proposal is responsive to this petition.

The Bluewater Network petition was submitted to EPA on June 7, 2002.

See 69 FR 16188, March 29, 2004.

1. Fuel Economy Labels Could Be Improved To Better Reflect Real-World Driving

First, it is important to stress that the EPA city and highway mpg ratings are estimates—they are not intended to give consumers an exact indication of the fuel economy they will achieve. The complete range of consumer fuel economy experience can not be represented perfectly by any one estimate. Fuel economy varies based on a wide range of factors, which we have discussed above. There will always be consumers that achieve real-world fuel economy both better and worse than a given estimate.

In the past few years, there have been a number of studies, conducted by a variety of sources, suggesting that there is often a shortfall between the EPA estimates and real-world fuel economy. Several organizations have provided consumers with their own fuel economy estimates, which in some cases vary from EPA's estimates. For example, Consumer Reports utilizes on-road driving to measure fuel economy under a variety of conditions. They derive city, highway, and overall fuel economy estimates, and their methods clearly demonstrate the large degree of variation across vehicles. While their city fuel economy estimates fall on average below the EPA label values, their highway estimates are, on average, higher than the EPA label values. Consumer Reports' overall fuel economy estimates range from 27 percent below to 20 percent above the EPA overall rating. The Automobile Association of America (AAA) likewise publishes the

fuel economy results they achieve in their annual auto guide for new cars and trucks. In their 2004 auto guide, about half of their estimates were below the EPA combined city/highway value, and about one half were above the EPA city/highway combined value. Their estimates ranged from 40 percent lower than EPA's to 22 percent higher, again reflecting a great deal of vehicle-to-vehicle variation. Other sources of fuel economy data include Edmunds.com, the Department of Energy's (DOE) “Your MPG” database on the fueleconomy.gov Web site, and DOE's FreedomCar program.

Each of these studies differs in its test methods, driving cycles, sampling of vehicles, and methods of measuring fuel economy. There are strengths and weaknesses of each study, which we discuss further in Section II and in the Draft Technical Support Document. Collectively, these studies indicate there are many cases where real-world fuel economy falls below the EPA estimates. The studies also indicate that real-world fuel economy varies significantly depending on the conditions under which it is evaluated. Nevertheless, taken as a whole, these studies reflect a wide range of real-world driving conditions, and show that fuel economy can be much lower than EPA's estimates if more real-world conditions are considered.

The fundamental problem with the current fuel economy estimates is that the test procedures on which they are based are missing a number of critical factors that exist in real-world driving and have a significant impact on fuel economy. The following section discusses the limitations of our existing fuel economy test procedures.

2. Today's Fuel Economy Tests Do Not Represent the Full Range of Driving Conditions

The current city and highway fuel economy tests do not represent the full range of real-world driving conditions. The 1985 adjustment factors were designed to ensure that the fuel economy estimates across the vehicle fleet reflected the average impacts of a number of conditions not represented on the tests. However, as we noted earlier, many changes have occurred since then that make it once again a reasonable time to reevaluate the fuel economy test methods. Given the significant degree of variation that is apparent across vehicles, we believe it is important to reconsider the approach of “one-size-fits-all” adjustment factors and instead move to an approach that more directly reflects the impacts of fuel economy on individual vehicle models.

The city fuel economy estimate is based on the Federal Test Procedure (FTP), which was designed to measure a vehicle's tailpipe emissions under urban driving conditions. The driving cycle used for the FTP is called the LA-4, which was developed in the mid-1960's to represent home-to-work commuting in Los Angeles. The FTP is also one of the tests used to determine emissions compliance today. The FTP includes a series of accelerations, decelerations, and idling (such as at stop lights). It also includes starting the vehicle after it has been parked for an extended period of time (called a “cold start”), as well as a start on a warmed-up engine (called a “hot start”). The total distance covered by the FTP is about 11 miles and the average speed is about 21 mph, with a maximum speed of about 56 mph.

The highway fuel economy estimate is based on the Highway Fuel Economy Test (HFET), which was developed by EPA in 1974 and was designed to represent a mix of interstate highway and rural driving. It consists of relatively constant higher-speed driving, with no engine starts or idling time. The HFET covers a distance of about 10 miles, at an average speed of 49 mph and a top speed of about 60 mph.

There are several key limitations in the FTP and HFET tests that cause them to not adequately reflect real-world driving today. First, most consumers understandably think “highway” fuel economy means the fuel economy you can expect under freeway driving conditions. In fact, the highway test has a top speed of only 60 mph, since the test was developed more than 20 years ago to represent more rural driving conditions at a time when the national speed limit was 55 miles per hour. The national speed limit since has been eliminated, states have established speed limits of 65 to 70 miles per hour, and much driving is at even higher speeds. Recent real-world driving studies indicate that about 28 percent of driving (vehicle miles traveled, or VMT) is at speeds of greater than 60 mph. (This analysis is detailed in the Draft Technical Support Document). These studies also show that 33 percent of real-world driving VMT falls outside the FTP/HFET speed and acceleration activity region. Thus, a substantial amount of high speed driving is not captured at all in today's FTP or HFET tests. This is a critical weakness in our current fuel economy test procedures. Since higher speed driving has a negative impact on fuel economy, incorporating these higher speed driving conditions into the fuel economy tests would lower the fuel economy estimates.

Second, the maximum acceleration rates of both the FTP and HFET tests are a relatively mild 3.3 miles-per-hour per second, considerably lower than the maximum acceleration rates seen in real-world driving. Recent real-world driving studies indicate that maximum acceleration rates are as high as 11 to 12 mph/sec and significant activity occurs beyond 3.3 mph/sec. Even at the time these tests were first developed, the real-world accelerations were higher than 3.3 mph/sec, but the test cycle's acceleration rates needed to be constrained to the mechanical limitation of the dynamometer test equipment. These constraints no longer exist with today's dynamometers, so we now have the ability to incorporate higher maximum acceleration rates that more closely reflect those of actual driving. In fact, we have incorporated higher acceleration rates into a test recently developed for emissions compliance, which we discuss in the next section. As with high speed driving, higher acceleration rates have a negative impact on fuel economy; thus, if these higher accelerations were factored into our fuel economy methods, the estimates would be lower.

The maximum deceleration rate of the FTP and HFET tests is important to consider as well, because it relates to the regenerative breaking effect of hybrid electric vehicles. The FTP and HFET tests include a mild maximum deceleration rate of −3.3 mph/sec; yet in recent real-world driving rates as high as −11 to −17 mph/sec were recorded. Under higher deceleration rates, the effects of regenerative breaking for hybrid electric vehicles are diminished, thereby lowering fuel economy. In this regard, today's FTP and HFET tests result in better fuel economy, which is seldom achieved under actual driving conditions.

Third, both tests are run at mild ambient conditions (approximately 75 degrees Fahrenheit), while real-world driving occurs at a wide range of ambient temperatures. Fuel economy is lower at temperatures colder or warmer than the 75 degree F test temperature. Only about 20 percent of VMT occurs between 70 and 80 degrees F—approximately 15 percent of VMT occurs at temperatures above 80 degrees F, and 65 percent occurs below 70 degrees F. Moreover, neither the FTP nor HFET tests are run with accessories operating, such as air conditioners, heaters, or defrosters. These accessories, most notably air conditioning, can have a significant impact on a vehicle's fuel economy.

Finally, there are many factors that affect fuel economy that cannot be replicated on dynamometer test cycles in a laboratory. These include road grade, wind, vehicle maintenance (e.g., tire pressure), snow/ice, precipitation, fuel effects, and others. It is not possible to develop a test cycle that captures the full range of factors impacting fuel economy. However, it is clear that the FTP and HFET tests alone are missing some critical elements of real-world driving. All of these factors have a negative impact on fuel economy. This largely explains why our current estimates often do not reflect consumers' real-world fuel economy experience. However, since the 1985 adjustment factors were established, EPA has adopted several new test cycles for emission compliance purposes, which collectively represent a much broader range of in-use driving conditions than those captured by the FTP and HFET tests. These additional emission tests, discussed below, can be brought into the fuel economy estimate calculations.

3. Additional Emissions Tests Reflect a Broader Range of Real-World Driving Conditions

Since 1984 when we last updated the fuel economy estimate methodology, EPA has established several new test cycles for emissions certification. EPA was concerned that the FTP omitted many critical driving modes and conditions that existed in actual use, and that emissions could be substantially higher during these driving modes compared to the FTP. Manufacturers were frequently designing their vehicles' emission control systems to meet the specified FTP test conditions, and actual emission levels could be quite different under the broader range of real-world “off-cycle” conditions.

The need for these actions was recognized by Congress, in the passage of Sections 206(h) and 202(j) of the Clean Air Act Amendments of 1990 (CAAA).

Section 206(h) required EPA to study and revise as necessary the test procedures used to measure emissions, taking into consideration the actual current driving conditions under which motor vehicles are used, including conditions relating to fuel, temperature, acceleration, and altitude. Section 202(j) of the CAAA required EPA to establish emission standards for carbon monoxide under cold (20 deg. F) temperature conditions.

See 42 U.S.C. 7525(h), 42 U.S.C. 7521(j).

In 1992, EPA published rules implementing the 202(j) cold temperature testing requirement, acknowledging that the ambient temperature conditions of the FTP test (run between 68 and 86 °F) do not represent the full range of ambient temperature conditions that exist across the United States and that cold temperature had different emissions effects on different vehicle designs.

EPA's cold temperature emission regulations required manufacturers to conduct FTP testing at 20 °F. By promulgating this new test procedure and associated emission standard, EPA sought to encourage manufacturers to employ better emission control strategies that would improve ambient air quality across a wider range of in-use conditions.

See 57 FR 31888, July 17, 1992.

In fulfillment of the 206(h) CAAA requirement, EPA published a report in 1993 which concluded that the FTP cycle did not represent the full range of urban driving conditions that could impact the in-use driving emission levels.

Consequently, EPA promulgated a rule in 1996 that established two new test procedures, with associated emission standards, that addressed certain shortcomings with the current FTP.

Known as the “Supplemental FTP,” or “SFTP,” these procedures, similar to the cold temperature FTP, encouraged the use of the better emission controls across a wider range of in-use driving conditions in order to improve ambient air quality.

U.S. Environmental Protection Agency. Federal Test Procedure Review Project: Preliminary Technical Report. U.S. Environmental Protection Agency, No. EPA420-R-93-007, May 1993. Website:

http://www.epa.gov/otaq/sftp.htm.

See 61 FR 54854 published on October 22, 1996.

One of the new test cycles, the US06, was designed to address high speed, aggressive driving behavior (with more severe acceleration rates and speeds) as well as rapid and frequent speed fluctuations. The US06 test contains both lower-speed city driving and higher-speed highway driving modes.

Its top speed is 80 mph, and average speed is 48 mph. The top acceleration rate exceeds eight mph per second. The other new SFTP test, the SC03, was designed to address air-conditioner operation under a full simulation of high temperature (95 °F), high sun-load, and high humidity. The SC03 drive cycle was designed to represent driving immediately following a vehicle startup, and rapid and frequent speed fluctuations.

Its top speed is about 55 mph and average speed is 22 mph. The top acceleration rate is about five mph per second.

See 40 CFR Part 86 Appendix I (g).

Ref. 40 CFR Part 86 Appendix I (h).

The basis for the SFTP rulemaking was a study of real-world driving in four cities, Baltimore, Spokane, Atlanta and Los Angeles, where driving activity was measured on instrumented vehicles as well as by chase cars.

17 18

At that time, it was found that 18 percent of the driving (in Baltimore) occurred outside of the speed/acceleration distribution of the FTP drive schedule. More recent real-world driving activity data indicates that driving has become even more aggressive than it was in 1992. Recent real-world activity data collected in California and Kansas City found that about 28 percent of driving (vehicle miles traveled) is at speeds greater than 60 mph. Further, about 33 percent of recent real-world driving falls outside of the FTP/HFET speed and acceleration activity region.

19 20 21 22

This is based on extensive chase car studies in California and instrumented vehicle studies in Kansas City. Our assessment of these recent real-world driving activity studies is described in detail in the Draft Technical Support Document.

Final Technical Report on Aggressive Driving Behavior for the Revised Federal Test Procedure Notice of Proposed Rulemaking, 1995. Website:

http://www.epa.gov/otaq/sftp.htm.

U.S. Environmental Protection Agency. Federal Test Procedure Review Project: Preliminary Technical Report. U.S. Environmental Protection Agency, No. EPA420-R-93-007, May 1993. Website:

http://www.epa.gov/otaq/sftp.htm.

Sierra Research, Inc., “Task Order No. 2 SCF Improvement—Field Data Collection,” Sierra Report No. SR02-07-04, July, 2002.

U.S. EPA Draft Technical Support Document “Fuel Economy Labeling of Motor Vehicles: Revisions to Improve Calculation of Fuel Economy Estimates,” December, 2005.

Brzezinski, D., E. Nam, J. Koupal, G. Hoffman. Changes in Real World Driving Behavior: Analysis of Recent Driving Activity Data. Proceedings of the 15th Coordinating Research Council On Road Vehicle Emissions Workshop, 2005.

Eastern Research Group. Late Model Vehicle Emissions and Fuel Economy Characterization Study: Addendum to the Kansas City Exhaust Characterization Study-Draft Report. ERG No. 0133.18.004.001, September 26, 2005.

Clearly, the FTP and HFET tests alone do not fully capture the broad range of real-world driving conditions. In order for EPA's fuel economy tests to be more representative of key aspects of real-world driving, it is critical that we consider the test conditions represented by these additional emission tests.

4. Fuel Economy on Driving Modes Represented by Additional Emissions Tests is Lower for Many Vehicles

As discussed above, there are several key conditions missing from the current fuel economy test procedures that are prevalent in real-world driving. These conditions—higher speeds, faster

accelerations, air conditioning operation, and cold temperatures—have already been incorporated into our test procedures for emissions compliance, as a result of our finding in the 1990's that they have a significant impact on emissions. Our analysis below demonstrates that these additional driving conditions can also have a significant impact on fuel economy—and that these impacts vary widely from vehicle to vehicle. Thus, we believe that these factors need to be included in our fuel economy test methods.

We analyzed fuel economy data collected by manufacturers for emissions certification purposes in the 2003, 2004 and 2005 model years. This analysis included data from all five tests used for emissions compliance today, including the FTP, HFET, US06, SC03, and Cold Temperature FTP. The fuel economy measured on the standard fuel economy tests (FTP and HFET) was compared to the fuel economy on the other emissions certification tests (US06, SC03, and Cold FTP) in order to assess the impact of these factors on fuel economy. The analysis includes data from more than 400 vehicles. Comparisons were made to the unadjusted city and highway fuel economy test results, and the findings are summarized below. Because so many other factors bear on real-world consumer experience, it is important to point out that these comparisons are not intended to indicate the exact impact of a given factor on real-world fuel economy. However, comparing these different test results is informative because we establish the relative magnitude of the impacts and of the variation across vehicles. The entire report of this analysis is in the docket for this rulemaking.

U.S. Environmental Protection Agency, Office of Transportation and Air Quality, “Vehicle Fuel Economy Labeling and The Effect of Cold Temperature, Air-Conditioning Usage and Aggressive Driving on Fuel Economy,” Draft Staff Report, August 2005.

Cold Temperature Operation.

To assess the impact of cold temperature operation on fuel economy, we compared the fuel economy measured over the Cold FTP test directly to that over the standard FTP test. The driving cycles in these two tests are identical (i.e., the LA4 cycle). Both tests include both cold and hot starts at their respective ambient temperatures, and both tests are generally run with accessories turned off. The difference in fuel economy should therefore be entirely due to the difference in ambient temperature: 20 °F versus 75 °F.

On average, fuel economy over the Cold FTP was about 12 percent lower than over the standard FTP. There was wide vehicle-to-vehicle variation, with the loss in fuel economy due to the cold conditions as much as 40 percent. Figure I.B-1 below shows the range of cold temperature impacts. Hybrid vehicles tended to show the greatest sensitivity to cold temperature. Of the six vehicles showing a cold temperature impact of greater than 30 percent, five are hybrids. Overall, conventional gasoline vehicles averaged a cold temperature effect of about −11 percent, while the impact on hybrid vehicles averaged about −32 percent.

EP01FE06.000

Air Conditioning.

To assess the impact of air conditioning on fuel economy, we compared the fuel economy measured over the SC03 test to a comparable portion of the FTP. The SC03 test is run with the air-conditioning turned onto its maximum setting in a test cell set at 95 °F with strong sun load and moderate humidity. On average, air conditioner operation at 95 °F reduced fuel economy by about 21 percent. The impact of air conditioning ranged from −41 percent to −25 percent for more than a third of the vehicles. Similar to the cold temperature impacts, there was a great deal of vehicle-to-vehicle variation in the impact of air conditioning on fuel economy. Figure I.B-2 shows the distribution of the percentage differences (negative numbers indicate lower fuel economy over SC03). As can be seen in the figure, the vast majority of vehicles show an impact of −27.5

percent to −7.5 percent. Hybrid vehicles tended to show greater sensitivity to air conditioning operation than conventional vehicles. The effect of air conditioning operation reduced hybrid fuel economy by 31 percent, 50 percent greater than the 20 percent impact on conventional vehicle fuel economy.

EP01FE06.001

Aggressive and High-Speed Driving.

The US06 test was designed to address aggressive driving behavior, such as high acceleration rates and high speeds. The US06 test contains both lower-speed but aggressive urban driving and higher-speed highway driving modes. Because of the different driving modes contained on the US06 test, for the purpose of assessing the impacts of high speed and aggressive driving we developed a combination of the city and highway tests which is roughly comparable to that contained in the US06 cycle.

On average, the fuel economy over the US06 cycle was almost 30 percent lower than over the composite FTP and HFET fuel economy. The observed impacts ranged from −44 percent to −25 percent for more than 80 percent of the vehicles. Figure I.B-3 shows the distribution of per vehicle impacts due to the aggressive driving of the US06 cycle. Hybrid vehicles showed a slightly greater impact of aggressive driving on fuel economy than conventional gasoline vehicles (33 percent versus 29 percent, respectively).

EP01FE06.002

Conclusions.

Many of the vehicles whose fuel economies were most affected by these driving conditions were hybrids and other high mile-per-gallon vehicles. In general, high mpg vehicles will be more sensitive to changes in driving conditions for two reasons. One, because they use relatively little fuel in the first place, any increase in fuel consumption will show up as a relatively larger percentage fuel consumption increase. Two, because of the non-linearity of fuel economy with respect to fuel consumption, an increase in fuel consumption will lower the fuel economy of a high mpg vehicle much more than it will lower the fuel economy of a low mpg vehicle. For example, the fuel consumption increase associated with a 35 mpg rating that actually achieves 30 mpg in the real-world is the same as a 15 mpg rating that actually achieves 14 mpg.

Hybrids, most of which achieve relatively high mpg and therefore share the issues discussed above, also face some additional challenges. Hybrids may well be the most significant powertrain technology innovation driven to market commercialization primarily because of its fuel economy potential. In addition, the nature of hybrid technology (the addition of a battery as a second source of on-board power, sophisticated control systems, sometimes a smaller engine) suggests that fuel economy will likely be more sensitive to certain conditions such as high acceleration and deceleration rates, cold ambient temperatures, etc. Finally, by industry standards, hybrids are a relatively young technology, and there is every reason to believe that as the technology matures, hybrid vehicle fuel economy will become much more robust over a broader range of driver behavior and climate conditions.

This analysis clearly shows that the driving conditions represented by US06, SC03 and Cold FTP tests can have substantial, measurable negative impact on fuel economy. There also is a large amount of vehicle-to-vehicle variation—that is, different vehicles are impacted differently by these factors. These findings call into question the appropriateness of the continued use of the current “one-size-fits-all” 10 and 22 percent adjustment factors applied, respectively, to FTP and HFET fuel economy test results. The FTP and HFET tests clearly do not adequately reflect the broad range of conditions that exist in today's real-world driving. The additional emission test cycles incorporate several critical factors that are present in real-world driving, and that can have a significant impact on fuel economy. Thus, these additional emission test cycles need to be brought into the fuel economy test methods, so that the estimates themselves will be more representative of the fuel economy consumers can expect to achieve in the real-world.

C. What New Requirements Are We Proposing?

We are proposing to revise and improve the methods used to determine the city and highway fuel economy estimates by incorporating fuel economy results over a broader range of driving conditions. An overview of this proposal is provided below. Section II provides a detailed explanation of the proposed new test methods, as well as the data and analysis upon which it is based.

In addition, we are proposing minor changes to revise the format and content of the fuel economy label to make the information more useful to consumers. We also are proposing minor changes related to the fuel economy information program, including revising the comparable vehicle classes and adding a new provision for the electronic distribution of the annual Fuel Economy Guide. An overview of each of these proposals follows.

1. Revised Test Methods for Calculating City and Highway Fuel Economy Estimates

Today's proposal would revise the test methods by which the city and highway fuel economy estimates are calculated. We are proposing to replace the current method of adjusting the city (FTP) test result downward by 10 percent and the highway (HFET) test result downward by 22 percent. Instead, we are proposing a new approach that incorporates additional test methods that address factors that impact fuel economy, but are missing from today's tests—specifically, higher speeds, more aggressive driving (e.g., higher acceleration rates), the use of air conditioning, and the effect of cold temperature. The proposed test methods

would bring into the fuel economy estimates the test results from the five emissions tests in place today: FTP, HFET, US06, SC03, and Cold FTP. Thus, we refer to this as the “5-cycle” method. Under our proposal, rather than basing the city mpg estimate solely on the adjusted FTP test result, and the highway mpg estimate solely on the adjusted HFET test result, each estimate would be based on a “composite” calculation of all five tests, weighting each appropriately to arrive at new city and highway mpg estimates. The new city and highway estimates would each be calculated according to separate city and highway “5-cycle” formulae that are based on fuel economy results over these five tests. The conditions represented by each test would be “weighted” according to how much they occur over average real-world city or highway driving. For example, we have derived weightings to represent driving cycle effects, trip length, air conditioner compressor-on usage, and operation over various temperatures. This methodology is described in detail in Section II.

We also are proposing a downward adjustment to account for effects that are not reflected in our existing five test cycles. There are many factors that impact fuel economy, but are difficult to account for in the test cell on the dynamometer. These include roadway roughness, road grade (hills), wind, tire pressure, heavier loads, hills, snow/ice, effects of ethanol in gasoline, larger vehicle loads (e.g., trailers, cargo, multiple passengers), and others. Current data indicates that these impacts can lower fuel economy from 9 to 13 percent. Thus, we need to account for these factors in our new test methods, as they will lower a driver's fuel economy beyond those factors we are accounting for from our existing test cycles. We are proposing an 11 percent downward adjustment to account for these non-dynamometer effects. Our basis for this downward adjustment factor is detailed in Section II.C.3 and the Draft Technical Support Document.

The 5-cycle approach, including this 11 percent downward adjustment factor to account for non-dynamometer effects, will result in city and highway estimates that reflect average fuel economy. We are proposing to continue to set the city and highway mpg estimates at the average, or mean, level. However, we understand that many drivers expect to achieve or exceed the fuel economy indicated by these mpg estimates. By continuing to set the estimates at the average level, by definition, half of drivers will get worse fuel economy than the label values. We seek comment on whether the city and highway estimates should be set a level that is lower than average—for example, to ensure that 75 percent, or even more, of drivers achieve or exceed the label values.

Because the 5-cycle method is inherently vehicle-specific, the difference between today's values and the new fuel economy estimates could vary widely from vehicle to vehicle. Today's proposed approach would result in city fuel economy estimates that are between 10 to 20 percent lower than today's labels for the majority of conventional vehicles. For vehicles that achieve generally better fuel economy, such as gasoline-electric hybrid vehicles, new city estimates would be about 20 to 30 percent lower than today's labels. The new highway fuel economy estimates would be 5 to 15 percent lower for the majority of vehicles, including hybrids.

Today's proposal would greatly improve the EPA fuel economy estimates, so that they come closer to the fuel economy that consumers achieve in the real-world. However, as discussed previously in this notice, these are still estimates. Even with the improved fuel economy test methods proposed today, some consumers will continue to get fuel economy that is higher or lower than the new estimates.

Under this new 5-cycle approach, some auto manufacturers have expressed concern about the potential for increased test burden. The three additional emission tests that we propose to include in the fuel economy calculation are run today on a much more limited number of vehicle groups than are the FTP and HFET tests. Typically, for every 3-4 FTP and HFET tests conducted, only one US06 or SC03 test is run, and cold FTP testing is even more limited. If we were to require full 5-cycle testing across all vehicle types, the testing demands for the auto industry could increase dramatically, and could trigger the need for a major expansion of their testing facilities.

Thus, we are proposing to implement the new fuel economy test methods in a way that gives the auto industry sufficient lead time to plan for their increased testing needs. This enables us to implement an improved fuel economy label methodology as soon as possible—in the 2008 model year. We also are implementing an approach that mitigates the testing burden where warranted. We have done this in two key ways.

First, for the first three model years (2008 through 2010), we would provide manufacturers with the option of using a scale of adjustments based on an analysis of data developed from the 5-cycle method. This approach, called the mpg-based approach, incorporates the effects of higher speed/aggressive driving, air conditioning use, and colder temperatures, but less directly than the 5-cycle vehicle-specific method. The mpg-based adjustments were derived by applying the 5-cycle formulae to a data set of recent fuel economy test data, and developing a regression line through the data. (See Section II for a full description of this approach). These adjustments differ based on the mpg a vehicle obtains over the FTP (City) or HFET (Highway) tests. In other words, every vehicle with the same mpg on the FTP test would receive the same adjustment for its city fuel economy label. Likewise, every vehicle with the same mpg on the HFET test would receive the same adjustment for its highway fuel economy label. This method of adjustment would not require any testing beyond the FTP/HFET tests already performed today, thus, it can be implemented sooner than the 5-cycle approach as an interim improvement to our fuel economy test methods. However, during this timeframe, manufacturers may choose to run full 5-cycle testing for any of their vehicle models. This approach would provide consumers with more accurate estimates, while allowing the industry the necessary lead time to prepare for the necessary testing under the 5-cycle approach.

Second, when we move to the 5-cycle vehicle-specific approach in model years 2011 and beyond, we are proposing criteria that would select specific vehicle groups for full 5-cycle testing, rather than requiring complete 5-cycle data generation for every vehicle. We believe this approach would result in fuel economy estimates that are generally as accurate as they would be under full 5-cycle testing. In other words, we are only requiring full 5-cycle testing where we can predict with reasonable certainty that the fuel economy results under the 5-cycle method would yield a significantly different result than the mpg-based adjustments.

We propose to establish a tolerance band around the mpg-based city and highway adjustment lines. Manufacturers would be required to calculate a 5-cycle fuel economy estimate for each vehicle group for which 5-cycle data exists for emissions purposes. If the 5-cycle fuel economy estimate for this vehicle group falls below the respective tolerance band around the mpg adjustment line, then the manufacturer would be eligible to use the mpg-based adjustments for each

vehicle configuration represented by that set of 5-cycle data. That is, the 5-cycle vehicle group may include within it several vehicle groupings, or specific vehicle model types, for which additional FTP/HFET data is available. The manufacturer would be able to use the MPG line to determine the fuel economy label adjustments for each of these model types with associated FTP/HFET test data. Fuller 5-cycle testing would be required for all vehicles represented by a vehicle group for which the 5-cycle fuel economy is below the tolerance bands. Section II further describes the level of these tolerance bands and how this concept would be implemented. A full discussion of our proposed methodology and results is contained in Section II.

2. Revised Label Format

To make the label more easily understood by consumers, we are also proposing changes to the fuel economy label format specified in the regulations. The proposed changes include updating the look of the label, simplifying its contents, and improving its graphics, among others. The purpose of these changes is to present the fuel economy information in a manner that is easier for the consumer to understand and use. The proposed changes are discussed in detail in Section IV.

3. Revised Comparable Vehicle Classes

The comparable vehicle classes are currently defined in EPA's fuel economy regulations. They are needed to fulfill the EPCA statutory requirement to provide fuel economy information about comparable vehicles on the label.

These classes were last revised in 1984. Since that time, there have been some significant changes to vehicle designs which warrant changes to the defined classes. Briefly, we are proposing to add SUV and Minivan classes, and to consolidate some classes which have become less prevalent in the market. This is discussed in more detail in Section V.

See 49 U.S.C. 32908(b)(1)(C).

4. Minor Changes in Certain Test Procedures

We are proposing minor procedural changes in certain test procedures. First, the US06 drive cycle contains elements of both city and highway types of driving, yet the exhaust sample is collected in only one “bag,” yielding one overall fuel economy result. In order to more accurately reflect the city portion of the drive cycle into the city fuel economy estimate, and the highway portion of the US06 into the highway fuel economy estimate, we are proposing a revised test protocol that would require collecting the exhaust sample into two bags, thus providing separate results from the city and highway portions. This has the benefit of more accurately capturing how a vehicle's fuel economy would be impacted over the various types of driving reflected in the cycle, but with very minimal cost impact.

Second, today diesel vehicles are not required to run the cold FTP test since they are currently exempt from the cold carbon monoxide standard. We are proposing that diesel vehicles be required to run this test for 5-cycle fuel economy purposes.

Finally, the current cold FTP test gives manufacturers the option, but does not require them to, run the heater or defroster while performing this test at 20 degrees F. We expect that in most cases in the real world, consumers would indeed be running these accessories in colder temperatures, which will impact their fuel economy. We also understand that some, but not all, manufacturers today do run these accessories during the test. Therefore, to ensure this test most accurately reflects real-world conditions, and to ensure these conditions are run uniformly across manufacturers, we are seeking comment on requiring manufacturers to run the heater and defroster while performing the cold FTP test.

5. Other Fuel Economy-Related Topics

In addition to the proposed fuel economy label calculations and label formats, we are proposing a few other changes related to the fuel economy labels and annual fuel economy booklet. These topics are discussed in Section V.

D. Today's Proposal Does Not Impact or Change CAFE Test Procedures

Today's proposal does not alter the FTE and HFET driving cycles, the measurement techniques or the calculation methods used to determine CAFE. EPCA requires that CAFE be determined from the EPA test procedures in place as of 1975 (or procedures that give comparable results), which are the city and highway tests of today, with a few small adjustments for minor procedural changes that have occurred since 1975.

Today's proposal will not adjust the CAFE calculations; the new method for calculating fuel economy label estimates will fall under regulations that are separate from the CAFE regulations (currently, the regulations for calculating CAFE are in 40 CFR 600.501-85 through 513-91).

See 49 U.S.C. 32904(c).

E. When Will the New Fuel Economy Estimates Take Effect?

We want the public to benefit from the improved information provided by the new fuel economy estimates as soon as possible. Therefore, we propose that these new regulations take effect with the 2008 model year, which will be available for sale at dealers in the fall of 2007. We believe this is the earliest possible date for implementation, since some manufacturers typically begin certifying model year 2008 vehicles as early as late 2006. We also encourage manufacturers to voluntarily utilize these new methods sooner, and are therefore proposing that manufacturers may voluntarily comply with the new regulations as soon as the final regulations are published.

F. How Will EPA Communicate to the Public the Transition Between the Old Label Values and New?

To ensure that the public understands the relationship between the old estimates and the new, EPA plans to conduct extensive public outreach concurrent with the implementation of a final rule. We will provide information about the new estimates and how to use them via web-based information, fact sheets, and other communication methods. This information will be designed to explain all aspects of any new calculation methods, including their impact on label estimates from previous model years.

G. Statutory Provisions and Legal Authority

1. EPCA

The statutory authority for today's proposal is provided by the Energy Policy and Conservation Act (EPCA). Most of the labeling provisions applicable to vehicle labeling and information are found at 49 U.S.C. 32908. This section restricts EPA's requirements for fuel economy labeling to automobiles rated at no more than 8,500 pounds gross vehicle weight. It requires manufacturers of automobiles to attach a fuel economy label to a prominent place on each automobile manufactured in a model year and also requires the dealers to maintain the label on the automobile.

EPCA requires EPA to promulgate regulations to measure and calculate fuel economy.

To the extent practicable, EPCA requires that fuel

economy tests be carried out with emissions tests performed under section 206 of the Clean Air Act (42 U.S.C. 7525).

EPA's resulting fuel economy regulations are found in 40 CFR Part 600. EPA has broad discretion in determining how to measure and calculate fuel economy for purposes of labeling under 49 U.S.C. 32908(b).

The fact that EPA's current fuel economy labeling regulations includes the reporting of separate “city” and “highway” fuel economy is a result of a series of EPA regulations as discussed in Section I.A. above. Thus, in developing today's proposal (discussed in Section III below), we considered, but ultimately are not proposing, other methodologies for reporting fuel economy.

See 49 U.S.C. 32908(b)(1).

See 49 U.S.C. 32904(c).

Id.

EPCA places testing restrictions on corporate average fuel economy (CAFE), discussed below. Today's proposal does not impact those restrictions.

EPCA imposed some specific requirements for the information to be included on the fuel economy label.

Today's proposal retains these items:

See 49 U.S.C. 32908(b)(2)(A) through (F).

a. The fuel economy of the automobile.

b. The estimated annual fuel cost of operating the automobile.

c. The range of fuel economy of comparable automobiles of all manufacturers.

d. A statement that a booklet is available from the dealer to assist in making a comparison of fuel economy of other automobiles manufactured by all manufacturers in that model year.

e. The amount of the automobile fuel efficiency tax imposed on the sale of the automobile under section 4064 of the Internal Revenue Code of 1986 (26 U.S.C. 4064).

f. Other information required or authorized by the Administrator that is related to the information required [within items a. through d.]

EPCA also defines “fuel economy” as the average number of miles traveled by an automobile for each gallon of gasoline (or equivalent amount of other fuel) used, as determined by EPA.

Thus, today's proposal retains the requirement to report fuel economy as miles-per-gallon.

See 49 U.S.C. 32901(a)(10).

EPCA requires EPA to prepare a fuel economy booklet containing information that is “simple and readily understandable.”

It further instructs DOE to publish and distribute the booklet. EPA is required to “prescribe regulations requiring dealers to make the booklet available to prospective buyers.”

This booklet is more commonly known as the annual “Fuel Economy Guide.”

See 49 U.S.C. 32908(c).

Id.

EPCA also contains statutory provisions for average fuel economy (known widely as “Corporate Average Fuel Economy,” or CAFE).

Under these provisions, EPA is required to prescribe testing and calculation procedures to measure fuel economy for each model and calculate average fuel economy for a manufacturer, using the same procedures that were used for 1975 model year passenger automobiles (weighted 55 percent urban cycle and 45 percent highway cycle), or procedures that give comparable results.

This requirement does not apply to the fuel economy information manufacturers apply to the fuel economy label required in 49 U.S.C. 32908(b).

See 49 U.S.C. 32902-32904.

See 49 U.S.C. 32904(c).

Id.

EPA is also required to consult with the Federal Trade Commission (FTC), DOT and DOE in carrying out the fuel economy information requirements in EPCA.

See 49 U.S.C. 32908(f).

2. Energy Policy Act of 2005

Section 774 of the Energy Policy Act of 2005 (EPAct) directs EPA to “update or revise the adjustment factors in sections 600.209-85 and 600.209-95, of the Code of Federal Regulations, CFR Part 600 (1995) Fuel Economy Regulations for 1977 and Later Model Year Automobiles to take into consideration higher speed limits, faster acceleration rates, variations in temperature, use of air conditioning, shorter city test cycle lengths, current reference fuels, and the use of other fuel depleting features.”

See Pub. L. 109-58, 119 Stat. 835 (2005).

In today's proposal, the 5-cycle approach changes the adjustment factors by establishing a new method to calculate fuel economy estimates that uses fuel economy results from additional test procedures combined with a changed adjustment factor. The mpg-based approach uses the same test methods as the current fuel economy program (i.e., the FTP and HFET tests), but changes the adjustment factors applied to those test results. These options satisfy the EPAct provisions as follows.

First, the 5-cycle method proposed today directly includes the effects of higher speed limits, faster acceleration rates, variations in temperature, and use of air conditioning by including fuel economy measured during tests that incorporate these features. The mpg-based approach also takes these factors into consideration, but less directly, as it incorporates the effects of these factors by basing the adjustment factor on an analysis of data developed from the 5-cycle method. Under our proposal, we use the mpg-based approach as an interim option to establish an appropriate period of lead time for manufacturers. We also allow its continued use only where the average effects reflected under the mpg-based adjustments (of higher speed/acceleration, air conditioning, and cold temperature) on a specific vehicle configuration would be representative of those measured under actual 5-cycle testing.

Second, we interpret the statute's reference to “shorter city test cycle lengths” to mean shorter than the current FTP cycle used to determine city fuel economy. We have addressed that concern in the proposal by weighting in updated factors for “cold starts” and “hot starts” (where the engine is not warmed up or has been parked for a brief amount of time and then restarted) into the equation for determining city fuel economy. This simulates shorter city test cycle lengths where a vehicle's engine is more frequently shut down and restarted than in the current FTP test. Also, the US06 and SC03 test cycles are physically shorter in length than the FTP (the FTP is about 11 miles in length, whereas the US06 is about 8 miles, and the SC03 is about 3.6 miles.)

Third, we interpret the statutory reference to “current reference fuels” to mean the laboratory fuels used to perform the fuel economy tests, and that the underlying concern of Congress was that the high-quality lab fuels would give higher fuel economy than the typical fuel used by consumers. The quality of the laboratory test fuel is specified in EPA regulations for emission compliance.

The test gasoline fuel is roughly equivalent to premium, high-octane fuel available at the pump. It is necessary that all vehicles use the same grade of fuel to provide a level playing field for manufacturers to compare the emission compliance results to the federal emission standards, since certain fuel specifications can have an impact on tailpipe emissions. The impact of the higher-octane test fuel on fuel economy is less significant but there are other real-world fuel differences that can have a noticeable impact, as discussed in Section II. For instance, ethanol has a lower energy content than gasoline, and

when blended with gasoline, with all other things being equal, will slightly lower fuel efficiency. Other seasonal variations in fuel composition (

e.g.

, oxygenates in winter fuel) may also cause a slight reduction in fuel economy. EPA is proposing an adjustment factor to account for fuel differences and other fuel-depleting features as described further in Section II.

See 40 CFR 86.113-94.

3. Relationship of Today's Proposal With Other Statutes and Regulations

Automobile Disclosure Act

. A provision in EPCA (at 49 U.S.C. 32908(b)(2)) allows the fuel economy information to be included on the window sticker label of vehicle manufacturing and price information required by the Automobile Disclosure Act at 15 U.S.C. 1232 (the so-called “Monroni” label.). To that end, the Federal Trade Commission issued a “Fuel Guide” concerning the fuel economy advertising for new automobiles, published in the

Federal Register

at 16 CFR Part 259. This guide refers back to EPA's fuel economy regulations and specifically to how manufacturers are permitted to advertise the city and highway fuel economy of their vehicles.

Internal Revenue Code.

This code contains the provisions governing the administration of the Gas Guzzler Tax.

It contains the table of applicable taxes and defines which vehicles are subject to the taxes. The IRS code specifies that the fuel economy to be used to assess the amount of tax will be the combined city and highway fuel economy as determined by using the procedures in place in 1975, or procedures that give comparable results (similar to EPCA's requirements for determining CAFE). Today's proposal does not impact these procedures.

See 26 U.S.C. 4064.

Clean Air Act.

Reference is made in EPCA to the Clean Air Act statute. Specifically, EPCA states that fuel economy shall to the extent practicable include the emissions tests required under Section 206 of the Clean Air Act.

Today's proposal incorporates three additional types of emissions tests required under the Clean Air Act for fuel economy testing, as discussed in detail in Section II. We also propose to make several changes to existing emissions tests. These changes are being proposed under the statutory authority of Section 206 of the Clean Air Act, which permits the Administrator to define, and to revise from time to time, the test procedures used to determine compliance with applicable emission standards.

See 49 U.S.C. 32904(c).

Additional Provisions in the Energy Policy Act of 2005 and Transportation Equity Act of 2005.

This action is expected to have no impact on the alternative motor vehicle federal income tax credits the Internal Revenue Service (IRS) is establishing under Section 1341 of the Energy Policy Act of 2005. IRS is in the process of preparing the final guidance for these new federal income tax credits for consumers who purchase new hybrid, diesel, dedicated alternative fuel, or fuel cell vehicles beginning on January 1, 2006. The Energy Policy Act of 2005 requires EPA to coordinate with and support IRS' implementation of these new tax credits, and EPA is providing input on a number of technical issues. EPA anticipates that the fuel economy values used to help determine tax credit eligibility for light-duty vehicles will be “unadjusted” laboratory city fuel economy test values. Accordingly, the changes being proposed today are anticipated to have no impact on the tax credit program.

Similarly, this action is expected to have no impact on the “HOV Facilities” regulations EPA is establishing under section 1121 of the Transportation Equity Act of 2005. EPA is in the process of developing proposed regulations to identify low emission and energy-efficient vehicles for the purpose of assisting states administering high-occupancy lane transportation plans. EPA anticipates that the fuel economy values used to identify these vehicles will be the “unadjusted” FTP-based fuel economy test values. Accordingly, the changes proposed today are anticipated to have no impact on the HOV facilities program.

II. Description of the Proposed Fuel Economy Label Methodology

The current fuel economy label values utilize measured fuel economy over city and highway driving cycles and adjust these values downward by 10 and 22 percent, respectively, to account for a variety of factors not addressed in EPA's vehicle test procedures. These factors include differences between the way vehicles are driven on the road and over the test cycles, air conditioning use, widely varying ambient temperature and humidity, varying trip lengths, wind, precipitation, rough road conditions, hills, etc. The purpose of the new formulae for city and highway fuel economy labels is to widen the base for the labels to include actual vehicle testing over a wider range of driving patterns and ambient conditions than is currently covered by the FTP and HFET tests.

For example, vehicles are often driven more aggressively and at higher speeds than is represented in the FTP and HFET tests. The incorporation of measured fuel economy over the US06 test cycle into the fuel economy label values would make the label values more realistic. Drivers often use air conditioning in warm, humid conditions, while the air conditioner is turned off during the FTP and HFET tests. The incorporation of measured fuel economy over the SC03 test cycle into the fuel economy label values would reflect the added fuel needed to operate the air conditioning system. Vehicles also often are driven at temperatures below 75 degrees Fahrenheit (F), at which the FTP and HFET tests are performed. The incorporation of measured fuel economy over the cold temperature FTP test into the fuel economy label values would reflect the additional fuel needed to start up a cold engine at colder temperatures.

The proposed vehicle-specific, 5-cycle approach to fuel economy label estimation would incorporate estimates of the fuel efficiency of each vehicle during high speed, aggressive driving, air conditioning operation and cold temperatures into each vehicle's fuel economy label. It would combine measured fuel economy over the two current fuel economy tests, the FTP and HFET, as well as that over the US06, SC03 and cold FTP tests into estimates of city and highway fuel economy for labeling purposes. The test results from each cycle (and in some cases, portions of cycles or emission “bags”)

would be weighted to represent the contribution of each cycle's attributes to onroad driving and fuel consumption. The vehicle-specific, 5-cycle approach would eliminate the need to account for the effect of aggressive driving, air conditioning use and colder temperatures on fuel economy through generic factors (as done today) which may not reflect that particular vehicle's sensitivity to these factors. A generic adjustment would still be necessary to

account for factors not addressed by any of the five dynamometer tests. The magnitude of such an adjustment is comparable to today's 10 and 22 percent generic adjustments. Overall, under the vehicle specific 5-cycle approach, each vehicle's label fuel economy would better reflect the capabilities of that vehicle on the road.

The FTP consists of two parts, referred to in the regulations as the “cold start” test and the “hot start” test. Each of these parts is divided into two periods, or “phases': A “transient” phase and a “stabilized” phase. Because the stabilized phase of the hot start test is assumed to be identical to the stabilized phase of the cold start test, only the cold start stabilized phase is typically run. These “phases” are often called “bags,” terminology that results from the sample bags in which the exhaust samples are collected. The phases are run in the following order: Cold start transient (Bag 1), cold start stabilized (Bag 2), and hot start transient (Bag 3).

Currently, the US06, SC03 and cold FTP tests are only performed on a sub-set of new vehicle configurations. In contrast, for fuel economy purposes, FTP and HFET tests are performed on many more vehicle configurations. In order to minimize the number of additional US06, SC03 and cold FTP tests resulting from this proposal, we are proposing that manufacturers be allowed to estimate the fuel economy over these three tests for vehicle configurations that are not normally tested for emission compliance purposes using the fuel economy measurements that are normally available. This is currently done on a more limited basis for both the FTP and HFET, and is referred to as analytically derived fuel economy (ADFE).

We are also proposing that manufacturers be allowed to use the interim approach to fuel economy label estimation, the mpg-based approach, indefinitely when the available 5-cycle fuel economy data indicate that a vehicle's specific 5-cycle fuel economy is very close to that estimated by the mpg-based curve.

EPA's current policy for analytically derived fuel economy estimates for the FTP and HFET tests is contained in the EPA memorandum entitled, “Updated Analytically Derived Fuel Economy (ADFE) Policy for 2005 Model Year,” March 11, 2004, CCD-04-06 (LDV/LDT).

Even with these policies, we expect that some manufacturers would have to perform some additional US06, SC03, or cold FTP tests to address differences in vehicle designs which are not covered by the analytical derivation methodology. Other manufacturers may decide to perform additional tests simply to improve accuracy over the analytical derivation methodology. Depending on how manufacturers choose to apply this method, this additional testing could involve the construction of additional test facilities. (Test burden issues are discussed further in Section VI of this preamble.) Therefore, in order to allow sufficient lead-time for the construction of these facilities, we are proposing to allow manufacturers the option of using an alternative, interim set of adjustments through the 2010 model year until the 5-cycle approach becomes mandatory with the 2011 model year. However, a manufacturer can still use the 5-cycle formula prior to the 2011 model year for specific vehicle models, if it so desires.

The interim set of adjustments is termed the “mpg-based” adjustment. (See Figure II-1 in the following section for a graphical depiction of these adjustments.) The mpg-based approach is a sliding scale of adjustments which varies according to a vehicle's measured fuel economy over the FTP and HFET tests. The mpg-based adjustment factors were developed from applying the 5-cycle formulae to 423 recent model year vehicles and determining the average difference between the 5-cycle and current city and highway fuel economies. Thus, because the data used to develop the average adjustment factors were derived from 5-cycle fuel economies, the mpg-based adjustment factors include the effect of high speeds, aggressive driving, air conditioning, and colder temperatures. However, they do so based on the impact of these factors on the average vehicle, not the individual vehicle, which is the case with the 5-cycle formulae. For example, for vehicles with FTP fuel economy of 20-30 mpg, the mpg-based approach would adjust the FTP fuel economy downward by 22-24 percent, versus today's 10 percent downward adjustment. Thus, city fuel economy label values under the mpg-based approach tend to be about 13-15 percent lower than today's label values. For vehicles with HFET fuel economy of 25-35 mpg, the mpg-based approach would adjust the HFET fuel economy downward by 29 percent, versus today's 22 percent downward adjustment. Thus, highway fuel economy label values under the mpg-based approach would tend to be about 9 percent lower than today's label values.

As mentioned above, the mpg-based equations described above were developed from the 5-cycle fuel economy estimates for 423 2003-2005 model year vehicles. We propose to update the mpg-based curves periodically using all of the available 5-cycle fuel economy estimates for the previous three or more model years. These revised mpg-based equations would be issued through the publication of an EPA guidance document. EPA would publish the mpg-based equations by January 1 of the calendar year prior to the model year to which the equations first apply (e.g., for model year 2010 fuel economy calculations the equations would be made available before January 1, 2009). In order to keep the mpg-based equations up-to-date and based on recent technology vehicles, EPA would update these equations periodically, but no more than on an annual basis. However, rather than publish the equations applicable to 2008 model year vehicles via guidance, the proposed regulations contain the equations that would be applicable to 2008 model year vehicles, as well as the components of the equations to be utilized for future model year vehicles. We request comment on this updating of the mpg-based equations.

In addition to proposing the mpg-based adjustment factors for the 2008-2010 model years, as mentioned above, we propose to allow use of this method of label estimation to be used for 2011 and later model years for those vehicles which meet certain criteria (discussed in detail below) that indicate that the full 5-cycle testing would not likely result in significantly different fuel economy label values. Each year, a number of vehicles are tested over all five dynamometer test cycles for emission certification purposes (i.e., emission data vehicles). The fuel economy data for the five dynamometer test cycles for each emission data vehicle can be inserted into the 5-cycle formulae and the 5-cycle city and highway fuel economy values determined. Emission data vehicles also undergo testing over the FTP and HFET. Thus, the mpg-based city and highway fuel economy values for each emission data vehicle can also be determined using the available FTP and HFET fuel economy values. The 5-cycle city and highway fuel economy values can be compared to the mpg-based city and highway fuel economy values, respectively, for each emission data vehicle.

The mpg-based line represents the effects of high speed, high acceleration, air conditioning, and colder temperatures of the average new vehicle. Therefore, we believe that it is reasonable to allow continued use of the mpg-based line when the available 5-cycle fuel economy data (from emissions certification testing) indicates that the particular vehicle design reflects at least these average effects. To accomplish this, we defined the lower bound of a tolerance band around the mpg-based line as the criteria for whether the mpg-based line could be used or whether 5-cycle testing would be required. We chose four and five percent as the tolerance bands for the 5-cycle city and 5-cycle highway fuel economy values, respectively. Mathematically, the tolerance line is defined by Y × mpg-based fuel economy, where Y is 0.96 for city fuel economy and 0.95 for highway fuel economy. In other words, if the 5-cycle city fuel economy value is greater than 0.96 times the mpg-based city fuel economy, all the vehicle configurations

represented by the emission data vehicle (i.e., all vehicles within the vehicle test group) would be eligible to use the mpg-based approach. Similarly, when the 5-cycle highway fuel economy is less than the mpg-based highway fuel economy minus five percent, all vehicle configurations represented by the emission data vehicle would be required to use the vehicle-specific 5-cycle approach. This could be done using ADFE estimates, when appropriate. This approach is appropriate because those vehicles above the upper tolerance band that used the mpg-based line would simply be reducing their fuel economy down to the average level, even though the 5-cycle data indicated better than average performance was likely for that vehicle group. Because of the better-than-average performance, we expect that most manufacturers will want to do complete 5-cycle testing for vehicles likely to be above the upper tolerance band. However, we request comment on whether there may be some inherent variability regarding all outliers above

and

below the tolerance band that would make it desirable to require 5-cycle testing in all of these cases.

If the 5-cycle city fuel economy fell below the mpg-based city fuel economy by more than four percent, but the 5-cycle highway fuel economy did not fall below the mpg-based highway fuel economy by more than five percent, all the vehicle configurations represented by the emission data vehicle would be required to use the vehicle-specific 5-cycle approach for both city and highway fuel economy, since fuel economy values for all five cycles are important in estimating 5-cycle city fuel economy. However, if the 5-cycle highway fuel economy was less than the mpg-based highway fuel economy by more than five percent, but the 5-cycle city fuel economy was not more than four percent lower than the mpg-based city fuel economy, all the vehicle configurations represented by the emission data vehicle would use mpg-based approach to estimate the city fuel economy label. For highway label estimation, all the vehicle configurations represented by the emission data vehicle would use an approximate 5-cycle formula for highway fuel economy which includes vehicle-specific fuel economy measurements for the FTP, HFET and US06 tests, but the values for the SC03 and cold FTP tests could be estimated based on relationships developed from other vehicles. This is appropriate because the impact of the cold FTP test on highway fuel economy is not vehicle-specific, but modeled. Also the impact of the SC03 test on highway fuel economy is very small, particularly compared to that for the US06 test.

The proposed criteria for long term use of the mpg-based approach (5-cycle city fuel economy above −4.0 percent and 5-cycle highway fuel economy above −5.0 percent) are based on the balance of three factors. One, we designed them to be sufficiently large so that simple test-to-test variability would not cause an emission data vehicle to fail the criteria. This was a greater concern for the highway fuel economy comparison, due to the dominance of the US06 fuel economy (which inherently has greater test-to-test variability than the other tests) in the 5-cycle formula. Two, we desired to minimize the potential error in the fuel economy label. Label fuel economy values are rounded to the nearest one mpg. Thus, we desired to keep the difference between the 5-cycle and mpg-based fuel economy values within roughly one mpg, if possible. Three, we desired to avoid additional fuel economy testing that had little impact on the label values.

The four percent tolerance band for city fuel economy is equivalent to roughly 0.6-0.7 mpg on average. Due to the contribution of a number of independent fuel economy measurements in the 5-cycle city fuel economy formula, the effect of test-to-test variability should be much lower than 4.0 percent. Based on the 5-cycle test results of 423 recent model year vehicles, we estimate that 90 percent of all emission data vehicles would meet the 4.0 percent. Thus, we believe that this criterion adequately satisfies the three factors mentioned above.

The five percent tolerance band for highway fuel economy is equivalent to roughly 1.1 mpg on average. Thus, it is slightly higher than the typical error associated with rounding. However, due to the dominant contribution of the US06 fuel economy in the 5-cycle highway fuel economy formula, and the fact that this test tends to have relatively high variability, we are concerned that test-to-test variability could be on the order of 3.0 percent in the 5-cycle highway fuel economy formula. We estimate that 75 percent of all emission data vehicles would meet the 5.0 percent. Thus, again, we believe that this criterion adequately satisfies the three factors mentioned above.

Overall, allowing the continued use of the mpg-based approach would reduce the number of additional SC03 and cold FTP tests by about 90 percent and reduce the number of additional US06 tests by about 75 percent indefinitely. We request comment on the continued use of the mpg-based approach beyond the 2010 model year and on the 4.0 and 5.0 percent criteria for its use.

Section II.A presents the proposed interim mpg-based formulae and the proposed vehicle-specific 5-cycle formulae for city and highway fuel economy label values. Section II.B describes how these formulae would be applied to develop labels for specific grouping of vehicles. Section II.C describes how the 5-cycle formulae were derived. Section II.D describes how the mpg-based formulae were derived. Section II.E describes how the current city and highway fuel economy values would change under the proposed formulae.

A. Proposed Fuel Economy Label Formulae

Currently, manufacturers test their vehicles over two dynamometer tests in order to develop their fuel economy label values: the FTP or city test and the HFET or highway test. Fuel economies measured over these two tests are multiplied by 0.90 and 0.78, respectively. These “adjusted” fuel economies are then sales-weighted using procedures outlined in Subpart D of Part 600 of Title 40 of the Code of Federal Regulations (CFR) to develop fuel economy label values by model type.

Under today's proposal, we would replace the 0.90 and 0.78 factors with new factors which are not simply constants. For model years 2008-2010, a manufacturer would have the option of using two distinct methodologies to calculate the city and highway fuel economy values for any specific test vehicle. One approach is called the mpg-based approach or formula, since the city and highway label values are based on the fuel economy (or MPG) measured over the FTP and HFET, respectively. The other approach is called the vehicle-specific 5-cycle approach, since the city and highway label values are based on the test results of five test cycles, the FTP, HFET, US06, SC03 and cold FTP. Beginning with the 2011 model year, we propose that manufacturers would use the vehicle-specific 5-cycle method, but that the mpg-based approach could still be used by qualifying vehicles. Below we present the specific equations under the two approaches which would be used to convert fuel economies measured over the dynamometer cycles into city and highway fuel economy values prior to sales weighting. We are not proposing any changes to the methods for combining city and highway fuel economy values for specific vehicles into label values for a model type.

The formulae for the 5-cycle approach are, as indicated by its name, based on the fuel economy measurements over the five test cycles (FTP, HFET, US06, SC03 and cold FTP). Both approaches also include an additional downward adjustment to represent effects impossible to incorporate in laboratory dynamometer testing. However, the formulae for the mpg-based approach are also based on fuel economy measurements over the five test cycles. The difference is the set of 5-cycle fuel economy measurements that are used. Under the vehicle-specific 5-cycle approach, the fuel economy measurements over the 5 dynamometer test cycles would all be performed on (or estimated for) a specific vehicle in the current model year. Under the mpg-based approach, historic fuel economy data over the 5 test cycles would have been analyzed to produce a fleet-wide average relationship between (1) FTP fuel economy and 5-cycle city fuel economy, and (2) HFET fuel economy and 5-cycle highway fuel economy. Under the mpg-based approach, a specific vehicle's city and highway fuel economy labels are based on this fleet-wide average relationship, as opposed to that vehicle's own results over the 5 test cycles. In other words, every vehicle with the same measured FTP fuel economy would receive the same city fuel economy label value. Likewise, every vehicle with the same measured HFET fuel economy would receive the same highway fuel economy label value. Figure II-1 shows the 5-cycle city fuel economy for 423 recent model year vehicles and the mpg-based city fuel curve which has been developed from these data. The horizontal axis is the measured FTP fuel economy.

EP01FE06.003

Application of the 5-cycle approach to these vehicles would have produced the city fuel economy values indicated by the diamonds in the plot. (The nine hybrid vehicles are indicated by large squares.) Application of the mpg-based formula to these vehicles would have produced city fuel economy values by reading a number off of the curved line in the plot.

Figure II-2 shows the 5-cycle highway fuel economy for the same 423 recent model year vehicles and the mpg-based highway fuel economies which have been developed from these data. The horizontal axis is the measured HFET fuel economy.

EP01FE06.004

Both Figure II-1 and II-2 include several data points which are represented by large squares. These are vehicles which incorporate hybrid technology. Hybrids appear to fall well below the mpg-based curve for city fuel economy, but not for highway fuel economy. This issue will be discussed in more detail below.

Given that both approaches utilize the 5-cycle fuel economy formulae in some fashion, it is useful to begin this section with a description of how the fuel economy measured over the 5 test cycles are combined to represent onroad city and highway fuel economy. Then we will describe how the fleet-average formulae for the mpg-based approach were derived from these 5-cycle fuel economy estimates.

The 5-cycle formulae are derived from extensive data on real-world driving conditions, such as driving activity, temperatures, air conditioner operation, trip length, and other factors. In this section and in the Draft Technical Support Document, we fully describe the basis for developing these formulae. We seek comment on all aspects of the formulae and the underlying data upon which they are based. We also encourage interested parties to submit any additional data that would be relevant in our final analysis. Further, we want to ensure the 5-cycle approach continues in future years to reflect updated conditions impacting real-world fuel economy. Therefore, we encourage the public to submit any such data in the future so that EPA may assess such new information and evaluate the need for changes to this approach over time.

Since our goal is to develop a consistent, objective approach that applies to all vehicles, we have assumed that all types of vehicles are driven and maintained similarly, and we have proposed to weight the five driving cycles and apply non-dynomometer adjustments in the same way for all types of vehicles. However, if data showed that a specific type of vehicle is driven or maintained very differently, and this impacted fuel economy significantly (

e.g.

, an unusually low incidence of aggressive driving, A/C usage, etc.), then one might consider different weights or adjustment factors on this basis. We seek comment on any data that would inform whether unique weighting factors or non-dynomometer adjustments should be considered for specific vehicle technologies (

e.g.

, hybrids or diesels). For example, hybrids may be purchased preferentially by people whose driving patterns take advantage of their performance characteristics, and hybrid owners may be more conscious of driving techniques (such as mild braking) that improve fuel economy. Even if this were the case today, this difference would not necessarily persist as hybrids become more prevalent in the fleet. Moreover, it is not clear how such vehicle technology-specific factors can or should be reflected in EPA's fuel economy test methods or calculations. We seek comment on the contribution of such factors to the on-road fuel economy experience of consumers, and on the relevance of these factors to the fuel economy label. We also seek comment on the extent to which such unique factors might reduce the perceived objectivity of the fuel economy estimates if they presume differences in driving behavior.

1. MPG-Based Approach (Available in 2008-2010 Model Years)

Under the mpg-based approach, the city fuel economy value would be calculated as follows:

EP01FE06.005

where

FTP FE = the fuel economy in miles per gallon of fuel during the FTP test conducted at an ambient temperature of 75 °F.

This value is normally a sales-weighted average of the vehicle models included in the “fuel economy grouping” (e.g., model type) as defined in 40 CFR 600.002-93.

Likewise, the highway fuel economy value would be calculated as follows:

EP01FE06.006

where

HFET FE = fuel economy in mile per gallon over the HFET test.

This value is normally a sales-weighted average of the vehicle models included in the “fuel economy grouping” (e.g., model type) as defined in 40 CFR 600.002-93.

The rationale for the various constants in Equations (1) and (2) is described in Section II.B.

2. Vehicle-Specific 5-Cycle Approach (Applicable to 2011 and Later Model Years and Optional in Prior Model Years)

Under the vehicle-specific 5-cycle approach, the city fuel economy value would be calculated as follows:

EP01FE06.007

, where

EP01FE06.008

where,

EP01FE06.009

or,

EP01FE06.010

where

Bag y FE

= the fuel economy in miles per gallon of fuel during the specified bag of the FTP test conducted at an ambient temperature of 75 ° or 20°

F. The rationale for the various constants in the equations is described below in Section II.B. Likewise,

EP01FE06.011

where

US06 FE = fuel economy in mile per gallon over the US06 test,

HFET FE = fuel economy in mile per gallon over the HFET test,

SC03 FE = fuel economy in mile per gallon over the SC03 test.

Vehicles tested over a 4-bag FTP would substitute the fuel economy over Bag 4 for Bag 2 in the above equation.

Under the vehicle-specific 5-cycle formula, the highway fuel economy value would be calculated as follows:

EP01FE06.012

, where

EP01FE06.013

EP01FE06.014

where the various symbols have the same definitions as described under the formula for the vehicle-specific 5-cycle city fuel economy value.

B. Application of the Formulae To Develop Fuel Economy Labels for Specific Vehicles

We are not proposing any major changes to the way that vehicle configurations are grouped for fuel economy labeling purposes. For model years 2008-2010, when the mpg-based formulae are applicable, there would be no change in the procedure by which specific vehicle labels are developed.

Since the mpg-based formulae are based solely on the current fuel economy test cycles, no additional tests would need to be conducted. Only the effective adjustment factors would be modified.

See 40 CFR 600 and relevant EPA guidance.

Starting with the 2011 model year, vehicle manufacturers would first utilize their available 5-cycle fuel economy testing of emission data vehicles to determine which test groups could utilize the mpg-based approach and which would have to use the vehicle-specific 5-cycle approach. The test groups for which their emission data vehicles passed the 4.0 percent and 5.0 percent criteria described above would face no additional testing requirements. Just as in 2008-2010, the mpg-based formulae would be applied to fuel economy values measured over the FTP and HFET already being performed and city and highway label values determined.

Figure II-3 shows how the 4.0 percent criterion would work for city fuel economy.

EP01FE06.015

The upper line in the figure is the mpg-based formula for city fuel economy. The lower line represents a difference of 4.0 percent from city fuel economy based on the mpg-based formula. The points shown in Figure II-3 represent city fuel economy of emission data vehicles estimated by the 5-cycle fuel economy formula. The model types represented by emission data vehicles whose 5-cycle city fuel economy values fall above the lower line would be allowed to use the mpg-based approach for that model year. The model types represented by emission data vehicles whose 5-cycle city fuel economy values fall below the lower bounding line would be required to use the 5-cycle approach for that model year. Implicit in this proposal is that manufacturers would be allowed to use the mpg-based approach for a particular test group if the 5-cycle fuel economy for an emission data vehicle exceeded the mpg-based curve by more than the 4.0 or 5.0 percent criteria on the high side, since this would result in a lower fuel economy label value.

The test groups for which their emission data vehicles did not pass the 4.0 percent and 5.0 percent criteria described above could face some additional testing requirements. All the vehicle sub-configurations contained in these test groups would require fuel economy values over all five cycles for

use in the 5-cycle city and highway fuel economy formulae. The city and highway label values produced by the 5-cycle fuel economy formulae would then be averaged and sales-weighted just as they are today. However, the fuel economy values over the five test cycles could be generated in either of two ways in most instances. One way would be to test the vehicle over the US06, SC03 and cold FTP tests (the FTP and HFET tests already being performed under current requirements). The other way would be estimate fuel economy values over the US06, SC03 and cold FTP tests analytically (i.e., ADFEs) from testing of a similar vehicle over these three cycles. Specifically, we propose to allow manufacturers to estimate the effect of differences in inertial test weight, road load horsepower and N/V ratio (the ratio of engine revolutions to vehicle speed when the vehicle is in its highest gear). A procedure to estimate the effect of these three vehicle parameters on FTP and HFET fuel economy has already been developed. We plan to work with manufacturers to develop analogous formulae for the US06, SC03 and cold FTP tests. We would implement these estimation procedures using agency guidance, as is currently done for FTP and HFET fuel economy.

It is possible for the 5-cycle fuel economy values to meet the above criteria for either city or highway fuel economy, but not the other. If the 5-cycle fuel economy values for a specific emission data vehicle are more than four percent below the mpg-based estimate for city fuel economy, but no more than five percent below the mpg-based estimate for highway fuel economy, all the vehicle configurations represented by that emission data vehicle would be required to use the 5-cycle formulae in complying with the fuel economy label requirements for both city and highway fuel economy. All five cycles play a significant role in the 5-cycle city fuel economy formula. Once the five tests have been performed for the city estimate, there is little reason not to use the same information to derive the highway fuel economy estimate.

We propose a different approach for the opposite situation. If the 5-cycle fuel economy values for a specific emission data vehicle are no more than four percent below the mpg-based estimate for city fuel economy, but more than five percent below the mpg-based estimate for highway fuel economy, all the vehicle configurations represented by that emission data vehicle would be allowed to use the mpg-based formulae in deriving the city fuel economy label value. The highway fuel economy value, however, would be based on an alternative, simplified 5-cycle formula as opposed to the full 5-cycle highway fuel economy formula. This alternative 5-cycle highway formula would be based on fuel economy values over the FTP, HFET and US06 tests. The impact of the SC03 and cold FTP tests is relatively small in the 5-cycle highway fuel economy formula, as explained in the Draft Technical Support Document.

This approach requires that we develop a simplified 5-cycle highway fuel economy formula which is consistent with the full 5-cycle formula. We developed this simplified formula using estimates of the average impact of the SC03 and cold FTP test results on 5-cycle highway fuel economy. In both cases, we estimated this average impact by regressing the impact of these test cycles on the 5-cycle highway fuel economy for the 423 vehicles in our certification database against fuel economy values which would be available from FTP, HFET and US06 testing. This analysis (described in detail in the Draft Technical Support Document) results in the following alternative calculation for highway fuel economy.

EP01FE06.016

EP01FE06.017

EP01FE06.018

EP01FE06.019

We expect that the continued use of the mpg-based approach and the development of analytical estimation procedures for US06, SC03 and cold FTP fuel economy would allow manufacturers to avoid the vast majority of additional tests that would have been required if every vehicle currently tested over the FTP and HFET tests had to be tested over the US06, SC03 and cold FTP tests. The option to use the mpg-based approach after 2010 should alone eliminate 90 percent of the potential need for additional SC03 and cold FTP testing and 75 percent of the potential need for US06 testing. At the same time, we expect that there would be some need for additional testing when the available estimation procedures mentioned above do not apply. For example, the current estimation procedures for FTP and HFET fuel economy address changes in axle ratio, tractive road load horsepower and inertia test weight. Differences involving changes in transmission design, engine displacement, turbo-charging, etc., require actual testing. We expect that a similar situation would exist with the estimation of US06, SC03 and cold FTP fuel economy.

We request comment on the appropriateness of the continued use of

the mpg-based approach beyond the 2010 model year. We also request comment on the appropriateness of the 4.0 and 5.0 percent tolerance bands for city and highway fuel economy, respectively. We also seek comment on alternative approaches that may employ concepts similar to the tolerance band, or other ways of extrapolating fuel economy test results to a broader group of vehicle configurations. We specifically request comment on an approach which would employ tighter criteria (

e.g.

, a tolerance of 3 percent) that would allow the use of the mpg-based approach beyond 2010 model year, but which would include other aspects which would avoid full 5-cycle testing of all the model types which failed to pass the criteria. For example, failing the initial criteria might require the manufacturer to generate fuel economy data over the US06, the least expensive of the three additional cycles. City and highway fuel economy values could then be calculated using three cycles (the FTP, HFET, and US06), and tested with additional criteria (

e.g.

, comparison to a tolerance band around the appropriately generated mpg-based line) to assess whether the mpg-based approach could be used or whether full 5-cycle testing would be required.

C. Derivation of the Proposed 5-cycle Fuel Economy Formulae

1. Five-Cycle Fuel Economy Estimates

The purpose of the 5-cycle fuel economy formulae is to best represent city and highway fuel economy in the U.S. using the test results from the 5 test cycles. To the fullest extent possible, we desire to account for the effect of seasonal and geographical variations on automotive fuel economy, as well as the different driving habits of individual drivers. As described in Section I., we chose to base the fuel economy label values on 5 vehicle emission and fuel economy tests which are already being performed. This maximizes the use of fuel economy information that is already currently being collected, while at the same time minimizes the costs associated with the proposal, as described in more detail below in Section VI. The five current emission and fuel economy tests and their key aspects are described below in Table II-1. Actual second by second descriptions of these driving cycles can be found in Section 86 of Title 40 of the Code of Federal Regulations.

Table II-1.—Key Features of the Five Current Emission and Fuel Economy Tests

Test

Driving

Ambient

temperature

Engine start

Accessories

FTP

Low speed

75 °F

Cold and hot

None.

HFET

Mid-speed

75 °F

Hot

None.

US06

Aggressive; low and high speed

75 °F

Hot

None.

SC03

Low speed

95 °F

Hot

A/C on.

Cold FTP

Low speed

20 °F

Cold and hot

None.

We have highlighted in bold the distinctive features of the five current vehicle tests. The FTP, HFET and US06 are all performed at an ambient temperature of 75 °F. Each test consists of a distinctive driving pattern. In addition, the FTP test consists of three distinct measurements, called bags. Bags 1 and 3 consist of the exact same driving pattern, but Bag 2 consists of a different pattern. Given that separate emission measurements are already made for each bag, we considered each bag of the FTP to be its own driving cycle. In addition, as discussed in Section V, the US06 cycle includes both low and high speed driving. We are proposing that separate emission measurements be made for these two types of driving, again providing separate estimates of fuel use for these two driving patterns. Therefore, we have available fuel economy estimates for five distinct driving patterns:

(1) Bags 1 and 3 of the FTP,

(2) Bag 2 of the FTP,

(3) HFET,

(4) the city portion of US06 and

(5) the highway portion of US06.

We propose to combine the results of these five tests to represent typical city and highway driving patterns. (The separation of the US06 test into two distinct sections is discussed further below.)

The FTP and the cold FTP are the only tests which include a cold start (i.e., an engine start after an overnight soak); the fuel needed to warm up the engine at 75 °F is taken from the FTP results. The SC03 test is the only test to be performed with the air conditioning system operational. Therefore, its results are used to augment the fuel economy from the five driving pattern tests for the fuel needed to operate air conditioning. The cold FTP is the only test performed at a temperature below 75 °F. Therefore, its results are used to represent the additional fuel needed to warm up an engine after a cold start, as well as any fuel needed to operate a warmed up engine, at colder temperatures.

As implied above, we estimate the fuel needed to start and warm up the engine separately from fuel used to operate the engine after start-up, or running fuel use. This is consistent with the approach taken in EPA emission models, such as MOBILE6.2 and MOVES. In terms of a mathematical formulae,

Total fuel use = start fuel use + running fuel use

and,

EP01FE06.020

We describe the estimation of start fuel use in Section II.B.1 and the estimation of running fuel use in Section II.B.2. In Section II.B.3, we discuss other aspects of driving which are not addressed by the dynamometer tests and which are addressed by applying an overall, or off-test adjustment factor to the city and highway fuel economy formulae. The reader is referred to Chapter II of the Draft Technical Support Document for a more detailed discussion of each of the inputs to the fuel economy formulae.

1. Start Fuel Use

For a specific vehicle, the fuel needed to warm up the engine depends primarily on two factors:

(1) The ambient temperature at which the vehicle has been sitting, and

(2) the length of time which the vehicle has been sitting since it was last used (commonly referred to as soak time).

Emissions during engine start up have been studied for some time. Most recently, estimates of start fuel use as a function of ambient temperature were made for use in EPA's new emission inventory model, MOVES (

tor

ehicle

mission inventory

ystem).

The relationship between start fuel use relative to that at 75 °F at other ambient temperatures is as follows:

A draft of MOVES2004 was released for public comment on Dec. 31, 2004.

Koupal, J., and L. Landman, E. Nam, J. Warila, C. Scarbro, E. Glover, R. Giannelli. MOVES2004 Energy and Emissions Report—Draft Report. U.S. Environmental Protection Agency, No. EPA420-P-05-003, March 2005, pp 57-63. Web site:

http://www.epa.gov/otaq/models/ngm/420p05003.pdf.

Start Fuel Use Relative to that at 75 °F =

1 + 0.01971 × (Ambient Temperature − 75) + 0.000219 × (Ambient Temperature − 75)

As will be seen below, we do not need an absolute estimate of start fuel use, simply an estimate of start fuel use relative to some specified ambient condition, such as 75 °F, which is the nominal temperature of the FTP test.

MOVES does not yet include the effect of soak time on start fuel use. Therefore, we obtained a relationship between start fuel use and ambient temperature which was developed by the California Air Resources Board for use in their emission inventory model, EMFAC2000.

EPA utilizes the results of this study in our current emission model, MOBILE6.2, to estimate the effect of soak time on regulated emissions during start-up. The equation for fuel use versus soak time (in minutes) relative to the fuel use after a 12 hour soak is as follows:

California Air Resources Board. Public Meeting to Consider Approval of Revisions to the State's On-Road Motor Vehicle Emissions Inventory—Technical Support Document. California Environmental Protection Agency, March 2000. See Section 6.7 (Start Correction Factors). Web site:

http://www.arb.ca.gov/msei/on-road/doctable_test.htm.

For soaks of 90 minutes or less:

Start Fuel Use = 0.00433672 × Soak Time − 0.000002393 × (Soak Time)

For soaks greater than 90 minutes:

Start Fuel Use = 0.25889542+0.0014848 × Soak Time − 0.0000006364 × (Soak Time)

As is assumed in EMFAC2000 and MOBILE6.2, we assumed that these relationships are independent of ambient temperature.

In order obtain the combined effect of ambient temperature and soak time, we multiplied the two above equations together, as follows:

For soaks of 90 minutes or less:

Start Fuel Use = ⌊0.00433672 × Soak Time − 0.000002393 × (Soak Time)

⌋×[1+0.01971 × (Ambient Temperature − 75)+0.000219 × (Ambient Temperature − 75)

]

For soaks greater than 90 minutes:

Start Fuel Use = ⌊0.25889542+0.0014848 × Soak Time − 0.0000006364 × (Soak Time)

⌋×[1+0.01971 × Ambient Temperature − 75)+0.000219 × (Ambient Temperature − 75)

]

The hot and cold starts contained in the standard and cold temperature FTP tests occur after 10 minute and 12 hour soaks, respectively. The above equations relating the effect of soak time on start fuel use indicate that the start fuel use after a 10 minute soak is only 4 percent of that after a 12 hour soak. The above equation relating the effect of temperature on start fuel use indicates that start fuel use at 20 °F is 2.75 times that at 75 °F. Combining these effects, the start fuel use after a 10 minute soak at 20 °F is about 11 percent that of a 12 hour soak at 75 °F. Thus, the start fuel use after the hot starts of both standard and cold temperature FTP tests are relatively small compared to that of a cold start at 75 °F.

In contrast to the cold start after a 12 hour soak, the hot starts for Bag 3 of the standard and cold temperature FTP tests and the US06, SC03 and HFET tests occur after only a 10 minute soak. The above equation indicates that the fuel use for a hot start is only 4 percent of that for a cold start.

In order to estimate start fuel use throughout the U.S. under average ambient conditions, we need estimates of the soak times for typical vehicle operation, as well as the ambient temperature at start up. The amount of time a vehicle has sat prior to start up varies dramatically depending on the time of day at which it is started. For example, for vehicles started up at 6 a.m., nearly all have sat idle overnight. However, for vehicles started at noon, most have been driven in the past 4-5 hours. Ambient temperature varies significantly during the day. Thus, it is more accurate to evaluate start fuel use by hour of the day rather than simply at the daily average temperature. Ambient temperatures also vary dramatically across the U.S., as does the distribution of vehicle miles traveled (VMT). Therefore, we combined estimates of vehicle starts and prior soak times by hour of the day with estimates of ambient temperature and VMT by county in order to reflect the effects of both soak time and ambient temperature on start fuel use.

We obtained estimates of each of these input parameters from EPA's MOBLE6.2 and MOVES emission models. The draft MOVES2004 model includes estimates of ambient temperature by hour of the day for each month of the year for each county in the U.S. These estimates were obtained from the National Weather Service and represent 30-year averages. The draft MOVES2004 model includes estimates of vehicle miles traveled (VMT) by vehicle type for every county in the U.S. during 2002. We used these estimates to determine the percentage of VMT by cars and light trucks in each county. MOBILE6.2 includes estimates of the frequency distributions of vehicle soak times by time of day, as well as the frequency distribution of vehicle starts by hour of the day. Draft MOVES2004 also includes estimates of VMT by month of the year for the nation as a whole.

We first estimated the effect of soak time on start fuel use by hour of the day. These estimates ranged from a low of 0.25 of an overnight soak at 2 p.m. to a high of 0.68 of an overnight soak at 6 a.m. This makes sense, as most vehicles being started at 6 a.m. in the morning have sat overnight, while most vehicles being started in the middle of the afternoon have been used in the past few hours. These estimates are independent of temperature, because the temperature during any particular hour is assumed to be constant.

In order to estimate start fuel use across the nation throughout the year, we calculated the start fuel use for each hour of the day by month for each county in the U.S. and then weighted each estimate by the relative number of starts occurring in each hour of the day and by the relative amount VMT in each month and county. Finally we summed the weighted start fuel use estimates across all hours of the days, months and counties and found the average.

The average start fuel use resulting from this process was 0.4665 of an overnight soak at 75 °F. We can simulate this average start fuel use with a variety of combinations of hot and cold starts at 20 °F and 75 °F. For example, the level of start fuel use is equal to a 0.4665 weighting of the cold start fuel use in Bag 1 of the FTP at 75 °F and no weighting of the start fuel use at 20 °F.

Or, this level of start fuel use is also equal to a lower weighting of the cold start fuel use in Bag 1 of the FTP at 20 °F and no weighting of the start fuel use at 75 °F. In order to select a single combination which best incorporated the measured start fuel use at both 20 °F and 75 °F, we evaluated start fuel use only as a function of soak time and time of day, assuming temperature was constant throughout the day. We found that the typical start fuel use was 0.330 times that of a cold start (12 hour soak). We then determined that a weighting of 0.24 for a cold start at 20 °F and 0.76 for a cold start at 75 °F, combined with an overall weighting of 0.330 for cold starts produced the same level of start fuel use as 0.4665 times a cold start at 75 °F, or the average level of start emissions estimated to occur in-use.

In terms of the use of the FTP test results, Bag 3 contains the start fuel use after a 10-minute soak, and Bag 1 contains the start fuel use after a 12 hour soak. Other aspects of Bag 1 and Bag 3 are the same (i.e., the vehicle is driven exactly the same, only the soak time prior to start up differs). As indicated above, however, the start fuel use after a 10 minute soak can be assumed to be negligible compared to that after the 12 hour soak.

This means that the difference between fuel use in Bag 1 and Bag 3 is the start fuel use following a 12 hour soak. Thus, the average start fuel use in the U.S. is 0.24 times 0.330 times the difference between fuel use in Bag 1 and Bag 3 of the cold temperature FTP plus 0.76 times 0.330 times the difference between fuel use in Bag 1 and Bag 3 of the standard FTP at 75 °F.

The Draft MOVES2004 model also assumes that start fuel use after a hot start is negligible.

Hybrids are tested over what is commonly referred to as a 4-bag FTP test, with Bag 4 consisting of a Bag 2 repeated after Bag 3. In this case, the cold start fuel use would be determined exactly as described above. However, these four bags can also be combined into two bags, with Bag 1 consisting of a typical Bag 1 and Bag 2 and Bag 2 consisting of a typical Bag 3 and Bag 4. In this case, cold start fuel use would be determined from the difference in fuel use between Bags 1 and 2 of the 2-bag FTP test.

This estimate of start fuel use is in terms of total fuel use per start. In order to combine this with running fuel use in terms of gallons per mile, start fuel use must be divided by the average trip length. We based our estimate of the average trip length in the U.S. on the National Household Travel Survey (NHTS). The NHTS was performed in 2001 and statistically surveyed approximately 26,000 households in the U.S. This survey represents the sixth in a series of surveys dating back to 1969. (The name of the survey has changed a few times and the precise survey methods have varied to some degree.) NHTS found that the average trip taken using a personal vehicle in the U.S. was 9.8 miles long. This estimate excludes very long trips, such as those taken on vacations, as well as commercial trips, such as those by taxi cabs. Based on the survey questionnaire, we believe that the survey also excludes brief stops (e.g., those at gas stations or convenience stores), as well as extremely short trips (e.g., moving a vehicle out of a driveway to allow another vehicle to exit, moving from one shopping center to another just across the street). Using trip information from instrumented vehicles in Baltimore and Spokane (described in more detail below), about 27 percent of all trips fall into one of these two categories. Thus, we believe that a more precise estimate of trip length, and one that is more consistent with our estimate of the fraction of cold starts described above, is 7.7 miles (9.8 miles divided by 1.27).

This trip length of 7.7 miles includes all driving, both city and highway oriented. NHTS does not attempt to split driving into city and highway categories. Therefore, additional information was needed to perform this split. As will be described in more detail below, we estimate that 43 percent of all U.S. driving falls under our definition of city driving, while 57 percent falls into the highway driving category. The highway fuel economy label assumes no cold starts (i.e., it is based solely on the HFET, which is a hot start test), except insofar that the effect of a cold start is included in the 22 percent adjustment factor. Since even long trips have a beginning and often begin with a cold start, we assumed that the average highway trip had a length of 60 miles. This is somewhat arbitrary. However, once trip length is over 20 miles, start fuel use has very little impact on fuel economy. Still, the inclusion of some start fuel use in the highway fuel economy estimate makes this estimate more realistic. Assuming an average trip length of 60 miles for highway driving, the average length of a city trip must be 3.5 miles for the overall average to be 7.7 miles. Using these two estimates of average trip length allows us to convert fuel use per engine start into fuel use per mile.

The total volume of fuel used in either Bag 1 or Bag 3 of the FTP can be determined by dividing the number of miles of driving during these portions of the test (3.59 miles for either bag) by the fuel economy measured during that bag. Thus, the equation for fuel use per start at either 20 °F or 75 °F is as follows:

For vehicles tested over either a 3-Bag FTP or 4-Bag FTP:

EP01FE06.021

For vehicles tested over either a 2-Bag FTP:

EP01FE06.022

where x is either 20 °F or 75 °F.

The equation for start fuel use in terms of gallons per mile is:

For city driving:

EP01FE06.023

For highway driving:

EP01FE06.024

2. Running Fuel Use

Running fuel use depends primarily on how the vehicle is driven and the use of fuel to power accessories. Of the latter, air conditioning is the most significant and the primary accessory addressed in the emission and fuel economy dynamometer tests. Once the vehicle is warmed up, ambient temperature has only a modest effect on fuel use.

The five dynamometer tests include four distinct driving cycles, or patterns of driving. In addition, the FTP and US06 cycles (the latter as proposed to be modified) each include two distinct driving patterns. Two basic characteristics of these driving patterns are depicted in Table II-2: average speed and a basic measure of the average power required by the engine.

Table II-2.—Driving Characteristics of the Current Dynamometer Tests

Cycle

Average speed

Average power

FTP (Bags 2 and 3)

19.6

40.9

FTP: Bag 3

25.6

53.6

FTP: Bag 2

16.1

33.8

HFET

48.2

34.9

US06

48.0

104.3

US06: City Bag

21.5

152.9

US06: Highway Bag

61.0

78.2

SC03 (run with air conditioning on)

21.4

49.2

Cold Temperature FTP (same driving cycle as FTP)

19.6

40.9

Power defined as velocity times the change in velocity per second during cruise or accelerations. Power is set equal to zero during decelerations and not considered in the determination of average power.

The FTP and the cold temperature FTP both involve the same driving cycle, just at different ambient temperatures. Thus, their average speeds and power are identical, both for the total cycle and for each bag of emissions measured. The FTP and SC03 involve distinct, but similar driving cycles. Both are low speed cycles having similar average speeds and power levels. As the SC03 test is only run with the air conditioning on and all the other tests are run with air conditioning off, it is not possible to isolate the effect of the driving cycle differences between the FTP and SC03 tests directly. Thus, this leaves five distinct driving patterns which can be used to represent typical U.S. driving: Bag 2 of the FTP, Bag 3 of the FTP, HFET, City Bag of US06 and Highway Bag of US06.

As shown in Table II-2, both Bags 2 and 3 of the FTP are low speed cycles, but their average power requirements differ by a factor of 1.7. As will be seen below, it is useful to consider each bag separately in simulating typical city and highway driving.

The current US06 test currently consists of 600 seconds of driving and the emissions are collected in one bag (i.e., one single collection of pollutants emitted during the test). Thus, the fuel economy result is over the entire cycle. The US06 driving cycle consists of 5 hills, or 5 driving segments which begin and end with the vehicle at idle. All but the second and third hills consist of relatively low speed driving, while the second hill reaches 71 mph and the third hill reaches 80 mph. Therefore, in terms of predicting fuel economy, it is useful to separate the low speed driving from the high speed driving. For practical reasons, when separating the city into “city” and “highway” portions, we grouped the second hill with the four low speed hills in the city bag and the highway bag consists of the relatively long third hill. Overall, seconds 0-131 and 496-600 of the cycle would comprise the city bag and seconds 132-495 would comprise the highway bag. The description of the hills within US06 and their designation is summarized in Table II-3 below.

Table II-3.—Split of US06 Cycle Into City and Highway Portions

Hill

Portion of driving cycle (cumulative seconds)

Maximum speed (mph)

Designation

0-43

44.2

City.

44-134

70.7

City.

134-499

80.3

Highway.

500-563

29.8

City.

564-600

51.6

City.

As described in the Introduction, driving at an average speed below 45 mph is defined as city driving, while that above 45 mph is defined as highway driving. We obtained a description of average U.S. driving from the Draft MOVES2004 motor vehicle emissions model. This description included a distribution of vehicle speeds and levels of vehicle specific power. Using the definition of city and highway driving, we separated the MOVES description of driving into city and highway categories. We then performed a linear regression to estimate what two combinations of the five driving cycles or bags best fit average U.S. city and highway driving patterns, respectively. The results are two sets of cycle combinations in terms of time spent driving. These are shown in Table II-3. We then used the average speeds of the various cycles and bags to convert these to combinations to a mileage basis. The combinations of cycles found to best represent onroad driving in terms of both time spent driving and mileage driven are shown in Table II-4.

Table II-4.—Weighting Factors for the Five Dynamometer Cycles (Percent)

Cycle

City driving

Time

(percent)

Mileage

(percent)

Highway driving

Time

(percent)

Mileage

(percent)

Bag 3 FTP

Bag 2 FTP

HFET

US06 City

US06 Hwy

From the results shown in Table II-4, over 90 percent of the time spent in city driving, and nearly 90 percent of the mileage, is best explained by Bags 2 and 3 of the FTP cycle. Roughly 80 percent of both driving time and mileage of highway driving is best explained by the highway portion of the US06 cycle. These findings confirm that the FTP (the current basis for the city fuel economy label) is still generally representative of most low speed driving in the U.S. However, the relatively low speed and mild accelerations of the HFET (the current basis for the highway fuel economy label) is not representative of higher speed driving in the U.S.

These results also confirm the separation of the two types of driving contained in the US06 cycle. Only the city portion of US06 appears in the description of city driving and only the highway portion of US06 appears in the description of highway driving. At the same time, the relative weights for Bags 2 and 3 in the description of city driving are similar to that implicit in the FTP, which is 52 percent and 48 percent, respectively.

As mentioned above, the fuel use over the three dynamometer cycles, when combined using these weighting factors, best matches the fuel use which would occur during typical city and highway driving. The weighting is performed in terms of fuel use, or fuel consumption per mile. For example, fuel use during city driving is 0.48 times the multiplicative inverse of the fuel economy measured over Bag 2 of the FTP cycle plus 0.41 times the multiplicative inverse of the fuel economy measured over Bag 3 of the FTP cycle plus 0.11 times the multiplicative inverse of the fuel economy measured over the city bag of the US06 cycle.

EP01FE06.025

EP01FE06.026

These estimates of running fuel use accounts for a wider variety of city and highway driving patterns than the FTP and HFET cycles alone. However, these combinations of fuel use still do not include any fuel use related to air conditioning or cold temperature. Fuel use related to air conditioning is estimated using the SC03 test. As shown in Table II-2, the driving pattern contained in the SC03 test is similar to that of the FTP, but not identical.

Using the MOVES2004 methodology for modeling fuel use, we estimated the combination of Bags 2 and 3 of the FTP which would match the fuel use over the SC03 cycle with the air conditioning turned off. This combination is 0.39 times the fuel consumption over Bag 2 and 0.61 times the fuel consumption over Bag 3. Thus, we propose to estimate the incremental fuel use due to the operation of the air conditioner as the difference in fuel use measured over the SC03 versus this combination of fuel use over Bags 2 and 3 of the standard FTP.

This difference in fuel use between the two tests provides a direct estimate of the impact of air conditioning use for the conditions present during the SC03 test. The SC03 test is performed at 95 °F and 40 percent relative humidity. The test only lasts 10 minutes and the vehicle is pre-heated with radiant lamps for 10 minutes prior to the test. Thus, the air conditioning compressor is generally engaged throughout the entire test. As shown in Table II.-2., the speed of the vehicle during the SC03 test is also relatively low, at an average speed of 21.5 mph. Of course, onroad, vehicles operate at different speeds and ambient temperatures and the compressor may not be engaged 100 percent of the time, particularly during longer trips. All three of these factors can affect the impact of air conditioning on fuel economy. We therefore adjust the estimate of the impact of air conditioning on fuel use from the SC03

test in three ways to account for these three factors.

The largest factor is portion of driving time during which the compressor is actually engaged to cool inlet air to the vehicle. The Draft MOVES2004 model contains an algorithm which estimates the percentage of time which the compressor is engaged as a function of ambient temperature and humidity. This algorithm was developed from the direct measurement of air conditioning operation of 20 vehicles in Phoenix, Arizona during the summer and fall of 1992.

The algorithm considers both the frequency that the system is turned on by the driver and the frequency that the compressor is engaged once the system is turned on. We combined this algorithm with long term average meteorological conditions for each county in the U.S. to estimate the percentage of driving time during which the compressor was engaged under those conditions. We considered both diurnal and seasonal temperature variations, as well as variations in the amount of driving performed throughout the day and across seasons. We estimate that drivers have the air conditioning turned on 23.9 percent of the time on average across the U.S., and the compressor is engaged 15.2 percent of the time.

Koupal, J. W. Air Conditioning Activity Effects in MOBILE6 (M6.ACE.001). U.S. Environmental Protection Agency, No. EPA420-R-01-054, November 2001.

Website:

http://www.epa.gov/otaq/models/mobile6/r01054.pdf

We then adjusted this latter percentage to account for reduced compressor loads at temperatures less than 95 °F and higher loads above 95 °F.

Again this was done for each county in the U.S., accounting for diurnal and seasonal temperature and driving differences. From this, we estimate that the average load of the air conditioning compressor in-use is about 87 percent of that at 95 °F (i.e., during the SC03 test). Thus, the average load of the compressor in-use is the same as 13.3 percent (15.2 percent × 0.87) of the load experienced during the SC03 test.

Nam, Edward K., “Understanding and Modeling NO

Emissions From Air Conditioned Automobiles,” 2000, SAE #2000-01-0858.

Finally, the impact of air conditioning on fuel economy varies with vehicle driving pattern. Most air conditioning compressors are belt-driven by the engine. The efficiency of both the engine and compressor varies with engine speed and load. This variation is difficult to model, as the speed and load of engines in various vehicles varies dramatically based on the vehicle's drivetrain design, even over the same driving cycle. Therefore, we assume that the efficiency of the engine and air conditioning compressor implied in the SC03 test applies to other types of driving, as well. However, a more basic effect related to driving pattern is that the faster a vehicle is moving, the shorter the amount of time that the vehicle needs to be cooled while it travels a specific distance. Other factors being equal, this reduces the amount of energy needed to cool the vehicle per mile of travel. Therefore, for a specific set of ambient conditions, we assume that the impact of air conditioning on fuel use is constant with driving time (i.e., fuel use in terms of gallons per hour is constant). This means that the excess fuel use due to operating the air conditioner varies inversely proportional to vehicle speed. In other words, at low vehicle speeds, like that of the SC03 test, excess fuel use is relatively high on a per mile basis. At high vehicle speeds, like that of highway driving, the excess fuel use due to operating the air conditioner is relatively low on a per mile basis. We confirmed this assumption by testing five vehicles over a variety of test cycles at EPA's Ann Arbor laboratory with both the air conditioning turned on and off. The results of this test program and an analysis of the data are described in the Draft Technical Support Document.

The air conditioning compressor is also often engaged when the defroster is turned on to keep the windshield from fogging up. The air conditioning dehumidifies the air and excesses the effectiveness of the defroster. Today's proposal does not include a specific weighting for demisting activity. We lack a direct estimate of the frequency that the defroster is turned on or the compressor is engaged during demisting. Due to the fact that the defroster tends to be operated at lower ambient temperatures than the air conditioner, the load on the engine is generally much lower than that during summertime air conditioning. Thus, the impact of demisting on fuel economy is likely much smaller than that of summertime air conditioning.

Given the above, the impact of air conditioning on running fuel use is estimated as 13.3 percent of the difference between fuel use per mile over the SC03 and a combination of Bags 2 and Bag 3 of the FTP times 21.5 mph and divided by the average speed of either city or highway driving. Based on the descriptions of city and highway driving from Draft MOVES2004, the average speeds are 19.9 mph and 57.1 mph, respectively. Thus, the excess fuel use due to air conditioning operation is:

EP01FE06.027

EP01FE06.028

Finally, we have to add the impact of colder ambient temperatures on running fuel use. We can obtain a direct estimate of the impact of colder ambient temperatures on running fuel use by comparing the fuel use over the standard and cold temperature FTP tests. By focusing on Bag 2 of each FTP test, we exclude the impact of cold temperature on start up fuel use, which was already addressed in Section II.B.1 above. For hybrid vehicles, which are tested over the bag 2 driving cycle twice (the first time as Bag 2 and the second time as Bag 4), we propose to harmonically average the fuel economies from Bags 2 and 4.

We considered including Bag 3 in the determination of the effect of cold temperature on running fuel use. Bag 3 includes some higher speed driving, so its inclusion broadens the overall driving pattern included in the estimate. This would particularly improve the representativeness of the estimate for highway driving. However, Bag 3 begins with a hot start, unlike Bag 2 which simply follows directly after Bag 1 with no engine shut-off and restart in between. At 75 °F, a hot start requires a negligible volume of additional fuel use. However, at 20 °F, even a hot start can require some excess fuel use. Thus, including the difference between Bag 3 fuel use at 20 and 75 °F in the estimate of the impact of cold temperature on running fuel use could also include some excess fuel use related to engine warm up, as well. Available data indicate that the relative impact of operation at 20 °F versus 75 °F is nearly identical for the two bags (10 percent for Bag 2 and 11 percent for Bag 3). However, the fuel economy over Bag 3 is lower than over Bag 2, so the absolute difference in fuel use between 20 °F and 75 °F is actually lower in Bag 3 than Bag 2. We request comment on whether the impact of cold temperature on running fuel use should only involve Bag 2 or should involve both Bags 2 and 3.

Neither MOBILE6.2 nor MOVES2004 include correlations of the effect of ambient temperature on running fuel use. However, as just described, the impact of colder ambient temperatures on running fuel use is small (i.e., 10 percent over a drop in temperature of 55 °F). We believe that the additional fuel use is primarily due to the loss of heat to the cooler ambient air, higher friction in the slightly cooler moving parts, as well as slight changes in the properties of the cooler intake air and air fuel mixture during combustion. All of these changes are expected to be gradual and fairly linear. Therefore, we assume that the excess fuel use increases linearly as temperatures decrease below 75 °F. Above 75 °F, we assumed that there was no further reduction in running fuel use. (This latter assumption was confirmed as part of the five vehicle test program described above.) We also assume that the excess fuel use is independent of driving pattern. In other words, the excess fuel use is the same for city and highway driving on an absolute basis. We request comment on assuming that the excess running fuel use due to colder temperatures is independent of driving pattern on a relative basis (i.e., in percentage terms).

Using the same meteorological and VMT inputs described above related to start fuel use, we estimate the average temperature in the U.S. at which driving occurs is 58.7 °F. This temperature is 70 percent of the way from 75 °F to 20 °F. Thus, any excess fuel use associated with operation at 20 °F should be weighted by 100 percent minus 70 percent, or 30 percent.

Given the fact that over 80 percent of city driving is represented by Bags 2 and 3 of the FTP, we decided to use the fuel economy measured during Bags 2 and 3 of the cold FTP directly to represent the fuel economy of city driving at 20 °F. We repeated the regression of the VSP distribution of city driving from Draft MOVES2004 against the VSP distributions of just Bags 2 and 3. The best fit produced a 50/50 weighting of the two bags. Thus, we propose to represent the fuel economy of city driving at 20 °F by a 50/50 harmonic average of the fuel economy over Bags 2 and 3 of the cold FTP. Mathe- matically, then, for city driving:

EP01FE06.029

Highway driving occurs at higher speeds than those typical of the cold FTP. We conducted a detailed review of past test programs which evaluated the impact of colder temperatures on fuel economy at highway driving speeds. This review is described in the Draft Technical Support Document. There, we concluded that the effect of cold temperature on fuel economy at city driving speeds could overestimate the effect at higher speeds. Thus, we decided not to use the fuel economy measured over the cold FTP directly to represent the impact of cold temperature on highway fuel economy. Instead, we believe that it is more prudent at this time to simply assume that running fuel use at 20 °F at highway speeds is 4 percent greater than that at 75 °F. Thus, mathematically, for highway driving:

EP01FE06.030

Combining the estimates of running fuel use at 75 °F without the air conditioning system running with the estimate of excess fuel use of running the air conditioning system and the estimate of excess fuel use due to colder ambient temperatures produces the following formulae for running fuel use:

For city driving:

EP01FE06.031

For highway driving:

EP01FE06.032

3. Adjustment Factor for Non-Dynamometer Effects

Fuel economy estimated using the five current dynamometer tests can account for many factors, including vehicle design, driving pattern, trip length, cold temperature and air conditioning. However, there are still a large number of factors which affect vehicle fuel economy that cannot be addressed by dynamometers tests. These include roadway roughness, road grade (hills), fuel quality, large vehicle loads (e.g., trailers, cargo, multiple passengers), wind, precipitation, to name just a few. Even when a factor is addressed by a dynamometer test, such as driving pattern or air conditioning, the effect can only be approximated, as all realistic driving patterns cannot possibly be included in a test having a reasonable length of time. Nor can all the possible ambient conditions affecting air conditioner operation be tested. Thus, any estimate of in-use fuel economy derived from the five dynamometer tests is necessarily approximate, both with respect to factors addressed directly by the tests and those which are not.

The impacts of a number of these factors on onroad fuel economy relative to that measured on a dynamometer is possible to estimate, while others are difficult to estimate. One factor which can be estimated is fuel quality. EPA's certification test fuel contains no oxygenates, while commercial gasoline contains significant volumes of ethanol and methyl tertiary butyl ether (MTBE). Both ethanol and MTBE contain less energy per gallon, so vehicles operating on fuel containing these oxygenates tend to achieve lower fuel economy, generally in proportion to the reduction in the energy content of the finished gasoline. For example, the driver of a vehicle operating on gasoline containing ten percent ethanol by volume would experience a 3.5 percent decrease in fuel economy compared to gasoline not containing any ethanol or other oxygenate. We expect the nation's gasoline supply to contain roughly 5.4 billion gallons of ethanol by 2008. This is equivalent to 37 percent of the nation's gasoline supply containing 10 percent ethanol by volume. Thus, by 2008, we expect commercial gasoline on average to contain about 1.2 percent less energy per gallon than EPA test fuel. Thus, this difference in energy content means that onroad fuel economy will be about 1.2 percent less than that estimated using the 5-cycle formulae described in the previous section. This effect could increase beyond 2008 as more ethanol is used in the nation's gasoline supply.

Another factor which can be estimated is tire pressure. In February 2001, NHTSA conducted a survey of the tire pressure of in-use vehicles. Tire pressures were measured on over 11,500 vehicles at 24 locations throughout the U.S. The results of the study and our analysis of the data are described in the Draft Technical Support Document. We found that the tires of the average car were under-inflated by 1.1 pounds per square inch (psi), while those on light trucks were under-inflated by 1.9 psi. Using estimates of the effect of tire pressure on fuel economy presented by NHTSA, we estimate that the fleet-wide effect of under-inflation is 0.5 percent.

Another factor which can be estimated, though more approximately, is wind. Wind affects vehicular fuel economy in two ways. First, aerodynamic drag is proportional to the square of vehicle speed (i.e., the higher the vehicle speed, the faster aerodynamic drag increases for a given increase in speed). Thus, increasing wind speed by 1 mph increases aerodynamic drag, and thus, reduces fuel economy, more than the effect of decreasing wind speed by 1 mph. Second, both the effective area of a vehicle and its drag coefficient increases as the true wind direction moves to either side from head-on. Basically, vehicles are designed to move forward through the air, not sideways. Thus, any side wind increases drag and decreases fuel economy. Based on a distribution of wind speeds (yielding an average wind speed in the U.S. of 9.4 mph), we estimate that these two effects reduce onroad fuel economy on average by 5-6 percent.

Several other factors are still relevant to a 5-cycle fuel economy estimate, namely altitude, road grade, road surface, road curvature, brake drag, wheel alignment, tire switching, and vehicle load. EPA estimated the impact of these factors to be 8 percent at the time of the 1984 label adjustment rule.

We have reduced the impact of road surface from 4 percent to 1-3 percent due to increased urbanization and road paving which has occurred since that time. Thus, we estimate these other factors to reduce onroad fuel economy by 5-7 percent. Combining this estimate with those of fuel quality, tire pressure and wind produces an overall downward effect of 11-15 percent.

As described further in Section II.E below, we also compared the 5-cycle fuel economy values to fleet-wide estimates of fuel economy made by FHWA for 2002 and 2003, after we made several adjustments to improve the comparability of the two estimates. The 5-cycle fuel economy values best match the FHWA-based estimates when we include a factor of 0.88-0.91 in the 5-cycle fuel economy formulae (i.e., a reduction of 9-12 percent due to factors not addressed by the 5-cycle formulae). We propose to average these two ranges (i.e., the 9-12 percent range based on FHWA, and the 11-15 percent range based on the analysis of non-dynamometer effects discussed above) and account for these factors by including a factor of 0.89 in the 5-cycle city and highway formulae (i.e., a reduction of 11 percent in both city and highway fuel economy).

D. Derivation of the MPG-Based Approach

The mpg-based approach to fuel economy label adjustments utilizes the results of applying the 5-cycle formulae to all vehicles for which we were able to gather fuel economy data for all five dynamometer cycles. We requested that all manufacturers submit to us all their available fuel economy data for vehicles which had been tested over at least one of the US06, SC03 or cold FTP tests. We combined this data with our own fuel economy data to develop a database of 423 recent model year vehicles which had been tested over all five cycles. We applied the above 5-cycle formulae to these vehicles. We then developed a relationship between the 5-cycle city and highway fuel economies and the city and highway fuel economies using the current adjustment factors, respectively.

We evaluated two options for developing this relationship. One option plotted 5-cycle fuel economy versus fuel economy using the current adjustment factor. The other option plotted the inverse of 5-cycle fuel economy (i.e., fuel consumption) versus the inverse of fuel economy using the current adjustment factor. As indicated from the description of the 5-cycle fuel economy formulae, most of the modeling of fuel economy is performed in terms of fuel consumption (i.e., gallons of fuel burned per mile versus miles traveled per gallon of fuel burned). While both types of plots produce relationships with a high degree of correlation, the plots in terms of fuel consumption are linear, while those in terms of fuel economy are non-linear. Given that the linear relationship is simpler and the degrees of correlation are essentially the same, we are proposing to base the mpg-based adjustments on the correlations in terms of fuel consumption. However, the label values themselves would remain in terms of fuel economy, as required by EPCA. We request comment on the use of the correlations performed in terms of fuel consumption versus those performed in terms of fuel economy. Both approaches are described in detail in the Draft Technical Support Document.

Figures II-5 and II-6 show the relationship between the inverse of 5-cycle city (or highway) fuel economy (i.e., fuel consumption) versus the inverse of FTP (or HFET) fuel economy. Figure II-5 shows city fuel consumption, while Figure II-6 shows highway fuel consumption.

EP01FE06.033

EP01FE06.034

The results of regressing 5-cycle fuel consumption versus fuel consumption over the FTP or HFET are shown in the above figures. In terms of fuel economy:

EP01FE06.035

EP01FE06.036

The standard deviation of the difference between the mpg-based equations and the 5-cycle fuel economies are 2 percent for city and 5 percent for highway. These differences are roughly equivalent to 0.5 mpg for city fuel economy and 1-2 mpg for highway fuel economy. Thu

This text is long and has been trimmed here. Open the source document for the complete record.

This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.