The Safer Affordable Fuel-Efficient (SAFE) Vehicles Rule III for Model Years 2022 to 2031 Passenger Cars and Light Trucks

The previous section has discussed, at a high level, how NHTSA generates the technology inputs and assumptions used in the CAFE Model. The process for generating these inputs and assumptions involves NHTSA using engineering judgment to evaluate and synthesize data from a variety of sources, including data submitted by vehicle manufacturers; consolidated publicly available data, such as press materials, marketing brochures, and other information; data from collaborative research, testing, and modeling with other Federal agencies and laboratories; data from research, testing, and modeling with independent organizations; data and assumptions from work done for prior rules; and feedback from stakeholders on prior rules and meetings conducted prior to the commencement of this rulemaking, to the extent it is still relevant and applicable.

This section discusses the specific technology pathways, effectiveness, and cost inputs and assumptions used in the compliance analysis. As an example, NHTSAs explained in the previous section that the starting point for estimating technology costs is an estimate of the DMC--the component and assembly costs of the physical parts and systems that make up a complete vehicle--for any particular technology.

After spending over a decade refining the technology pathways, effectiveness, and cost inputs and assumptions used in successive CAFE Model analyses, NHTSA has developed guiding principles to ensure that the CAFE Model's compliance analysis reflects impacts reasonably expected in the real world. These guiding principles are as follows:

Technologies have complementary or non-complementary interactions with the full-vehicle technology system. The fuel economy improvement from any individual technology must be considered in conjunction with the other fuel economy-improving technologies applied to the vehicle, because technologies added to a vehicle do not result in a simple additive fuel economy improvement from each individual technology. In particular, NHTSA expects this result from engine and other powertrain technologies that improve fuel economy by allowing the ICE to spend more time operating at efficient engine speed and load conditions or from combinations of engine technologies that work to reduce the effective displacement of the engine.

The effectiveness of a technology depends on the type of vehicle to which the technology is being applied. When discussing “vehicle type” in the analysis, NHTSA is referring to the ten vehicle technology classes (e.g., small car, small car performance, medium car, medium car performance, small SUV, small SUV performance, medium SUV, medium SUV performance, pickup truck, or pickup truck high towing (HT)). A small car and a medium performance SUV that use the exact same technology have very different starting fuel economy values; when the exact same technology is added to both of those vehicles, the technology provides different effectiveness improvements for the vehicles.

The cost and effectiveness values for each technology are reasonably representative of what can be achieved across the entire industry. Each technology model employed in the analysis is designed to be representative of a wide range of specific technology applications used in industry. Some manufacturers' systems may perform better or worse than the modeled systems and some may cost more or less than the modeled systems; however, employing this approach ensures that, on balance, the analysis captures a reasonable level of costs and benefits that would result from any manufacturer applying the technology.

A consistent reference point for cost and effectiveness values must be identified before assuming that a cost or effectiveness value could be employed for any individual technology. For example, this analysis uses a set of engine map models developed by starting with a small number of engine configurations, and then, in a systematic and controlled process, adding specific well-defined technologies to create a new map for each unique technology combination. Again, providing a

consistent reference point to measure incremental technology effectiveness values ensures that NHTSA is capturing accurate effectiveness values for each technology combination.

The following sections discuss the engine, transmission, hybridization, mass reduction, aerodynamic, tire rolling resistance, and other vehicle technologies considered in this analysis. The following sections discuss:

How NHTSA defines technology in the CAFE Model; \157\

\157\ Note: Due to the diversity of definitions industry employs for technology terms, or in describing the specific application of technology, the terms defined here may differ from how the technology is defined in some parts of the industry.

How NHTSA assigns technology to vehicles in the analysis fleet used as a starting point for this analysis;

Any adoption features applied to the technology, so the analysis better represents manufacturers' real-world decisions;

Technology effectiveness values; and

Technology cost.

Note that the following technology effectiveness sections provide examples of the range of effectiveness values that a technology could achieve when applied to the entire vehicle system, in conjunction with the other fuel economy-improving technologies already in use on the vehicle. To see the incremental effectiveness values for any particular vehicle moving from one technology key to a more advanced technology key, see the CAFE Model Fuel Economy Adjustment Files that are installed as part of the CAFE Model Executable File, and not in the input/output folders. Similarly, the technology costs provided in each section are examples of absolute costs seen in specific model years, for specific vehicle classes. The Technologies Input File contains all absolute technology costs used in the analysis across all model years. 1. Engine Paths

ICE vehicles convert chemical energy in fuel to useful mechanical power. The chemical energy in the fuel is released and converted to mechanical power by being oxidized, or burned, inside the engine. The air/fuel mixture entering the engine and the burned fuel/exhaust by- products leaving the engine are the working fluids in the engine. The engine power output is a direct result of the work interaction between these fluids and the mechanical components of the engine.\158\ The generated mechanical power is used to perform useful work, such as vehicle propulsion.\159\

\158\ Heywood, J., Internal Combustion Engine Fundamentals, Chapter 1, McGraw-Hill Education: Columbus, OH (2018) (hereinafter, “Heywood (2018)”).

\159\ Ibid, containing a complete discussion on fundamentals of engine characteristics, such as torque, torque maps, engine load, power density, brake mean effective pressure (BMEP), combustion cycles, and components.

NHTSA classifies the extensive variety of light-duty vehicle ICE technologies into discrete Engine Paths. These paths are used to model the most representative characteristics, costs, and performance of the fuel economy-improving engine technologies most likely available during the rulemaking timeframe. The paths are intended to be representative of the range of potential performance levels for each engine technology. In general, the paths are tied to ease of implementation of additional technology and how closely the technologies are related. The technology paths are presented in Chapter 3.1.1 of the Final TSD.

The Engine Paths have been selected and refined over a period of more than 10 years, based on engines in the market, stakeholder comments, and engineering judgment, subject to the following factors: the included technologies are those most likely available during the rulemaking timeframe and within the range of potential performance levels for each technology, and excluded technologies are those unlikely to be feasible in the rulemaking timeframe, unlikely to be compatible with U.S. fuels, or for which there was not appropriate data available to allow the simulation of effectiveness across all vehicle technology classes in this analysis.

The Engine Paths begin with one of the three base engine configurations: DOHC engines have two camshafts per cylinder head (one operating the intake valves and one operating the exhaust valves), single overhead camshaft (SOHC) engines have a single camshaft, and overhead valve engines, which have a single camshaft located inside of the engine block (beneath the valves rather than overhead) connected to a rocker arm through a pushrod that actuates the valves. DOHC and SOHC engine configurations are common in the light-duty fleet.

The next step along an Engine Path is the Basic Engine Path technologies. These include VVL, SGDI, and a basic level of DEAC. VVL dynamically adjusts how far the valve opens and reduces fuel consumption by reducing pumping losses and optimizing airflow over a broader range of engine operating conditions. Instead of injecting fuel at lower pressures and before the intake valve, SGDI injects fuel directly into the cylinder at high pressures allowing for more precise fuel delivery while providing a cooling effect and allowing for an increase in the CR, more optimal spark timing for improved efficiency, or both. DEAC disables the intake and exhaust valves and turns off fuel injection and spark ignition (SI) on select cylinders, which effectively allows the engine to operate temporarily as if it were smaller while also reducing pumping losses to improve efficiency. For the proposal and now for this final rule, NHTSA's analysis has integrated variable valve timing (VVT) technology in all non-diesel engines, so there is not a separate box for it on the Basic Engine Path. VVL, SGDI, and DEAC can be applied to an engine individually or in combination with each other.

Moving beyond the Basic Engine Path technologies are the “advanced” engine technologies, which are technologies that require significant changes to the structure of the engine or an entirely new engine architecture. The advanced engine technologies represent the application of alternate combustion cycles, various applications of forced induction technologies, or advances in-cylinder deactivation.

Advanced cylinder deactivation (ADEAC) systems, also known as rolling or dynamic cylinder deactivation systems, allow the engine to vary the percentage of cylinders deactivated and the sequence in which cylinders are deactivated. Depending on the engine's speed and associated torque requirements, an engine might have most cylinders deactivated (e.g., low torque conditions, as with slower speed driving) or it might have all cylinders activated (e.g., high-torque conditions, as with merging onto a highway).\160\ An engine operating at low-speed/ low-torque conditions can save fuel by operating at a fraction of its total displacement. NHTSA models two ADEAC technologies: advanced cylinder deactivation on a single overhead camshaft engine (ADEACS) and advanced cylinder deactivation on an ADEACD.

\160\ See Tula Technology, Inc., Dynamic Skip Fire, last revised: 2026, available at: https://www.tulatech.com/combustion-engine/ (accessed: May 28, 2026), discussing how the company's proprietary cylinder deactivation technology operates in real-world situations. NHTSA's modeled ADEAC system is not based on this specific system, and therefore the effectiveness improvement is different in NHTSA's analysis than with this system; however, the theory still applies.

Forced induction gasoline engines include both supercharged and turbocharged downsized engines, which can pressurize or force more air into an engine's intake manifold when higher power output is needed. The raised pressure results in an increased amount

of airflow into the cylinder to support combustion, increasing the specific power of the engine. The first-level turbocharged downsized technology (TURBO0) engine represents a basic level of forced air induction technology being applied to a DOHC engine. A cooled exhaust gas recirculation (CEGR) system takes engine exhaust gases, passes them through a heat exchanger to reduce their temperature, then mixes them with incoming air in the intake manifold to reduce peak combustion temperature, thereby improving fuel efficiency and emissions. NHTSA models the base TURBO0 turbocharged engine with the addition of cooled exhausted recirculation (TURBOE), basic cylinder deactivation (TURBOD), variable valve lift (TURBO1), and advanced cylinder deactivation (TURBOAD). Advancing further down the Turbo Engine Path leads to an engine with a higher brake mean effective pressure (BMEP), which is a function of displacement and power. In other words, the higher the BMEP, the higher the power density of the engine. NHTSA models an advanced turbocharging technology (TURBO2) that runs increasingly higher turbocharger boost levels, burning more fuel and making more power for a given displacement. This analysis pairs turbocharging with engine downsizing, meaning that the turbocharged downsized engines improve vehicle fuel economy by using less fuel to power the smaller engine while maintaining vehicle performance.

The technology pathways represent an increase in the level or combinations of technologies being applied, with lower levels at the top and higher levels at the bottom of the path. Chapter 3.1.1 of the Final TSD shows the technology pathways for visualization purposes; however, the CAFE Model could apply any cost-effective combinations of technologies from those given pathways. Levels of improvement are dependent upon the vehicle class and the technology combinations. Again, in general, the paths are tied to ease of implementation of additional technology and how closely the technologies are related. An example of how this applies to the TURBO family of technologies is described below. The pathways are not aligned from “least effective” to “most effective” because assuming so would ignore several important considerations, including how technologies interact on a vehicle, how technologies interact on vehicles of different sizes that have different power requirements, and how hardware changes may be required for a particular technology. For example, the scenario below describes how, once a manufacturer downsizes an engine accompanying the application of a turbocharger, it would most likely not re-upsize the engine to add a less advanced turbocharger. The interaction of these technology combinations is discussed in more detail in Final TSD Chapter 2.

While TURBO0 is modeled with cooled EGR (TURBOE) and with DEAC (TURBOD), these technologies do not apply to TURBO1 or TURBO2; this decision is intentional. NHTSA defines TURBO1 in the analysis by adding VVL to the TURBO0 engine, and TURBO2 is the highest turbo downsized engine with a high BMEP. The benefits of cooled EGR and DEAC on TURBO1 and TURBO2 technologies would occur at high engine speeds and loads, which do not occur on the two-cycle tests. Because NHTSA measured technology effectiveness in this analysis based on the delta in improvements in vehicles' two-cycle test fuel consumption values, adding cooled EGR and DEAC to TURBO1 and TURBO2 would provide little effectiveness improvement for the corresponding increase in cost, a technology decision that the agency does not believe manufacturers would adopt in the real world. NHTSA's modeling effectively captures these complex interactions among technologies--an example of why effectiveness values from different technologies cannot simply be added together.\161\ This potential for added costs with limited efficiency benefit is also an example of why the CAFE Model technology tree is not ordered from least to most effective technology and why particular technologies are included on the technology tree while others are not. Final TSD Chapter 2 provides more discussion on interactions among individual technologies in the full-vehicle simulations.

\161\ NHTSA-2021-0053-0007-A3, at 15; NHTSA-2021-0053-0002-A9, at 21-23.

Consistent with the approach of preventing moving backward in the technology tree, the Model does not allow a vehicle assigned a TURBO2 technology to adopt a TURBOE technology. A vehicle in the analysis fleet that is assigned the TURBO2 technology indicates a manufacturer has made the decision to either skip over or move on from lower levels of force induction technology. Moving backwards on the technology tree from TURBO2 to any of the lower turbo technologies would require the engine to be upsized to meet the same performance metrics as the analysis fleet vehicle. As discussed further in Section II.C.2.c, NHTSA ensures the vehicles in this analysis meet similar performance levels after the application of fuel economy-improving technology as they did before the application of the technology, because the agency's objective is to measure the costs and benefits of manufacturers responding to CAFE standards in this analysis and not the costs or benefits related to changing performance metrics in the fleet. Moving from a higher to a lower turbo technology works counter to saving fuel as the engine would grow in displacement, requiring more fuel, adding frictional losses, and increasing weight and cost. Accordingly, the agency believes that the Turbo engine pathway appropriately captures the ways manufacturers might apply increasing levels of turbocharging technology to their vehicles.

In this analysis, HCR engines represent a class of engines that achieve a higher level of fuel efficiency by implementing a high geometric CR with varying degrees of late intake valve closing (LIVC) (i.e., closing the intake valve later than usual) using VVT, and without the use of an electric drive motor.\162\ These engines operate on a modified Atkinson cycle, allowing for improved fuel efficiency under certain engine load conditions while still offering enough power not to require an electric motor; however, there are limitations on how HCR engines can apply LIVC and the types of vehicles that can use this technology. The way that each individual manufacturer implements a modified Atkinson cycle is unique, as each manufacturer must balance not only fuel efficiency considerations, but also emissions, on-board diagnostics, and safety considerations, which include the vehicle being able to operate responsively to the driver's demand.

\162\ LIVC is a method manufacturers use to reduce the effective compression ratio and allow the expansion ratio to be greater than the compression ratio resulting in improved fuel economy but reduced power density. Further technical discussion on HCR and Atkinson engines are discussed in Final TSD Chapter 3.1.1.2.3. The 2015 NAS Report, Appendix D, includes a short discussion on thermodynamic engine cycles.

NHTSA defines HCR engines as being naturally aspirated, gasoline, SI, using a geometric CR of 12.5:1 or greater \163\ and able to apply various levels of LIVC dynamically based on load demand. An HCR engine uses less fuel for each

engine cycle, which increases fuel economy but decreases power density (or torque). Generally, during high loads--when more power is needed-- the engine will use variable valve actuation to reduce the level of LIVC by closing the intake valve earlier in the compression stroke (leaving more air/fuel mixture in the combustion chamber), increasing the effective CR, reducing over-expansion, and sacrificing efficiency for increased power density.\164\ However, there is a limit to how much the air-fuel mixture can be compressed before ignition in the HCR engine due to the potential for engine knock.\165\ Engine knock can be mitigated in HCR engines with higher octane fuel; however, the fuel specified for use in most vehicles is not higher octane fuel. Conversely, at low loads, the engine will typically increase the level of LIVC by closing the intake valve later in the compression stroke, reducing the effective CR, increasing the over-expansion, and sacrificing power density for improved efficiency. By closing the intake valve later in the compression stroke (i.e., applying more LIVC), the engine's displacement is effectively reduced, which results in less air and fuel for combustion and a lower power output.\166\ Varying LIVC can be used to mitigate, but not eliminate, the low power density issues that can constrain the application of an Atkinson-only engine.

\163\ Note that even if an engine has a compression ratio of 12.5:1 or greater, it does not necessarily mean it is an HCR engine in NHTSA's analysis, as discussed below. NHTSA looks at a number of factors to perform baseline engine assignments.

\164\ Variable valve actuation is a general term used to describe any single or combination of VVT, VVL, and variable valve duration used to dynamically alter an engine's valvetrain during operation.

\165\ Engine knock in spark ignition engines occurs when combustion of some of the air/fuel mixture in the cylinder does not result from propagation of the flame front ignited by the spark plug rather, one or more pockets of air/fuel mixture explode outside of the envelope of the normal combustion front.

\166\ Power = (force x displacement)/time.

The phrase “low power density issues” translates to a low torque density,\167\ meaning that the engine cannot create the torque required at necessary engine speeds to meet load demands. To the extent that a vehicle requires more power in a given condition than an engine with low power density can provide, that engine would experience issues like engine knock for the reasons discussed above; more importantly, an engine designer would not allow a particular engine design to be used in conditions where the engine has the potential to operate in unsafe conditions. Instead, a manufacturer could significantly increase an engine's displacement (i.e., size) to overcome those low power density issues,\168\ or could add an electric motor and battery pack to provide the engine with more power; however, a far more effective pathway would be to apply a different type of engine technology, like a downsized, turbocharged engine.\169\ Because of these limitations with HCR engines, NHTSA restricts the Model from applying this technology to vehicles that would be negatively impacted by the technology, like pickup trucks.\170\

\167\ Torque = radius x force.

\168\ 2024 EPA Automotive Trends Report at 54 (“As vehicles have moved towards engines with a lower number of cylinders, the total engine size, or displacement, is also at an all-time low.”). The discussion below describes why NHTSA does not believe manufacturers will increase the displacement of HCR engines to make the necessary power because of the negative impacts it has on fuel efficiency.

\169\ See Toyota, 2024 Toyota Tacoma Makes Debut on the Big Island, Hawaii, last revised: May 19, 2023, available at: https://pressroom.toyota.com/2024-toyota-tacoma-makes-debut-on-the-big-island-hawaii/ (accessed: May 28, 2026). The 2024 Toyota Tacoma comes in eight “grades,” all of which use a turbocharged engine.

\170\ Final TSD Chapter 3.1.1.2.3 includes more discussion on HCR and HCR restrictions.

Vehicle manufacturers' intended performance attributes for a vehicle--like payload and towing capability, features for off-road use, and other attributes that affect aerodynamic drag and rolling resistance--dictate whether an HCR engine can be a suitable technology choice for that vehicle.\171\ As vehicles require higher payloads and towing capacities,\172\ experience higher road load forces from larger all-terrain tires or less aerodynamic designs, or experience driveline losses for AWD and 4WD configurations, more engine torque is required at all engine speeds. When more engine torque is required, the application of HCR technology becomes less effective and more limited.\173\ For these reasons, and to maintain a performance-neutral analysis, NHTSA limits non-hybrid and non-plug-in-hybrid HCR engine application to certain categories of vehicles.\174\

\171\ Supplemental Comments of Toyota, Notice of Proposed Rulemaking: Safer Affordable Fuel-Efficient Vehicles Rule, Docket ID No: NHTSA-2018-0067 and Docket No. EPA-HQ-OAR-2018-0283, at 6; Feng, R. et al. Investigations of Atkinson Cycle Converted from Conventional Otto Cycle Gasoline Engine, SAE Technical Paper 2016- 01-0680, SAE International: Warrendale, VA (2016), available at: https://www.sae.org/publications/technical-papers/content/2016-01-0680/ (accessed: May 28, 2026).

\172\ See Tucker, S., What Is Payload: A Complete Guide. Kelly Blue Book, last revised: Feb. 2, 2023, available at: https://www.kbb.com/car-advice/payload-guide/#link3 (accessed: May 28, 2026). (“Roughly speaking, payload capacity is the amount of weight a vehicle can carry, and towing capacity is the amount of weight it can pull. Automakers often refer to carrying weight in the bed of a truck as hauling to distinguish it from carrying weight in a trailer or towing.”).

\173\ See Supplemental Comments of Toyota, Docket No. NHTSA- 2018-0067 and Docket No. EPA-HQ-OAR-2018-0283, at 6, 8 (Mar. 25, 2019), available at: https://www.regulations.gov/comment/NHTSA-2018-0067-12376 (accessed: May 28, 2026) (Supplemental Toyota Comments) (“Tacoma has a greater coefficient of drag from a larger frontal area, greater tire rolling resistance from larger tires with a more aggressive tread, and higher driveline losses from 4WD. Similarly, the towing, payload, and off-road capability of pick-up trucks necessitate greater emphasis on engine torque and horsepower over fuel economy. This translates into engine specifications such as a larger displacement and a higher stroke-to-bore ratio . . . . Tacoma's higher road load and more severe utility requirements push engine operation more frequently to the less efficient regions of the engine map and limit the level of Atkinson operation . . . . This endeavor is not a simple substitution where the performance of a shared technology is universal. Consideration of specific vehicle requirements during the vehicle design and engineering process determine the best applicable powertrain.”).

\174\ To maintain performance neutrality when sizing powertrains and selecting technologies, NHTSA performs a series of simulations in Autonomie, which are further discussed in the Final TSD Chapter 2.3.4 and in the CAFE Analysis Autonomie Documentation. The concept of performance neutrality is discussed in detail above in Section II.C.2.c, Technology Effectiveness Values, and additional reasons why NHTSA maintains a performance neutral analysis are discussed in Section II.C.2.f, Technology Applicability Equations and Rules.

NHTSA includes three HCR Engine Path technology options in this analysis: (1) a first-level Atkinson-enabled engine (HCR) with VVT and SGDI; (2) an Atkinson-enabled engine with CEGR (HCRE); and (3) an Atkinson-enabled engine with DEAC (HCRD). This updated family of HCR engine map models also reflects the statement in NHTSA's May 2, 2022 final rule that a single engine that employs an HCR, CEGR, and DEAC “is unlikely to be utilized in the rulemaking timeframe based on comments received from the industry leaders in HCR technology application.” \175\

\175\ 87 FR 25796 (May 2, 2022).

These three HCR Engine Path technology options (HCR, HCRE, HCRD) should not be confused with the hybrid and plug-in hybrid electric pathway options that also utilize HCR engines in combination with a P2 hybrid powertrain (e.g., P2HCR, P2HCRE, PHEV20H, and PHEV50H); those hybridization path options are discussed in Section II.D.3 below. In contrast, Atkinson engines in NHTSA's power-split hybrid powertrains (SHEVPS, PHEV20PS, and PHEV50PS) run the Atkinson Cycle full time but are connected to an electric motor. The full-time Atkinson engines are also discussed in Section II.D.3.

The Miller cycle is another alternative combustion cycle that effectively uses an extended expansion stroke, similar to the Atkinson cycle but with the application of forced induction to

improve fuel efficiency. Miller cycle-enabled engines have a similar trade-off in power density as Atkinson engines; the lower power density requires a larger volume engine in comparison to an Otto cycle-based turbocharged system for similar applications.\176\ To address the impacts of the extended expansion stroke on power density during high- load operating conditions, the Miller cycle operates in combination with a forced induction system. In NHTSA's analysis, the first-level Miller cycle-enabled engine includes the application of variable turbo geometry technology (VTG), or what is also known as a variable-geometry turbocharger. VTG technology allows for the adjustment of key geometric characteristics of the turbocharging system, thus allowing adjustment of boost profiles and response based on the engine's operating needs. The adjustment of boost profile during operation increases the engine's power density over a broader range of operating conditions and increases the functionality of a Miller cycle-based engine. The use of a variable geometry turbocharger also supports the use of CEGR. NHTSA's second level of VTG engine technology (VTGE) is an advanced Miller cycle-enabled system that includes the application of at least a 40V- based electronic boost system. An electronic boost system has an electric motor added to assist the turbocharger; the motor assist mitigates turbocharger lag and low boost pressure by providing the extra boost needed to overcome the torque deficit at low engine speeds.

\176\ National Research Council, Assessment of Technologies for Improving Fuel Economy of Light-Duty Vehicles--2025-2035, The National Academies Press: Washington, DC (2021), available at: https://doi.org/10.17226/26092 (accessed: May 28, 2026) (hereinafter, “2021 NAS report”).

Variable compression ratio (VCR) engines work by changing the length of the piston stroke of the engine to optimize the CR and improve thermal efficiency over the full range of engine operating conditions. Engines that use VCR technology are currently in production as small-displacement, turbocharged, in-line four-cylinder, high BMEP applications.

Diesel engines have several characteristics that result in better fuel efficiency over traditional gasoline engines, including reduced pumping losses due to lack of (or greatly reduced) throttling, high- pressure direct injection of fuel, a combustion cycle that operates at a higher CR, and a very lean air/fuel mixture relative to an equivalent-performance gasoline engine. However, diesel technologies require additional systems to control nitrogen oxide (NOX) emissions, such as a NOX adsorption catalyst system or a urea/ammonia selective catalytic reduction system. NHTSA included two levels of diesel engine technology in the analysis: the first-level diesel engine technology (Advanced Diesel Engine (ADSL)) is a turbocharged diesel engine, and the more advanced diesel engine (DSLI) adds DEAC to the ADSL engine technology. The diesel engine maps are new for this analysis and are based on a modern 3.0L turbo-diesel engine.

Finally, compressed natural gas (CNG) systems are ICE vehicles that run on natural gas as a fuel source. The fuel storage and supply systems for these engines differ tremendously from gasoline, diesel, and flexible-fuel vehicles.\177\ The CNG engine option has been included in past analyses; however, the light-duty analysis fleet does not include any dedicated CNG vehicles. As with the last analyses, CNG engines are included as an analysis fleet-only technology and are not applied to any vehicle that did not already include a CNG engine.

\177\ Flexible-fuel vehicles (FFV) are designed to run on gasoline or gasoline-ethanol blends of up to 85 percent ethanol.

There are other vehicle technologies that work in various ways to improve fuel efficiency, such as turbo compounding, negative valve overlaps in-cylinder fuel reforming (NVO), passive prechamber combustion (PPC), and high energy ignition, which are not included in NHTSA's analysis. The International Council on Clear Transportation (ICCT) also provided examples of technology we do not use in the analysis such as NVO, PPC, and high energy ignition.\178\ Though suitable explanations for their exclusion could be that these technologies are in various stages of development and some, like PPC, are in very limited production, the primary reason NHTSA opted not to include them in the analysis is that the agency does not have information suggesting that these technologies will gain significant adoption during the rulemaking timeframe. This topic was discussed in detail in the 2022 final rule,\179\ and the agency has not found evidence of significant development for mass market production across multiple vehicle lines since then that would indicate manufacturers are now pursuing these costly technologies within the same standard-setting years. NHTSA will monitor these technologies as time progresses as part of NHTSA's continuous efforts to improve its modeling for any future analysis.\180\

\178\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 2-9.

\179\ 87 FR 25784 (May 2, 2022).

\180\ NHTSA is aware of the PPC technology from two different manufacturers that is planned to be included on a relatively low volume of vehicles for MY 2027.

The first step in assigning engine technologies to vehicles in the analysis fleet is to use data for each manufacturer to determine which vehicle platforms share engines. Within each manufacturer's fleet, NHTSA develops and assigns unique engine codes based on configuration, technologies applied, displacement, compression ratio, and power output. NHTSA also assigns engine technology classes, which are codes that identify engine architecture (i.e., how many cylinders the engine has, whether it is a DOHC or SOHC, and so on) to account accurately for engine costs in the analysis.

When assigning engine technologies to vehicles in the analysis fleets, it is important to consider the actual technologies on a manufacturer's engine and compare them to the engine technologies in the analysis. NHTSA has over 250 unique engine codes in the light-duty analysis fleet, meaning that the technologies present on those engines in the real world must be identified and matched to the 29 engine map models (and therefore engine technology on the technology tree) \181\ that best represents those real-world engines. When considering how best to fit each of those 250 engines to the 29 engine technologies and engine map models, NHTSA uses specific technical elements contained in manufacturer publications, press releases, vehicle benchmarking studies, technical publications, manufacturer's specification sheets, occasionally CBI, and engineering judgment. The information NHTSA reviews includes specific technologies such as cylinder deactivation and direct injection, along with engine architecture (DOHC or SOHC), engine displacement, compression ratio, and horsepower, which help NHTSA to appropriately assign a modeled engine to an engine in the analysis fleet. For example, an engine having a 13.0:1 CR is a good indication that the engine would be considered an HCR engine. Some engines that achieve a slightly lower CR (e.g., 12.5) may also be considered an HCR engine depending on other technology on the engine, such as the inclusion of SGDI, increased engine displacement compared to other

competitors, reduction of engine parasitic losses through variable or electric oil and water pumps, or the combination of these technologies. Importantly, engine technologies are never assigned based on one factor alone but rather data and engineering judgment are used to assign complex real-world engines to their corresponding engine technologies in the analysis. NHTSA believes that the initial characterization of the fleet's engine technologies reasonably captures the current state of the market while maintaining a reasonable amount of analytical complexity. Also, in addition to the 29 engine map models used in the Engine Pathways, there are 16 additional potential powertrain technology assignments available in the Hybridization Pathways.

\181\ NHTSA assigns each engine code technology that most closely corresponds to an engine map; for most technologies, one box on the technology tree corresponds to one engine map that corresponds to one engine code.

Engine technology adoption in the Model is defined through a combination of technology path logic, refresh and redesign cycles, phase-in capacity limits, and SKIP logic. Path logic defines technology adoption by preventing an engine design from moving from one advanced engine tree to another. Once in an advanced engine tree, it must stay there. For example, any light-duty basic engine can adopt one of the TURBO engine technologies, but vehicles that have turbocharged engines in the analysis fleet stay on the Turbo Engine Path to prevent unrealistic engine technology change in the short timeframe considered in the rulemaking analysis. This is included to represent real-world considerations of stranded capital, which is when manufacturers amortize research, development, and tooling expenses over many years. Besides technology path logic, which applies to all manufacturers and technologies, NHTSA places additional constraints on the adoption of VCR and HCR technologies.

VCR technology requires a complete redesign of the engine and, in the analysis fleet, Nissan is the only manufacturer (including the Infiniti brand) to incorporate this technology. VCR engines are complex, costly by design, and address many of the same efficiency losses as mainstream technologies like turbocharged downsized engines. This makes it unlikely that a manufacturer that has already started down an incongruent technology path would adopt VCR technology. Because of these issues, VCR engine technology adoption is limited to OEMs that have already employed the technology and their partners. NHTSA does not believe any other manufacturers will invest in developing and marketing this technology in their fleet in the rulemaking timeframe.

As recognized in past analyses,\182\ HCR engines excel in lower power applications for lower load conditions, such as driving around a city or steady State highway driving without large payloads. Thus, their adoption is more limited than some other technologies. Accordingly, HCR engines are subject to three limitations.

\182\ The discussions at 83 FR 43038 (Aug. 24, 2018), 85 FR 24383 (Apr. 30, 2020), 86 FR 49658 and 49661 (Sept. 3, 2021), and 87 FR 25786 and 25790 (May 2, 2022) are incorporated here by reference.

First, vehicles with 405 or more HP, and (to simulate parts sharing) vehicles that share engines with vehicles with 405 or more HP, are not allowed to adopt HCR engines due to their prescribed power needs being more demanding and likely not supported by the lower power density found in HCR-based engines.\183\ Because LIVC essentially reduces the engine's displacement, to make more power and keep the same levels of LIVC, manufacturers would need to increase the displacement of the engine to make the necessary power. NHTSA does not believe manufacturers will increase the displacement of their engines to accommodate HCR technology adoption, because as displacement increases, so do friction, pumping losses, and fuel consumption. This bears out in industry trends: total engine size (or displacement) is at an all-time low, and trends show that industry focus on turbocharged downsized engine packages are leading to their much higher market penetration.\184\ Separately, as seen in the analysis fleet, manufacturers generally use HCR engines in applications where the vehicle's power requirements fall significantly below the agency's HCR HP threshold. In fact, the average HP for the sales-weighted average of vehicles in the analysis fleet that use HCR Engine Path technologies is 194 HP, demonstrating that HCR engine use has indeed been limited to lower HP applications, and well below the 405 HP threshold. In fringe cases where a vehicle classified as having higher load requirements does have an HCR engine, it is coupled to a hybrid system.\185\

\183\ Heywood (2018) at Chapter 5.

\184\ See 2024 EPA Automotive Trends Report at 54, 85.

\185\ See the Market Data Input File. As an example, the reported total system horsepower for the Ford Maverick HEV is also 191 HP, well below the 405 HP threshold. See also the Lexus LC/LS 500h: the Lexus LC/LS 500h also uses premium fuel to reach this performance level.

Second, to maintain a performance-neutral analysis,\186\ pickup trucks and (to simulate parts sharing) \187\ vehicles that share engines with pickup trucks are excluded from receiving HCR engines that are not accompanied by a hybrid powertrain. In other words, pickup trucks and vehicles that share engines with pickup trucks can receive HCR-based engine technologies only in the Hybridization Pathways of technologies. Pickup trucks and vehicles that share engines with pickup trucks are excluded from receiving HCR engines not accompanied by a hybrid powertrain because these often-heavier vehicles have higher low- speed torque needs, higher base road loads, increased payload and towing requirements,\188\ and have powertrains sized and tuned to perform this additional work beyond what passenger cars are required to conduct. Vehicle manufacturers' intended performance attributes for a vehicle--like payload and towing capability, intention for off-road use, and other attributes that affect aerodynamic drag and rolling resistance--dictate whether an HCR engine can provide a reasonable fuel economy improvement for that vehicle.\189\ For example, road loads are

composed of aerodynamic loads, which include vehicle frontal area and its drag coefficient, along with tire rolling resistance, all of which contribute to higher engine loads as vehicle speed increases.\190\ NHTSA assumes that a manufacturer intending to apply HCR technology to their pickup truck or vehicle that shares an engine with a pickup truck would do so in combination with an electric system to assist with the vehicle's load needs.

\186\ As discussed in detail in Sections II.C.2.c and II.C.2.f above, NHTSA maintains a performance-neutral analysis to capture only the costs and benefits of manufacturers adding fuel economy- improving technology to their vehicles in response to CAFE standards.

\187\ See Section II.C.2.f.

\188\ See SAE, Performance Requirements for Determining Tow- Vehicle Gross Combination Weight Rating and Trailer Weight Rating, SAE Standard J2807_202411, SAE International: Warrendale, PA (2024), available at: https://doi.org/10.4271/J2807_202411 (accessed: May 28, 2026); Reed, T., SAE J207 Tow Tests--The Standard, MotorTrend (2015), available at: https://www.motortrend.com/how-to/1502-sae-j2807-tow-tests-the-standard/ (accessed: May 28, 2026). When stating “increased payload and towing requirements,” NHTSA is referring to a defined set of requirements that manufacturers follow to ensure the manufacturer's vehicle can meet a set of performance measurements when building a tow vehicle to give consumers the ability to “cross-shop” between different manufacturers' vehicles. As discussed in detail above in Sections II.C.2.c and II.C.2.f, NHTSA maintains a performance-neutral analysis to ensure that the analysis is only accounting for the costs and benefits of manufacturers adding technology in response to CAFE standards. This means that adoption features, like the HCR application restriction, are applied to a vehicle that begins the analysis with specific performance measurements, like a pickup truck, where application of the specific technology would likely not allow the vehicle to meet the manufacturer's baseline performance measurements.

\189\ ICCT asked NHTSA to stop quoting a 2019 Toyota comment explaining why NHTSA does not allow HCR engines in pickup trucks, stating that Toyota's purpose in explaining that the Tacoma and Camry achieve different effectiveness improvements using their HCR engines is being misinterpreted. See NHTSA-2018-0067-12387 NHTSA disagrees. Toyota's comment is still relevant for this final rule as the limitations of the technology have not changed, which Toyota describes in the context of comparing why the technology provides a benefit in the Camry that one should not expect to see in the Tacoma. See Supplemental Toyota Comments at 6, 8. Note that Toyota also submitted a second set of supplemental comments (NHTSA-2018- 0067-12431) that confirms NHTSA's understanding of the most important concept to support NHTSA's decision to limit HCR adoption on pickup trucks, which is that Atkinson operation is limited on pickup trucks. See Supplemental Comments of Toyota Motor North America, Inc., in the NHTSA Docket No. NHTSA-2018-0067-12376-A1, at 8-9 in Regulations.gov. See Supplemental Comments of Toyota, Docket No. NHTSA-2018-0067 and Docket No. EPA-HQ-OAR-2018-0283, at 2-3 (July 15, 2019), available at: https://www.regulations.gov/comment/NHTSA-2018-0067-12431 (accessed: May 28, 2026).

\190\ 2015 NAS Report, at pp. 207-42.

Finally, HCR engine application is restricted for some heavily performance-focused manufacturers that have demonstrated a significant commitment to power-dense technologies such as turbocharged downsizing,\191\ such that their fleets use nearly 100 percent turbocharged downsized engines. This means that no vehicle manufactured by these manufacturers can receive an HCR engine. Again, this adoption feature is implemented to avoid an unquantified amount of stranded capital that would be realized if these manufacturers switched from one technology to another.

\191\ Three manufacturers that meet the criteria (near 100 percent turbo downsized fleet, and future hybrid systems are based on turbo downsized engines) described and are excluded: BMW, Mercedes-Benz, and Jaguar Land Rover.

Note that these adoption features apply only to vehicles that receive HCR engines that are not accompanied by a hybrid powertrain. A P2 hybrid system that uses an HCR engine overcomes the low-speed torque needs using the electric motor and thus has no restrictions or SKIPs applied.

ICCT commented on the application of HCR technologies, stating that, “NHTSA inappropriately prevents the application of HCR engine on engines with 405 horsepower, pickup trucks and vehicles that share engines with pickup trucks, or performance-focused manufacturers.” \192\ ICCT has provided similar comments on previous CAFE rulemakings, but it has not provided data to support its claims beyond pointing to its prior comments, to which NHTSA has previously responded. To avoid repetition, previous discussions located in prior related documents are incorporated here by reference.\193\ The agency notes that HCR engines have yet to be applied to the use cases identified by ICCT, and we will continue to assess technology improvements and refine our modeling efforts based on the best available data.

\192\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 9-10.

\193\ 86 FR 74236 (Dec. 29, 2021), 87 FR 25710 (May 2, 2022), Final Br. for Resp'ts, Nat. Res. Def. Council v. NHTSA, Case No. 22- 1080, ECF No. 2000002 (D.C. Cir. May 19, 2023).

NHTSA realizes that engine technology, vehicle type, and their applications are always evolving.\194\ The Hyundai Santa Cruz, a unibody pickup truck with a 4-cylinder HCR engine, is one example of a pickup truck with a non-hybrid HCR engine. However, the Santa Cruz is not comparable in capability to other pickup models like the Tacoma, Colorado, and Canyon, and it therefore cannot be assumed that those pickup models should be able to adopt non-hybrid HCR technology as well. Small unibody pickup trucks like the Santa Cruz and the Ford Maverick do not have the same capabilities and functionality as a mid- size body-on-frame pickup like the Toyota Tacoma.\195\ NHTSA believes that its current restrictions for HCR are reasonable and appropriate, and the agency has not been presented with any new information that would suggest otherwise. NHTSA's stance on this issue is also borne out in real-world trends. Manufacturers who currently offer HCR engines in their fleets and therefore had the potential to introduce HCR technologies on recently redesigned vehicles that previously used high- displacement NA engines (such as Toyota Tacoma or Chevrolet Colorado) or TURBO technologies (such as the Mazda CX-90 replacing CX-9) have instead opted to introduce or continue to pursue turbocharged or hybrid engines. NHTSA does not believe HCR in its current state can provide enough fuel efficiency benefit to support removing the current HCR restrictions; however, this by no means precludes manufacturers from developing and deploying HCR technology for future iterations of their pickup trucks.

\194\ ICCT has disagreed with NHTSA's HCR restrictions in the past, and while NHTSA has made attempts to better explain its position on HCR technology and where NHTSA believes it is appropriate, NHTSA's justification has remained the same. NHTSA does not believe the HCR technology is applicable to these types of vehicles because of the nature of how the technology works, and removing the restrictions would present an unrealistic pathway to compliance for manufacturers that is not maximum feasible.

\195\ The specification of 2024 Ford Maverick, Toyota Tacoma, and Hyundai Santa Cruz are in the docket accompanying this final rule.

NHTSA also emphasizes that, in the real world, manufacturers are not required to follow the technology pathways to compliance that the agency models in the standard-setting analysis but can instead take their own pathway based on their respective business models, technology availability, market share, and other considerations. The CAFE Model simulates an example of a low-cost compliance pathway, and no manufacturer is required to comply with the pathway as it has been modeled. Instead, manufacturers are free to choose their own path to compliance. NHTSA has added features and restrictions into the CAFE Model to make the compliance simulation more representative of how manufacturers make decisions about technology adoption in the real world. This is to ensure that the CAFE Model does not simulate unrealistic compliance pathways. For example, if the CAFE Model simulated manufacturers abandoning one technology in favor for another, particularly with respect to HCR technology for pickup trucks and high HP vehicles, the results and corresponding costs and benefits would be unrealistic and could lead to NHTSA setting standards that are more stringent than maximum feasible. For this and other reasons, the agency endeavors to model the most realistic and low-cost pathway to compliance. NHTSA's standard-setting analysis is also restricted in ways that manufacturers are not, which increases the likelihood that manufacturers will not follow the technology pathways projected in the standard-setting analysis.\196\

\196\ 49 U.S.C. 32902(h).

Lucid Group, Inc. (Lucid) commented on how NHTSA maintains performance neutrality in its modeling, stating that, “[t]he technical analysis links fuel economy improvements to reductions in peak horsepower without incorporating vehicle mass and power to weight. Real-world efficiency strategies optimize engine output relative to mass rather than relying on absolute horsepower reductions alone.” \197\ The commenter misunderstands the agency's approach. NHTSA uses peak HP for the sizing of powertrains to maintain performance neutrality but also takes into account vehicle mass to ensure that the modeling is capturing all of the complex interactions among numerous vehicle

attributes, including mass, acceleration, and technology combinations, among many others. This is discussed in detail in Final TSD Chapter 3 and the Autonomie documentation.

\197\ Lucid, Docket No. NHTSA-2025-0491-6043-A1, at 10.

How effective an engine technology is at improving a vehicle's fuel economy depends on several factors, such as the vehicle's technology class and any additional technology added or removed from the vehicle in conjunction with the new engine technology, as discussed in Section II.C above. The Autonomie model's full-vehicle simulation results provide most of the effectiveness values that are used as inputs to the CAFE Model. Chapter 2.4 of the Final TSD and the CAFE Analysis Autonomie Documentation provide a full discussion of the Autonomie modeling. The Autonomie modeling uses engine map models as the primary inputs for simulating the effects of different engine technologies.

Engine maps provide a three-dimensional representation of engine performance characteristics at each engine speed and load point across the operating range of the engine. Engine maps have the appearance of topographical maps, typically with engine speed on the horizontal axis and engine torque, power, or BMEP on the vertical axis. A third engine characteristic, such as brake-specific fuel consumption (BSFC), is displayed using contours overlaid across the speed and load map. The contours provide the values for the third characteristic in the regions of operation covered on the map. Other characteristics typically overlaid on an engine map include engine emissions, engine efficiency, and engine power. The engine maps developed to model the behavior of the engines in this analysis are referred to as engine map models.

The engine map models used in this analysis are representative of technologies currently in production or expected to be available in the rulemaking timeframe. The engine map models are developed to be representative of the performance achievable across the industry for a given technology, and they are not intended to represent the performance of a single manufacturer's specific engine. NHTSA targets a broadly representative performance level because the same combination of technologies produced by different manufacturers will differ in performance, due to manufacturer-specific designs for engine hardware, control software, and emissions calibration. Accordingly, the agency expects that the engine maps developed for this analysis will differ from engine maps for manufacturers' specific engines. However, it is intended and expected that the incremental changes in performance modeled for this analysis, due to changes in technologies or technology combinations, will be similar to the incremental changes in performance observed in manufacturers' engines for the same changes in technologies or technology combinations.

IAV developed most of the engine map models used in this analysis. IAV is one of the world's leading automotive industry engineering service partners with an over 35-year history of performing research and development for powertrain components, electronics, and vehicle design.\198\ SwRI developed the light-duty diesel engine maps for this analysis. SwRI has been providing automotive science, technology, and engineering services for over 70 years.\199\ Both IAV and SwRI developed these engine maps using GT-POWER. GT-POWER is a commercially available industry-standard engine performance simulation tool. GT- POWER can be used to predict detailed engine performance characteristics, such as power, torque, airflow, volumetric efficiency, fuel consumption, turbocharger performance and matching, and pumping losses.\200\

\198\ IAV GmbH, IAV, available at: https://www.iav.com/ (accessed: May 28, 2026).

\199\ Southwest Research Institute, Southwest Research Institute, available at: https://www.swri.org (accessed: May 28, 2026).

\200\ This weblink has additional information on the GT-POWER tool: Gamma Technologies, GT-POWER: Industry Leading Engine Simulation Software, available at: https://www.gtisoft.com/gt-power/ (accessed: May 28, 2026).

Just like Argonne optimizes a single vehicle model in Autonomie following the addition of a singular technology to the vehicle model, these engine map models were built in GT-POWER by incrementally adding engine technology to an initial engine--built using engine test data, component test data, and manufacturers' and suppliers' technical publications--and then optimizing the engine to consider real-world constraints like heat, friction, and knock. One of the basic assumptions the agency makes when developing these engine maps is using 87 octane Tier 3 gasoline because it is the most common octane rating on which engines are designed to operate, and it is the test fuel manufacturers will have to use for EPA fuel economy testing.201 202 203 A small number of initial engine configurations with well-defined BSFC maps are used, and then, in a systematic and controlled process, specific well-defined technologies are added to optimize a BSFC map for each unique technology combination. This could theoretically be done through engine or vehicle testing, but such an approach would require conducting tests on a single engine, and each configuration would require physical parts and associated engine calibrations to assess the impact of each technology configuration. This is impractical for the rulemaking analysis because of the extensive design, prototype part fabrication, development, and laboratory resources that are required to evaluate each unique configuration. Both NHTSA and the automotive industry use modeling as an approach to assess an array of technologies with more limited physical testing. Modeling offers the opportunity to isolate the effects of individual technologies by using a single or small number of initial engine configurations and incrementally adding technologies to those initial configurations. This provides a consistent reference point for the BSFC maps for each technology and for combinations of technologies that enable us to identify and quantify carefully the differences in effectiveness among technologies.

\201\ 79 FR 23414 (Apr. 28, 2014).

\202\ DOE, Selecting the Right Octane Fuel, available at: https://www.fueleconomy.gov/feg/ octane.shtml#:~:text=You%20should%20use%20the%20octane%20rating%20req uired%20for,others%20are%20designed%20to%20use%20higher%20octane%20fu el (accessed: May 28, 2026).

\203\ It is also important to note that regulation of fuels used for determining CAFE compliance is outside the scope of NHTSA's authority. 49 U.S.C. 32904(c).

Before its use in the Autonomie analysis, both IAV and SwRI validated the generated engine maps against a global database of benchmarked data, engine test data, single-cylinder test data, prior modeling studies, technical studies, and information presented at conferences.\204\ IAV and SwRI also validated the effectiveness values from the simulation results against detailed engine maps produced from the

Argonne engine benchmarking programs, as well as published information from industry and academia.\205\ This ensures reasonable representation of simulated engine technologies. Additional details and assumptions that are used in the engine map modeling are described in detail in Chapter 3.1 of the Final TSD and the CAFE Analysis Autonomie Model Documentation chapter titled “Autonomie--Engine Model.”

\204\ Friedrich, I. et al., Automatic Model Calibration for Engine-Process Simulation with Heat-Release Prediction, SAE Technical Paper 2006-01-0655, SAE International: Warrendale, VA (2006), available at: https://doi.org/10.4271/2006-01-0655 (accessed: May 28, 2026); Rezaei, R. et al., Zero-Dimensional Modeling of Combustion and Heat Release Rate in DI Diesel Engines, SAE International Journal Of Engines. Vol. 5(3) at 874-85 (2012), available at: https://doi.org/10.4271/2012-01-1065 (accessed: May 28, 2026); Berndt, R. et al., Multistage Supercharging for Downsizing with Reduced Compression Ratio, MTZ Worldwide. Vol. 76: pp. 10-11 (2015), available at: https://doi.org/10.1007/s38313-015-0036-4 (accessed: May 28, 2026); Neukirchner, H. et al., Symbiosis of Energy Recovery and Downsizing, MTZ Worldwide, Vol. 75: pp. 4-9 (2014), available at: https://doi.org/10.1007/s38313-014-0219-4 (accessed: May 28, 2026).

\205\ Bottcher, L., & Grigoriadis, P., ANL-BSFC map prediction Engines 22-26, National Highway Traffic Safety Association: Washington, DC (2019), available at: https://lindseyresearch.com/wp-content/uploads/2021/09/NHTSA-2021-0053-0002-20190430_ANL_Eng-22-26-Updated_Docket.pdf (accessed: May 28, 2026); Reinhart, T., Engine Efficiency Technology Study, Final Report, SwRI Project No. 03.26457, Southwest Research Institute: San Antonio, TX (2022), available at: https://downloads.regulations.gov/EPA-HQ-OAR-2022-0829-0230/attachment_17.pdf (accessed: May 28, 2026).

Note that absolute BSFC levels are never applied from the engine maps to any vehicle model or configuration for the rulemaking analysis; only the absolute fuel economy values from the full-vehicle Autonomie simulations are used to determine incremental effectiveness for switching from one technology to another technology. The incremental effectiveness is then applied to the absolute fuel economy or fuel consumption value of vehicles in the analysis fleet, which are based on CAFE compliance data. For subsequent technology changes, NHTSA applies incremental effectiveness changes to the absolute fuel economy level of the previous technology configuration. Therefore, for a technically sound analysis, it is most important that the differences in BSFC among the engine maps be accurate and not the absolute values of the individual engine maps.

While the fuel economy improvements for most engine technologies in the analysis are derived from the database of Autonomie full-vehicle simulation results, the analysis incorporates a handful of what the agency refers to as “analogous effectiveness values.” These are used when an engine map model is not available for a particular technology combination. To generate an analogous effectiveness value, data from analogous technology combinations for available engine map models are used by conducting a pairwise comparison to generate a data set of emulated performance values for adding technology to an initial application. Analogous effectiveness values are used only for four SOHC technologies. NHTSA has determined that the effectiveness results using these analogous effectiveness values provided reasonable results. This process is discussed further in Chapter 3.1.4.2 of the Final TSD.

The engine technology effectiveness values for all vehicle technology classes can be found in Chapter 3.1.4 of the Final TSD. These values show the calculated improvement for upgrading the listed engine technology for a given combination of other technologies. The range of effectiveness values listed for each specific technology (e.g., TURBO1) represents the addition of the TURBO1 technology to every technology combination that could select the addition of TURBO1. These values are derived from the Argonne Autonomie Results Dataset and the righthand side Y-axis shows the number of Autonomie simulations that achieve each percentage effectiveness improvement point. The dashed line and gray shading indicate the median and 1.5X interquartile range (IQR), which is a helpful metric to identify outliers. After comparing these histograms to the box and whisker plots presented in prior CAFE program rule documents, the number of effectiveness outliers is extremely small.

ICCT commented on the application of the engine sizing algorithm and when it is applied in relation to vehicle road load improvement technologies stating that, “NHTSA continues to only downsize engines for large changes in tractive load,” which they assume artificially increases the overall performance of the fleet.\206\ The commenter misunderstands the agency's analytical approach. Final TSD Chapter 2.3.4 discusses NHTSA's approach of sizing powertrains, which iteratively goes through both low and high-speed acceleration performance loops and adjusts powertrain size as needed based on the performance neutrality requirements.\207\

\206\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 3.

\207\ CAFE Analysis Autonomie Documentation chapters titled “Vehicle and Component Assumptions” and “Vehicle Sizing Process.”

ICCT also implies that the analysis should require engine resizing for every technology change on a vehicle platform. NHTSA does not resize the engine for every technology change on a vehicle platform because doing so would artificially inflate effectiveness relative to cost. Manufacturers have repeatedly and consistently conveyed that the costs for redesign and the increased manufacturing complexity resulting from continually resizing engine displacement for small technology changes preclude them from doing so. It would not be reasonable or cost effective to expect resizing powertrains for every unique combination of technologies, and even less reasonable and less cost effective for every unique combination of technologies across every vehicle model due to the extreme manufacturing complexity that would be required.\208\ NAS stated in its 2011 report, “[f]or small (under 5 percent [of curb weight]) changes in mass, resizing the engine may not be justified, but as the reduction in mass increases (greater than 10 percent [of curb weight]), it becomes more important for certain vehicles to resize the engine and seek secondary mass reduction opportunities.” \209\ NHTSA's analysis evaluates engine resizing with mass changes of 10 percent or greater.\210\

\208\ For more details, see comments and discussion in the 2020 Rulemaking preamble Section VI.B.3.(a)(6) Performance Neutrality.

\209\ National Research Council, Assessment of Fuel Economy Technologies for Light-Duty Vehicles, The National Academies Press: Washington, DC, p. 107 (2011), available at: https://doi.org/10.17226/12924 (accessed: May 28, 2026) (hereinafter, “2011 NAS report”).

\210\ See Final TSD Chapter 3.4 on mass reduction for further discussion on engine resizing with respect to mass reduction.

ICCT also commented regarding the validity of the continued use of NHTSA's engine map models. ICCT stated that, “[a]lthough NHTSA scales its MY 2010 hybrid Atkinson engine map to match the thermal efficiency of the MY 2017 Toyota Prius, this appears to have been the only update made to the several engine maps that underpin all base and advanced engine technologies. The remaining engine maps are still primarily based on outdated engines (e.g., from MY 2011, 2013 and 2014 vehicles). Even with the updated hybrid engine, the newest Toyota Prius demonstrates an additional 10 percent improvement over the outgoing variant, due in part to improvements in engine efficiency.” \211\ ICCT also took issue with NHTSA not using two of EPA's engine map models and commented on the lack of effectiveness benefit they perceived for adding cylinder deactivation technology to turbocharged and HCR engines.

\211\ ICCT, Docket No. NHTSA-2025-0491-5240-A1, at 3-4 and -A2, at 2.

In response to this comment, NHTSA emphasizes that many of the engine maps were developed specifically to support analysis for the current rulemaking timeframe. The engine map models encompass engine technologies present in the analysis fleet and technologies that could be applied in the rulemaking timeframe. In many cases, those engine technologies are

mainstream in the baseline fleet and will continue to be mainstream during the rulemaking timeframe. For example, the engines on some MY 2024 vehicles in the analysis fleet have technologies that were introduced ten or more years ago. Ensuring we use engine maps that are representative of those technologies is important for the analysis. The most basic engine technology levels also provide a useful consistent starting point for the incremental improvements for other engine technologies. The timeframe for the testing or modeling used to generate a given engine map is unimportant because the mere passage of time does not affect the validity of engine map data. A given engine or model will produce the same BSFC map regardless of when testing or modeling is conducted. Eliminating engine maps based on temporal considerations alone would arbitrarily result in discarding useful and valid technical information.

ICCT also commented that the hybrid engine map models are outdated and that the hybrid effectiveness values exceed reasonable thermal efficiency.212 213 This issue is further discussed in Section III.D.3 of this preamble. NHTSA previously responded to ICCT's criticisms for not employing EPA's engine map models for the 2020 final rule setting MY 2021-2026 standards, demonstrating that the modeled engines provided similar incremental effectiveness values as the EPA engine map models.\214\ In any event, the relevant question is not whether NHTSA's engine map models are similar to those of the EPA (or any other analysis), but whether the models reasonably approximate engine performance to enable an assessment of technological effectiveness, and NHTSA's engine maps do just that. ICCT have not submitted information demonstrating otherwise. Notwithstanding their apparent preference for EPA's engine map models, ICCT did not provide information demonstrating that models used in the analysis are not reasonably similar to those of the EPA.\215\

\212\ Supplemental Comments of Toyota, Notice of Proposed Rulemaking: Safer Affordable Fuel-Efficient Vehicles Rule, Docket No. NHTSA-2018-0067 and Docket No. EPA-HQ-OAR-2018-0283.

\213\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 3-4.

\214\ 85 FR 24397-8 (Apr. 30, 2020).

\215\ In some instances, such as with the benchmarked Honda hybrid engine, the models used in this final rule analysis show better effectiveness improvements than EPA's engine map models. As both models--like any two different models--seek to simulate real world performance through slightly different approaches to quantitative analysis, the results will inherently vary.

Regarding engine effectiveness modeling, ICCT commented that “[t]he modeled benefit of adding cylinder deactivation (DEAC) to turbocharged and HCR engines appears to be only about 25 percent of the benefit of adding DEAC to the base engine. While DEAC added to turbo or HCR engines will have lower pumping loss reductions than when added to base naturally aspirated engines, DEAC can still be expected to provide significant pumping loss reductions while enabling the engine to operate in a more thermally efficient region of the engine map.” \216\

\216\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 2-3.

As described in numerous previous rulemakings and repeated in the NPRM, fuel-efficiency technologies (such as adding DEAC to a turbocharged engine) have complex interactions, and the effectiveness values of various technologies cannot be simply added together.\217\ Turbocharging and DEAC both work to reduce engine pumping losses, and when working together, they often provide a fuel-efficiency improvement greater then when they are working independently. But much of the improvement attributable to each technology occurs in the same regions of engine operation where one or the other technology has a dominant effect that overshadows the benefits of the other. In other words, the benefits of the technologies are overlapping in the similar regions where the engine operates. These complex interactions among technologies are captured in the engine modeling used in the analysis for this rule, as described in greater detail in TSD Chapter 3.

\217\ 88 FR 56167 (Aug. 17, 2023). This example is also given in Section II.C.2.c of this preamble.

ICCT commented that NHTSA's technology costs and effectiveness values should be updated to reflect the most recent data available, but it did not provide or reference a comprehensive alternative dataset.\218\ NHTSA's technology costs are based on the most recent, comprehensive cost data available that represent the discrete costs for each technology used in the analysis. NHTSA continually seeks out updated costs that are representative of the specific technology being modeled and makes use of learning curves to help capture how technology costs change over time. NHTSA has not received any updated comprehensive cost data for technology since publishing the NPRM and, therefore, continues to use the cost sources cited in this analysis that were used for the NPRM. The engine costs in NHTSA's analysis are the product of engine DMCs, RPE, and the learning effect, updated to a consistent dollar year. Engine DMCs are obtained from multiple sources but primarily from the 2015 NAS report.\219\ For VTG and VTGE technologies (e.g., Miller Cycle), NHTSA uses cost data from a FEV technology cost assessment performed for ICCT,\220\ which is aggregated using individual component and system costs from the 2015 NAS report. Costs from the 2015 NAS report that have referenced a Northeast States Center for a Clean Air Future 2004 report \221\ are considered, but NHTSA believes the reference material from the FEV report provides more updated cost estimates for the VTG technology.

\218\ ICCT, Docket No. NHTSA-2025-0491-5240-A1, at 3-5.

\219\ Table S.2, at pp. 7-8 of National Research Council, Cost, Effectiveness, and Deployment of Fuel Economy Technologies for Light-Duty Vehicles, The National Academies Press: Washington, DC (2015), available at: https://doi.org/10.17226/21744 (accessed: May 28, 2026) (hereinafter, “2015 NAS report”).

\220\ Isenstadt A. et al., Downsized, Boosted Gasoline Engines, Draft, International Council on Clean Transportation (2016), available at: https://theicct.org/publication/downsized-boosted-gasoline-engines-2/ (accessed: May 28, 2026).

\221\ NESCCAF, Reducing Greenhouse Gas Emissions from Light-Duty Motor Vehicles, Final Report, NESCCAF: Boston, MA (2004), available at: https://www.nesccaf.org/documents/rpt040923ghglightduty.pdf (accessed: May 28, 2026).

All engine technology costs start with a base engine cost, and then additional technology costs are based on cylinder and bank count and configuration; the DMC for each engine technology is a function of unit cost multiplied by either the number of cylinders or number of banks, based on how the technology is applied to the system. The total costs for all engine technologies in all model years across all vehicle classes can be found in the Technologies Input File. 2. Transmission Paths

Transmissions transmit torque generated by the engine from the engine to the wheels. Transmissions primarily use two mechanisms to improve fuel efficiency: (1) a wider gear range, which allows the engine to operate longer at higher efficiency speed-load points and (2) improvements in friction or shifting efficiency (e.g., improved gears, bearings, seals, pumps, and other components), which reduce parasitic losses.

NHTSA models only automatic transmissions (AT) in the light-duty analysis. The three subcategories of ATs that are modeled in this analysis include traditional ATs, dual-clutch transmissions (DCT), and continuously variable transmissions (CVT and

eCVT).\222\ The agency also includes high efficiency gearbox (HEG) technology improvements as options to the transmission technologies (designated as L2 or L3 in the analysis to indicate level of technology improvement).\223\ There has been a significant reduction in manual transmissions (MT) over the years, and they make up less than one percent of the vehicles produced in MY 2024.\224\ Due to the declining trend of MTs and their current low production volumes, NHTSA has removed MTs from this analysis and assigned vehicles using MTs as DCTs in the analysis fleet.

\222\ Note that eCVT transmissions are only coupled with hybrid electric drivetrains and are therefore not included as a standalone transmission option on the CAFE Model's technology pathways.

\223\ See 2015 NAS Report at p. 191. HEG improvements for transmissions represent incremental advancements in technology that improve efficiency, such as reduced friction seals, bearings and clutches, super finishing of gearbox parts, and improved lubrication. These advancements are all aimed at reducing frictional and other parasitic loads in transmissions to improve efficiency. NHTSA considers three levels of HEG improvements in this analysis based on the NAS 2015 recommendations and CBI data.

\224\ 2024 EPA Automotive Trends Report.

To assign transmission technologies to vehicles in the analysis fleets, NHTSA identifies which Autonomie transmission model is most like a vehicle's real-world transmission, considering the transmission's configuration, costs, and effectiveness. As with engines, data from manufacturers' CAFE reports and publicly available information are used to assign transmissions to vehicles and determine which platforms share transmissions. Transmission codes that include information about the manufacturer, drive configuration, transmission type, and number of gears are used to link shared transmissions in a manufacturer's fleet. Just as manufacturers share transmissions in multiple vehicles, the CAFE Model treats transmissions as “shared” if they share a transmission code and transmission technologies will be adopted together.

While identifying an AT's gear count is fairly easy, identifying HEG levels for ATs and CVTs is more difficult. NHTSA reviews the age of the transmission design, relative performance versus previous designs, and technologies incorporated to assign a HEG level. There are no HEG Level 3 ATs in the analysis fleet. NHTSA finds all 7-speed, all 9- speed, all 10-speed, and some 8-speed ATs to be advanced transmissions operating at HEG Level 2 equivalence. The agency assigns eight-speed ATs and CVTs newly introduced for the light-duty market in MY 2016 and later as HEG Level 2. All other ATs are assigned to their respective transmission's initial technology level (e.g., AT6, AT8, and CVT). For DCTs, the number of gears in the assignments usually match the number of gears listed by the data sources, with some exceptions (dual-clutch transmissions with seven and nine gears are assigned to DCT6 and DCT8, respectively). NHTSA assigns any vehicle in the light-duty analysis fleet with a power-split hybrid (SHEVPS) powertrain an electronic continuously variable transmission (eCVT). Finally, the limited number of MTs in the light-duty fleet are assigned as DCTs, as MTs are not modeled in Autonomie for this analysis.

Most transmission adoption features are instituted through technology path logic (i.e., decisions about how less advanced transmissions of the same type can advance to more advanced transmissions of the same type). Technology pathways are designed to prevent “branch hopping”--changes in transmission type that would correspond to significant changes in transmission architecture--for vehicles that are relatively advanced on a given pathway. For example, any automatic transmission with more than five gears cannot move to a DCT. NHTSA also prevents “branch hopping” as a proxy for stranded capital, which is discussed in more detail in Section II.C and Chapter 2.6 of the Final TSD.

The automatic transmission path precludes adoption of other transmission types once a platform progresses past an AT8. This restriction is used to avoid the significant level of stranded capital loss that could result from adopting a completely different transmission type shortly after adopting an advanced transmission, which would occur if a different transmission type has been adopted after AT8 in the rulemaking timeframe. Vehicles that did not start out with AT7L2 transmissions cannot adopt that technology in the Model. It is likely that other vehicles will not adopt the AT7L2 technology, as vehicles that have moved to more advanced ATs have overwhelmingly moved to 8-speed and 10-speed transmissions.\225\

\225\ 2024 EPA Automotive Trends Report, at p. 79, Figure 4.24.

Vehicles that do not originate with a CVT or vehicles with multispeed transmissions beyond AT8 in the analysis fleet cannot adopt CVTs. Vehicles with multispeed transmissions greater than AT8 demonstrate increased ability to operate the engine at a highly efficient speed and load. Once on the CVT path, the platform is allowed to apply only improved CVT technologies. Due to the limitations of current CVTs, discussed in Final TSD Chapter 3.2, this analysis restricts the application of CVT technology on light-duty vehicles with greater than 300 lb.-ft of engine torque. This is because of the higher torque (load) demands of those vehicles and CVT torque limitations based on durability constraints. NHTSA believes the 300 lb.-ft restriction represents an increase over current levels of torque capacity that is likely to be achieved during the rulemaking timeframe. This restriction aligns with CVT application in the analysis fleet, in that CVTs are seen only on vehicles with under 280 lb.-ft of torque.\226\ In addition, this restriction is used to avoid stranded capital. Finally, the analysis allows vehicles in the analysis fleet that have DCTs to apply an improved DCT and allows vehicles with an AT5 to consider DCTs. Drivability and durability issues with some DCTs have resulted in a low relative adoption rate over the last decade. This is also broadly consistent with manufacturers' technology choices.\227\

\226\ Market Data Input File.

\227\ 2024 EPA Automotive Trends Report, at p. 79, Figure 4.24.

Autonomie models transmissions as a sequence of mechanical torque gains. The torque and speed are multiplied and divided, respectively, by the current ratio for the selected operating condition. Furthermore, torque losses corresponding to the torque/speed operating point are subtracted from the torque input. Torque losses are defined based on a three-dimensional efficiency lookup table that has the following inputs: input shaft rotational speed, input shaft torque, and operating condition. NHTSA populates transmission template models in Autonomie with characteristics data to model specific transmissions.\228\ Characteristics data are typically tabulated data for transmission gear ratios, maps for transmission efficiency, and maps for torque converter performance, as applicable. Different transmission types require different quantities of data. The characteristics data for these models come from peer-reviewed sources, transmission and vehicle testing programs, results from simulating current and future transmission configurations, and confidential data obtained from OEMs and suppliers.\229\ HEG improvements

are modeled via improvements to the efficiency map of the transmission. As an example, the AT8 model data comes from a transmission characterization study.\230\ The AT8L2 has the same gear ratios as the AT8; however, gear efficiency map values are increased to represent application of the HEG level 2 technologies. The AT8L3 models the application of HEG level 3 technologies using the same principle, further improving the gear efficiency map over the AT8L2 improvements. There are 13 transmissions in the analysis, and each transmission is modeled in Autonomie with defined gear ratios, gear efficiencies, gear spans, and unique shift logic for the technology configuration to which the transmission is applied. These transmission maps are developed to represent the gear counts and span, shift and torque converter lockup logic, and efficiencies that can be seen in the fleet, along with upcoming technology improvements, all while balancing key attributes, such as drivability, fuel economy, and performance neutrality. This modeling is discussed in detail in Chapter 3.2 of the Final TSD and the CAFE Analysis Autonomie Documentation chapter titled “Autonomie-- Transmission Model.”

\228\ Autonomie Input and Assumptions Description Files.

\229\ Argonne National Laboratory, Downloadable Dynamometer Database, last revised: 2025, available at: https://www.anl.gov/taps/downloadable-dynamometer-database (accessed: May 28, 2026); Kim, N. et al., Advanced Automatic Transmission Model Validation Using Dynamometer Test Data, SAE 2014-01-1778, Presented at SAE 2014 World Congress & Exhibition, Apr. 8, 2014, Detroit, MI (2014), available at: https://www.sae.org/publications/technical-papers/content/2014-01-1778/ (accessed: May 28, 2026); Kim, N. et al., Development of a model of the dual clutch transmission in autonomie and validation with dynamometer test data, International Journal of Automotive Technologies, Vol. 15: pp. 263-71 (2014), available at: https://doi.org/10.1007/s12239-014-0027-5 (accessed: May 28, 2026).

\230\ CAFE Analysis Autonomie Documentation chapter titled “Autonomie--Transmission Model.”

The effectiveness values for the transmission technologies, for all technology classes, are shown in Chapter 3.2.4 of the Final TSD. Note that the effectiveness for the AT5 and eCVT technologies is not shown. The eCVT transmissions do not have standalone effectiveness values because those technologies are implemented only as part of hybrid- electric powertrains. The AT5 has no effectiveness values because it is a reference-point technology against which all other transmission technologies are compared.

NHTSA's transmission DMCs come from the 2015 NAS report and studies cited therein. The costs are taken almost directly from the 2015 NAS report adjusted to the current dollar year or for the appropriate number of gears. Chapter 3.2 of the Final TSD discusses the specific 2015 NAS report costs used to generate these transmission cost estimates, and all transmission costs across all model years can be found in the CAFE Model's Technologies Input File. NHTSA has used the 2015 NAS report transmission costs for the last several light-duty CAFE Model analyses (since re-evaluating all transmission costs for the 2020 final rule) and has not received comments or feedback on these costs. 3. Hybridization Paths

The hybridization paths each include a set of technologies that share common hybrid powertrain components, like batteries and electric motors, for certain vehicle functions that were powered solely by ICEs traditionally. While all vehicles (including conventional ICE vehicles) use batteries and electric motors in some form, some component designs and powertrain architectures contribute to greater levels of hybridization than others, allowing the vehicle to use less gasoline or other fuel.

As explained elsewhere, NHTSA endeavors to model how manufacturers could apply technology to respond to CAFE standards. Hybrid technologies can improve fuel economy, and NHTSA believes that the inputs and assumptions selected to represent hybrid technologies are reasonable to use in NHTSA's CAFE Model. NHTSA provides details of the inputs and assumptions in the Final TSD accompanying this final rule and provides more information regarding the agency's rationale and approach throughout Section II and III of this preamble.

Unlike with other technologies in the analysis, Congress placed specific limitations on how NHTSA considers the fuel economy of alternative fueled vehicles, which includes not only BEVs and FCEVs but also dual-fueled vehicles like PHEVs.\231\ For PHEVs and other hybrid technologies, which are discussed in this section, NHTSA restricts its analysis by using fuel economy values that assume “charge sustaining” (gasoline-only) operation only.\232\ The fuel economies of BEVs and FCEV technologies are excluded entirely from NHTSA's standard-setting analysis.\233\ Final TSD Chapter 2.2 contains discussion of NHTSA's consideration of PHEVs, BEVs, and FCEVs in the Final SEIS analysis.

\231\ 49 U.S.C. 32902(h)(1) and (2). In determining maximum feasible fuel economy levels, “the Secretary of Transportation--(1) may not consider the fuel economy of dedicated automobiles; [and] (2) shall consider dual fueled automobiles to be operated only on gasoline or diesel fuel.”

\232\ NHTSA has estimated two sets of technology effectiveness values using the Argonne full-vehicle simulations: one set does not include the electrification portion of PHEVs, and one set includes the combined fuel economy for both ICE operation and electric operation. Final TSD Chapter 3.3 has more information.

\233\ CAFE Model Documentation at S4.6 Technology Fuel Economy Improvements.

Among the simpler configurations with the fewest hybrid components is micro HEV technology (SS12V), which uses a 12-volt system that simply restarts the engine from a stop. Mild HEVs use a 48-volt belt integrated starter generator (BISG) system that restarts the engine from a stop and provides some regenerative braking functionality.\234\ Mild HEVs are often also capable of minimal electric assist to the engine during take-off.

\234\ See 2015 NAS Report, at p. 130 (“During braking, the kinetic energy of a conventional vehicle is converted into heat in the brakes and is thus lost. An electric motor/generator connected to the drivetrain can act as a generator and return a portion of the braking energy to the battery for reuse. This is called regenerative braking. Regenerative braking is most effective in urban driving and in the urban dynamometer driving schedule (UDDS) cycle, in which about 50 percent of the propulsion energy ends up in the brakes (NRC 2011, 18).”).

Strong hybrid-electric vehicles (SHEVs) have higher system voltages compared to mild hybrids with BISG systems and are capable of engine stop/start, regenerative braking, electric motor assist of the engine at higher speeds and power demands with the ability to provide limited all-electric propulsion. Common SHEV powertrain architectures, classified by the interconnectivity of common hybrid vehicle components, include both a series-parallel architecture by power-split device (SHEVPS) as well as a parallel architecture (SHEVP2). SHEVP2s-- though enhanced by the electric components, including just one electric motor--remains fundamentally similar to a conventional powertrain.\235\ In contrast, SHEVPS powertrains are considerably different than a conventional powertrain, as they use two electric motor/generators, which allows the use of a lower power-density engine. This results in a higher potential for fuel economy improvement compared to a SHEVP2, though the SHEVPS engine power density is lower.\236\ Put another way, “[a] disadvantage of the power-split architecture is that when towing or driving under other real-world

conditions, performance is not optimum.” \237\ In contrast, “[o]ne of the main reasons for using parallel hybrid architecture is to enable towing and meet maximum vehicle speed targets.” \238\ This is an important distinction to understand why NHTSA allows certain types of vehicles to adopt SHEVP2 powertrains and not SHEVPS powertrains.

\235\ Kapadia, J. et al., Powersplit or Parallel--Selecting the Right Hybrid Architecture, SAE International Journal of Alternative Power, Vol. 6(1): pp. 68-76 (2017), available at: https://doi.org/10.4271/2017-01-1154 (accessed: May 28, 2026) (hereinafter, “Kapadia et al. (2017)”). Parallel hybrids architecture typically adds the electrical system components to an existing conventional powertrain.

\236\ Id.

\237\ 2015 NAS Report, at p. 134.

\238\ Kapadia et al. (2017).

PHEVs utilize a combination gasoline-electric powertrain, like that of a SHEV, but have the ability to plug into the electric grid to recharge the battery, like that of a BEV; this contributes to all- electric mode capability in both blended and non-blended PHEVs.\239\ The analysis includes PHEVs with an AER of 20 and 50 miles to encompass the range of PHEV AER in the baseline fleet. Final TSD Chapter 3.3 contains more information on every hybrid technology considered in the analysis, including common acronyms and a brief description of each hybrid technology. For brevity, NHTSA refers to technologies by their acronyms in this section.

\239\ Some PHEVs operate in charge-depleting mode (i.e., “electric-only” operation--depleting the high-voltage battery's charge) before operating in charge-sustaining mode (similar to strong hybrid operation, the gasoline and electric powertrains work together), while other (blended) PHEVs switch between charge- depleting mode and charge-sustaining mode during operation.

As with previous CAFE analyses, there are a number of engine options available for SHEVs and PHEVs. These engines better represent the variety of different hybrid architectures and engine options available in the real world for SHEVs and PHEVs while still maintaining a reasonable level of analytical complexity.

ICCT commented that NHTSA did not include additional mild hybrid technology such as more capable, higher output 48-volt mild hybrid systems beyond P0 mild hybrids, such as P2, P3, or P4 configurations \240\ which offer additional benefits of electric power take-offs (i.e., launch assist or “a short power boost to the drivetrain”) \241\ or “slow-speed electric driving” \242\ on the vehicle's drive axle(s).

\240\ John German, Docket No. NHTSA-2023-0022-53274-A1, at 6-7, referenced by ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 4.

\241\ MECA, Docket No. NHTSA-2025-0491-5331-A1, at 10.

\242\ ICCT, Docket No. NHTSA-2023-0022-54064-A1, at 20, referenced by ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 6.

In response, NHTSA acknowledges that these mild hybrid configurations, such as P2 (mild) and P4, could offer improvements compared to P0 mild hybrids. Non-P0 powertrains, however, require significant changes and would require a higher capacity battery--both leading to increased powertrain cost; this is similar to what the agency observed in past rulemakings with the (P1) crank integrated starter generator system, with the non-P0 mild hybrid not being a cost- effective way for manufacturers to meet standards in the rulemaking timeframe. For this reason, NHTSA did not include additional mild hybrid technology for this final rule but may consider mild hybrid advancements in future analysis if they become more prevalent in the U.S. market.

MECA also commented on strong hybrid market penetration, noting that, “the NPRM anticipates only single digit increases in full hybrid technology penetration” \243\ and that hybrid powertrains exist in the light-duty and medium-duty vehicle space without relying on full electrification (i.e., EVs).\244\ Zero Emission Transportation Association (ZETA) similarly commented, “NHTSA's perception of consumers' acceptance of strong hybrids is likewise out of date and inconsistent with recent sales trends and evaluations in the trade press.” \245\ ICCT added, “Removing HEV application restrictions and improving fuel consumption improvement values would more accurately lead to more vehicles adopting this highly cost-effective technology.” \246\

\243\ MECA, Docket No. NHTSA-2025-0491-5331-A1, at 12.

\244\ MECA, Docket No. NHTSA-2025-0491-5331-A1, at 4-5.

\245\ ZETA, Docket No. NHTSA-2025-0491-6039-A2, at 35-36.

\246\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 4.

Regarding MECA's comment about the projection of hybrid penetration, NHTSA notes that in preamble Section IV, in both the NPRM and this final rule, NHTSA projects increases in the penetration of hybrids, which is in line with MECA's assessment that hybrids will be a contributing technology for adoption across the fleet. Regarding purported phase-in caps for SHEV technologies, no such restriction on adoption applies--the Model would allow 100 percent if the technology were cost-effective. Though strong hybridization is allowed on all vehicle types, NHTSA allows different types of strong hybrid powertrains to be applied to different types of vehicles for the reasons discussed below and believes that allowing SHEVPS and SHEVP2 powertrains to be applied subject to the base vehicle's performance and utility requirements is a reasonable approach to maintaining a performance-neutral analysis. NHTSA has explored a sensitivity case that restricts SHEV adoption and its impacts on this analysis; for more information on this and other sensitivity cases, see FRIA Chapter 9.

NHTSA also received comments requesting that NHTSA include extended-range electric technologies (EREVs) in its analysis, with both MECA and ICCT indicating that industry is retooling and has plans to launch EREV pickup trucks.\247\ Though NHTSA is aware of recent manufacturer announcements of plans to introduce EREVs,\248\ NHTSA does not currently have a method through which to estimate the fuel economy improvements for adopting the technology because true EREVs do not exist in the baseline fleet.\249\ In this analysis, as well as in previous CAFE rulemaking analyses, NHTSA has modeled a variant of EREV to represent PHEV50PS using a downsized ICE with larger battery pack and electric motors.\250\ PHEV powertrain architectures, like SHEVs, include both series-parallel power-split and parallel hybrid architectures, but do not currently include series hybrid architectures (commonly known as extended-range electric vehicles or EREVs). NHTSA will continue to monitor these technology variants for future analyses as viable pathways for mass market adoption.

\247\ NACAA, Docket No. NHTSA 2025-0491-5884, at 13; MECA, Docket No. NHTSA-2025-0491-5331-A1, at 13; AVE, Docket No. NHTSA- 2025-0490-0033-A1, at 2; ICCT, Docket No. NHTSA 2025-0491-5240-A2, at 21-24.

\248\ Ram Trucks, The Ram 1500 REV: Range-extended Electric Truck, last revised: 2026, available at: https://www.ramtrucks.com/electric/1500-rev.html (accessed: May 26, 2026).

\249\ Further, if an EREV existed that only used the engine to charge its battery and could not be operated in a charge-sustaining mode (i.e., could not be only operated on gasoline or another fuel), it would be an electric vehicle and would not be considered in NHTSA's standard-setting analysis per the limits in 32902(h).

\250\ CAFE Analysis Autonomie Documentation at p. 254.

NACAA requested that NHTSA conduct a comprehensive technology assessment that includes “advanced engine stop-start systems.” \251\ NHTSA notes that the analysis already includes micro hybrid (SS12V) and mild hybrid (BISG) technologies, and NACAA did not identify or expand on any other specific advanced stop-start system technology that it believes should be included in the analysis.

\251\ NACAA, Docket No. NHTSA-2025-0491-5884, at 13.

ICCT commented that if SHEVPS restrictions were removed for smaller pickup trucks and SUVs, this would

better reflect what is in the baseline fleet and improve the SHEVPS application within the fleet.\252\ Though anomalous examples exist in the fleet,\253\ NHTSA has not removed the SHEVPS restrictions for smaller pickup trucks and SUVs because, in the real world, performance vehicles with certain powertrain configurations cannot adopt the technologies listed above and maintain vehicle performance without a cost-effective pathway to redesigning the entire powertrain, as described below.

\252\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 4.

\253\ Such as the Ford Maverick and Ford Escape hybrids, mentioned in ICCT's comment.

As described in Final TSD Chapter 3.3, NHTSA assigns hybrid technologies to vehicles in the analysis fleet \254\ using manufacturer-submitted CAFE compliance information, publicly available technical specifications, marketing brochures, articles from reputable media outlets, and data subscriptions.\255\ Final TSD Chapter 3.3.2 shows the penetration rates of hybrid technologies in the standard- setting analysis fleets. Over half the analysis fleet has some level of hybridization, with the vast majority--over 50 percent of the fleet-- being micro hybrids. Like the other technology pathways, as the CAFE Model adopts hybrid technologies for vehicles, more advanced levels of hybrid technologies will supersede all prior levels, while certain technologies within each level are mutually exclusive. The only adoption feature applicable to micro (SS12V) and mild (BISG) hybrid technology is path logic; vehicles may adopt micro and mild hybrid technology only if the vehicle did not already have a more advanced level of hybridization.

\254\ The standard-setting analysis fleet does NOT contain BEVs or FCEVs; the Final SEIS fleet considers all technologies, including BEVs and FCEVs.

\255\ Wards Intelligence, U.S. Car and Light Truck Specifications and Prices, `22 Model Year (2022), available at: https://omdia.tech.informa.com/om132144/us-car-and-light-truck-specifications-and-prices-22-model-year (accessed: May 28, 2026).

The adoption features that NHTSA applies to strong hybrid technologies include path logic, powertrain substitution, and vehicle class restrictions. Per the technology pathways, SHEVPS, P2x, P2TRBx, and the P2HCRx technologies are considered mutually exclusive. When the Model applies one of these technologies, the others are immediately disabled from future application. However, all vehicles on the strong hybrid pathways can still advance to one or more of the plug-in technologies, when applicable in the modeling scenario (i.e., allowed in the Model).

When the Model applies any strong hybrid technology to a vehicle, the transmission technology on the vehicle is superseded; regardless of the transmission originally present, P2 hybrids adopt an advanced 8- speed automatic transmission (AT8L2), and power split hybrids adopt a continuously variable transmission via power-split device (eCVT). When the Model applies the P2 technology, the Model can consider various engine options to pair with the P2 architecture according to existing engine path constraints--taking into account relative cost effectiveness. For SHEVPS technology, the existing engine is replaced with a full-time Atkinson cycle engine.\256\ For P2s, NHTSA picks the 8-speed automatic transmission to supersede the vehicle's incoming transmission technology. This is because most P2s in the market use an 8-speed automatic transmission,\257\ therefore it is representative of the fleet now. NHTSA also believes that 8-speed transmissions are representative of the transmissions that will continue to be used in these hybrid vehicles, as NHTSA anticipates manufacturers will continue to use these “off-the-shelf” transmissions based on availability and ease of incorporation in the powertrain. The eCVT (power-split device) is the transmission for SHEVPSs and is therefore the technology NHTSA has picked to supersede the vehicle's prior transmission when adopting the SHEVPS powertrain.

\256\ This engine type is designated as Eng26 in the list of engine map models used in the analysis. Final TSD Chapter 3.1.1.2.3 provides more information.

\257\ NHTSA is aware that some Hyundai vehicles use six-speed transmissions, and some Ford vehicles use 10-speed transmissions, but NHTSA has observed that the majority of P2s use eight-speed transmissions.

SKIP logic is also used to constrain adoption of SHEVPS and PHEVx0PS technologies. These technologies are “skipped” for vehicles with engines \258\ that meet one of the following conditions: the engine belongs to an excluded manufacturer; \259\ the engine belongs to a pickup truck (i.e., the engine is on a vehicle assigned the “pickup” body style); the engine's peak HP is more than 405 HP; or the engine is on a non-pickup vehicle but is shared with a pickup. The reasons for these conditions are similar to those for the SKIP logic that NHTSA applies to HCR engine technologies, discussed in more detail in Section II.D.1.

\258\ This refers to the engine assigned to the vehicle in the 2024 analysis fleet.

\259\ Excluded manufacturers include BMW, Daimler, and Jaguar Land Rover.

It may be helpful to understand why NHTSA does not apply SKIP logic to P2s but does apply SKIP logic to SHEVPSs. Note the difference between SHEVP2 and SHEVPS architectures: P2 architectures are better for “larger vehicle applications because they can be integrated with existing conventional powertrain systems that already meet the additional attribute requirements” of large-vehicle segments.\260\ No SKIP logic applies to P2s because NHTSA believes that this type of hybrid powertrain is sufficient to meet all the performance requirements for all types of vehicles. Manufacturers have proven this with vehicles like the Ford F-150 Hybrid and Toyota Tundra Hybrid.\261\ If NHTSA were to size (in the Autonomie simulations) the SHEVPS motors and engines to achieve “not optimum” performance, the electric motors would be unrealistically large (on both a size and cost basis), and the accompanying engine also would have to be a very large displacement engine, which is not characteristic of how vehicle manufacturers apply SHEVPS to ICE vehicles in the real world. Instead, for vehicles that have particular performance requirements--which the analysis defines as vehicles with engines that belong to an excluded manufacturer, engines belonging to a pickup truck or shared with a pickup truck, or engines with a peak HP of more than 405 HP--the model allows for the adoption of SHEVP2 architectures that should be able to handle the vehicle's performance requirements.

\260\ Kapadia et al. (2017).

\261\ Buchholz, K., 2022 Toyota Tundra: V8 Out, Twin-Turbo Hybrid Takes Over, SAE International: Warrendale, VA, last revised: Sept. 22, 2021, available at: https://www.sae.org/articles/2022-toyota-tundra-v8-out-twin-turbo-hybrid-takes-sae-ma-06782 (accessed: May 28, 2026); Visnic, B., Hybridization the Highlight of Ford's All-New 2021 F-150, last revised: June 30, 2020, available at: https://www.sae.org/articles/hybridization-highlight-fords-new-2021-f-150-sae-ma-03885 (accessed: May 28, 2026).

As mentioned above, while strong hybridization is allowed on all vehicle types, NHTSA allows different types of strong hybrid powertrains to be applied to different types of vehicles for the reasons discussed above. NHTSA believes that allowing SHEVPS and SHEVP2 powertrains to be applied subject to the base vehicle's performance requirements is a reasonable approach to maintaining a performance- neutral analysis.

The engine and transmission technologies on a vehicle are superseded when PHEV technologies

are applied. For example, the Model applies an AT8L2 transmission with all PHEV20T/50T plug-in technologies, and the Model applies an eCVT transmission for all PHEV20PS/50PS and PHEV20H/50H plug-in technologies in the fleet; Final TSD Chapter 3.3 provides more details on different system combinations of hybridization. A vehicle adopting PHEV20PS/50PS receives a hybrid full-time Atkinson cycle engine, and a vehicle adopting PHEV20H/PHEV50H receives an HCR engine. For PHEV20T/50T, the vehicle receives a TURBO1 engine.

Autonomie determines the effectiveness of each hybridized powertrain type by modeling the basic components, or building blocks, for each powertrain and then combining the components modularly to determine the overall efficiency of the entire powertrain. The components, or building blocks, which contribute to the effectiveness of a hybridized powertrain in the analysis include the vehicle's battery, electric motors, power electronics, and accessory loads. Autonomie identifies components for each hybridized powertrain type and then interlinks those components to create a powertrain architecture. Autonomie then models each hybridized powertrain architecture and provides an effectiveness value for each architecture. For example, Autonomie determines a PHEV's efficiency in part by considering the efficiencies of the battery (including charging efficiency), the electric traction drive system (ETDS) (the electric machine and power electronics), and mechanical power transmission devices.\262\ Autonomie further combines the modeled hybrid components of the hybrid powertrain to include the ICE and related power for transmission components.\263\ Argonne uses data from their Advanced Mobility Technology Laboratory (AMTL) to develop Autonomie's hybrid powertrain models. The modeled powertrains are not intended to represent any specific manufacturer's architecture but act as surrogates predicting representative levels of effectiveness for each hybrid technology. NHTSA discusses the procedures for modeling each of these subsystems in detail in the Final TSD and in the CAFE Analysis Autonomie Documentation and provides a summary below.

\262\ Iliev, S. et al., Vehicle Technology Assessment, Model Development, and Validation of a 2021 Toyota RAV4 Prime, DOT HS 813 356, NHTSA: Washington, D.C. (2023), available at: https://downloads.regulations.gov/NHTSA-2023-0022-0010/attachment_6.pdf (accessed: May 28, 2026).

\263\ See the CAFE Analysis Autonomie Documentation.

NHTSA received a comment from Anderson Economic Group (AEG) stating that NHTSA omits the real-world costs of heat loss during fueling for BEVs and only considers these losses for ICE vehicle refueling.\264\ Contrary to the commenter's assertion, Autonomie modeling captures charging efficiency and considers the associated losses of the modern- day charging technology. For more information on these assumptions, see CAFE Analysis Autonomie Documentation.\265\ Further discussion on this comment with regard to the unconstrained analysis can be found in SEIS Appendix D.

\264\ AEG, Docket No. NHTSA-2025-0491-5981-A1, at 6-7.

\265\ CAFE Analysis Autonomie Documentation at p. 226.

NHTSA received comments about the hybrid effectiveness values used in the analysis. ZETA commented that DOE/EPA's data on hybrid effectiveness “is consistent with NHTSA's effectiveness estimates.” \266\ ICCT commented extensively on hybrid effectiveness values. ICCT noted that though NHTSA's SHEVPS effectiveness is acceptable, ICCT recommends including “at least one future (power-split) hybrid system improvement beyond that which is already modeled.” \267\ ICCT commented further that NHTSA's SHEVP2 effectiveness is too low and noted that NHTSA's modeling constrains SHEVP2 motor power to below 45 kW \268\ for future model years, citing ICCT's 2025 HEV study, which it says shows specifications for primary drive motors and that the SHEVP2 motor rated between 34kW and 42 kW in 2024 and between 135 kW and 175 kW in 2030 and beyond.\269\

\266\ ZETA, Docket No. NHTSA-2025-0491-6039-A2, at 36-37.

\267\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 4.

\268\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 17.

\269\ ICCT, Docket No. NHTSA-2025-0491-5240-A9, at 17, Table 4.

In response to the comments on a purported need to align NHTSA's effectiveness values with industry and the perspective on future hybrid component specifications, NHTSA notes that NHTSA generally models effectiveness values only with components that exist in the current fleet--not concept powertrain components that might exist in the market years from now that have drastically different power ratings compared to the present day. There are ways to optimize electric motors, engines, and battery packs and integrate them in the powertrain, but in the analysis this is limited to discrete packages that NHTSA has modeled with Autonomie, as discussed further in Section II.C.2.c. For example, ICCT projected hypothetical electric motor and hybrid battery sizes in MY 2030 do not appear to even be based on any OEM announcements, component teardowns, or concept vehicles within the rulemaking timeframe. Without more concrete data about these future developments, NHTSA cannot accurately simulate the effectiveness and costs associated with their adoption. In contrast, NHTSA modeled ten SHEVP2 variants from basic to advanced engines that cover the span of hybrid technologies present in the current fleet.\270\ Though NHTSA has not made any changes to the analysis in response to ICCT's comment, NHTSA will continue to monitor for these improvements and will continue to update its analyses.

\270\ Autonomie models SHEV P2 with Eng1, Eng5b, Eng12, Eng13, Eng18, Eng32, Eng33, Eng36, Eng37, and Eng41.

The fundamental components of a hybrid powertrain's propulsion system--the electric motor and inverter--ultimately determine the vehicle's performance and efficiency. For this analysis, Autonomie employs a set of electric motor efficiency maps created by Oak Ridge National Laboratory (ORNL), one for a traction motor and an inverter, the other for a motor/generator and inverter.\271\ The electric motor efficiency maps, created from production vehicles like the 2007 Toyota Camry hybrid and the 2011 Hyundai Sonata hybrid, represent electric motor efficiency as a function of torque and motor rotations per minute (RPM). These efficiency maps provide nominal and maximum speeds, as well as a maximum torque curve. Argonne uses the maps to determine the efficiency characteristics of the motors, which include some of the losses due to power transfer through the electric machine.\272\ Specifically, Argonne scales the efficiency maps, specific to powertrain type, to have total system peak efficiencies ranging from 96 to 98

percent \273\--such that their peak efficiency value corresponds to the latest state-of-the-art technologies, as opposed to retaining dated system efficiencies (90 to 93 percent).\274\

\271\ Burress, T. et al., Evaluation of the 2007 Toyota Camry Hybrid Synergy Drive System, ORNL: Washington, DC (2008), available at: https://doi.org/10.2172/928684 (accessed: May 28, 2026) (hereinafter, “Burress et al. (2008)”); Olszewski, M., Annual Progress Report for the Power Electronics and Electric Machinery Program, ORNL/TM-2011/263, Oak Ridge National Laboratory: Washington, DC (2011), available at: https://info.ornl.gov/sites/publications/files/Pub31483.pdf (accessed: May 28, 2026) (hereinafter, “Olszewski (2011)”).

\272\ CAFE Analysis Autonomie Documentation chapter titled “Vehicle and Component Assumptions--Electric Machines--Electric Machine Efficiency Maps.”

\273\ CAFE Analysis Autonomie Documentation chapter titled “Vehicle and Component Assumptions--Electric Machines--Electric Machine Peak Efficiency Scaling.”

\274\ Burress et al. (2008); Olszewski (2011).

Beyond the powertrain components, Autonomie also considers electric accessory devices that consume energy in 2-cycle testing and how they affect overall vehicle effectiveness, such as radiator fans, engine control units, transmission control units, cooling systems, and safety systems. In real-world driving and operation, the electrical accessory load on the powertrain varies depending on how the driver uses certain features and the condition in which the vehicle is operating, such as night driving or hot weather driving. However, for regulatory test cycles related to fuel economy, the electrical load is repeatable because the fuel economy regulations control these factors. Accessory loads during test cycles vary by powertrain type and vehicle technology class, since distinctly different powertrain components and vehicle masses consume different amounts of energy.

The analysis fleets consist of different vehicle types with varying accessory electrical power demand. For instance, vehicles with different motor and battery sizes require different sizes of electric cooling pumps and fans to manage component temperatures optimally. Autonomie has built-in models that can simulate these varying subsystem electrical loads. However, for this analysis, NHTSA uses a fixed (by vehicle technology class and powertrain type), constant power draw to represent the effect of these accessory loads on the powertrain on the 2-cycle test. NHTSA expects that fixed accessory load values will, on average, have similar impacts on effectiveness as found on actual manufacturers' systems. This process is in line with the past analyses.275 276 NHTSA aggregates electrical accessory load modeling assumptions for the different powertrain types (hybridized and conventional) and technology classes from data from the 2016 Draft Technical Assessment Report (TAR), and the 2016 EPA Proposed Determination,\277\ data from manufacturers,\278\ research and development data from DOE's Vehicle Technologies Office,279 280 281 and DOT-sponsored vehicle benchmarking studies completed by Argonne's AMTL.

\275\ Technical Assessment Report at Chapter 5 (2016).

\276\ EPA Proposed Determination TSD at pp. 2-270 (2016).

\277\ Id.

\278\ Alliance of Automobile Manufacturers (now Auto Innovators) Comments on Draft TAR, at p. 30.

\279\ DOE, Electric Drive Systems Research and Development, last revised: 2025, available at: https://www.energy.gov/eere/vehicles/electric-drive-systems-research-and-development (accessed: May 28, 2026).

\280\ Argonne National Laboratory, Advanced Mobility Technology Laboratory (AMTL), last revised: 2025, available at: https://www.anl.gov/taps/advanced-mobility-technology-laboratory (accessed: May 28, 2026).

\281\ DOE's lab years are 10 years ahead of manufacturers' potential production intent (e.g., 2020 lab year is MY 2030).

Certain technologies' effectiveness for reducing fuel consumption requires optimization through the appropriate sizing of the powertrain. Autonomie uses sizing control algorithms based on data collected from vehicle benchmarking,\282\ and the modeled hybrid components are sized based on performance neutrality considerations. This analysis iteratively minimizes the size of the powertrain components to maximize efficiency while enabling the vehicle to meet multiple performance criteria. The Autonomie simulations use a series of resizing algorithms that contain “loops,” such as the acceleration performance loop (0-60 mph), which automatically adjusts the size of certain powertrain components until a criterion, like the 0-60 mph acceleration time, is met. As the algorithms examine different performance or operational criteria that must be met, no single criterion can degrade; once a resizing algorithm completes, all criteria will be met, and some may be exceeded as a necessary consequence of meeting others.

\282\ CAFE Analysis Autonomie Documentation chapter titled “Vehicle Sizing Process--Vehicle Powertrain Sizing Algorithms-- Light-Duty Vehicles--Conventional Vehicle Sizings Algorithm.”

Autonomie applies different powertrain sizing algorithms depending on the type of vehicle considered because different types of vehicles not only contain specific, optimized components, but they must also operate in varying driving modes. While the conventional powertrain sizing algorithm must consider only the power of the engine, the more complex algorithm for hybridized powertrains must simultaneously consider multiple factors, which could include the engine power, electric machine power, battery power, and battery capacity. Also, while the resizing algorithm for all vehicles must satisfy the same performance criteria, the algorithm for some electric powertrains must also allow those hybridized vehicles to operate in certain driving cycles, like the US06 cycle (a high acceleration aggressive driving schedule), without assistance of the combustion engine and ensure the electric motor/generator and battery can handle the vehicle's regenerative braking power, all-electric mode operation, and intended range of travel.

To establish the effectiveness of the technology packages, Autonomie simulates the vehicles' performance on compliance test cycles.\283\ For vehicles with conventional powertrains and micro hybrid powertrains, Autonomie simulates the vehicles using the 2-cycle test procedures and guidelines.\284\ For mild HEVs and strong HEVs, Autonomie simulates the same 2-cycle test, with the addition of repeating the drive cycles until the final state-of-charge (SOC) is approximately the same as the initial SOC, a process described in SAE J1711; SAE J1711 also provides test cycle guidance for testing specific to PHEVs.\285\ PHEVs have a range of modeled effectiveness during “standard-setting” CAFE Model runs, in which the PHEV operates under a “charge sustaining” (gasoline-only) mode--similar to how SHEVs function.

\283\ EPA, How Vehicles are Tested, last revised: 2025, available at: https://www.fueleconomy.gov/feg/how_tested.shtml (accessed: May 28, 2026); Good, D., EPA Test Procedures for Electric Vehicles and Plug-in Hybrids, Draft Summary, EPA: Washington, DC (2017), available at: https://www.fueleconomy.gov/feg/pdfs/EPA%20test%20procedure%20for%20EVs-PHEVs-11-14-2017.pdf (accessed: May 28, 2026); CAFE Analysis Autonomie Documentation, chapter titled “Test Procedure and Energy Consumption Calculations.”

\284\ 40 CFR part 600.

\285\ PHEV testing is broken into several phases based on SAE J1711: charge-sustaining on the city and HWFET cycle, and charge- depleting on the city and HWFET cycles.

The Alliance for Vehicle Efficiency (AVE) submitted a comment emphasizing the need for harmonized test procedures across Federal regulatory programs when evaluating hybrid and electrified powertrains. AVE further stated that differences in test procedures result in inconsistent performance valuation of hybrid vehicle technologies.\286\

\286\ AVE, Docket No. NHTSA-2025-0490-0033-A1, at 7.

In response to AVE's comment, NHTSA emphasizes that its analysis is based on the 2-cycle testing prescribed by EPA's regulation for calculating CAFE compliance, in accordance with EPCA.\287\ We discuss the use of 2-cycle

testing for compliance in Section VI of this preamble.

\287\ See 49 U.S.C. 32904(c) (“Testing and calculation procedures . . . . [T]he Administrator shall use the same procedures for passenger automobiles the Administrator used for model year 1975 (weighted 55 percent urban cycle and 45 percent highway cycle), or procedures that give comparable results.”).

Chapters 2.4 and 3.3 of the Final TSD and the CAFE Analysis Autonomie Documentation chapter titled “Test Procedure and Energy Consumption Calculations” discuss the components and test cycles used to model each hybrid powertrain type; please refer to those chapters for more technical details on each of the modeled technologies discussed in this section.

The range of effectiveness for the hybrid technologies used in this analysis is a result of the interactions between the components listed above and how the modeled vehicle operates on its respective test cycle. This range of values results in some modeled effectiveness values being close to real-world measured values and some modeled values departing from measured values, depending on the level of similarity between the modeled hardware configuration and the real- world hardware and software configurations. The range of effectiveness values for the hybrid technologies applied in the fleet is shown in Final TSD Figure 3-23 and Figure 3-24.

Some advanced engine technologies indicate low effectiveness values when paired with hybrid architectures. The low effectiveness results from the application of advanced engines to existing P2 architectures. This effect is expected and illustrates the importance of using the full-vehicle modeling to capture interactions between technologies and to capture instances of both complementary technologies and non- complementary technologies. In developing its hybrid powertrains, NHTSA considers the engine maps, engine technologies, electric motor power, and battery pack size. The hybrid powertrains are calibrated to operate in their respective hybrid architecture most effectively and to allow the electric machine to provide propulsion or assistance in regions of the engine map that are less efficient. As the Model sizes the powertrain for any given application, it considers all these parameters as well as performance neutrality metrics to provide the most efficient solution. In this instance, the P2 powertrain improves fuel economy, in part, by allowing the engine to spend more time operating at efficient engine speed and load conditions. This reduces the advantage of adding advanced engine technologies, which also improve fuel economy by broadening the range of speed and load conditions for the engine to operate at high efficiency. This redundancy in fuel-saving mechanisms results in a lower effectiveness when the technologies are added together.

The technology effectiveness values are developed specifically to support analyses for a rulemaking timeframe. For example, the hybrid Atkinson engine peak thermal efficiency was updated based on 2017 Toyota Prius engine data.\288\ As mentioned above, Argonne scales the efficiency maps, specific to powertrain type, to have total system peak efficiencies ranging from 96 to 98 percent \289\--such that their peak efficiency value corresponds to the latest state-of-the-art technologies, as opposed to retaining dated system efficiencies (90 to 93 percent).\290\ The 2016 maps scaled to peak efficiency are equivalent to (if not exceed) efficiencies seen in vehicles in the current fleet and in the future. Though the base references for these technologies are from a few years ago, NHTSA has worked with Argonne to update individual inputs to reflect the latest improvements. Accordingly, NHTSA has made no changes to the electric machine efficiency maps for this final rule analysis.

\288\ Atkinson Engine Peak Efficiency is based on 2017 Prius peak efficiency scaled up to 41 percent. CAFE Analysis Autonomie Documentation at p. 138. See ANL--All Assumptions_Summary_NPRM_022021.xlsx, ANL--Summary of Main Component Performance Assumptions_NPRM_022021.xlsx, Argonne Autonomie Model Documentation_NPRM.pdf and ANL--Data Dictionary_NPRM_022021.XLSX, which can be found in the rulemaking docket (NHTSA-2023-0022) by filtering for Supporting & Related Material.

\289\ See CAFE Analysis Autonomie Documentation, chapter titled “Electric Machine Peak Efficiency Scaling.”

\290\ Burress et al. (2008); Olszewski (2011).

When the CAFE Model turns a vehicle powered by an ICE into a hybridized vehicle, it must remove the parts and costs associated with the ICE (and, potentially, the transmission depending on the hybridization level and powertrain type) and add the costs of a battery pack and other non-battery hybridization components, such as the electric motor and power inverter. To estimate battery pack costs for this analysis, NHTSA needs an estimate of how much battery packs cost (i.e., a “base year” cost) and estimates of how that cost could reduce over time (i.e., the “learning effect”). The general concept of learning effects is discussed in detail in Section II.C and in Chapter 2 of the Final TSD, while the specific learning effect NHTSA applied to battery pack costs in this analysis is discussed below. NHTSA estimates base year battery pack costs for most hybrid technologies using BatPaC, which is an Argonne model designed to calculate the cost of hybrid battery packs.

Traditionally, a user would use BatPaC to cost a battery pack for a single vehicle, and the user would vary factors such as battery cell chemistry, battery power and energy, battery pack interconnectivity configurations, battery pack production volumes, charging constraints, or combinations of these factors, to name a few, to see how those factors would increase or decrease the cost of the battery pack. However, several hundreds of thousands of simulated vehicles in the analysis have hybridized powertrains, meaning that NHTSA would have to run individual BatPaC simulations for each full-vehicle simulation that requires a battery pack. This would have been computationally intensive and impractical. Instead, Argonne staff builds “lookup tables” with BatPaC that provide battery pack manufacturing costs, battery pack weights, and battery pack cell capacities for vehicles with varying power requirements modeled in these large-scale simulation runs.

Just like with other vehicle technologies, the specifications of different vehicle manufacturers' battery packs are extremely diverse. NHTSA, therefore, endeavored to develop battery pack costs that reasonably encompass the cost of battery packs for vehicles in each technology class.

In conjunction with the agency's partners at Argonne working on the CAFE analysis Autonomie modeling, NHTSA references assessment and outlook reports,\291\ vehicle teardown reports,\292\ and stakeholder discussions \293\ to determine common

battery pack chemistries for each modeled hybrid technology. The CAFE Analysis Autonomie Documentation chapter titled “Battery Performance and Cost Model--BatPaC Examples From Existing Vehicles in the Market” includes more detail about the reports referenced for this analysis.\294\ For mild hybrids, NHTSA uses the lithium iron phosphate (LFP)-G \295\ chemistry because power and energy requirements for mild hybrids are very low, the charge and discharge cycles (or need for increased battery cycle life) are high, and the battery raw materials are much less expensive than a nickel manganese cobalt (NMC)-based cell chemistry. NHTSA uses NMC622-G \296\ for all other hybrid vehicle technology base (MY 2022) battery pack cost calculations. NHTSA believes that, based on available data,\297\ NMC622 is more representative for the MY 2022 base year battery costs than LFP, and any additional cost reductions from manufacturers switching to LFP chemistry-based battery packs in years beyond 2022 are accounted for in the battery cost learning effects. The learning effects estimate potential cost savings for future battery advancements (a learning rate applied to the battery pack DMC); this final rule includes a dynamic NMC/LFP cathode mix over each future model year (for PHEVs). The battery chemistry that NHTSA uses is intended to represent reasonably what is used in the MY 2022 U.S. fleet, which is the DMC base year for the BatPaC calculations.\298\

\291\ Rho Motion, EV Battery subscriptions, available at: https://rhomotion.com/ (accessed: May 28, 2026); BNEF, Electric Vehicle Outlook 4Q 2023: Growth Ahead, last revised: Jan. 4, 2024, available at: https://about.bnef.com/insights/clean-transport/electrified-transport-market-outlook-4q-2023-growth-ahead/ (accessed: May 28, 2026); Benchmark Mineral Intelligence, Cathode, Anode, and Gigafactories subscriptions, available at: https://benchmarkminerals.com/ (accessed: May 28, 2026); International Energy Agency, Global EV Outlook 2022: Securing Supplies For an Electric Future, International Energy Agency: Paris, France (2022), available at: https://iea.blob.core.windows.net/assets/ad8fb04c-4f75-42fc-973a-6e54c8a4449a/GlobalElectricVehicleOutlook2022.pdf (accessed: May 28, 2026).

\292\ Hummel, P. et al., UBS Evidence Lab Electric Car Teardown--Disruption Ahead?, UBS: Zurich, Switzerland (2017), available at: https://neo.ubs.com/shared/d1ZTxnvF2k (accessed: May 28, 2026); A2Mac1, Automotive Benchmarking (proprietary data), available at: https://portal.a2mac1.com/ (accessed: May 28, 2026).

\293\ See Docket Submission of Ex Parte Meetings Prior to Publication of the Corporate Average Fuel Economy Standards for Passenger Cars and Light Trucks for Model Years 2027-2032 and Fuel Efficiency Standards for Heavy-Duty Pickup Trucks and Vans for Model Years 2030-2035 Notice of Proposed Rulemaking memorandum, which can be found in the rulemaking docket (NHTSA-2023-0022) by filtering for Supporting & Related Material.

\294\ CAFE Analysis Autonomie Documentation chapter titled “Battery Performance and Cost Model--BatPac Examples From Existing Vehicles in the Market.”

\295\ Lithium iron phosphate (LiFePO4) cathode and graphite anode.

\296\ Lithium nickel manganese cobalt oxide (LiNiMnCoO2) cathode and graphite anode.

\297\ Rho Motion, EV Battery subscriptions, available at: https://rhomotion.com/ (accessed: May 28, 2026); International Energy Agency, Global EV Outlook 2023: Catching Up with Climate Ambitions, International Energy Agency: Paris, France (2023), available at https://www.iea.org/reports/global-ev-outlook-2023 (accessed: May 28, 2026).

\298\ For this analysis, 2021$ costs have been updated to 2024$; this is not reflected directly in the base Battery Cost csv file, however, as this conversion was performed external to the file itself.

The Attorneys General \299\ stated that NHTSA's analysis assumed that all strong hybrids rely on NMC battery chemistries and countered that cheaper LFP batteries have quickly become the dominant global battery chemistry since 2022, meaning NHTSA's chemistry assumption overstates actual market costs.

\299\ Attorneys General, Docket No. NHTSA-2025-0491-6064-A2, at 62.

NHTSA does not apply LFP batteries to SHEVs in its analysis because no strong hybrid in the American automotive market uses the LFP battery chemistry for its battery packs. As for battery chemistry assumptions in the analysis, NHTSA accounts for the increasing prevalence of LFP displacing NMC cathodes in the U.S. market, applies a “composite correlation equation” for PHEVs,\300\ and models a dynamic, shifting mix of NMC and LFP chemistries through MY 2035.

\300\ This is also the case for battery-electric vehicles (BEVs) in the EIS analysis.

The Attorneys General \301\ also commented that NHTSA used 2022 as the base year to estimate the manufacturing cost of hybrid battery packs. The Attorneys General point out that material costs and the relative intensity of certain mineral uses dropped significantly between 2022 and 2024, meaning the 2022 baseline overestimates DMCs.

\301\ Attorneys General, Docket No. NHTSA-2025-0491-6064-A2, at 62.

While the base DMC is anchored to MY 2022, NHTSA does not assume that battery costs will remain constant at 2022 levels. To reflect how battery costs will decrease over the analytical timeframe, NHTSA applies specific “learning rates” to the BatPaC-generated DMCs, developed by Argonne's BatPaC team.\302\ Final TSD Chapter 3.3 explicitly notes that “NHTSA accounts for the potential cost savings for future battery cell chemistries using a learning rate applied to the battery pack DMC.” The anticipated drops in material costs and advancements in manufacturing are built into the future projections via these learning curves, and the analysis does not require a shift in base year. NHTSA did not update the battery cost base year for the final rule analysis.

\302\ Argonne National Laboratory, Cost Analysis and Projections for U.S.-Manufactured Automotive Lithium-Ion Batteries, ANL/CSE-24/ 1, Argonne National Laboratory: Lemont, IL (2024), available at: https://publications.anl.gov/anlpubs/2024/01/187177.pdf (accessed: May 28, 2026) (hereinafter, “ANL/CSE-24/1”).

NHTSA also looks at vehicle sales volumes for MY 2022 to determine a reasonable base production volume assumption.\303\ In practice, a single battery plant can produce packs using different cell chemistries with different power and energy specifications, as well as battery pack constructions with varying battery pack designs--different cell interconnectivities (to alter overall pack power end energy) and thermal management strategies--for the same base chemistry. However, in BatPaC, a battery plant is assumed to manufacture and assemble a specific battery pack design, and all cost estimates are based on one single battery plant manufacturing only that specific battery pack. For example, if a manufacturer has more than one PHEV in its vehicle lineup and each uses a specific battery pack design, a BatPaC user would include manufacturing volume assumptions for each design separately to represent each plant producing each specific battery pack. NHTSA has examined battery pack designs for vehicles sold in MY 2022 to determine a reasonable manufacturing plant production volume assumption. NHTSA considers each assembly line designed for a specific battery pack and for a specific PHEV as an individual battery plant. Since battery technologies and production are still evolving, it is likely to be some time before battery cells can be treated as commodities where the specific numbers of cells are used for varying battery pack applications and all other metrics remain the same.

\303\ See Chapter 2.2.1.1 of the Final TSD for more information on data NHTSA uses for sales volumes.

Similar to previous rulemakings, NHTSA uses sales as a starting point to analyze potential base modeled battery manufacturing plant production volume assumptions. Since actual production data for specific battery manufacturing plants are extremely hard to obtain and the battery cell manufacturer is not always the battery pack manufacturer,\304\ NHTSA calculates an average production volume per manufacturer metric to approximate hybrid vehicle production volumes for this analysis. This metric is calculated by taking an average of all of one hybrid vehicle type (for example, all PHEVs) battery energies reported in a vehicle manufacturer's pre-MY 2022 reports \305\ and dividing by the averaged sales-weighted energy per-vehicle; the resulting volume is then rounded to the nearest 5,000. Manufacturers are not required to report gross battery pack sizes for the pre-model year or mid-model year compliance reports, so NHTSA estimates pack size for each vehicle based on proprietary data and publicly available data, like a

manufacturer's published or announced specifications. This process is repeated for all hybrid vehicle technologies. NHTSA believes this provides a reasonable base year plant production volume--especially in the absence of actual production data--since the compliance report data from manufacturers already includes accurate related data, such as vehicle model and estimated sales information metrics.\306\ The final battery manufacturing plant production volume assumptions for different hybrid technologies are as follows: mild hybrid and strong hybrids are manufactured assuming 200,000 packs and PHEVs are manufactured assuming 20,000 packs.

\304\ Zhou, Y. et al., Lithium-Ion Battery Supply Chain for E- Drive Vehicles in the United States: 2010-2020, ANL/ESD-21/3, Argonne National Laboratory: Argonne, IL (2021), available at: https://publications.anl.gov/anlpubs/2021/04/167369.pdf (accessed: May 28, 2026); Gohlke, D. et al., Quantification of Commercially Planned Battery Component Supply in North America Through 2035, Final Report, ANL-24/14, Argonne National Laboratory: Alexandria, VA (2024), available at: https://publications.anl.gov/anlpubs/2024/03/187735.pdf (accessed: May 28, 2026).

\305\ 49 CFR 537.7.

\306\ NHTSA uses publicly available range and pack size information and linked the information to vehicle models.

As mentioned above, the BatPaC Lookup Tables provide $/kWh battery pack costs based on vehicle power and energy requirements. As the total cost of a battery pack increases the higher the power/energy requirements, the cost per kWh decreases. This represents the cost of hardware that is needed in all battery packs but is deferred across more kW/kWh in larger packs, which reduces the per kW/kWh cost. Table 3-78 in Final TSD Chapter 3.3.5 shows an example of the BatPaC Lookup Tables for SHEVPS technology classes.

Note that the values in the table discussed above should not be considered the total battery $/kWh costs that are used for vehicles in the analysis in future model years. As detailed below, battery costs are also projected to decrease over time as manufacturers improve production processes, shift battery chemistries, and make other technological advancements. In addition, select modeled tax credits further reduce the estimated costs; additional discussion of those tax credits is located throughout this preamble, Final TSD Chapter 2.3, FRIA Chapters 8 and 9, and preamble Section II.C.2.e.

The CAFE Analysis Autonomie Documentation details other specific assumptions that Argonne used to simulate battery packs and their associated base year costs for the full-vehicle simulation modeling, including updates to the battery management unit costs and the range of power and energy requirements used to bound the lookup tables.\307\ CAFE Analysis Autonomie Documentation and Chapter 3.3 of the Final TSD provide further information about how NHTSA used BatPaC to estimate base year battery costs. The full range of BatPaC-generated battery DMCs is in the file ANL--Summary of Main Component Performance Assumptions_NPRM_2206.\308\ Note again that these charts represent the DMC using a dollar per kW/kWh metric; absolute battery costs used in the analysis by technology key can be found in the CAFE Model Battery Costs File.

\307\ CAFE Analysis Autonomie Documentation chapter titled “Battery Performance and Cost Model--Use of BatPac in Autonomie for FRM runs.”

\308\ The DMCs in the Argonne file are in 2021$ (from the 2024 final rule).

The DOE and Argonne developed battery cost correlation equations from BatPaC for use in the 2024 CAFE final rule analysis--cost equations that continue to be used in this analysis.\309\ These cost equations--developed for use through MY 2035--are tailored for different vehicle segments,\310\ different levels of hybridization,\311\ and anticipated plant production volumes.\312\ These equations represent cost improvements achieved from advanced manufacturing, pack design, and cell design with current and anticipated future battery chemistries,\313\ design parameters, forecasted market prices, and vehicle technology penetration. Argonne's Cost Analysis and Projections for U.S.-Manufactured Automotive Lithium- ion Batteries report contains a detailed discussion of the inputs and assumptions used to generate these cost equations.\314\

\309\ ANL/CSE-24/1.

\310\ The vehicle classes considered in this project include compact cars, mid-size cars, mid-size SUVs, and pickup trucks.

\311\ The levels of hybridization considered in this project include light-duty micro HEVs, mild HEVs, strong HEVs, and PHEVs.

\312\ Production volumes were determined for each vehicle class and type for each model year. See ANL/CSE-24/1 at Equation 1 and Table 13.

\313\ Battery cathode chemistries considered in this project include nickel-based materials (NMC622, NMC811, NMC95, and LMNO) as well as lower cost LFP cathodes; varying percentages of silicon content (5%, 15%, and 35%) within a graphite anode were considered, as well.

\314\ ANL/CSE-24/1.

The Attorneys General \315\ argue that NHTSA's evaluation of economic feasibility relies on cost models that fail to account for dramatically falling battery costs, thereby artificially inflating the cost of strong hybrid vehicles. They assert that NHTSA's inaccurate assumptions affect the CAFE Model results by inflating projected regulatory costs and distorting the projected makeup of the fleet. In addition to comments on battery base year and chemistry assumptions that were discussed above, the Attorneys General argued that NHTSA's attempt to account for battery technology advancements by applying a 1.5-percent learning rate is inadequate to keep pace with current battery technology advancements. They cite evidence suggesting that EV battery manufacturing has recently demonstrated a 7.5-percent learning rate, which is five times higher than the rate NHTSA applied.

\315\ Attorneys General, Docket No. NHTSA-2025-0491-6064-A2, at 62-63.

In response to these comments, NHTSA emphasizes that the 1.5- percent learning rate applies to later years--well beyond those years subject to this regulation--when battery technology matures, not in the near term. The Final TSD clarifies that NHTSA specifically uses the 1.5-percent learning rate for battery packs from MY 2036 and beyond, which is appropriate because the agency expects battery technology to be largely mature by that time. The near-term battery learning rates differ across each vehicle class, specific hybrid powertrain technology, and across each year (through MY 2035), with year-over-year improvements spanning from above four percent year-over-year to below one percent year-over-year.\316\

\316\ ANL/CSE-24/1.

Though batteries and relative battery components are the biggest cost drivers of hybridization, non-battery hybridization components, such as electric motors, power electronics, and wiring harnesses, also add to the total cost required to electrify a vehicle. Different levels of hybrid vehicles have variants of non-battery hybridization components and configurations to accommodate different vehicle classes and applications with respective designs. For instance, some SHEVs may be engineered with only one electric motor, while other SHEVs may be engineered with two or even three electric motors within their powertrains to provide AWD functionality. In addition, some hybrid vehicle types still include conventional powertrain components, like an ICE and transmission.

For all hybrid vehicle powertrain types, NHTSA groups non-battery hybridization components into four major categories: electric motors, power electronics (generally including the DC-DC converter, inverter, and power distribution module), charging components (charger, charging cable, and high-voltage cables), and thermal management systems. NHTSA further groups the components into those composing the ETDS, and all other components. Though each manufacturer's ETDS and power electronics vary between the same hybrid vehicle types and between different hybrid vehicle types, NHTSA

considers the ETDS for this analysis to be composed of the electric motor and inverter, power electronics, and thermal system.

When researching costs for different non-battery hybridization components, NHTSA finds that different reports vary in components considered and cost breakdown. This is not surprising, as vehicle manufacturers use different non-battery hybridization components in different vehicle systems, or even in the same vehicle type, depending on the application. For each of the component categories discussed above, NHTSA examines cost teardown studies and uses the best available cost estimate for each component from these different reports. These reports capture components in most manufacturers' systems but not all, and NHTSA believes that this is a reasonable approach for this analysis, given the non-standardization of hybrid powertrain designs and subsequent component specifications. Other sources NHTSA uses for non-battery hybridization component costs include an EPA-sponsored FEV teardown of a 2013 Chevrolet Malibu ECO with eAssist for some BISG component costs,\317\ which were validated against a 2019 Dodge Ram eTorque system's publicly available retail price,\318\ and the 2015 NAS report.\319\ Broadly, the total BISG system cost, including the battery, fairly matches these other cost estimates. Component cost information is discussed further in Final TSD Chapter 3.3.5.

\317\ FEV, Inc., Light Duty Vehicle Technology Cost Analysis: 2013 Chevrolet Malibu ECO With eAssist BAS Technology Study, FEV P311264, Contract No. EP-C-12-014, WA 1-9, EPA: Washington, DC (2014), available at: https://www.regulations.gov/document/EPA-HQ-OAR-2015-0827-0342 (accessed: May 28, 2026).

\318\ Colwell, K., The 2019 Ram 1500 eTorque Brings Some Hybrid Tech, if Little Performance Gain, to Pickups, Car and Driver, last revised: Mar. 14, 2019, available at: https://www.caranddriver.com/reviews/a22815325/2019-ram-1500-etorque-hybrid-pickup-drive (accessed: May 28, 2026).

\319\ 2015 NAS Report, at p. 305.

NHTSA received a number of comments about hybrid powertrain costs. Stellantis argued that NHTSA's cost assumptions were too low, while other commenters argue that NHTSA's cost estimates are too high. Stellantis \320\ argued that NHTSA's powertrain cost assumptions for SHEV technology are significantly underestimated regarding what is economically and commercially feasible, especially for large vehicles. Stellantis urges NHTSA to adjust SHEV cost estimates upward specifically for large vehicles and not assume a similar cost across small cars, crossovers, SUVs, and full-size pickup trucks--noting that correcting these cost assumptions will alter the technology penetration rates assumed in the modeling.

\320\ Stellantis, Docket No. NHTSA-2025-0491-5968-A1, at 10.

In contrast to the comment from Stellantis, NHTSA received several comments stating that the agency overestimates hybrid costs compared to current markets.321 322 These commenters argue that NHTSA's cost estimates for strong hybrid electric vehicles (SHEVs) are excessively high and outdated and contradict current market prices.323 324 ICCT asserted that the agency overestimates cost by comparing price premiums for hybrids and their non-hybrid counterparts.\325\ ZETA commented similarly, referencing a paper asserting the differences in price premiums between the current market prices and NHTSA's analysis, and incorrectly concluding that NHTSA inflated hybrid powertrain cost estimates.\326\

\321\ ICCT, Docket No. NHTSA-2025-0491-5240-A1, at 3-4.

\322\ ZETA, Docket No. NHTSA-2025-0491-6039-A1, at 15 and -A2, at 34-35.

\323\ ICCT, Docket No. NHTSA-2025-0491-5240-A1, at 3-4.

\324\ ZETA, Docket No. NHTSA-2025-0491-6039-A1, at 15 and -A2, at 34-35.

\325\ ICCT, Docket No. NHTSA-2025-0491-5240-A1, at 5.

\326\ ZETA, Docket No. NHTSA-2025-0491-6039-A2, at 34-37.

As with previous rulemakings, NHTSA continues to reject the use of vehicle MSRPs to estimate component costs, as this oversimplifies the technology cost walk that the Model performs between powertrains in a given model year.\327\ Final TSD Chapter 2.4 describes NHTSA's technology cost estimation methodology and clarifies that the agency estimates technology costs from the ground-up and scales for factors such as electric motor power and does reference vehicle prices to estimate technology costs. The Vehicle Report Output File and read related CAFE Model Documentation describe the actual technology costs used in the model. NHTSA is not making any changes to hybrid vehicle costs from the NPRM analysis for this final rule.

\327\ A technology cost walk is the step-by-step cost transaction the model performs when a vehicle adopts technology, allowing cost changes to be attributed to specific technology changes.

For the non-battery electrification component learning curves, NHTSA uses cost information from Argonne's 2016 Assessment of Vehicle Sizing, Energy Consumption, and Cost Through Large-Scale Simulation of Advanced Vehicle Technologies report.\328\ The report provides estimated cost projections from the 2010 lab year to the 2045 lab year for individual vehicle components.\329\ NHTSA considers the component costs used in HEVs and determines the learning curve by evaluating the year over year cost change for those components. Argonne published a 2020 and a 2022 version of the same report; however, those versions did not include a discussion of the high- and low-cost estimates for the same components.\330\ The learning estimates generated using the 2016 report align in the middle of the high- and low-cost estimates from the Argonne reports, and therefore NHTSA continues to apply the learning curve estimates based on the 2016 report. There are many sources that NHTSA could have picked to develop learning curves for non-battery electrification component costs; however, given the uncertainty surrounding extrapolating costs out to MY 2050, NHTSA believes these learning curves provide a reasonable estimate.

\328\ Moawad, A. et al., Assessment of Vehicle Sizing, Energy Consumption and Cost Through Large Scale Simulation of Advanced Vehicle Technologies, ANL/ESD-15/28, Argonne National Laboratory: Argonne, IL (2016), available at: https://doi.org/10.2172/1245199 (accessed: May 28, 2026).

\329\ DOE's lab year equates to 5 years after a model year (e.g., DOE's 2010 lab year equates to MY 2015). ANL/ESD-15/28 at p. 116.

\330\ Islam, E. et al., Energy Consumption and Cost Reduction of Future Light-Duty Vehicles Through Advanced Vehicle Technologies: A Modeling Simulation Study Through 2050, ANL/ESD-19/10, Argonne National Laboratory: Lemont, IL (2020), available at: https://publications.anl.gov/anlpubs/2020/08/161542.pdf (accessed: May 28, 2026); Islam, E. et al., A Comprehensive Simulation Study to Evaluate Future Vehicle Energy and Cost Reduction Potential, ANL/ ESD-22/6, Argonne National Laboratory: Lemont, IL (2022), available at: https://publications.anl.gov/anlpubs/2023/11/179337.pdf (accessed: May 28, 2026).

ICCT commented that NHTSA's electric motor and inverter costs are too high for SHEVs that utilize greater than 57 kW \331\ and attached a related 2025 HEV study they published.\332\

\331\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 17.

\332\ ICCT, Docket No. NHTSA-2025-0491-5240-A9, at 15-19.

NHTSA interprets ICCT's comment as taking issue with learning rates associated with electric motors in the future and disagrees with ICCT's claim that the cost of electric motors greater than 57 kW is too high. NHTSA determined that the non-battery component learning rates used in this analysis reflect the trajectory of economies of scale and production capabilities predicted by industry. NHTSA has not updated the non-battery component learning rates for the final rule.

In summary, NHTSA calculates the total hybrid powertrain costs by

summing individual component costs, which ensures that all technologies in a hybrid powertrain appropriately contribute to the total system cost. NHTSA combines the costs associated with the ICE (if applicable) and transmission, non-battery hybridization components like the electric machine, and battery pack to create a full-system cost. Chapter 3.3.5.4 of the Final TSD presents the total costs for each hybrid powertrain option, broken out by the components NHTSA discussed throughout this section. In addition, the section discusses where to find each of the component costs in the CAFE Model's various input files. 4. Road Load Reduction Paths

No car or truck uses energy (whether gas or otherwise) 100 percent efficiently when it is driven down the road. If the energy in a gallon of gas is thought of as a pie, the amount of energy ultimately available from that gallon to propel a car or truck down the road would only be a small slice. Instead, most of the energy is lost due to thermal and frictional losses in the engine and drivetrain and drag from ancillary systems (e.g., the air conditioner, alternator generator, or various pumps). The rest is lost to what engineers call road loads. For the most part, road loads include wind resistance (or aerodynamics), drag in the braking system, and rolling resistance from the tires. At low speeds, aerodynamic losses are very small, but as speeds increase these losses rapidly become dramatically higher than any other road load. Drag from the brakes in most cars is practically negligible. Tire rolling resistance losses can be significant: at low speeds rolling resistance losses can be more than aerodynamic losses. Whatever energy is left after these road loads is spent on accelerating the vehicle anytime its speed increases. This is where reducing the mass of a vehicle is important to efficiency because the amount of energy to accelerate the vehicle is always directly proportional to a vehicle's mass. All else being equal, reduce a car's mass and better fuel economy is guaranteed. However, at freeway speeds, aerodynamics plays a more dominant role in determining fuel economy than any other road load or vehicle mass.

NHTSA includes three road load reducing technology paths in this analysis: the Mass Reduction (MR) Path, Aerodynamic Improvements (AERO) Path, and Low Rolling Resistance Tires (ROLL) Path. For all three paths, NHTSA assigns vehicles in the analysis fleet technologies and identifies adoption features based on the vehicle's body style. The light-duty fleet body styles NHTSA includes in the analysis are convertible, coupe, sedan, hatchback, wagon, SUV, pickup, minivan, and van. Figure II-3 shows the light-duty fleet body styles used in the analysis. [GRAPHIC] [TIFF OMITTED] TR30SE26.078

As expected, the road load forces described above operate differently based on a vehicle's body style, and the technology adoption features and effectiveness values reflect this. The following sections discuss the three Road Load Reduction Paths. 5. Mass Reduction

Mass reduction is a relatively cost-effective means of improving fuel economy, and vehicle manufacturers are expected to apply various mass reduction technologies to meet fuel economy standards. Vehicle manufacturers can reduce vehicle mass through several different techniques, such as modifying and optimizing vehicle component and system designs, part consolidation, and adopting materials that are conducive to mass reduction (e.g., advanced high strength steel), aluminum, magnesium, and

plastics, including carbon fiber reinforced plastics).

For this analysis, NHTSA considers five levels of mass reduction technology (MR1-MR5) that include increasing amounts of advanced materials and mass reduction techniques applied to the vehicle's glider.\333\ The subsystems that may make up a vehicle glider include the vehicle body, chassis, interior, steering, electrical accessory, brake, and wheels systems. NHTSA accounts for mass changes associated with powertrain changes separately.\334\ The agency's estimates of how manufacturers could reach each level of mass reduction technology, and a discussion of advanced materials and mass reduction techniques can be found in Chapter 3.4 of the Final TSD.

\333\ Note that in the previous analysis associated with the MYs 2024-2026 final rule, there was a sixth level of mass reduction available as a pathway to compliance. For this analysis, this pathway was removed because it relied on extensive use of carbon fiber composite technology to an extent that is only found in purpose-built racing cars and a few hundred road legal sports cars costing hundreds of thousands of dollars. Final TSD Chapter 3.4 provides additional discussion on the decision to include five mass reduction levels in this analysis.

\334\ Glider mass reduction can sometimes enable a smaller engine while maintaining performance neutrality. Smaller engines typically weigh less than bigger ones. NHTSA captures any changes in the resultant fuel savings associated with powertrain mass reduction and downsizing via the Autonomie simulation. Autonomie calculates a hypothetical vehicle's theoretical fuel mileage using a mass reduction to the vehicle curb weight equal to the sum of mass savings to the glider plus the mass savings associated with the downsized powertrain.

Reaching the highest level of mass reduction considered in this analysis, MR5, requires a blend of aluminum and carbon fiber components. Achieving MR5 with aluminum exclusively is unlikely to be achievable by manufacturers during the rulemaking timeframe. Though aluminum technology can be an effective mass reduction pathway, it has its limitations. First, aluminum does not have a fatigue endurance limit. That is, with aluminum components there is always some combination of stress and cycles when failure occurs. Automotive design engineering teams will dimension highly stressed cross sections to provide an acceptable number of cycles to failure. But this often comes at mass savings levels that fall short of what would be expected purely based on density specific strength and stiffness properties for aluminum.

Looking at real data, the mostly aluminum (cab and bed are made from aluminum) 2021 Ford F150 achieves less than a 14-percent mass reduction compared to its 2014 all-steel predecessor.\335\ This is an especially pertinent comparison because both vehicles have the same footprint within a 2-percent margin and presumably were engineered to similar duty cycles given that they both came from the same manufacturer. Per the agency's regression analysis, the Ford F-150 achieves MR3. As mentioned in the Final TSD Chapter 3.4, achieving MR4 using aluminum only would require that the body in white structure be made almost entirely from aluminum. It may be possible to achieve MR5 without the use of carbon fiber, but the resultant vehicle would not achieve performance parity with customer expectations in terms of crash safety, noise and vibration levels, and interior content. The discontinued Lotus Elise is an example of an aluminum and fiberglass car that achieved MR5 but represents an extremely niche vehicle application that is unlikely to translate to mainstream, high-volume models. Therefore, it is entirely reasonable to assume that carbon fiber “hang on” panels and closures would be necessary to achieve MR5 at performance parity.

\335\ Ford, 2021 F-150 Technical Specifications (2021), available at: https://www.fromtheroad.ford.com/content/dam/fordmediasite/us/en/library/2021/specs/2021-F-150-Technical-Specs.pdf (accessed: May 28, 2026); Ford, Used 2014 F-150--Specs & Features, available at: https://www.edmunds.com/ford/f-150/2014/features-specs/ (accessed: May 28, 2026).

In past rules, commenters have noted that the NAS study relies on very little application of carbon fiber technology to achieve their highest level of mass reduction technology. NHTSA notes that the NAS study espouses a maximum level of mass reduction of approximately 14.5 percent using composites (e.g., fiberglass) and carbon fiber technology only in closures structures (e.g., doors, hoods, and decklids) and hang-on panels (e.g., fenders). This is the “alternative scenario 2” in the NAS study and is a similar light-weighting technology application strategy to what the analysis roughly associates with MR5, but MR5 requires a 20-percent mass reduction. In this scenario, NHTSA is allotting more mass reduction potential for the same carbon fiber technology application than the NAS study does.

CALSTART, Inc. (CALSTART) commented on mass reduction, stating, “. . . mass reduction is often treated as a secondary pathway rather than a primary technology lever. This underestimates the true `maximum feasible' frontier.” Offering support for the approach NHTSA takes with mass reduction, the American Chemistry Council (ACC) commented, “ACC supports NHTSA standards that recognize mass reduction as an effective strategy to achieve both improved safety and fuel economy outcomes.” \336\ CALSTART appears to misunderstand how the CAFE model analyzes and applies mass reduction, since mass reduction is a primary, not secondary, technology pathway, as explained in the CAFE Model Documentation and Final TSD Chapter 3.4. For this final rule, NHTSA did not make any changes to the mass reduction pathway in response to comments.\337\

\336\ ACC, Docket No. NHTSA-2025-0491-5049-A1, at 2.

\337\ For this final rule, NHTSA did correct an error in the NPRM market data input file that effectively skipped light trucks under 50k vehicles by nameplate. NHTSA has since corrected the market data file such that SKIP is applied to light trucks as discussed in Final TSD Chapter 3.4.3.

ICCT commented on specific mass reduction technology levels, stating that, “Nearly half of the baseline fleet is assumed to be at mass reduction level MR2 or MR3” and that, “NHTSA effectively limits further minor reduction in mass.” \338\ Fleet assignments for mass reduction, along with the regression analysis used to establish MR0 for each vehicle technology class, is discussed in Final TSD Chapter 3.4. Vehicles assigned MR2 or MR3 in the analysis fleet are lighter than the modeled vehicle for a given vehicle technology class and have already incurred the costs of MR2 or MR3 as a result of this mass reduction. Because these vehicles are already lighter than their respective vehicle technology class MR0 and early stages of mass reduction have already been achieved, continuing to remove mass from these vehicles will cost more. If NHTSA were to re-baseline MR0 continually, it would no longer account for the level of mass reduction already applied to vehicles in the analysis fleet, overestimating the available levels of mass reduction and underestimating the associated costs. The regression analysis used in this rulemaking appropriately captures the levels of mass reduction in the analysis fleet and further potential mass reduction throughout the rulemaking timeframe. NHTSA will revisit the regression analysis for future rulemaking and make appropriate updates.

\338\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 10.

NHTSA assigns mass reduction levels to vehicles in the analysis fleet by using regression analyses that consider a vehicle's body design \339\ and body style,

in addition to several vehicle design parameters, like footprint, horsepower, bed length (for pickup trucks), and battery pack size (if applicable), among other factors. NHTSA has been improving on the light-duty regression analysis since the 2016 Draft TAR and continues to find that it reasonably estimates mass reduction technology levels of vehicles in the analysis fleet. Chapter 3.4 of the Final TSD contains a full description of the regression analyses used for the analysis fleet and examples of results of the regression analysis for select vehicles.

\339\ The body design categories NHTSA uses are 3-box and 2-box pickup trucks. A 3-box has a box in the middle for the passenger compartment, a box in the front for the engine and a box in the rear for the luggage compartment. A 2-box has a box in front for the engine and then the passenger and luggage box are combined into a single box.

There are several ways NHTSA ensures that the CAFE Model considers mass reduction technologies in the way that manufacturers might apply them in the real world. Given the degree of commonality among the vehicle models built on a single platform, manufacturers do not have complete freedom to apply unique technologies to each vehicle that shares the same platform. Though some technologies (e.g., low rolling resistance tires) are very nearly “bolt-on” technologies, others involve substantial changes to the structure and design of the vehicle and therefore often necessarily affect all vehicle models that share that platform. In most cases, mass reduction technologies are applied to platform level components and therefore the same design and components are used on all vehicle models that share the platform. Each vehicle in the analysis fleet is associated with a specific platform family. A platform “leader” in the analysis fleet is a vehicle variant of a given platform that has the highest level of mass reduction technology in the analysis fleet. As the Model applies technologies, it “levels up” all variants on a platform to the highest level of mass reduction technology on the platform. For example, if a platform leader is already at MR3 in MY 2024, and a “follower” starts at MR0 in MY 2024, the follower will get MR3 at its next redesign (unless the leader is redesigned again before that time and further increases the mass reduction level associated with that platform, then the follower would receive the new mass reduction level).

In addition to leader-follower logic for vehicles that share the same platform, NHTSA also restricts MR5 technology to platforms that represent 50,000 vehicles or fewer. The CAFE Model does not apply MR5 technology to platforms representing high-volume sales, like a Chevrolet Traverse, for example, where hundreds of thousands of units are sold per year. NHTSA also restricts MR5 technology from being applied to low volume non-passenger vehicles with a focus on durability and utility, such as the Ineos Grenadier and Ram 1500 Classic.\340\ NHTSA uses the combination of the leader-follower logic, the 50,000- unit threshold, and the high-utility vehicle restriction to make the simulation of mass reduction technologies more realistic. This is because NHTSA assumes that MR5 would require carbon fiber technology.\341\ There is high global demand from a variety of industries for a limited supply of carbon fibers; specifically, aerospace, military/defense, and industrial applications demand most of the carbon fiber currently produced. Currently, only about 10 percent of the global dry carbon fiber supply is allocated to the automotive industry, limiting the global supply base to supporting approximately 70,000 vehicles.\342\ In addition, the production process for carbon fiber components is significantly different than for traditional vehicle materials. NHTSA uses this adoption feature as a proxy for stranded capital (i.e., when manufacturers amortize research, development, and tooling expenses over many years) from leaving the traditional processes and to represent the significant paradigm change to tooling and equipment that would be required to support molding carbon fiber panels. There are no other adoption features for mass reduction in the analysis.

\340\ See the Final TSD Chapter 3.4 for more information on MR5 restrictions.

\341\ See the Final TSD for CAFE standards for MYs 2024-2026 and Chapter 3.4 of the Final TSD accompanying this rulemaking for more information about carbon fiber.

\342\ Sloan, J., Carbon Fiber Suppliers Gear Up for Next Generation Growth, last revised: Feb. 11, 2020, available at: https://www.compositesworld.com/articles/carbon-fiber-suppliers-gear-up-for-next-gen-growth (accessed: May 28, 2026).

In the Autonomie simulations, mass reduction technology is simulated as a percentage of mass removed from the specific subsystems that make up the glider. The mass of subsystems that make up the vehicle's glider is different for every technology class, based on glider weight data from the A2Mac1 database \343\ and two NHTSA- sponsored studies that examined light-weighting a passenger car and light truck. NHTSA accounts for mass reduction from powertrain improvements separately from glider mass reduction. Autonomie considers several components for powertrain mass reduction, including engine downsizing and fuel tank, exhaust systems, and cooling system light- weighting.\344\ With regard to the light-duty vehicle fleet, the 2015 NAS report suggested an engine downsizing opportunity exists when the glider mass is light-weighted by at least 10 percent. The 2015 NAS report also suggested that 10-percent light-weighting of the glider mass alone would boost fuel economy by 3 percent and any engine downsizing following the 10-percent glider mass reduction would provide an additional 3-percent increase in fuel economy.\345\ The NHTSA light- weighting studies applied engine downsizing (for some vehicle types but not all) when the glider weight was reduced by 10 percent. Accordingly, the analysis limits engine resizing to several specific incremental technology steps; important for this discussion, engines in the analysis are resized only when mass reduction of 10 percent or greater is applied to the glider mass or when one powertrain architecture replaces another architecture. A summary of how the different mass reduction technology levels improve fuel consumption is shown in Final TSD Chapter 3.4.4.

\343\ A2Mac1, Automotive Benchmarking (proprietary data), available at: https://portal.a2mac1.com/ (accessed: May 28, 2026). The A2Mac1 database tool is widely used by industry and academia to determine the bill of materials (a list of the raw materials, sub- assemblies, parts, and quantities needed to manufacture an end- product) and mass of each component in the vehicle system.

\344\ Though NHTSA does not account for mass reduction in transmissions, NHTSA does reflect design improvements as part of mass reduction when going from, for example, an older AT6 to a newer AT8 that has similar if not lower mass.

\345\ 2015 NAS Report.

Regarding mass reduction costs, ICCT commented, “[a]lthough NHTSA improved its methodology for calculating mass reduction costs (the `bracketed' approach), its costs for M0-M4 levels of mass reduction remain based on outdated studies.” \346\ ICCT did not provide more recent comprehensive mass reduction cost estimates. NHTSA prices mass reduction technologies using a series of studies completed by EDAG in 2012 and 2018.347 348 These prices are then multiplied by an RPE factor to account for indirect costs

associated with applying and developing the mass reduction technologies, such as R&D, personnel, and facilities. To account for ongoing improvements in manufacturing, learning, and efficiencies, the prices are reduced in a linear fashion year after year. Accordingly, NHTSA concludes that the current mass reduction costs and learning curves are still representative of mass reduction levels during the rulemaking timeframe and has not made changes to mass reduction costs for this final rule.

\346\ ICCT, Docket No. NHTSA-2025-0491-5240-A2, at 3.

\347\ Singh, H. et al., Mass Reduction of Light-Duty Vehicles for Model Years 2017-2025: Peer Review Comments Log, DOT HS 812 487, NHTSA: Washington, DC (2018), available at: https://www.nhtsa.gov/sites/nhtsa.gov/files/documents/13250f-peer_review_comment_resolution_document-final-112818-v3-tag.pdf (accessed: May 28, 2026).

\348\ Singh, H. et al., Mass Reduction for Light-Duty Vehicles for Model Years 2017-2025, Final Report, DOT HS 811 666, NHTSA: Washington, DC (2012), available at: https://static.nhtsa.gov/nhtsa/downloads/CAFE/2017-25_Final/811666.pdf (accessed: May 28, 2026).

NHTSA's mass reduction costs are based on two NHTSA light-weighting studies--the teardown of a MY 2011 Honda Accord and a MY 2014 Chevrolet Silverado pickup truck \349\--and the 2021 NAS report.\350\ The costs for MR1-MR4 rely on the light-weighting studies, while the cost of MR5 references the carbon fiber costs provided in the 2021 NAS report. Unlike the other technologies in this analysis that have a fixed technology cost (for example, it costs about $3,000 to add an AT10L3 transmission to a light-duty SUV or pickup truck in MY 2027), the cost of mass reduction is calculated on a dollar per pound saved basis based on a vehicle's starting weight. Put another way, for a given vehicle platform, an initial mass is assigned using the aforementioned regression model. The amount of mass to reach each of the five levels of mass reduction is calculated by the CAFE Model based on this number and then multiplied by the dollar per pound saved figure for each of the five mass reduction levels. The dollar per pound saved figure increases at a nearly linear rate going from MR0 to MR4. However, this figure increases steeply going from MR4 to MR5 because the technology cost to realize the associated mass savings level is an order of magnitude larger. This dramatic increase is reflected by all three studies NHTSA relied on for mass reduction costing, and NHTSA believes that it reasonably represents what manufacturers would expect to pay for using increasing amounts of carbon fiber on their vehicles.

\349\ Singh, H. et al., Mass Reduction for Light-Duty Vehicles for Model Years 2017-2025, Final Report, DOT HS 811 666, NHTSA: Washington, DC (2012), available at: https://static.nhtsa.gov/nhtsa/downloads/CAFE/2017-25_Final/811666.pdf (accessed: May 28, 2026); Singh, H. et al., Mass Reduction for Light-Duty Vehicles for Model Years 2017-2025: Peer Review Comments Log, DOT HS 812 487, NHTSA: Washington, DC (2018), available at: https://downloads.regulations.gov/NHTSA-2021-0053-0011/attachment_5.pdf (accessed: May 28, 2026).

\350\ This analysis applied the cost estimates per pound derived from passenger cars to all passenger car segments, and the cost estimates per pound derived from full-size pickup trucks to all light-duty truck and SUV segments. The cost estimates per pound for carbon fiber (MR5) were the same for all segments.

Like past analyses, NHTSA considers several options for mass reduction technology costs. The agency has determined that the NHTSA- sponsored studies accounted for significant factors the agency believes are important to include in this analysis, including materials considerations (material type and gauge, while considering real-world constraints such as manufacturing and assembly methods and complexity), safety (including the Insurance Institute for Highway Safety's (IIHS) small overlap tests), and functional performance (including towing and payload capacity and noise, vibration, and harshness), and gradeability in the pickup truck study.\351\

\351\ Final TSD Chapter 7.3 has additional detail on this analysis.

First, NHTSA limits application of MR5 in the analysis to represent the limited volume of available dry carbon fiber and the resultant high costs of the raw materials. This constraint is described above and in more detail in Final TSD Chapter 3. The CAFE Model assumes that there is not enough carbon fiber readily available to support vehicle platforms with more than 50,000 vehicles sold per year. NHTSA believes this volume constraint does more to limit the application of MR5 technology in the analysis than does its high price. Even if a lower price is used, the dominant constraint would still be volume. Second, NHTSA does not believe that a lower price would prove to be a competitive pathway to compliance with exotic materials technology compared to other less expensive technologies with higher effectiveness. The MR5 effectiveness as applied to vehicles in this analysis considers the total effect of reducing that level of mass from the vehicle, from the vehicle's starting mass reduction level. As an example, while the cost of going from MR0 or MR1 to MR5 may be slightly overstated (but still limited in total application by the volume cap), the cost of going from MR4 to MR5 is not. NHTSA continues to consider the balance of carbon fiber and other advanced materials for mass reduction to meet MR5 levels and may update that value in future rules. 6. Aerodynamic Improvements

The energy required for a vehicle to overcome wind resistance, or more formally what is known as aerodynamic drag, ranges from minimal drag at low speeds to extremely significant drag at highway speeds.\352\ Reducing a vehicle's aerodynamic drag is, therefore, an effective way to reduce the vehicle's fuel consumption. Aerodynamic drag is characterized as proportional to the frontal area (A) of the vehicle and a factor called the coefficient of drag (Cd). The coefficient of drag (Cd) is a dimensionless value that represents a moving object's resistance against air, which depends on the shape of the object and flow conditions. The frontal area (A) is the cross-sectional area of the vehicle as viewed from the front. Aerodynamic drag of a vehicle is often expressed as the product of the two values, CdA, which is also known as the drag area of a vehicle. The force imposed by aerodynamic drag increases with the square of vehicle velocity, accounting for the largest contribution to road loads at higher speeds.\353\

\352\ 2015 NAS Report, at p. 207.

\353\ See, e.g., Pannone, G., Technical Analysis of Vehicle Load Reduction Potential for Advanced Clean Cars, Final Report, California Air Resources Board and The California Environmental Protection Agency: Sacramento, CA (2015), available at: https://ww2.arb.ca.gov/sites/default/files/2020-04/13_313_ac.pdf (accessed: May 28, 2026). The graph on p. 20 shows how the aerodynamic force becomes the dominant load force at higher speeds.

Manufacturers can reduce aerodynamic drag either by reducing the drag coefficient or reducing vehicle frontal area, which can be achieved by passive or active aerodynamic technologies. Passive aerodynamics refers to aerodynamic attributes that are inherent to the shape and size of the vehicle. Passive attributes can include the shape of the hood, the angle of the windscreen, or even overall vehicle ride height. Active aerodynamics refers to technologies that variably deploy in response to driving conditions. Examples of active aerodynamic technologies are grille shutters, active air dams, and active ride height adjustment. Manufacturers may employ both passive and active aerodynamic technologies to improve aerodynamic drag values.

There are four levels of aerodynamic improvement (over AERO0, the first level) available in the analysis (AERO5, AERO10, AERO15, AERO20). Refer to Figure II-3 for a visual of each body style considered in the analysis. Each AERO level is associated with 5-, 10-, 15-, or 20- percent aerodynamic drag improvement values over a reference value computed for each vehicle body style. These levels, or bins, respectively correspond to the level of aerodynamic drag reduction over the reference value (e.g., “AERO5” corresponds to the 5-percent aerodynamic drag improvement value over the reference value). While

each level of aerodynamic drag improvement is technology neutral--that is, manufacturers can ultimately choose how to reach each level by using whatever technologies work for the vehicle--NHTSA estimates a pathway to each technology level based on data from a National Research Council of Canada-sponsored wind tunnel testing program. The program included an extensive review of production vehicles utilizing aerodynamic drag improvement technologies and of industry comments.\354\ NHTSA's example pathways for achieving each level of aerodynamic drag improvement are discussed in Chapter 3.5 of the Final TSD.

\354\ Larose, G. et al., Evaluation of the Aerodynamics of Drag Reduction Technologies for Light-Duty Vehicles: A Comprehensive Wind Tunnel Study, SAE International Journal of Passenger Cars-- Mechanical Systems, Vol. 9(2): pp. 772-84 (2016), available at: https://doi.org/10.4271/2016-01-1613 (accessed: May 28, 2026).

NHTSA assigns aerodynamic drag reduction technology levels in the analysis fleets based on vehicle body styles.\355\ NHTSA computes an average coefficient of drag based on vehicle body styles, using coefficient of drag data from the MY 2015 analysis fleet. Different body styles offer different utility and have varying levels of form drag. This analysis considers both frontal area and body style as unchangeable utility factors affecting aerodynamic forces; therefore, the analysis assumes all reductions in aerodynamic drag forces come from improvements in the drag coefficient. Then NHTSA uses drag coefficients for each vehicle in the analysis fleet to establish an initial aerodynamic technology level for each vehicle. NHTSA compares the vehicle's drag coefficient to the calculated drag coefficient by body style mentioned above to assign initial levels of aerodynamic drag reduction technology to vehicles in the analysis fleets. NHTSA can find most vehicles' drag coefficients in manufacturers' publicly available specification sheets; however, in cases where this information cannot be found, NHTSA uses engineering judgment to assign the initial technology level.

\355\ These assignments do not necessarily match the body styles that manufacturers use for marketing purposes. Instead, NHTSA makes these assignments based on engineering judgment and the categories used in the modeling, considering how this affects a vehicle's AERO and vehicle technology class assignments.

NHTSA looks at vehicle body style and vehicle HP to determine which types of vehicles can adopt different aerodynamic technology levels. For this analysis, AERO15 and AERO20 cannot be applied to minivans, and AERO20 cannot be applied to convertibles, pickup trucks, and wagons. In addition, NHTSA does not allow application of AERO15 and AERO20 technology to vehicles with more than 780 HP. This threshold is informed by information about performance of ICE vehicles. NHTSA recognizes that manufacturers tune aerodynamic features on these vehicles to provide desirable downforce at high speeds and to provide sufficient cooling for the powertrain, rather than reducing drag, resulting in middling drag coefficients despite advanced aerodynamic features. Therefore, manufacturers may have limited ability to improve aerodynamic drag coefficients for high performance ICE vehicles without reducing HP. This threshold for performance vehicles only limits the application of aerodynamic technologies on 2,518 units of sales volume in the analysis fleet.\356\

\356\ See the Market Data Input File.

The aerodynamic technology effectiveness values that show the potential fuel consumption improvement from AERO0 technology are found and discussed in Chapter 3.5.4 of the Final TSD. For example, the AERO20 values represent the range of potential fuel economy improvement that could be achieved through the replacement of AERO0 technology with AERO20 technology for every technology key that is not restricted from using AERO20. NHTSA uses the change in fuel consumption values between entire technology keys and not the individual technology effectiveness values. Using the change between whole technology keys captures the complementary or non-complementary interactions among technologies.

NHTSA has carried forward the established AERO technology costs previously used in the 2020 final rule for the MY 2024-2026 standards analysis,\357\ and the 2024 rulemaking and has updated those costs to the dollar-year used in this analysis. For light-duty AERO improvements, the cost to achieve AERO5 is relatively low, as manufacturers can make most of the improvements through body styling changes. The cost to achieve AERO10 is higher than AERO5, due to the addition of several passive aerodynamic technologies, and consecutively the cost to achieve AERO15 and AERO20 is much higher than AERO10 due to use of both passive and active aerodynamic technologies. The cost estimates are based on CBI submitted by the automotive industry in advance of the 2018 CAFE NPRM and on the agency's assessment of manufacturing costs for specific aerodynamic technologies. The 2018 FRIA contains discussion of the cost estimates.\358\ NHTSA has not received additional information from stakeholders regarding the AERO costs since conducting the analysis for the 2018 final rule and has continued to use those cost estimates for subsequent analysis, including for this analysis. Final TSD Chapter 3.5 contains additional discussion of aerodynamic improvement technology costs, and costs for all technology classes across all model years are in the Technologies Input File.

\357\ Note the FRIA accompanying the 2020 final rule, Chapter VI.C.5.e.

\358\ Note the PRIA accompanying the 2018 NPRM, Chapter 6.3.10.1.2.1.2 for a discussion of these cost estimates.