3

Certification, Compliance, and Enforcement

3.1 INTRODUCTION

As with any regulatory program for mobile sources, the issues of certification, compliance, and enforcement are critical to the success of the program. For medium- and heavy-duty vehicles (MHDVs), designing and implementing effective certification, compliance, and enforcement is particularly challenging given the nature of this sector, including the multiple components of a truck that could be subject to certification, compliance, and enforcement measures (e.g., whole vehicle, engine, powertrain, trailer, etc.), the diversity of potential testing regimes (engine certification, whole vehicle simulation, in-use enforcement, etc.), varying duty cycles, and the variety and heterogeneity of the vehicles subject to any regime.

The committee and its predecessor (the “NRC Phase One” committee) have provided extensive comments on certification, compliance, and enforcement in their first two reports (NRC, 2010, 2014). This chapter expands on those previous findings and recommendations by addressing the following topics. Section 3.2 of this chapter provides an overview of the certification, compliance, and enforcement approaches adopted by the National Highway Traffic Safety Administration (NHTSA) and the Environmental Protection Agency (EPA) in the Phase I standards and Phase II standards. Section 3.3 provides our findings and recommendations with respect to improvements and challenges with the current pre-sale certification and compliance models and tools used by NHTSA and EPA in their rulemakings. Section 3.4 addresses in-use compliance and enforcement.

3.2 SUMMARY OF CERTIFICATION APPROACHES IN PHASE I AND PHASE II RULES

Perhaps the most anticipated aspects of the Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty Engines and Vehicles—Phase II rules (EPA and NHTSA, 2016) are the changes to the certification structure. Considerable changes have been made in the Phase II Greenhouse Gas Emissions Model (GEM) for heavy-duty vehicles. In general, the Phase II GEM features more manufacturer-specified inputs and fewer default inputs compared to Phase I GEM, and more real-world route choices versus previous standardized routes. An engine test remains an element of the certification of the vehicles, and the fuel maps used in the engine certification can be directly used by the manufacturer for use in GEM modeling. This section summarizes the major changes made to the certification regulations in the Phase II rule for MHDVs. The section is divided by regulated classes: first, heavy-duty vehicles; second, vocational vehicles; and third, trailers for heavy-duty tractors.

3.2.1 Heavy-Duty Combination Tractors

Certification in the Phase I rule was intentionally designed to be readily implementable. NHTSA and EPA wanted to quickly achieve a means to reduce greenhouse gas (GHG) emissions and improve vehicle efficiency without requiring the regulated manufacturers to invest sizable capital for test equipment and new technologies which were not currently available. The certification structure for Phase I has two major components. One is an engine dynamometer certification test to measure CO₂, CH₄, and N₂O emissions in grams per brake-horsepower-hour, which is aligned with what has been used for criteria pollutants for many years. (Emissions of CH₄ and N₂O are relatively low compared to CO₂; however, they are measured because on an equal-mass basis they have global warming potentials that are 28-36 and 265-280 times greater than CO₂, respectively [EPA, 2017]). The other is the use of the GEM, which simulates vehicle fuel efficiency and CO₂ emissions over three distinct driving cycles for prescribed vehicle configurations and weights, resulting in a single output expressed in fuel consumed in gallons per 1,000 ton-miles and an equivalent CO₂ output expressed as grams per ton-mile. This certification strategy was expedient at the time, but it is representative of vehicle performance over only certain routes and confines the inputs for the simulation model to minimize complexity—and therefore precision—of the measured emissions and efficiency of the certified vehicles.

After the promulgation of the Phase I rules, the comments from industry and recommendations in the NRC Phase One Report (NRC, 2010) encouraged NHTSA and EPA to improve the representativeness of GEM in the ensuing rulemaking. The committee in its GEM analysis encouraged these agencies to include more route choices. It argued in favor of more manufacturer-specific inputs in lieu of the default attributes for the vehicle in GEM and made specific recommendations for refinements to the GEM inputs.

Some original equipment manufacturer (OEM) presenters to the committee recommended that EPA consider eliminating the engine test aspect of the certification strategy and rely only on the full vehicle simulation, which would include CO₂ results from the use of a specified engine map. In general, the Phase II certification approach has not deviated from the original Phase I test strategy: the engine test for CO₂ emissions remains, as does the GEM simulation for the vehicle. The certification structure for the vehicle (GEM), however, has been modified considerably to address changes in expected available technology and improvements to the robustness and fidelity of the tests and simulation model.

NHTSA and EPA propose numerous changes in response to the recommendations made by the Committee in its first report (2014). Those recommendations are summarized below.

• Recommendation 3.2: GEM 3 should be compatible with manufacturer’s order entry system.
• Recommendation 3.3: Output from GEM 3 should include graphs and other presentation methods to help manufacturers gain greater insight into actions to improve GHG emissions and fuel efficiency.
• Recommendation 3.4: Develop a mechanism to determine accurate tire rolling resistance, and store it for public availability.
• Recommendation 3.5: Improve methods for determining aerodynamic performance to be more representative of real-world experience.
• Recommendation 3.6: Configure test cycles and routes to ensure that products are not optimized for the test, but optimized for real-world fuel economy and CO₂ reduction.
• Recommendation 3.7: Consider allowing OEMs to replace fixed code with their own validated models of components or systems which represent real-world operation.
• Recommendation 3.8: Determine whether the steady-state speed-torque test is sufficient to capture all efficiency measures taken (e.g., downsizing engine is an approach that GEM would not properly address).
• Recommendation 3.9: Measured values for vehicle weights should replace fixed values.
• Recommendation 3.10: GEM should accommodate the power-to-weight ratio of a vehicle if GEM permits the benefit accrued from integrating the powertrain and aftertreatment systems with interactive controls.
• Recommendation 3.11: Revise cycles and schedules to replicate over-the-road activities without becoming cumbersome.
• Recommendation 3.12: Determine whether GEM can be used to derive the fuel economy benefits from all trailers in combination with tractors.

NHTSA and EPA addressed all the recommendations made by the committee in its 2014 report; however, with regard to Recommendation 3.12, the agencies were unable to justify the regulation of all trailers (see the subsection “Trailers” below).

Another important change in the Phase II rule is the application of useful life. EPA has always stipulated that engines or vehicles certified for criteria pollutant emissions would meet the standard through a “useful life” period, and that the technology required to meet those certified values cannot be disassembled from the certified engine configuration during the useful life period. The Phase I fuel economy regulations promulgated by NHTSA (EPA and NHTSA, 2011) did not include useful life requirements. EPA applied the same useful life periods for other criteria pollutant emissions to the Phase I CO₂ standard for MHDVs. In the Phase II regulations, NHTSA incorporates the EPA useful life periods, and EPA has changed the useful life of the light heavy-duty engine category from 110,000 miles to 150,000 miles to reflect today’s experience while keeping useful life numbers for the other categories the same as in Phase I. The useful life compliance period for trailers is 10 years (see Table 3-1).

In addition to useful life compliance, the Phase II regulations include a production audit of five vehicles per manufacturer per year. These vehicles will be tested on a chassis dynamometer. NHTSA and EPA concede this is not a confirmatory test to certify results; however, these results can be used to refine a production audit process in the future.

As a final enforcement practice, EPA states in the Phase II rule that it will continue to conduct selective enforcement audits with the manufacturers and evaluate in-use compliance by testing Phase II in-use engines for CO₂ compliance. More detailed analyses of the GEM refinements for vehicle certification are included in other sections of this report.

3.2.2 Vocational Vehicles

The manufacture of the vocational vehicle is generally shared between the chassis manufacturer, which produces an incomplete or stripped chassis, and the final stage manufacturer, which assembles the body of the vehicle that accommodates the load and work-specific equipment. In some cases, the entire vehicle is built by a specialty manufacturer. The responsibility for certifying these vehicles falls to the chassis manufacturer in both Phase I and Phase II regulations. In Phase I, NHTSA and EPA considered only the vehicle weight class in grouping vehicles for certification: light heavy-duty (Class 2b through Class 5), medium heavy-duty (Class 6 and Class 7), and heavy heavy-duty (Class 8).

The NRC Phase One Report expressed concerns that test cycles and standards should reflect the performance of vocational vehicles over varied duty cycles. In response to this concern, NHTSA and EPA have widened the categories under which vocational vehicles are certified. In the Phase II rule, the three vocational vehicle weight classes are further divided into distinct categories reflecting driving patterns: regional, multipurpose, and urban. The CO₂ certification test cycle weighting factors are adjusted to reflect these driving cycles. Certain exemptions for emergency vehicles will apply.

TABLE 3-1 Phase II Useful Life Compliance Periods by Vehicle Class and Trailers

SOURCE: EPA and NHTSA (2016, pp. 74256, 74262).

3.2.3 Trailers

The Phase I rule did not regulate trailers. In its Phase Two First Report, the committee stated clearly its belief that trailers could and should be regulated. It pointed to the success of the EPA voluntary SmartWay program as a sound foundation for efficiently enforcing a Phase II performance standard nationwide which could result in anywhere from 3 to 11 percent improvement in fuel economy from aerodynamic treatment to the trailer (NRC, 2010).

The committee’s Phase Two First Report also made recommendations regarding trailers for the agency to consider in its Phase II rulemaking. These recommendations have generally been included in the Phase II rule:

• Recommendation 6.1: Require all new 53-foot and longer dry and refrigerated box vans to meet performance standards that will reduce fuel consumption and CO₂ emissions.
• Recommendation 6.2: Consider including other trailers such as pups, flatbeds, and container vans if practicable.
• Recommendation 6.3: Define an optimum full vehicle test and validate it against real-world experiences. NHTSA should assess if adding yaw loads would provide significantly increased value to the calculated C_d of the vehicle. NHTSA should also compile test data and fuel economy results and make it available to the public.
• Recommendation 6.4: NHTSA should prepare a white paper clarifying the minimum performance needed from a safety perspective.

For most of the dry box, refrigerator box trailers, and pup vans between 28 and 53 feet the agency was able to justify applying performance standards in the Phase II rule. Technologies to be applied to achieve these standards include a variety of aerodynamic features and low-rolling-resistance tires, primarily carried over from EPA’s SmartWay program, and the California Air Resources Board’s Greenhouse Gas Reduction Regulation. The advantage these programs provide is that these features have already been road tested and verified by the SmartWay Technology Program.

For some box trailers with field equipment prohibiting aerodynamic treatment to either end of the trailer, referred to in a category called non-aero box vans, NHTSA and EPA have concluded that the low-rolling-resistance tires and the use of automatic tire inflation (ATI) will be applicable. A list of heavy hauler tractor-trailers and specialty field trailers are exempt from the trailer standards.

Table 3-2 provides EPA’s summary of the major technical input changes which have been incorporated in GEM from the Phase I rule to the Phase II rule.

3.3 PRE-MARKET CERTIFICATION

3.3.1 Introduction

Pre-market certification has been favored for decades in the automotive world and may be directly based on test results (crash-worthiness, criteria pollutant emissions, ad noise) or through component specifications (lighting and braking systems), which in turn requires compliant component design or verification testing. Usually there is an assumption that the test or specification addresses real-world use sufficiently well that no subsequent compliance procedure is needed. Vehicle recalls may arise when premature component failures or a specific design flaw or malfunction exceed a threshold for concern, but these are not part of an organized after-sales measurement campaign. In-use enforcement of vehicle performance is complex because both the manufacturer and the owner may be implicated in deviations from an expected norm.

Several examples of in-use compliance exist. As one example, there is a program in the United States that examines in-use heavy-duty vehicle NO_x and particulate matter (PM) emissions using on-board equipment, which originated following the realization that certain diesel engine manufacturers controlled the engines so that brake-specific NO_x was substantially higher under steady load over the road than during the transient FTP pre-market certification. As another example, dynamometer-based inspection and maintenance programs were used to control

TABLE 3-2 GEM Input Comparisons by Category—Phase I and Phase II

SOURCE: Derived from email to committee member by EPA.

in-use automobile emissions in air districts that exceeded national ambient air quality standards. These have been replaced in many cases by reliance on the vehicle on-board diagnostics, a practice which is shown to have flaws. Safety checks and annual inspections of trucks, buses, and automobiles represent in-use compliance practice. California has a diesel exhaust smoke opacity test as a surrogate for transient PM production. However, in all of these cases, the in-use standards are less stringent than the pre-market standards because it is acknowledged that the performance is hard to measure accurately in the field, that components can deteriorate, and that the in-use operation or test circumstances may be substantially dissimilar from the original certification test. Thresholds also allow for accuracy of field grade measurement equipment. In this way, most in-use approaches do not represent a high standard and would not be suitable on their own for controlling vehicle design through pre-market certification.

An obverse view is that in-use performance represents the real truth, and that formalized pre-market testing fails to control adequately for all circumstances because it must be limited in nature. For example, real-world automobile fatality statistics cannot be predicted by crash-worthiness ratings alone (Elliott, 2009). Designing toward the compliance test exacerbates this disconnection. Where the design includes controls or components that can behave in a nonlinear fashion, a vehicle may prove to be highly fuel efficient during the pre-market certification test, but less so when exploring other operating regimes.

Vehicle models or simulations may be used to predict fuel use. The model employs maps or modules to describe major components in the drivetrain (engine and transmission) or the interaction of components with the outside world (tires, surfaces, and gravitational force). Internal to the model, variables such as torque or speed are exchanged between component modules, and engine fuel flow yields the numerator in the overall efficiency metric. The model must include the controller logic used in the actual vehicle and must simulate the behavior of a driver. By employing a vehicle simulation, rather than one or two prescribed tests, current and proposed MHDV pre-market certification approaches provide the benefit of determining the fuel efficiency over a variety of operating schedules and permit cost-effective examination of vehicles with different combinations of components. However, the simulation accuracy relative to the real world is no better than the representativeness that the component maps in the simulation would imply. Those component maps or modules are themselves obtained from test procedures, and these procedures cannot cover all operational circumstances.

It is argued below that simulation represents a more versatile approach than whole vehicle testing and is a reasonable short- and mid-term approach. However, the growing need to reduce fuel use in all types of operation, over a full vehicle life span, and in an increasingly interconnected world may require in-use methods, as discussed below. Ultimately the problem must be addressed as a system optimization of a society of interacting vehicles and control devices.

3.3.2 Engines

3.3.2.1 Engine Dynamometer Testing Adequacy

At present, MHDV fuel efficiency is determined by simulation and is determined separately from criteria pollutant values, which are still engine based.¹ However, the engine fuel efficiency is regulated in addition to the MHDV fuel efficiency and is determined using the same well-established test procedure that is used for criteria pollutant standards. Heavy-duty engine dynamometer testing procedures are well established because they have been employed for three decades to quantify the emissions of regulated criteria pollutants from engines during EPA and California Air Resources Board (CARB) certification. The procedure defines not only the dynamometer measurement of load and speed, but also the way the exhaust is captured, diluted, and analyzed. In 2007 the reduction of PM from diesel engines to levels that require exhaust filtration was accompanied by new procedures that sought a higher level of accuracy from the measurement process (EPA, 2005). Engine fuel consumption and carbon dioxide production can be measured independently during the certification process and can be compared with one another by using fuel composition data.

___________________

¹ Some Class 2b vehicles are vehicle tested.

The variability in test run inputs, and the inability of dynamometer tests to assess all possible MHDV operation modes, is a challenge to engine fuel consumption estimation and testing of individual components. Cycle-averaged dynamometer brake-specific fuel consumption measurements in a single laboratory typically are repeatable from hot run to hot run with a precision of a fraction of a percent, but the permissible accuracy of fuel flow measurements, engine speed measurements, and engine shaft torque measurements alone suggests that the brake-specific fuel consumption could vary by about 2 percent between laboratories.² It is therefore difficult to determine the relative merits of certain technology choices by differencing total fuel consumption between laboratories and complete engines, since many technology improvements are claimed to have an efficiency improvement smaller than the intertest variability. This is even more true because fuel consumption variability is within a narrow window limited by thermodynamic considerations. If an assumption is made that engines can reduce their fuel consumption by at most 10 percent (a brake efficiency improvement of about 5 percent), then the potential measurement variability would be as much as one-fifth of that improvement range. However, there is no clear pathway for substantial improvement in this accuracy.

The full scope of all possible real-world operational modes is addressable using any given test procedure. The efficiency of modern engines with sophisticated controls is strongly influenced by speed, load, and transient behavior. To evaluate the real-world benefits of engine technology improvement in reducing fuel use, it is important that the engine is exercised in a way that mimics real-world use adequately. Unfortunately, engine speed, load, and dynamic behavior in real-world operation vary widely due to both vehicle application and operational circumstances. The controlled approach that offers relatively high accuracy and very high precision in the test cell does not permit the evaluation of the engine’s efficiency during extremes of temperature, humidity, and altitude. The approach also cannot account for the variable cooling power demands placed on the engine as climate varies or road grade changes. Also, it is far more difficult to measure instantaneous fuel consumption than cycle-averaged fuel consumption accurately. If it is measured using carbon dioxide instantaneous measurement, the combustion products are delayed and diffused over time by the exhaust and measurement systems (Madireddy and Clark, 2006). Therefore, there is difficulty in employing test cell instantaneous fuel economy data to verify models used to predict instantaneous efficiency of an engine or vehicle under transient operation. The more divorced a test procedure becomes from the real-world operation, the more the engine is likely to be configured to the test rather than the end use. Fuel price and technology cost changes will influence this gap through the design and calibration processes.

The test procedure has the most stringent requirements for criteria pollutant emissions and for fuels used, neither of which reflect real-world performance. The engine must exhibit the highest-criteria pollutant reduction stringency during the dynamometer test, because in-use compliance is inherently more relaxed in restricting the pollutant level. In some cases there are identifiable causes for fuel use deviations between test cell and real-world emissions: with current-technology diesel engines these include inactive selective catalytic reduction (SCR) operation, the need for diesel particulate filter (DPF) regeneration in cold weather, and variable back pressure from the DPF. Some proposed technologies, such as waste-heat recovery, depend on engine load history and ambient temperature for performance: their benefits are hard to quantify generically. The test cell procedure also specifies the fuel to be used in the vehicle. This assists with repeatability, but real-world diesel energy content, and hence either mass or volume of fuel used, vary widely (Bacha et al., 2007) and differ from the test fuel. For example, the specific gravity of No. 2 diesel fuel can vary from 0.82 to 0.88 at 15.6°C, and this affects heating value.

Dynamometer fuel efficiency measurements and real-world fuel efficiency will vary in sympathy when the dynamometer test schedule simulates torque and speed values that are encountered on the road. An engine that has multiple applications, such as line haul operation and transient urban operation, and engines paired with boost-interrupting transmissions cannot be correctly simulated with a single test cycle. Current measurements average fuel consumption over the operation period in the test cell for a legacy transient test with a high proportion of operation near rated speed, and a set of steady-state points. The cycle-averaged value from the transient test or a weighting of the point efficiencies can be used directly as a standard for engine efficiency. The current approach in GEM uses individual set-point efficiencies to map the steady-state efficiency to predict engine efficiency during

___________________

² The standard estimate of the error for torque can be as high as 2 percent of the maximum torque, with additional statistical limitations for torque calibrations shown in 40 C.F.R. Part 1065.307, Table 1.

vehicle modeling. However, measurement and digital post-processing methods already exist to determine the instantaneous, transient fuel consumption of the engine and to relate this to the instantaneous torque and speed and past history of commands given to the engine controller. With these data, one is able to reconstruct the efficiency of an engine over a wide range of behaviors or cycles during a simulation, although still under constraint of the test cell conditions. Cold-start operation in the test cell can add to the sophistication of these predictions. In this way, the engine efficiency is more realistically modeled in a range of applications with little additional effort during the test cell procedures. Test cell in combination with modeling rewards care in matching the engine and transmission in cases where the engine and transmission are not tested as a combined powertrain unit.

Measured instantaneous fuel consumption data may be used in both whole vehicle modeling to test against a vehicle standard or in simulation of engine cycles to determine compliance of engines with efficiency standards for a variety of vocations. Ultimately, one may see the test cell as an input measurement device for engine and vehicle performance simulation, rather than a technique for gathering cycle-averaged data used in direct comparison against a standard. Current use of steady-state points to produce an efficiency map in GEM is a first step toward such a goal but is insufficient for future use, because the steady-state efficiency and transient efficiency may not be the same, even at the same instantaneous torque and speed. One cause is due to the varied intake and exhaust pressures that may occur at the same torque and speed when the turbocharger is rising or falling in speed, or changing in geometry. Another cause is that the injection control strategy may vary because in-cylinder versus SCR NO_x reduction strategies change during transient operation. Although the argument can be extended that test cell measurements with varying ambient conditions will extend the applicability of the data for real-world operation, these measurements would be costly and onerous.

3.3.2.2 Harmony with Criteria Emissions

A trade-off exists between engine efficiency and engine-regulated (criteria) emissions in most cases, except where reductions in internal energy losses benefit both. Maintaining a strong linkage between regulated emissions and efficiency measurements is important, first because both should reflect reality, and second because it is counterproductive to develop engines that are efficient in one zone of operation and clean in another zone. Currently both are determined in the test cell, using the same transient test and steady-state test points. Whereas instantaneous fuel efficiency and instantaneous pre-aftertreatment criteria pollutants can be measured in a test cell, both urea SCR and the DPF to reduce PM make it difficult to associate tailpipe or stack criteria pollutant levels directly and repeatedly with engine operation.

3.3.2.3 Powertrain Measurement

Integrated powertrain measurement has been presented as an important alternative to engine testing in cases where the engine and transmission have been closely integrated and are controlled as one unit. This is a viable alternative to separate engine and transmission characterization, with subsequent combination of the engine and transmission within a vehicle simulation. In the future it is likely that the engine, transmission, and aftertreatment may be optimized and commonly controlled to the point that operation of the engine alone will not be possible. Characterization of the whole powertrain package is also important in that it will encourage designs with the benefits of a high level of integration. Nevertheless, it is essential that the test cell characterization of the powertrain provides no advantage to the powertrain results other than to acknowledge the benefits of the integrated design and control in the resulting data. Noting the variety of vocations of trucks, it is difficult to select a powertrain test that will compare equitably with separate engine and transmission tests for a range of real-world cycles. Powertrain tests, and tests of a mule (powertrain plus drive axle), are philosophically as valid as engine tests in achieving vehicle efficiency goals, but will require very careful consideration to establish an equivalency. As with engine tests, a powertrain test will have difficulty in providing data on instantaneous tailpipe or stack emissions of criteria pollutants.

In the case of hybrid drivetrains, the instantaneous energy use by the engine does not necessarily correspond proportionally to the energy delivered by the vehicle to the road, nor do the criteria pollutant emissions. Energy

may be drawn from, or stored in, an electric, hydraulic, or mechanical storage system following the direction of a control algorithm. It is possible, from measurement or combined measurement and control simulation, to create a hybrid powertrain model, but the instantaneous fuel demand depends on history of operation as well as the current vehicle activity. Even when a complete hybrid drivetrain is not used, some technology approaches such as the supply of hotel loads through electro-turbo compounding will not allow the engine to be considered in isolation from other integrated components.

3.3.2.4 Findings and Recommendations

Finding: A new technology or design improvement is likely to cause incremental reductions in engine fuel use that challenge the accuracy of the test cell in measuring the difference in fuel consumed and shaft power absorbed. The variability of measurement of fuel use in a test cell is a substantial fraction of the foreseeable reductions in engine fuel use in the future, but there is no clear pathway to reduce this variability substantially.

Finding: Current test cell procedures do not measure engine behavior over all circumstances encountered in the real world, nor are the test schedules weighted to match all real-world combinations of torque, speed, and transient behavior. Even at the same instantaneous torque and speed, steady-state engine efficiency does not necessarily represent the efficiency during transient operation. However, measurement and data processing techniques exist to yield instantaneous fuel consumption models that can be used in the simulation of fuel use over unseen cycles.

Recommendation 3-1: Steady-state maps should be replaced with an advanced predictive fuel flow model, derived from transient test cell data, for use in transient engine, powertrain, and vehicle simulations.

Finding: Currently engines are characterized for both fuel use and regulated pollutant production in the same test cell. If fuel efficiency is determined using simulation and an engine model derived in the test cell, the test circumstances for fuel efficiency and criteria pollutants will differ.

3.3.3 Whole Vehicle Modeling

3.3.3.1 Adequacy and Possible Improvements

Whole vehicle modeling offers substantial advantages over whole vehicle testing in that vehicle design changes and operational differences may be examined rapidly and inexpensively. In addition, a limited period of whole vehicle testing, whether on road or off road, usually fails to capture a range of real-world effects. However, substantial effort is needed to ensure that the model has desired accuracy in predicting real-world performance. The software platforms needed for accurate simulation are already established, but reaching this goal demands complete and sophisticated input data, and the inclusion of all significant components and effects.

Environments that are suited to graphical programming and model-assisted design and optimization are available to support vehicle simulation. Many software packages, private and commercial, are based on these environments and offer predetermined architectures, structures, and information flow to support simulation. The GEM tool is one such package. Errors associated with digital approximation are typically small, and the accuracy of the simulation depends almost entirely on the representativeness of the modules describing each component.

Meeting future fuel efficiency standards will require the use of a combination of strategies and components that reduce fuel use each by a fraction of a percent. As a result, simulation must be able to deal with these small quantities accurately. It will not be satisfactory to develop correction factors for simulation based on whole vehicle tests, because these factors will vary from test schedule to test schedule.

Few components can be modeled so simply that they just require a fixed efficiency, although this is often used today. As an example, an axle incorporating reduction and a differential have bearing losses, rubbing losses, and stirring losses, and the loss model can change with reversal of rotation and torque. Moreover, the viscous losses depend on the lubricant temperature, which in turn depends on ambient temperature and recent history of operation.

Some components can be modeled acceptably with a steady-state map for efficiency, but complex assemblies such as automatic transmissions, engines, and powertrains need acknowledgment of transient behavior.

As an example of the need for an accurate transient model, consider a modern engine that employs exhaust-gas recirculation, injection management strategy, and urea SCR to achieve low NO_x emissions. The strategy is usually different during transient operation than steady-state operation, and the instantaneous efficiency may be lower. Use of a steady-state map would deny the distinction between use of cruise control and a driver who dithers the pedal. Once these engines incorporate systems such as waste-heat recovery, the model will also need to include a factor to describe recent history of operation. Far more complexity is introduced if the model is expected to perform over a wider weather envelope than the present-day temperature window.

The driver model used in the simulation can affect predictions. Real-world drivers differ in gear choice, shifting style, lane control, and pedal operation. A simulation should employ the style of a typical driver so that the true benefits of driver aids, cruise management, automated shifting, and look-ahead technologies, can be quantified.

Hybrid drivetrains are a major pathway to vocational efficiency improvement. It is necessary to simulate hybrid vehicles that include a hybrid drivetrain, rather than relegate them to a physical test requirement, to allow equitable comparison with conventional vehicles and to examine their benefit over a variety of test schedules. These test schedules must be selected with great care, because even the rearrangement of microtrips or duration of grades within a schedule can affect hybrid vehicle performance.

In addition to selecting test schedules that are representative to use in MHDV simulations, the power-to-weight ratio must also be considered carefully, so as not to drive the market toward over- or under-powering of trucks. This involves not only the mass assigned to the truck in the simulation, but also the payload granted to the truck if ton-mile reference is used.

Currently, simulations such as GEM provide an attractive opportunity to determine the fuel efficiency of a range of vehicle designs over different routes. However, the inputs to GEM are relatively simple, such as a single rolling resistance value for tires, and a wind-averaged drag coefficient that is not speed dependent. The engine maps are also steady-state based. When the demand for greater fidelity and the ability to include more innovations increases, GEM will become more complex, and the component performance verification will become critical and substantial. As vehicles become progressively more complex, integrated, controlled, and connected, simulation with sufficient accuracy and attention to detail may become impossible. Not only will the simulation itself become difficult, but the compliant and well-defined collection of component input data will demand excessive resources. It is likely that simulation will remain a strong mid-term solution. However, post-sale in-use data collection, either with or without a level of pre-market simulation, is a likely long-term solution. Therefore, in the future, in-use measurement may replace simulation as a result of the design and operational complexity of MHDVs.

3.3.3.2 Specific Findings and Recommendations

Finding: Accurate simulation of MHDVs relies on the accuracy of the component maps or modules within the vehicle simulation model. Components cannot be modeled accurately using a steady-state approach or limited independent variables if their behavior is complex or depends on operating history of the component or other components.

Recommendation 3-2: Transient maps or advanced models, as opposed to fixed efficiency or steady-state maps, should be employed for components with complex behavior within a vehicle simulation unless it is ensured that the simpler models support acceptable accuracy.

Finding: Selections of test schedules and test weights for a simulation will affect results substantially, and hybrid vehicles will be more sensitive to the ordering of events within a test schedule. The choice of the baseline driver model will also affect simulation results both absolutely and relative to simulations that include driver aids or vehicle automation.

Recommendation 3-3: The baseline driver model used in MHDV simulations should be chosen to represent average driver behavior, so that benefits of driver training, automation, and driver aids are realistically modeled.

3.3.4 Components

3.3.4.1 Tires

Tire rolling resistance represents the substantial portion of the road load for most vocational vehicles, for example, transit buses, package delivery vehicles, and refuse haulers. For vocational vehicles, it is as important to characterize the tire losses as accurately as the engine efficiency. For line haul, rolling losses are still a substantial factor.

Currently tire inputs to the GEM consist of an effective rolling diameter and rolling loss coefficient. The tire’s effective rolling diameter varies over its lifetime, with a change on the order of 2 percent. If modeling is based on wheel rotational speed, the real ground speed will be affected, and for reduced diameter the required engine torque required to propel the vehicle will be reduced. If ground speed is used as a reference, gearing will change over the tire’s lifetime, and engine speed will be affected. In addition, longitudinal slip will cause tires to have a larger number of rotations per unit distance under positive torque, and a smaller number of rotations per mile under negative torque (e.g., during braking). Longitudinal slip of drive wheels will increase as a smaller proportion of the load is on those drive wheels. These effects may appear proportionally small, but so are the fuel efficiency gains needed to meet certification.

Tires do not maintain the same rolling resistance over their lifetime. A new tire generally exhibits the highest rolling resistance, with some of the resistance attributed to tread design. The resistance then declines as the tire ages to a baseline value that is more indicative of the casing materials and design. Tires with the same rolling resistance when new may trend to different values of rolling resistance when worn. Low-rolling-resistance tire casings may be regrooved or retreaded, with further change in rolling resistance.

Tire rolling resistance is affected by the type of road surface, which may affect a truck owner’s choice of tires. Furthermore, tire rolling resistance is not accurately described as a constant fraction of the vertical load on the tire. It varies with that load, with the speed of operation, with temperature, and with tire pressure.

If modeling is employed solely for certification against a standard, the use of a new-tire simple rolling resistance coefficient and rolling diameter may be a sufficient insurance of performance. However, for comparisons of models to real-world operation, diameter change, rolling resistance change, speed and load coefficient dependency, pavement surface effects, and longitudinal slip are all factors that can impact fuel use. An evaluation is needed to see whether use of average values for these parameters is sufficient, and a method is needed for determining or measuring these average values. Furthermore, the method of measuring rolling resistance must yield values that represent real-world use on average pavement, and must be consistent between laboratories and of high quality.

Recommendation 3-4: For certification against a standard using vehicle simulation, it is essential that the rolling diameter and rolling resistance tests are standardized to meet the same data quality level as engine tests, and that they yield results that are sufficiently representative for modeling average real-world performance. For comparisons of simulation data with real-world operation, tires should be treated with a more sophisticated model than a combination of rolling diameter and rolling resistance.

3.3.4.2 Aerodynamics

Aerodynamic losses are the dominant power demand in high-speed trucking and require the same degree of accuracy as a model input as does engine efficiency. However, it is difficult to measure the product of aerodynamic drag coefficient and frontal area (C_dA) reliably and repeatedly. Methods include computational modeling, scaled wind-tunnel testing, coast-down testing, and constant-speed testing. With on-road or track testing methods, the aerodynamic loss must be decoupled from rolling losses, and yaw angle of the wind is hard to control and measure. Moreover, yaw angle causes side thrust on a truck, which is reacted by the steering angle, which in turn causes

elevated tire rolling losses. In a track test these additional losses would be interpreted as aerodynamic yaw angle effects, whereas in a wind tunnel or simulation, they may be neglected. With wind-tunnel testing, scaling issues exist, ground effects may not be well modeled, and simulation of yaw in a tunnel requires correction for lateral force on the truck. Computational methods have difficulty in modeling vehicles without employing calibration from base cases, and require high computing power to support fine grid sizes.

Frontal C_dA does not necessarily reflect performance under yaw conditions, and C_dA does vary as a function of vehicle speed. It is also possible to have the same relative wind speed and yaw angle at different wind conditions and truck speed, so that both truck speed and wind speed (relative to ground speed) and the wind direction (relative to the truck direction) are all important in computing instantaneous drag. Two trucks with the same frontal C_dA value may differ in their values under yaw. Just as grade has been added to time-speed schedules for simulations, so could wind speed and direction to challenge the vehicle performance without simply using an average value. Inability to measure C_dA in an accurate and representative fashion is a substantial barrier to accurate simulation. Additional discussion of real-world conditions that impact aerodynamic drag can be found in the section on the SmartWay program in Chapter 5.

Engines are acknowledged to have efficiencies of the order of 40 percent, so that about 60 percent of the fuel energy is lost. The power supplied by the engine is used primarily to overcome road load. During low-speed operation tire rolling resistance dominates the road load—an effect which increases as tires wear and their performance degrades—and at high speed the aerodynamic losses are dominant. In this way engine inefficiency, tire losses, and aerodynamic losses are all important contributors to fuel use. All merit the same level of attention and measurement accuracy in the quest to improve vehicle fuel efficiency.

Finding: Regulation has focused more upon engine efficiency compliance than separate tire or aerodynamic compliance, even though all three systems are associated with the same order of magnitude of energy in truck operation. If engines are regulated for fuel efficiency in addition to regulation of the whole MHDV, then there is a parallel argument that tires and the aerodynamic body might also be regulated separately in addition to the engine and vehicle. If engines are regulated for fuel efficiency in a manner that acknowledges their complexity, then there is a parallel argument that tires and the aerodynamic body might also be regulated with the same acknowledgment of complexity including the degradation of tire performance as they wear.

Recommendation 3-5: It is essential that the determination of C_dA is standardized to meet the same data quality level and repeatability as engine tests, and that the values of C_dA are available separately at different yaw angles for use in advanced simulations. Where the simulation is used to compare with real-world activity, dependence of C_dA on vehicle speed should be measured or simulated.

3.4 IN-USE COMPLIANCE AND ENFORCEMENT

3.4.1 Need for In-Use Compliance and Enforcement

The purpose of pre-sale certification is to demonstrate that the design of each engine and vehicle model is capable of meeting applicable emission and fuel consumption standards and requirements such as on-board diagnostics (OBD). The purpose of post-sale evaluation of performance is twofold. The first objective is to verify that under real-world conditions vehicles continue to be in compliance with emission and fuel consumption requirements throughout their statutory useful life. Should a specific vehicle model have in-use emissions or fuel consumption that exceeds standards, or have a systematic design defect, NHTSA and EPA may require the specific model vehicle to be recalled and modified to reduce emissions or fuel consumption. The second objective of post-sale evaluation is to provide insight into whether real-world emissions and fuel consumption of the fleet as a whole are consistent with expectations based on regulatory standards. If they are not, NHTSA and EPA may make changes to the pre-sale test and certification procedures so that future engine and vehicle designs perform on road as they do in certification testing.

The importance of a post-sale verification of emission and fuel consumption standards compliance has been demonstrated for light-duty vehicles. Several examples follow:

• In the 1980s and 1990s, in-use compliance testing by EPA of groups of similar passenger vehicles found widespread noncompliance with criteria emission standards. The cost and bad publicity of the many resulting recalls resulted in vehicle manufacturers improving the reliability of emission control systems. The result is that few criteria emission recalls are occurring today, contributing to lower emissions.
• Random testing of in-use passenger vehicles revealed the inadequacy of the certification emission test to represent many modes of operation, and resulted in addition of new pre-sale tests that better predicted actual on-road emissions and fuel consumption.
• Hyundai/Kia recently paid a large fine for not following the proper procedures in determining light-duty fuel economy, which was revealed by in-use compliance testing. This finding resulted in relabeling vehicles not yet sold with lower fuel economy values, and the companies sent refund checks to owners to compensate for the misleading high ratings.
• The International Council on Clean Transportation recently reported on the well-known discrepancy between the certified level of fuel consumption and the measured on-road fuel consumption of in-use passenger vehicles. In Europe the certification test results under-predicted real-world fuel consumption of 2001 models by 8 percent. The surprising result was that the amount of under-prediction has grown year by year, as more stringent GHG emission standards were implemented, to 38 percent for 2013 models (Tietge et al., 2015). In-use testing revealed this trend and may provide the basis for correcting the shortfall with improved certification processes.

These examples from light-duty vehicles demonstrate the importance of an in-use compliance program. While some important lessons for a potential MHDV in-use compliance program can be gained from the extensive experience with the light-duty program, it is important to note that in-use compliance for MHDVs differs from passenger vehicles. For passenger vehicles, the complete vehicle is tested pre-sale for both criteria emissions and fuel consumption. The shortfall between the certification test and real-world fuel consumption is corrected in the fuel economy label attached to each new passenger vehicle. The pre-sale test can be easily replicated in use by borrowing vehicles from owners and testing them at certified laboratories. Emissions and fuel consumption can also be measured on road using portable emission measurement systems. These approaches allow vehicle models with high emissions, or faulty emission control system designs, to be identified, which can lead to corrective action.

For the larger MHDVs,³ only the engine receives a pre-sale emission and fuel consumption test in the Phase I and Phase II rules. The fuel consumption of the complete vehicle in these larger truck classes is determined by the GEM simulation model, based on user inputs including characteristics of the vehicle such as the drag coefficient or rolling resistance, and, in the Phase II rule, on engine fuel maps and torque curves. No actual measurement of fuel consumption of the vehicle being certified is performed. The pre-sale certification approach for MHDVs thus limits how an in-use compliance program is structured in several ways:

• Determining if the engine is in compliance with fuel consumption (or criteria emission) standards requires the engine be removed from a truck and tested on an engine dynamometer. The cost and time involved are very high, and procurement of in-use trucks for engine removal is difficult. Thus such testing is limited compared to testing passenger vehicles.
• Determining if the complete truck (Class 4 and above) is in compliance is limited to verifying the inputs to the GEM model, since there is no pre-sale vehicle GHG emission or fuel consumption test to duplicate in use. At best, in-use testing using on-truck emissions or fuel consumption measurement devices can identify suspect groups of vehicles with high GHG emissions or fuel consumption, but regulatory action to remedy the problem, such as recall, is likely limited to engine removal for testing, or validation testing of inputs to the GEM simulation model.

___________________

³ Class 2b and 3 vehicles are usually certified using a chassis dynamometer.

Table 3-3 compares the current certification and in-use compliance requirements for passenger vehicles and MHDVs.

For these reasons, an in-use compliance program for MHDVs patterned after the in-use compliance program for passenger vehicles is not likely to be practical or successful in providing confirmation that MHDVs are meeting standards and requirements. Thus, a different approach is needed to provide assurance of in-use compliance.

Finding: An effective method of determining in-use compliance of trucks, important for the overall effectiveness of the regulatory program, does not currently exist. One is necessary to ensure that individual vehicle models are meeting fuel consumption and GHG emission standards (as well as criteria emission standards), and to identify opportunities to improve the effectiveness and reduce the cost of the program.

Recommendation 3-6: NHTSA, in coordination with EPA, should develop an effective in-use compliance method that would allow the overall performance of the regulatory program to be quantified, identify whether groups of in-use trucks may not be in compliance, and provide insight into truck operating conditions where GHG emissions and fuel consumption of future trucks could be further reduced.

3.4.2 Approaches to Determine In-Use Compliance for MHDVs

In general, the objective of an in-use compliance program should be to determine if real on-road emissions and fuel consumption match the results expected from the regulatory standards and requirements. It should also identify groups of trucks that appear to have higher emissions and fuel consumption than expected, in order to implement corrective action. In a broader sense, it should provide data that suggest how the regulatory program could be made more effective (i.e., the concept of continuous improvement).⁴ Depending on the comprehensiveness and effectiveness of the in-use compliance program, it may complement, or could replace, the pre-sale certification program. Several in-use compliance concepts that meet some or all of the above objectives are discussed next.

___________________

⁴ See Recommendation 2-5 on the need for NHTSA to conduct an interim assessment of the regulations.

3.4.2.1 On-Board Diagnostics and Warranty Reporting

Modern passenger vehicles are required to have an on-board diagnostic system that is closely correlated to criteria emissions such that a warning light and diagnostic code is set whenever emissions increase by about 50 percent above the emission standard. In the past few years, MHDVs have also become subject to OBD requirements, although the sensitivity of the OBD system to identify problems with particle filters and exhaust catalysts on diesel engines currently lags far behind the effectiveness of gasoline vehicle OBD. It may be possible to develop OBD monitoring for MHDVs that is more closely tied to GHG emissions and fuel consumption, and is sufficiently comprehensive.⁵ If successful, the OBD data could be transmitted electronically from the truck to the truck manufacturer and/or NHTSA and EPA, to assess in-use performance of GHG emissions and fuel consumption, as well as criteria air pollutants, both for individual trucks and groups of trucks as a whole.

Engine manufacturers are currently required to report warranty repairs to NHTSA and EPA, and these reports have been used to identify systemic defects in engines. This reporting system could be strengthened to more directly become a means of recalling engines (and could be extended to complete trucks) that show a pattern of emissions and fuel-related component failures, especially for components whose current OBD monitoring capabilities are limited. The emission warranty period, currently only 100,000 miles, would need to be extended as well, perhaps to the useful life (as much as 435,000 miles), since 100,000 miles is often driven in the first year of a Class 8 truck’s life.

A combined OBD and warranty reporting program could serve as a somewhat limited in-use compliance program. It would meet the objective of identifying groups of engines that are exceeding emission and fuel consumption standards, and cause their repair. However, it would not alone provide a quantitative assessment of on-road emissions, nor would it identify the need to make program improvements that would address high off-cycle emissions or fuel consumption. These limitations exist because pre-sale development of the OBD monitoring system is primarily tied to sensor signals generated during a standard test procedure, not to the direct measurement of emissions or fuel consumption.

3.4.2.2 In-Use Test Cycle

NHTSA and EPA could adopt an in-use test cycle representative of real-world driving that could be easily replicated on an in-use truck equipped with standard portable emission measurement instruments. The cycle could be added to the GEM. The regulation would specify that in-use compliance will be based on the results of on-road tests from a sample of in-use trucks. The tests could be conducted under any conditions, such as temperature or load, which the trucks might normally experience. It would be the truck manufacturer’s responsibility to ensure that its trucks comply with the standards under a broad range of conditions for the specified in-use test cycle. This approach would partially meet the first objective of determining real-world emissions and fuel consumption (outlier trucks such as those operating at extreme conditions or cycles would not be reflected in on-road emission assessments). It could also identify groups of similar trucks with high emissions and fuel consumption, but may not identify specific areas for program improvement since testing would still involve a single defined test cycle.

3.4.2.3 Compliance Based on Actual On-Road Emissions

Modern trucks use sensors and an on-board computer to determine the needs of the engine, such as the amount of fuel needed to achieve required performance. Since fuel consumption is one of the metrics of the standards adopted by NHTSA and EPA, it is plausible that fuel use, or possibly directly measured CO₂ if a low-cost CO₂ sensor could be developed, could be collected from every truck on a real-time basis, and stored and periodically reported to the agencies. Other information such as speed can also be collected and, for many fleet operators, is currently reported electronically for operational analysis. A regulatory requirement that such data be reported would meet the objectives of quantifying real-world emissions and fuel consumption for comparison to what was

___________________

⁵ For a discussion of the technical issues in collecting real-world fuel consumption data from OBD systems, see NRC (2015, p. 349).

expected from the adopted standards and requirements, and could also identify groups of trucks that appear to be performing abnormally. It might also provide insight into how to improve the regulatory program, such as by identifying modes of operation where GHG emissions and fuel consumption are higher than expected, which could lead to improvements in the pre-sale certification protocol.

A more significant change could be to require the truck builders to be responsible for the on-road GHG emissions and fuel consumption of the trucks they sell. Truck manufacturers could select a standard they will meet, subject to a regulatory maximum, with in-use compliance based on on-road data subsequently collected on their trucks. Truck manufacturers know how most of their trucks will be operated (based on their sophisticated simulation models) and can reflect in their selection of a standard the fraction of vehicles that operate at loads, speed, altitudes, etc., that differ significantly from the average. This approach provides a greater risk of noncompliance to the truck manufacturer than does the current certification approach that has minimal risk once the vehicle is sold. However, if in-use compliance is based on a large group of their vehicles, such as all of a manufacturer’s vocational trucks, those trucks that are outliers may not have a large effect on the average. To further reduce the risk to the truck manufacturer, the initial years of implementation could be phased in with minimal penalties until experience with the in-use data is gained. Furthermore, NHTSA and EPA could impose monetary nonconformance penalties, as currently set forth in the Clean Air Act, in lieu of the threat of a recall, especially where defective design of components is not involved.

This approach significantly differs from current in-use compliance approaches. It places the responsibility for actual in-use emissions and fuel consumption on the truck manufacturer, increases the risk of noncompliance because responsibility would include all operational conditions, some of which may not be precisely known by the manufacturer, and raises numerous issues which need to be assessed, such as knowing the load on the vehicle, processing large amounts of data, and many more. However, it is the only approach that fundamentally meets the criterion of providing assurance that actual, real-world emissions and fuel consumption are achieved, and thus is included in this list of possible in-use compliance approaches.

Table 3-4 summarizes how each possible in-use compliance approach meets the three objectives listed at the beginning of this section.

The in-use compliance programs discussed above are concepts designed to illustrate the degree of change and need for new thinking in developing an effective in-use compliance program for MHDVs. This is in contrast to the current, limited in-use compliance program which is based in part on the program structure used for passenger vehicles. As the color coding in Table 3-4 illustrates, the program concept involving the most significant change meets all the objectives of an in-use compliance program, whereas the OBD/warranty reporting concept, which relies on improving existing regulatory provisions, only addresses one objective.

Finding: With at least 5 years available before a notice of proposed rulemaking (NPRM) for possible Phase III standards might be issued, EPA and NHTSA have sufficient time to explore potential in-use compliance concepts and approaches. The available time should allow stakeholder involvement in a deliberative process, before the Phase III NPRM is issued, to establish the objectives to be achieved, and assess and develop an in-use compliance concept that best meets the objectives. The method should provide sufficient granularity to offer insight on how

the GHG and fuel consumption reduction program can be made more effective, and to identify potential groups of trucks that appear to be failing to achieve standards, and why.

Recommendation 3-7: NHTSA, in coordination with EPA, should undertake studies and establish a dialog with stakeholders to identify the most feasible method of determining on a continuous basis the effectiveness of the MHDV fuel consumption and GHG emission program, including in-use compliance. The agencies should also evaluate if actual in-use emissions and fuel consumption could become the principal metric for determining compliance, including the costs, complexities, and benefits of such an approach, and if implemented which regulated entity should be responsible for complying with emissions in use based on considerations of practicality, feasibility, and fairness. NHTSA and EPA should also determine whether a compliance program based on in-use emissions could supplant the need for engine emission test and simulation model specific standards now in place, and if in-use compliance could be extended for the full life cycle of vehicles.

3.4.3 Program Effectiveness

The OBD and in-use compliance test cycle approaches discussed above do not alone provide for a comprehensive database on which to assess program effectiveness and identify ways of improving program effectiveness. However, these two approaches could be supplemented if NHTSA and EPA undertook a comprehensive and continuous program of collecting actual on-road truck GHG and fuel consumption data. Such a program would have to test a large sample of trucks under varying conditions each year, and include pre-control vehicles for a baseline, as well as vehicles subject to the Phase I and Phase II standards and requirements. The committee has recommended in prior reports (NRC, 2010, 2014) the importance of establishing baselines of fleet fuel consumption and GHG emissions to allow assessment of program effectiveness. When combined with an in-use compliance program more removed from actual on-road emissions, more of the objectives could be met.

Finding: Comprehensive and ongoing testing of in-use MHDVs can provide data necessary to evaluate the effectiveness of the MHDV program and to help develop an effective in-use compliance program.

Recommendation 3-8: NHTSA and EPA should commit resources to collecting real-world GHG emission and fuel consumption data for a large and representative national sample of pre-control trucks and for each model year subject to the Phase I and Phase II standards, with priority given to those categories of trucks with the greatest fuel consumption. These data could be used to assess the effectiveness of the Phase I and early Phase II standards. If the resources for collecting these data exceed that available to the agencies, the agencies should consider requiring the regulated industry to provide the needed data or fund its collection.

3.4.4 Individual Vehicle Compliance

Many states require passenger vehicles to undergo periodic (usually biennial) smog inspections as a condition of re-licensing. Older cars without advanced OBD typically get a dynamometer exhaust test for criteria pollutants. Newer models get an electronic OBD check.

Typically, larger trucks are not included in these periodic tests, for a number of reasons which include the logistics of inspection due to the size of trucks, and the interstate travel and multistate registration of Class 8 highway trucks and trailers which complicates state enforcement based on licensing. With OBD available on most 2013 and newer diesel MHDV engines, roadside OBD inspections of trucks (such as at weigh stations near urban areas) are possible. These OBD inspections would be directed at identifying vehicles with excessive fuel consumption rates and/or high CO₂ emissions. Roadside inspections could also verify that truck and trailer aerodynamic devices are present and low-rolling-resistance tires are being used (when the Phase II trailer regulation is implemented).

Finding: Periodic inspections of the OBD system on 2013 and newer in-use MHDVs could identify individual vehicles with excessive fuel consumption and/or high CO₂ and criteria emissions.

Recommendation 3-9: NHTSA and EPA should evaluate the on-road data they are collecting (see Recommendation 3-8) and determine if there is a need for periodic inspections of individual MHDVs to ensure fuel consumption and GHG emissions are not excessive (and as a co-benefit criteria emissions remain low). If a need exists, the agencies should develop recommended inspection practices and work with states to establish inspection programs.

3.5 REFERENCES