Office of air quality management

Since both Ford and General Motors, as well as numerous other manufacturers already have available 6 speed automatic transmission manufacturing plants, there would be no need for additional investment resources for building new plants or redirection of new AMT capacity outside of North America. Numerous transmission suppliers such as ZF or Aisin will have such units available for purchase by manufacturers that would prefer not to make further investments in transmission plants. Costs for 6 speed automatic transmissions in our study were representative of such suppliers. Such costs include the transmission suppliers' costs for development and investment in facilities for building the units in large volumes to supply a number of manufacturers.

• 
  • The projected benefit for the turbocharged, downsized, direct-injected, cam-phasing engines is based on very aggressive assumptions about the specific output that is possible for these types of engines. The most unlikely of these assumptions is that the engines will use premium fuel instead of regular fuel. All of the AVL analysis for these engines is based on premium fuel. Without premium fuel, the specific output possible from these engines will be significantly reduced and the engine sizes will be overly optimistic due to selection of very low engine displacements driven by unrealistic Brake Mean Effective Pressure (BMEP) curve assumptions that depended on high boost levels and premium fuel usage.
  
• 
  • Typical turbocharger installations require an intercooler, which increases vehicle drag.
  
• 
  • There are significant discrepancies between the benefits projected by AVL for downsized turbocharged MPFI engines and downsized turbocharged GDI-S engines. AVL has indicated through a direct comparison of turbocharged MPFI versus turbocharged GDI-S DCP engine maps that engine fuel consumption differences between these two technologies are as much as 12% at typical Federal Test Procedure engine operation conditions. Such large differences in fuel consumption are unexplained by the relatively minor physical differences between the engine technologies. This discrepancy affects a technology package used to justify the emission standard in four of the five vehicle classes.
  
• 
  • AVL has confirmed that the application of aggressively downsized turbocharged engines did not include consideration of vehicle launch, drive quality, and transient engine/transmission/turbo response. The simulation results provided by AVL indicate that the vehicles configured with these engines will have serious drive quality problems. General Motors believes such deteriorations in performance are not acceptable, and they demonstrate that not enough verification of “equal performance” was done. Demonstration of sufficient vehicle launch, drive quality, and transient performance should be required prior to consideration of this and other “torque­ modifying” new powertrain technologies.
  

While addition of an intercooler may affect vehicle drag slightly, there are numerous other refinements to vehicles still available to restore the original aerodynamic drag or to further improve it. These include underbody shields, revised grille openings, improvements in wheelhouse configurations and many more.

Regarding the greater than expected benefits between GDI-S turbo and MPFI turbo, one of the biggest benefits is an approximate 10% increase in output from the GDI-S turbo. At constant 0-60 mph time, the higher specific output allowed a slightly larger downsizing of the GDI engine and a longer axle ratio, which together result in reduced CO2 emissions. GDI-S turbo technology allows a significantly higher compression ratio (2-3 ratios), which is very beneficial for reducing CO2 emissions at part loads. Finally, GDI combustion technology increases burn rate, providing improved thermodynamic efficiency and again lower CO2 emissions. It is important to note that the engine data and maps used for this project came from actual engines developed by AVL.

156. Comment: Portions of the analysis done by AVL appear to have included the assumption of premium fuel usage. AVL states that regular fuel was assumed for all of the engine configurations that used some form of variable vale actuation, but engine specific output levels taken directly from AVL output results match exactly with other premium fuel AVL work on variable valve actuation. Further investigation of this issue by AVL indicated that in most, but not all, cases their assumptions fell within very aggressive regular fuel specific output levels. Whether through an assumption of premium fuel usage or an overestimate of what is possible with regular fuel, the result is an over-estimate of the specific output possible with each of these technologies, which enables unrealistically aggressive engine downsizing – and fuel consumption reductions – to be simulated while maintaining equal performance. This discrepancy contributes to an over-assumption of the specific output capability (and thus the chosen engine size) of every DCP, DVVL, and CVVL engine in the AVL analysis. (General Motors)

Agency Response: Staff disagrees with the comment. AVL has considerable experience with turbocharged engines, both port fuel injected and direct injected. They have access to the engine maps associated with such engines in Europe as well as their own maps from engines actually developed by AVL. These maps are utilized by the CRUISE model to obtain their emission reduction estimates. Since GM has only one or two vehicle models that are turbocharged and they do not appear to incorporate direct injection, their limited experience may result in being unaware of improvements that have been incorporated in the most modern engines of this type. When questioned further about the GM comment, AVL maintained they were confident in their assessment that the results could be obtained using regular fuel.

157. Comment: The staff report’s analysis assumes the use of regular unleaded gasoline for technologies that require more expensive, higher-octane gasoline; once this inconsistency is corrected, those technologies do not provide the assumed economic benefit. (Alliance of Automobile Manufacturers)

Agency Response: Staff disagrees with the comment. The technologies do not require more expensive higher-octane gasoline. See response to Comment 156.

158. Comment: The AVL study used a computer simulation tool and consistent methodology. However, AVL has described their study as a generic study whose results can be used to compare relative differences between groupings of technologies, not for projecting specific consumption targets for specific vehicles. As a generic study, the AVL work did not cover some important details and constraints that are a reality for vehicle manufacturers:

0. 
   • All of the engine maps used in the simulation study were based on AVL’s most optimistic, upper-limit projections of the full capability of the engine technologies, assuming full application of technology without sufficient constraints which reflect real-world combustion system dilution tolerance, airflow capacity, piston-to-valve clearances, oil system capacity at low speeds, idle speed control techniques, and noise vibration and handling (NVH) concerns. The AVL engine maps assumed a best case for all of these aspects of engine design, and in several cases their “best-in-class” results were a smoothed composite of results from multiple engines – no individual engines represented the engine maps used for setting the standards. A study like this does not provide a quantitative target value that is suitable for setting fuel consumption regulations. The maps used by AVL to represent DCP, CCP, DVVL, and CVVL all had significant fuel consumption improvements at light loads where, in the real world, the improvements would be limited by combustion system dilution tolerance versus airflow capacity tradeoffs and by piston-to-valve clearance constraints.
   
• 
  • AVL has indicated that all of the vehicle/powertrain configurations chosen for the standard were chosen to maintain equal performance. However, seven of the ten configurations used for setting the near-term standard have worse 50-70 mph performance than their baseline cases; four of those cases (large truck 04, large truck 05, small truck 04, and minivan 04) are significantly worse and would be considered unacceptable when compared to the baseline.
  
• 
  • AVL did not consider any gradeability or drive quality metrics when choosing engine sizes. In nine of the ten configurations used for setting the near-term standard, the gradeability calculated by AVL was worse than the baseline gradeability; five of those cases (large truck 04, large truck 05, small truck 04, minivan 04, and minivan 05) showed significant degradation in gradeability to the point where they would likely be considered unacceptable. AVL made no explicit calculations concerning drive quality (the typical response to accelerator pedal inputs required by the driver) so it is impossible to quantify the impacts. Drive quality issues are frequently prevalent when the calculated gradeability is poor and when aggressive engine downsizing is attempted, so it is expected that there would be drive quality problems with several of the chosen configurations. Since the standards set by ARB were almost entirely based on configurations where drive quality problems are likely to occur, the standards should not be considered feasible unless more analysis validating acceptable drive quality is performed.
  
• 
  • The method used by AVL to input transmission shift patterns and torque converter lock patterns was explicit and well defined. However, the actual shift patterns and lock/unlock patterns were not chosen in a reproducible, consistent manner. There was no explicit test of the shift points to ensure that they were not too early (which would hurt drive quality, cause shift busyness problems, and exaggerate fuel economy benefits) or too late (which would help drive quality at the expense of fuel economy), and there was no tabulation of the number of shifts per test cycle (usually accepted as a fair indicator of shift busyness).
  
• 
  • The method used by AVL to adjust their baseline simulations to actual test vehicle performance and fuel economy results was to first “tweak” drivetrain efficiencies to dial-in vehicle 0-60 performance, and then “tweak” transmission shift and lock patterns to dial-in vehicle fuel economy. While a method such as this might produce a simulated fuel economy number that equals the test data, it does not result in a reliable baseline simulation. If, for example, the quoted engine power for the baseline engine was higher than actual (resulting in a “fast” 0-60 simulation result), the AVL method would artificially reduce the baseline drivetrain efficiency to match performance. Then, in order to match fuel economy numbers (assuming everything else about the simulation is in order), the AVL method would have to artificially make the shift/lock points too early. The result would be a baseline simulation result with unrealistic drivetrain efficiencies and shift/lock points.
  
• 
  • Given the observed degradations in gradeability and the well-defined but unvalidated transmission shift/lock methods used, it is inappropriate and overly optimistic for ARB to assume in Table 5.2-4 that all vehicles would benefit from additional aggressive shift logic and early torque converter lockup. The ARB report states that “drivability and acceleration concerns must be accounted for carefully in these alterations of shifting schedules.” This is true, but it was not done by AVL or ARB. The ARB report states “…care must be exercised to ensure smooth, responsive drivability and low noise, vibration, and harshness. AVL was conservative in its modeling of these features to ensure good drivability and minimum vibration.” As described above, no systematic aggressiveness test was performed. The Table 5.2-4 adjustments are not justified. ARB had access to a full-featured simulation at AVL, but chose not to use simulation results, instead multiplying an unsimulated, unrealistic adjustment by the AVL results.
  

In summary, this analysis runs a very high risk of overestimating benefits and underestimating costs by applying multiple new technologies that can have unexpected effects in combination, usually resulting in identification of additional constraints. This problem is compounded by the use of technologies that are still early in the development stage, which might not develop to fruition and which cannot be modeled with precision. (General Motors)

Agency Response: Staff disagrees with the comment. Concerning the first point, the modeling team used “mule” vehicles that represented the individual class norms and then made adjustments to them so that comparisons between various technologies could be made. A completely idealized vehicle could have been developed based on class attributes, but it was concluded that using a mule vehicle and adjusting its specifications would provide more identifiable results for setting greenhouse gas emission reduction standards.

Regarding the use of optimized engine maps, AVL used blended engine maps for only a few of the more traditional technologies. The blended maps essentially represented best in class attributes that could be achieved through good engineering practices. In practice, an individual engine might not achieve these results because of design compromises as can be seen by the scatter of results from similar engines in production. As a result, AVL acknowledged the ability to predict the results for a particular engine might be off slightly but typical engines should behave similarly. As a practical matter, ARB staff did not use any of the very few runs estimating the benefits of advanced technologies where a blended map was utilized by AVL, because the particular cases reliant on them did not incorporate enough new technology to improve CO2 emissions significantly. Therefore, this concern does not even apply to the ARB results.

Regarding performance characteristics, AVL focused primarily on achieving comparable or slightly better 0–60 mph performance than the baseline 2002 vehicle in the class. In the study, the 50-70 mph acceleration task was chosen to be an elasticity test, i.e., maximum acceleration with the transmission locked in top gear (and the torque converter locked as well). This gives a good indication of the engine’s full load torque curve but does not represent normal driving operation. In reality, the passing situation would yield a downshift for vehicles equipped with automatic or automated manual transmissions and unlocking the torque converter on the automatic transmissions. Generally when allowing for a normal transmission downshift and unlocking the torque converter, performance is only slightly worsened compared to the baseline acceleration task. This was deemed an acceptable tradeoff in order to avoid selecting a larger engine than necessary with its associated higher CO2 emissions. For example, in modeling the large truck case 04, the torque converter was not allowed to unlock during full throttle acceleration, to get a better comparison to automated manual transmissions. The unlocking of the torque converter allows the engine to speed up and develop more power, similar in effect to a downshift. The large truck case 04 was rerun in response to GM’s comment to model allowing the torque converter to unlock, as would be the case in a production vehicle. The resultant passing time increased by only 7% compared to the baseline case. The remaining discrepancies are due to a slightly lower overall transmission ratio as well as differences in engine full load curves and auxiliaries.

For gradeability, a situation very similar to the passing task is present. For the case of large truck case 04, the torque converter is locked at 50 mph in top gear. If the converter is unlocked, reflecting what would normally happen in use, the gradeability becomes almost the same as the baseline case (8.21% vs. 8.71%). Nonetheless, when comparing a 4 speed automatic transmission baseline vehicle with a six speed transmission vehicle simulation, of course top gear gradeability will be lessened when utilizing a transmission with a tall overdrive ratio for top gear – that is the one of the primary benefits from a higher number of gear ratios in a six speed transmission since it provides lower CO2 emissions. But in examining the model runs cited as inadequate, staff found that for the same engine rpm, the ability to climb a grade was generally the same or better for the advanced scenarios, not worse (even without considering unlocking the converter, where applicable). What this means is that instead of climbing a steep grade in top gear, the advanced vehicle would normally downshift to fifth gear and would then have generally better grade climbing ability than the baseline vehicle in top gear, based on the modeling results. Similarly, when engine downsizing is coupled with a transmission possessing more gear ratios, engine drivability or responsiveness is generally preserved since more gear ratios are available to better match the engine speed and torque output to a given driver command. GM’s competitors generally utilize smaller engine displacements and transmissions with more gear ratios that automotive reviewers seem to prefer in terms of responsive performance and powertrain refinement.

Finally, regarding the comment addressing AVL’s methods in adjusting their baseline simulations to actual test vehicle performance and fuel usage, AVL countered that their baseline model correction is dependent on the accuracy of performance data published by industry. The hypothetical case mentioned is possible if inaccurate data is provided. To avoid errors such as suggested, AVL confirmed drivetrain efficiencies with industry experts and verified that the chosen transmission shift and converter lock-up schedules were realistic compared to calibrations for similar production vehicles in the market place.

159. Comment: Integrating fuel economy technologies into a vehicle is a very complex process that almost always results in some benefit deterioration for each application of a fuel economy technology. Vehicle integration involves a balance of all the vehicle attributes. This balancing often results in fuel economy benefits of a technology described in the public literature, by component suppliers, or produced by sub-systems simulations being significantly reduced when the technology is actually integrated into the vehicle. A major reason for ARB’s overestimation of vehicle fuel economy potential is a disregard for this critical issue. (General Motors)

The NESCCAF study specifically intended to model combined technologies in a manner that preserved attributes that customers expect in new vehicles. Accordingly, every effort was made to consider noise, vibration, harshness, performance and other considerations. While staff agrees that unforeseen engineering issues may arise when trying to integrate several new technologies in a vehicle, the solutions need not reduce the projected benefits. For example, when incorporating cylinder deactivation in the Chrysler 5.7 liter Hemi engine, additional work was needed to reduce low frequency noise present during deactivation modes. Their solution was to incorporate a redesigned muffler and exhaust pipe connector that didn’t affect the CO2 reduction benefits of the technology. Honda encountered similar issues when designing the cylinder deactivation system in the Odyssey V6 minivan, but overcame engine vibration and noise issues by incorporating active engine mounts and noise cancellation technology. The engine also incorporates electronically controlled variable valve timing and lift seamlessly. Neither of these solutions would affect the CO2 reductions that could be achieved with these systems. On the other hand, General Motors tried to solve similar problems by adding an expensive control valve in the exhaust system to attenuate certain frequencies that could have affected performance and the CO2 reduction benefits of cylinder deactivation. That approach could also introduce durability/reliability concerns at higher mileage. General Motors apparently witnessed the better approaches and apparently delayed introduction of cylinder deactivation until they designed a better solution. When General Motors finally introduced cylinder deactivation more than a year later than they first announced, it incorporated attenuation features very similar to the Chrysler approach rather than GM’s initial approach and retained all of the potential benefits of the technology. Staff therefore expects that automotive engineers working with advanced concepts and electronic technologies currently available can solve unforeseen engineering issues while retaining the benefits of the basic approaches.

160. Comment: Four of the ten vehicle configurations used to set the near-term standard were combinations of OHV engine technologies that are unlikely to be applied in the real world. Minivan 04 applied CVVL along with CCP. Small truck 05, large truck 04, and large truck 05 all applied DeAct plus DVVL plus CCP. The application of either CVVL or DVVL to OHV engines is not realistic as the mechanisms that might provide such function (especially in combination with DeAct and CCP) do not exist and are not being considered for development. Two major roadblocks preventing the combination of these technologies are (1) the fact that DeAct technology already uses a dedicated valve lifter and lifter housing that would preclude adding a new mechanism in the lifter valley and (2) the strict packaging requirements currently met by OHV engine designs would be violated if a large new CVVL or DVVL mechanism were added to the top of the cylinder head. Because these technology combinations have not been demonstrated in any realistic form, they violate the statement by ARB that “the technologies being explored are currently available on vehicles in various forms or have been demonstrated by auto companies and/or vehicle component suppliers in at least prototype form.” (General Motors)