1. Institute for Internal Combustion Engines and Thermodynamics
Analysis of ACEA proposals for thresholds on kinematic parameters and on other boundary values S. Hausberger G. Silberholz S. Lipp
Stefan Hausberger
RDE test procedure 1

2. CONTENT
Data sources and results for
vehicle mileage shares urban/road/motorway
Road gradient shares in vehicle mileage
Average velocities
Idling time
Analysis of 95 percentile of v*a>0.1 as kinematic parameter

3. Sources used HBEFA, TREMOVE and national traffic data
Collection of PEMS tests on LDV
PEMS trips with 18 different LDV
Most of them driven by different drivers and on different routes
Total 857 trips (approx. 5% not useful)

4. Option for higher Urban weighting
Vehicle mileage shares
 % km
TREMOVE Data
CLEAR proposal 2014
ACEA Position 2015
Option for higher Urban weighting
Urban
26%
28%
30%
Road
59%
47%
44%
40%
Motorway
15%
25%
Sum
100%
 ACEA proposal supported by available data.
30% urban may reflect importance of urban air quality better
Classification by map important for understanding of results (shown later)
Details: TREMOVE data seems to heavily overestimate rural mileage:
region -->
GER
AUT
EU 27
CLEAR option
Source
FLEETS for 2005
HBEFA for 2012
TREMOVE 3.3.2
National
alternative V3
Urban
29.0%
28.5%
26.0%
28.8%
20.7%
26.5%
28%
Road
43.9%
39.9%
53.8%
37.2%
67.0%
58.7%
47%
Mortorway
27.1%
31.7%
20.3%
34.0%
12.3%
14.8%
25%

5. Road gradients  ACEA proposal supported by available data:
Germany
CH&AUT Fr
-6%
1.1%
3.8%
0.5%
-4%
4.6%
7.4%
2.6%
-2%
19.9%
21.6%
13.0% 0%
80.1%
78.4%
87.0% 2%
95.4%
92.6%
97.4% 4%
98.9%
96.2%
99.5% 6%
100.0%
 ACEA proposal supported by available data:
95 percentile of gradients < 3.5%

6. Average velocities Average velocities in PEMS trips analysed at TUG
Urban
Road MW
Urban 20-40km/h for vavg is ok with existing PEMS data.
If very dense urban traffic shall be included 15km/h can be argued as minimum (but dense traffic could be tested also e.g. with and
Motorway average velocity may be set > 120 km/h (e.g. 125km/h), vavg on AT highways = 118 km/h

7. Driving Dynamics How to obtain representative thresholds?
Analysis performed:
All trips from all vehicles used, approx. 5% eliminated (e.g. trips with jumps from speed signal)
Several classifications remain questionable, e.g.:
How to obtain representative thresholds?
Figures use a = (v(i+1) – v(i-1) )/(2x3.6) in all cases

8. Analysis of “normal conditions”
“95Perc. (v*a), a>0.1” from former picture sorted according to v*a
Findings:
Several driver interpret „aggressive“ rather as fast than as driving with frequent accelerations; similarly „Eco” as slow driving. Thus we see overlapping in the styles. rating of style in db is very subjective.
Data set from „Driver selected“ covers v*a values up to the 75% percentile of „intentional aggressive driving“. More dynamic driving obviously is found only in some cases when the driver was asked to use such a style.
The „Driver selected“ has 2 discontinuities at high v*a values which seem to differentiate between „moderate aggressive“ and „advanced aggressive“ style. The letter has approx. 10% of total values and starts above 24 m²/s³ in rural and motorway parts.
If we assume the „ advanced aggressive“ style not to be relevant for „normal driving definition“ according to the legislation, we can use these discontinuities or the 90percentile to define upper limits for v*a.

9. Options for thresholds for Driving Dynamics
Using the 90 or the 95-Percentile of the distribution of all driver selected trips seems to eliminate only a few outliers.  linear equations through these points.
Results show higher levels than ACEA proposal. The evaluation method applied for the WLTP data base by Heinz Steven gives approximately the average P95 (v*a) value of the 1Hz data from WLTP.
Option e.g.:
Extra urban:
11<95P(v*a)<24
Urban:
Define 2 lines: y= k*x + D
Figures use a = (v(i+1) – v(i-1) )/(2x3.6) in all cases

10. Alternative parameters to define lower threshold
Background:
Driving 5.1% of the time with (v*a) above threshold would be sufficient to pass 95 Percentile.
Option a) adjust percentile value to time shares <0.1 m/s² (see slide 13)
Option b) use RPA as parameter to define minimum average positive acceleration work
RPA separates aggressive and normal less pronounced than 95Perc (v*a)
Defining lower threshold for “normal driving” seems to be possible.
Attention: RPA calculated from a = (v(i) – v(i-1) )/(3.6)  approx. 10% higher results than with a = (v(i+1) – v(i-1) )/(7.2)

11. Alternative parameters to define lower threshold
Comparison RPA and 95 Percentile (v*a>0.1)
A trip with many lower accelerations lead to same RPA than a trip with fewer strong accelerations per km. Trips with stronger accelerations (if >5% of time) always lead to ahigher 95 percentile (v*a)
 RPA and 95 Percentile (v*a>0.1) do not correlate well
Figure shows one mark per trip part (urban, road, MW separated)
Attention: RPA calculated from a = (v(i) – v(i-1) )/(3.6)  approx. 10% higher results than with a = (v(i+1) – v(i-1) )/(7.2)

12. Option for variable v*a+ percentile value for MW to correct for permanent cruise control operation
definition of applicable percentile value for motorway part according to time share of positive accelerations (idea from Andreas Bittermann)
definition based on analysis of existing motorway trips
specific analysis with designed generic motorway trips necessary for validation of suggested definition
Bittermann version:
percentile = 1 - (0.015 / timeshare_a+)
(percentile = 95 for timeshare_a+ > 12.5%)
TUG version:
percentile = 9.5 * timeshare_a
(min = 0%; max = 95%)

13. Examples of NOx emissions EU6 #01, JRC test at EURO 6 diesel car without NOx aftertreatment
Measured NOx correlated with R² = 0.73 to 95 percentile of v*a<0.1 After CLEAR evaluation the correlation as well as the slope is lower but not eliminated
Note: SCM routes not shown, they have different trends

14. Examples of NOx emissions EU6 #02, JRC test at EURO 6 diesel car with SCR
Measured NOx correlated with R² = 0.48 to 95 percentile of v*a<0.1 After CLEAR evaluation the correlation as well as the slope is lower but not eliminated
Note: SCM routes not shown, they have different trends

15. Examples of NOx emissions EU6 #18, TUG test at EURO 6 diesel car with NOx storage catalyst
Measured NOx correlated with R² = 0.69 to 95 percentile of v*a<0.1 After CLEAR evaluation the correlation as well as the slope is lower but not eliminated