Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of...

20
The European Commission’s science and knowledge service Joint Research Centre Revision of RDE evaluation methods

Transcript of Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of...

Page 1: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

The European Commission’s science and knowledge service

Joint Research Centre

Revision of RDE

evaluation methods

Page 2: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

2

Objective of the revision

• In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased driving must be identified and

excluded) trip validity

• Establishing an emission calculation method that adjusts emissions to specific

trip work trip normality

• Evaluation methods should be: • Robust: properly differentiate normal from abnormal driving

• Unbiased: Always calculate more emissions when the trip work is higher

• Technology neutral

• Simple to understand and to use

• Should not invalidate valid trips (reduce test burden)

• Objective: understand how well/bad the current methods + some new

alternatives perform on a set of real PEMS tests covering a wide range

of vehicles and driving conditions input RDE4

Page 3: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

3

Boundary conditions (BCs) Annex IIIA,

Appendix 7a & 7b

- Trip composition: trip duration [90’-120’], distance per phase [> 16 km],

distance share per phase [U:29-44%, R:23-43, M:23-43], idle duration after 1st

ignition [< 15’’], vmean urban [15-40 km/h], v motorway [5’ above 100 km/h, <

3% above 145 km/h, vmax < 160 km/h]

- Temperature range: moderate/extended or beyond

- Altitude range: moderate/extended or beyond

- Altitude gain: cumulative elevation gain for total and urban [< 1200m /100 km],

start end trip altitude difference [<100 m]

- Trip dynamics: RPA (smooth driving) & v*apos95 (aggressive driving) per phase,

counts per speed bin [> 150]

- Stop time: urban stop share [6-30%]

- Cold start: vmean [15-40 km/h] & vmax [60 km/h] + idling time [< 90’’]

Page 4: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

4

Description of evaluation methods considered

1. BC + RAW + BCs from Appendix 5 (completeness (≥15% MAW per phase) +

normality (≥50% normal MAWs per phase)

2. BC + RAW/CO2 (As described in Appendix 7c for PHEVs) + BCs from Appendix 5

(completeness (≥15% MAW per phase) + normality (≥50% normal MAWs per

phase)

3. BC + EMROADv5.96B1 (MAW as described in Appendix 5) checks for completeness

(≥15% MAW per phase) + normality (≥50% normal MAWs per phase)

4. BC + MAW SF111 (MAW as described in Appendix 5 setting the scaling factor for

P1, P2, P3 to 1) checks for completeness (≥15% MAW per phase) + normality (≥50%

normal MAWs per phase)

Page 5: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

5

Description of evaluation methods

5. BC + MAW SF111 (MAW as described in Appendix 5 setting the scaling factor for

P1, P2, P3 to 1) + RSI (Relative Severity Index – ACEA proposal Sept2017) • validity checks for normality (≥50% normal MAWs per phase, tol1+ is always

25%) + RSI (≤20% in every phase)

6. BC + CLEARv2.0WLTC (PB as described in Appendix 6) • validity checks the correct coverage of power bins for Total and Urban trip

7. BC + ACEA MAW&PB combined proposal (MATLAB implementation 12/06/2017) • validity checks outliers (≤ 20%) and the Severity Index (≤ 20%) per phase

Page 6: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

6

PEMS database

ND: normal driving CS: charge sustaining

DD: dynamic driving CD: charge depleting

EPD: emission provoking driving BC: battery charge

LED: low emission driving

* loaded vehicle

Source VehicleID Fuel Transmission Rated power

[kW] Emission control

technologies

# trips

Designed as RDE-compliant

Other PEMS testing

TUG TUG01 Diesel Automatic 140 DPF+SCR 5 6Alpine: 3 EPD + 1

DD + 2 ND

TUG TUG02 Diesel Manual 90 DPF+SCR 5 9Alpine: 5 EPD, 4*,

2 DD + 2 ND

JRC PHEV Gasoline/ Electric

Automatic 110+75 TWC 1 (CS) 3 (2*CD + 1 BC)

JRC Veh01 Gasoline Manual 70 TWC 6 3 (DD)

JRC Veh02 Diesel Automatic 193 DOC+EGR+DPF+

SCR 4 3 (DD)

JRC Veh03 Diesel Automatic 110 EGR+LNT+DPF 7 2 (DD)

JRC Veh04 Gasoline Automatic 140 TWC 6 -

JRC Veh05 Gasoline Manual 51 TWC 6 2 (DD)

JRC Veh06 Diesel Manual 73 DOC+EGR+DPF+

SCR 4 2 (DD) + 2*

JRC Veh07 Diesel Automatic 130 EGR+DPF 3 -

OEM Veh08 Diesel Automatic 420 EGR+DPF 2 -

41 32 TWC: Three-way catalytic converter

DOC: diesel oxidation catalyst

DPF: diesel particle filter

SCR: selective catalytic reduction

EGR: exhaust gas recirculation

LNT: lean-NOx trap

Page 7: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

7

Overview of BCs fulfilment

VehicleID # trips designed

as RDE-compliant

valid trip composition

valid cumulative positive elevation

gain

valid trip dynamics

# trips with valid BCs

TUG01 5 3a,b 5 2* 5

TUG02 5 4c 5 5 5

PHEV 1 0d 1 1 0

Veh01 6 6 6 6 6

Veh02 4 4 4 4 4

Veh03 7 7 7 6* 6

Veh04 6 6 6 6 6

Veh05 6 6 6 6 6

Veh06 4 4 4 4 4

Veh07 3 3 3 3 3

Veh08 2 2 2 2 2

41 38 41 39 38 a too large urban stop time share b cold start vmean < 15 km/h !! May cause artifacts for tests starting with almost warm engine c too long idle cold start d cold start average speed to high !! May cause artifacts for tests starting with almost warm engine

* invalid v*a_pos95 (R/M)

When properly designed & driven, compliance with RDE BCs are almost guaranteed.

Usually COLD START-related BCs are the main cause of trip invalidation (as per Annex

IIIA).

Page 8: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

8

Fit for purpose of Appendix 7a – trip dynamics

How well the RPA & v*apos95 perform in terms of separating normal

from dynamic driving? • Based on JRC ad-hoc PEMS testing: same vehicle normal vs aggressive

driving

• RPA on all trips is valid

• V*a pos95 as reported by EMROAD

VehicleID

# trips Identified as

ND Identified as

DD

ND DD ND DD ND DD

Veh01 6 3 6 0 0 3

Veh02 4 3 4 0 0 3

Veh03 7 2 6 1 0 3

Veh05 6 2 6 0 2 0

Veh06 4 2 4 0 0 2

27 12

• Out of 27 normal trips 1 is

identified as dynamic

• Out of 12 dynamic trips, 2 are

identified as normal (both on a

small engine vehicle 51 kW)

v*a pos95 is a good indicator of

aggressive driving

Page 9: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

9

Does v*a pos95 eliminate Emission Provoking

driving?

• Based on TUG PEMS testing: same vehicle ND vs EPD driving

• RPA & V*a pos95 as reported by EMROAD

• Out of 6 EPD only 1 is identified as dynamic

Emission provoking driving can be achieved fulfilling the trip

dynamic criteria of Appendix 7a

Therefore EPD should be identified with the evaluation methods

TUG01

ND EPD

U RPA 0.23 0.19 0.17 0.15 0.17

R RPA 0.16 0.13 0.11 0.11 0.09

M RPA 0.07 0.06 0.05 0.09 0.03

U v*a 17.08 13.85 10.58 10.82 10.84

R v*a 20.71 17.99 11.61 13.08 10.06

M v*a 19.15 14.38 24.77 19.30 23.19

TUG02

ND EPD

0.17 0.21 0.16 0.19 0.18

0.14 0.16 0.12 0.19 0.15

0.10 0.10 0.05 0.07 0.09

12.93 16.86 12.09 15.05 12.93

19.38 24.12 16.55 24.20 19.92

24.86 26.64 14.60 23.40 16.22

Page 10: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

10

Trip validity. Performance of evaluation

methods to identify & exclude biased driving

RAW RAW/CO2 MAW

(EMROAD) MAW SF 1-1-1 MAW SF 1-1-1 + RSI PB (CLEAR) ACEA MAW+PB

Vehicle Route type Driving style Validity Validity Validity Validity Validity Validity Validity

TUG

01

Alpine ND Uc+Rn Uc+Rn Uc+Rn Uc+U&Rn U&Rn+U&R RSI M SI

Alpine ND Uc+Rn Uc+Rn Uc+Rn Uc+U&Rn U&Rn+U&R RSI

Alpine DD Mc+U&Rn Mc+U&Rn Mc+U&Rn Mc+U&Rn U&Rn+U&R RSI

Alpine EPD Un Un Un U&Rn U&Rn+U&R RSI

Alpine EPD Uc+U&Rn Uc+U&Rn Uc+U&Rn Uc+U&Rn U&Rn+U&R RSI Mout+U&R&M SI

Alpine EPD Uc+Un Uc+Un Uc+Un Uc+U&Rn U&Rn+U&R RSI U&R SI

Hilly ND

Hilly ND Un Un+U&M RSI

Hilly ND Mn25+U&M RSI Mout+M SI

Hilly ND Rn25+U&R&M RSI Mout+R&M SI

Hilly DD U&R RSI

Hilly LED

ND: normal driving

DD: dynamic driving

EPD: emission provoking driving

LED: low emission driving

c: completeness (>15% MAW)

n: normality (>50% normal MAW)

RSI: relative severity index (<20%)

out: outliers share (<20%)

25Normality check with tol1+ fixed at 25%

Page 11: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

11

Trip validity. Performance of evaluation

methods to identify & exclude biased driving

ND: normal driving

DD: dynamic driving

EPD: emission provoking driving

LED: low emission driving

* Loaded vehicle

RAW RAW/CO2 MAW

(EMROAD) MAW SF 1-1-1 MAW SF 1-1-1 + RSI

PB (CLEAR)

ACEA MAW+PB

Vehicle Route type Driving style Validity Validity Validity Validity Validity Validity Validity

TUG

02

Alpine ND Uc Uc Uc Uc+Un Un+U&R RSI

Alpine ND Uc+Mn Uc+Mn Uc+Mn Uc+U&Mn U&R25&Mn+U&R RSI

Alpine DD Uc+Rn Uc+Rn Uc+Rn Uc+U&Rn U&Rn+U&R RSI

Alpine DD U&Mc+Rn U&Mc+Rn U&Mc+Rn U&Mc+U&Rn U&Rn+U&R RSI

Alpine EPD Uc Uc Uc Uc R RSI

Alpine EPD* Uc Uc Uc Uc+Un U&Rn25+U&R RSI Mout + M SI

Alpine EPD* Uc+R&Mn Uc+R&Mn Uc+R&Mn Uc+U&R&Mn U&R&Mn+U&R&M RSI

Alpine EPD* U&Mc U&Mc U&Mc U&Mc R RSI

Alpine EPD* U&Mc U&Mc U&Mc U&Mc+Un Un+U&R RSI

Hilly ND

Hilly ND

Hilly ND R SI

Hilly ND

Hilly ND c: completeness (>15% MAW)

n: normality (>50% normal MAW)

RSI: relative severity index (<20%)

out: outliers share (<20%)

MAW variants are the only method effectively eliminating ALL EPD tests.

However, also some ND tests are eliminated

MAW identifies EPD tests not identified as DD by v*apos95 both checks

are needed

Page 12: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

12

Trip validity. Performance of evaluation

methods to identify & exclude biased driving

RAW RAW/CO2 MAW (EMROAD) MAW SF 1-1-1 MAW SF 1-1-1 + RSI PB (CLEAR) ACEA MAW+PB

Vehicle Route type Driving style Validity Validity Validity Validity Validity Validity Validity

TUG

02

Alpine ND

Alpine ND

Alpine DD

Alpine DD

Alpine EPD

Alpine EPD*

Alpine EPD*

Alpine EPD*

Alpine EPD*

Hilly ND

Hilly ND

Hilly ND

Hilly ND

Hilly ND

Page 13: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

13

TUG01 emission calculation overview Invalid test

Page 14: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

14

TUG01 emission calculation overview Invalid test

Page 15: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

15

TUG02 emission calculation overview Invalid test

Page 16: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

16

TUG02 emission calculation overview Invalid test

Page 17: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

17

PHEV methods

1. RAW emissions: mass NOx/kmRDE

• No validity criteria

2. RDE3 emissions: mass NOx/mass CO2 * WLTC CO2 on CS

• Validity criteria: 12 km on ICE on the Urban part of the trip

3. RDE3 emissions * kmICE/kmRDE

• No validity criteria

4. RDE3 emissions * (kmICE/kmRDE) /0.85

• No validity criteria

5. COMBINED MAW+PB (ACEA matlab). Emissions = NOx/CO2

• Validity criteria: outliers < 20% + SevIndex < 20%

Page 18: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

18

PHEV Trip validity &

emissions

RAW RDE3 ACEA MAW+PB

Vehicle Strategie Validity Validity Validity

PH

EV CD Urb ICE < 12 km Mout+M SI

CD Urb ICE < 12 km

CS Mout+M SI

BC U&R SI

Page 19: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

19

How does the MAW variations perform respect

the standard EMROAD?

• On the TUG dataset, standard EMROAD provides same robustness against EPD and less

invalidation of valid normal trips.

• However, with the JRC dataset there is a large amount of normal trips (~ 37%) that get

invalidated by the MAW normality criteria (>50% of normal MAWs per bin, usually in

the Urban phase).

VehicleID # trips Current SF

1.2/1.1/1.05 URB SF=1

1/1.1/1.05 ACEA proposal

1/1/1

Veh01 6 83.3 83.3 100.0

Veh02 4 50.0 100.0 100.0

Veh03 7 100.0 100.0 100.0

Veh04 6 16.7 50.0 100.0

Veh06 4 50.0 100.0 100.0

Veh07 3 66.7 100.0 100.0

Veh08 2 50.0 100.0 100.0

Total 38 62.5% 87.5% 100%

Minor modification of the current methodology (specifying updated values of the scaling factors) can significantly

reduce the number of invalid tests caused by the MAW evaluation.

Page 20: Revision of RDE evaluation methods - Europa · 2 Objective of the revision • In the context of RDE, evaluation methods aim at: • Checking the normality of trips (abnormal/biased

20

Setting the SFs to 1 does not modify much the

emission values