Evaluation of PEMS tests Veh 01 & Veh 02 with the CLEAR Method

Slides:



Advertisements
Similar presentations
WLTP-06-31e WLTP correction algorithms progress report from TUG (chassis dynamometer corrections) and TNO (coast down corrections) preliminary results.
Advertisements

WLTP drive trace normalization
1 Comparison of WLTP unified database distributions and WLTC rev2 distributions Heinz Steven WLTP WLTP-DHC
Instrumented Vehicle BAQ Instrumented In-Use-Vehicles, a Versatile Tool to Measure Emissions BAQ 2004 Agra, India Dec 2004 Instrumented In-Use-Vehicles,
1 Proposal for a downscaling procedure for the extra high speed phases of the WLTC for low powered vehicles within a vehicle class Technical justification.
WLTP-DHC Validation Phase I Results and Recommendations by India
RDE Working group Brussels, September 2015 Collection of NO x emissions data - First preliminary results RDE working group 14 September 2015 European Commission.
1 Comments on the Ste 3 gearshift calculation tool from validation 2 participants Heinz Steven WLTP WLTP-DHC
1 Proposals of WLTC versions for low powered vehicles Heinz Steven WLTP.
Proposed by Japan GRPE/WLTP-IG/ DHC subgroup
WLTP-12-17e Status report about the work of the gearshift issues task force.
RDE testing: how to define NTE emission limits?
Questions on cycle representativeness (French position) EU – WLTP 17 th of September 2013.
25 January 2016 European Commission - Joint Research Centre (JRC)
OICA CO 2 Impact through Weight Variation on Light duty Vehicles WLTP DTP6, Geneva OICA, Volkswagen WLTP-DTP-LabProcICE-076.
Submitted by the expert from Japan the secretariat
Bus and coach transport for greening mobility
State of the art of alternative vehicles performance
N. Ligterink, R. Cuelenaere
Improvement of Wind tunnel Measurement Process Status report
WLTP Modelling of fuel consumption and detection of driveability problems for “borderline” cars with different maximum speed caps. Heinz Steven
Random Cycle Generator
SiCo ACEA position RESULT slides proposal.
Institute for Internal Combustion Engines and Thermodynamics
Institute for Internal Combustion Engines and Thermodynamics
RDE Regulation Commission Meeting
HDV CO2 verification test (“SiCo”)
Status Januar Verification of test normality
RDE Task Force Meeting, 7th January 2014
Improvement of Family definitions
Development of the Japan’s RDE (Real Driving Emission) procedure
Questions on cycle representativeness
CLEAR Graz Stefan Hausberger, Nikolaus Furian
RDE Task Force Meeting, 28th November 2013, Brussels
28th of November 2013, Brussels
Comparison of different gearshift prescriptions
Input on wind tunnel criteria discussions
BRAKE PARTICLE EMISSIONS
Pems route and parameters
Institute for Internal Combustion Engines and Thermodynamics
absolutely essential first level
LDV Real Driving Emissions: - Drafting of physical PEMS protocol –
JAMA position on RDE Boundary Condition
Report Nr. I-25/10 Haus-Em 07/10/676 from
Additional RDE trip indicator(s)
Institute for Internal Combustion Engines and Thermodynamics
GTR Corrections, Open Points, Expert Proposals and Confirmations in GTR 15 1/2/2019.
Analyses related to dynamic effects in vehicle speed and NOx emission measurements by H. Steven
Emissions testing in the laboratory and on the road: Preliminary results for one Euro 6 diesel vehicle Pierre Bonnel Martin Weiss Joint Research Centre.
PN-PEMS Progress update
WLTP Correlation engine modeling
Japanese Fuel Efficiency Standard for Heavy Duty Vehicles
RDE testing of hybrid vehicles Contribution to the RDE working group
16th of November 2013, Brussels
Real Driving Emissions IWG Recommendations for Data Collection & EU Dataset
WLTP Correlation measurement
HDV CO2 Regulation in REPUBLIC OF KOREA
Reason for performance difference between LVW and GVW
WLTP Comparison of WLTP unified database distributions and WLTC rev2 distributions Heinz Steven
LDV Real Driving Emissions: - Assessment criteria and work plan -
Emissions testing with PEMS versus random laboratory driving cycles
RDE-LDV working group 13 April 2012 Pierre Bonnel Martin Weiss
J. Pavlovic, A. Marotta, B. Ciuffo WLTP 2nd Act March 15th, 2017
WLTP Modelling of fuel consumption and detection of driveability problems for “borderline” cars with different maximum speed caps. Heinz Steven
RDE-LDV working group 20 November 2012, Brussels, EU Pierre Bonnel
Comparison NEDC/WLTC Comparison of the influence of weighting factors as proposed by France on the validation 2 CO2 emission results for the WLTC By H.
RDE Task Force Meeting, 16th December 2013, Brussels
Informal document GRPE Rev.1
Comparison of key parameters of EU WLTP database and WLTC version 5
Presentation transcript:

Evaluation of PEMS tests Veh 01 & Veh 02 with the CLEAR Method Meeting Brussels, 28.11.2013 Stefan Hausberger, Silke Lipp

CONTENT Short reminder: Explanation of the CLEAR method + options for detailed evaluation and explanation of CO2-based CLEAR method Analysis for Vehicle#1 Vehicle data Difficulties Measured data Evaluation results vehicle#1 Analysis for Vehicle#2

“CLEAR” method x Pdrive De-normalise the generic Target Power Pattern (multiply with vehicle specific Pdrive) 2) Average measured instantaneous (1Hz) signals from PEMS test over 3s Bin the modal measured emissions into corresponding power class Check minimum number of valid emissions in each bin 4) Average emission value per power bin 5) Multiply avg. emission value with the corresponding time share of power class  weighted Emission value [g/h] for urban, road motorway and total trip Target Power Pattern x Pdrive Pnorm

“CLEAR” method Actual engine power of each vehicle is normalised by division by driving power demand of the vehicle to get representative power distribution for all cars. Pdrive = vref * [mref * aref + R0 + R1 * vref + R2 * vref²] With R0, R1, R2 ……road load mref……...kurb weight of the vehicle [kg] aref………reference acceleration [0.45 m/s²] vref………reference vehicle velocity [70 km/h = 19.4 m/s] Evaluation of WLTP data on 49 different cars  Generic Target Power Pattern for urban, road, motorway

“CLEAR” method using CO2 instead of Pe (1/2) Principal illustration char. Gas pedal The Problem Power signals from veh#1 and veh#2 were gained from gas pedal position which typically overestimates real propulsion power due to its characteristic line (> 50% over-estimation of real power, see later slides). This made the application of an alternative method necessary. Possible solution: Vehicle specific Willans lines are applied in WLTC correction functions, in MAC test procedure, in WLTC-NEDC correlation study,.. Logical to use them also in RDE From WLTC (or any other reasonable hot start test) a linear equation between power and CO2 emissions can be established

“CLEAR” method using CO2 instead of Pe (2/2) Additional steps in application of CLEAR: 1) Willans function of the vehicle test results in WLTC is produced CO2 [g/h] = D + k x Power [kW] 2) power is calculated from CO2 measured in PEMS trip as input for CLEAR. Go ahead like usual (details on results shown later). Willans-Coefficients: (CO2(t)– d) k P(t) = Etc. Additional condition: If v=0  P = 0 P_Willans

Results vehicle #1 Vehicle data: Kerb mass [kg] 1470 F0 [N] 79.19 F1 [N/(km/h)] 0.73 F2 [N//km/h)²] 0.03 Prated [kW] 109 Model year 2013 Emission Standard EURO 6 Engine type DIESEL

Test data available for vehicle #1 9 trips measured with vehicle no. 1

Difficulties with vehicle #1 As example details on power signals versus calculated power for Trip: ROM_01 Test time: 6780 s → chart shows power values between 900 s an 1100 s Calculated power by using the Willans-Coefficients is in the middle of measured and wheel power calculated from velocity and reference mass and road load data (no road gradients available). P_wheel = v*(m*a+R0+R1*v+R2*v²)

Difficulties with vehicle #1 Difficulties encountered with the test data 1st run: all trips were identified to be invalid by CLEAR due to too much deviation between measured and calculated power signal . Integral of measured power (from gas pedal position), lower value with motoring values added manually ca. 44000 kNm ca. 71000 kNm ca. 107000 kNm ca. 115000 kNm Example integral of power Trip: ROM_01 Calculated from measured 1Hz CO2 and Willans equation Calculated from measured 1Hz velocity with reference mass and road load

Difficulties with vehicle #1 Similar problems detected also for chassis dyno tests received for veh#1 After the detection of the inconsistencies discussion with JRC on data sources, road load settings etc. started. Many thanks to Theodoros Vlachos and Pierre Bonnel for the support in explaining contents and sources of data and especially for the extensive and good work in data collection! Conclusions TUG: Power signal calculated from measured gas pedal position are too inaccurate without knowing its characteristic line. Interpolation of power from existing fuel map of veh#1 failed due to missing data in low load areas Results from power signal shown only for information but are no valid evaluations! Option for CO2 based CLEAR evaluation was applied. Willans line had to be gained from chassis dyno test of same vehicle model at TUG! In second version of data provided by JRC emissions during cold start, DPF regeneration phases etc. have been set to zero by JRC if at test start. We excluded these “virtual zero emission phases” from evaluation. We assume there are no virtual zero emission events somewhere in the middle of test data (to be validated).

Vehicle no. 1 Willans-Line Creation of the Willans-Line from vehicle no.1: Dataset of a measured CADC-drive cycle for the determination of the Willans-Line: =Average values of positive wheel power Classification of different phases =Average values of the measured CO2 Vehicle no. 1 Willans-Line Coefficients of the Willans-Line for the vehicle no. 1 for calculation of the power in the next step (source: CADC measurement at TUG!) Trend line Average CO2 [g/h] Average positive wheel power [kW]

Results for vehicle #1 P_Willans Share of power for the different trips: Visualization of the different power values over time share. For CO2-based method the separation between idle and motoring needs to be improved „larger than“

Results for vehicle #1 P_Willans Note: Valid trips marked with dots, others miss measured emissions in one ore more power bins (only power bins with > 0.1% target time share tested for this criterion yet, otherwise more invalid trips) Lowest emission value from valid trip = 280 mg NOx/km  Vehicle#1 would fail, if CF< 3.5

Results for vehicle #1 P_Willans Note: Valid trips marked with dots, others miss measured emissions in one ore more poser bins Overview Vehicle#1 NOx emission results:

Area includes ~94% of NOx for this car P_Willans Some details for vehicle #1 Results per power bin for total trip show high variability of NOx at medium and high power areas Share in weighted NOx Fluctuation of the NOx-values from the different trips at higher power bins. Area includes ~94% of NOx for this car Examples for further details next slide „larger than“

Weighted engine speed gradient Weighted power gradient P_Willans Weighted NOx Weighted engine speed „larger than“ Weighted engine speed gradient Weighted power gradient

Some details for vehicle #1 P_Willans Some details for vehicle #1 Reasons for high and low NOx emissions in single power bins can be analysed based on power and rpm data, e.g.: Low engine speed level in “Eco” leads to full load phases where veh#1 obviously has no sufficient NOx control  high specific NOx (power bin 8 and 9 are empty -> trip avg. reduced) Higher engine speeds in “Agg” seem to have better NOx control. However, high derivates of P and rpm also lead to rather high specific NOx (in addition power bin 8 and 9 are filled) Route1 Eco “Power bin 6” Route1 Agg “Power bin 6” Route1 Eco “Power bin 7” Route1 Agg “Power bin 9”

Total weighted NOx [g/km] based on ~50% overestimated power! Results CLEAR with measured power signal P from gas pedal When power value gained from gas pedal position is used for the evaluation, power signals seem to be on average 50% too high.  If CLEAR still is applied, all trips are treated as if they have more than normal power distribution what results in a reduction of emissions: Total weighted NOx [g/km] based on ~50% overestimated power! Only validated power signals shall be allowed (e.g. +/- 3% accuracy) “Power at wheel hub” may be used as reference: measurable at chassis dyno; can be calculated with reasonable accuracy from Willans line; can be measured accurately with torque meter rim (in case of eventual disagreements between OEM and TA). OEM may use CAN signals for their tests (cost efficient, accurate) while TA, TS and anyone else can use the CO2 based CLEAR method. In case of doubts reference method could be power based method (distribution of driving resistances should be valid for all cars, “distribution of CO2” not necessarily)

Results vehicle #2 Vehicle data: Test mass [kg] 1700 R0 [N] 121 R1 [N/(km/h)] 0.8 R2 [N//km/h)²] 0.03 Prated [kW] 103 Vehicle data: Model year 2012 Emission Standard EURO 6 * Engine type DIESEL * in vehicle description it is noted as “EURO 5”, in file names EURO 6

Test data available for vehicle #2 17 trips where measured with vehicle no. 2

Difficulties with vehicle #2 Similar problems as for veh#1: Nor reliable power signal available. All evaluations based on “measured power” invalid due to too high deviations in measured and calculated cycle work Additionally Willans Lines gained from chassis dyno tests at JRC suffers also from not validated wheel power data.  power in WLTC calculated by TUG from mass and road load Eventually road load and mass was different in WLTC test than in NEDC? Results should be validated between TUG and JRC! With this task TUG is delayed due to the complications for veh#1  Willans line uncertain for veh#2  only preliminary results for internal discussion!

Willans line computed for vehicle #2 v… velocity m… test mass a… acceleration R0… road load coefficient R1… road load coefficient R2… road load coefficient Using calculated wheel power: Pwheel = v * (m*a + R0+R1*v+R2*v²) TUG-Willans Version seems to be too high (maybe the car was tested in WLTC with different mass and road load?) Other too low

Power-Data: Three different calculated data where considered for the calculation with CLEAR: P_Mean_corr P_Willans_JRC P_Willans_TU P_Mean_corr: The measured power values for each trip where taken with an addition for detecting drag power. Drag power was set on a value of -10 kW, when the acceleration a < 0 and the CO2 < CO2idle. Shown only for illustration! P_Willans_JRC: Willans-Coefficients where calculated by using the WLTC-data from JRC P_Willans_TU: Willans-Coefficients where calculated by using the wheel power

Comparison of the different Power-Data: Example for Trip: BS_2013_03_04 Time: 1000-1300 s Result: Different power values over time for each calculation.

Power-Data: Seems to be too high Result: Summation of power data results very different work values over trip time.

Results for vehicle #2 Total trip results evaluated based on measured power and “TUG-Willans power” P_Willans P_measured Correct results should be between these 2 evaluation results (TUG-Willans delivers rather too low power values, measured power seems to be too high)  further analysis after clarifications with JRC Note: Valid trips marked with dots, others miss measured emissions in one ore more power bins

Thank you very much for the attention!

Backup slides

Schematic picture „Option A“ “CLEAR” method using CO2 instead of Pe (1/2) Additional steps in application of CLEAR: 1) Willans function of the vehicle test results in WLTC are produced CO2 [g/h] = D + k x Power [kW] 2) calculate power from CO2 measured over PEMS trip as input for CLEAR. Go ahead like usual = actual option (details shown later) Options for a next CLEAR release: Convert x-axis in target power pattern from P [kW] to CO2 [g/h] (+ “if v=0  idling bin”) Bin emissions over CO2 instead of binning over power Calculate Power in CLEAR from CO2 measured with PEMS and allow binning for “Willans Power” as well as for directly measured power. Schematic picture „Option A“ x Pdrive  x (D + k x P) 