Correction of test cycle tolerances: assessing the impact on CO 2 results J. Pavlovic, A. Marotta, B. Ciuffo WLTP 2 nd Act November 10, 2016
Agenda Flexibilities of test cycle and laboratory procedures - Status in current NEDC and future WLTP; - Identification of flexibilities; - Methods to correct them Experimental work Results and conclusions
Flexibilities in NEDC and WLTP Some flexibilities in test procedures are allowed (e.g. test temperature, vehicle speed, etc.) and necessary in order to get valid test results. Why then to correct them? To increase repeatability (test-to-test variation r ) and reproducibility ( R ) in CO 2 results and to better reflect target cycle and procedure; To avoid possibilities that flexibility will be used with intention to artificially lower CO 2 results.
Flexibilities in NEDC and WLTP In NEDC no correction of flexibilities is foreseen; In WLTP flexibilities are greatly reduced and many loopholes are eliminated; In addition, European Task Force group for corrections of WLTP flexibilities was established and was working on their development and integration in the European legislation; Main research in developing correction algorithms has been carried out by Graz University of Technology (TUG) and Netherlands Organization for Applied Scientific Research (TNO).
Identification of test flexibilities Flexibilities NEDC WLTP Battery Status of Charge (SOC) fully charged battery equilibrium SOC - correction already implemented in WLTP Test Temperature 20-30⁰C 23 ± 3⁰C Target Road Load Vehicle speed Rotational mass accuracy ± 5% (± 10% at 20 km/h) ± 2 km/h; ±1.0 s < 0.5 s/occasion no correction for 1-axle tests accuracy ± 10N ± 2 km/h; ±1.0 s < 1.0 s/occasion 1.5% increased inertia for 1-axle tests
Corrections performed in this study SOC Correction - imbalance in the battery over the test is corrected; Temperature Correction - NEDC corrected for 25⁰C and WLTP 23⁰C; Road Load (RL) and Speed Corrections - actual speed profile and RLs are corrected to the target one; Distance Correction - offset from different brake behaviour of the driver is corrected.
Correction of flexibilities SOC correction E (Wh)=1/3600 x U x I(t) dt (1) CO 2 ( g km ) = 0.0036 x E 1 η alternator Willans factor 1 d Willans factor tabulated values from WLTP in g CO 2 /MJ (2) Battery current is monitored over the whole cycle and CO 2 emission results are corrected for the imbalances in the battery SOC; Minimum 5Hz data are used. Soak temperature correction CO 2 % = 0.18% t t C t m C t t and t m C are target and measured temperatures 0.18% is correction coefficient from Ligterink et al., 2014. NEDC tests are corrected for 25 ⁰C and WLTP for 23 ⁰C initial test temperature; Temperature of oil was recorded at the beginning of the tests.
Correction of flexibilities Road Load and Speed Corrections Actual power 2 P (a) = (F 0 a + F 1 a v a + F 2 a v a Target power + m (a) ) a a ) v (a) P (t) = (F 0 t + F 1 t v t + F 2 t v 2 t + m (t) ) a t ) v (t) F 0(a), F 1(a), and F 2(a) are calculated from coast downs performed immediately after the test Only positive wheel power is averaged v a - is the actual speed during the CO2 (g/s) 10 CO 2 (g/s) = k v x P + D test acquired at 10Hz. 8 6 K v is vehicle specific Willans coeff. (g/kws); 4 D is constant representing parasitic losses or 2 CO 2 emissions at zero power at the wheels 0 0 10 20 30 40 50 P _actual (kw)
Correction of flexibilities Distance Correction CO 2 d g km = CO 2 P corr g km d m (km) d t (km) d m is the distance measured during the test; d t is the target distance (23.27 km WLTP; 11.03 km NEDC) The final result CO 2 d is without the offset from different brake behaviour of the driver.
Experimental work VEH1 VEH2 VEH3 VEH4 Fuel Gasoline Diesel Gasoline Gasoline Emission standard EURO6 EURO5 EURO5 EURO5 Transmission Manual Automatic Automatic Manual Power (kw) 100 120 100 120 Test mass (kg) 1560 1520 1250 1360 NEDC NEDC Tests performed WLTP WLTP WLTP WLTP Laboratories LAB1 LAB6 LAB1 LAB6 LAB1 LAB1 involved Repeatability and reproducibility were analysed before and after the corrections
Results and discussion VEHICLE 1. NEDC Correction Results Repeatability increased in 2 out of 4 labs; Reproducibility increased and standard deviation dropped from 4.7 g/km to 2.0 g/km.
Results and discussion VEHICLE 1. WLTP Correction Results Repeatability increased in 1 out of 3 labs; Reproducibility decreased after corrections and standard deviation slightly increased from 2.3 g/km to 2.4 g/km.
VEHICLE 2. NEDC Correction Results Repeatability increased in 2 out of 4 labs and reproducibility slightly decreased after all corrections (from 1.2 to 1.4 g/km)
VEHICLE 2. WLTP Correction Results Repeatability increased in 2 out of 3 labs while reproducibility decreased after corrections.
VEHICLES 3 AND 4. WLTP Correction Results Repeatability increased for both vehicles and standard deviation dropped from ±0.9 to ±0.4 g/km for vehicle 3 and from ±0.8 to ±0.5 g/km for vehicle 4.
Conclusions Average influence of each correction step on CO 2 results Test cycle WLTP NEDC Phase Corrections SOC Speed+RL Distance Temperature Low 1.12 3.58 0.44 0.12 Medium 0.36 3.87 0.21 0.11 High 0.30 1.06 0.16 0.12 Extra-high 0.22 0.63 0.18 0.12 Total AVERAGE 0.44 0.57 0.19 0.11 Total MAX/MIN 0.89/0.01 1.44/0.01 0.34/0.00 0.16/0.00 UDC 3.26 0.37 1.50 0.46 EUDC 1.06 0.85 0.37 0.46 Total AVERAGE 2.11 0.47 0.79 0.46 Total MAX/MIN 5.76/0.20 1.78/0.10 1.71/0.13 0.67/0.34 The biggest influence had SOC correction for NEDC tests and the lowest temperature correction for the WLTP test due to more stringent WLTP initial test temperature requirements.
Conclusions All corrections steps performed in this study influenced the average CO 2 emissions by approximately 1.3% and 3.8%, for WLTP and NEDC respectively; This is an indication of the reduction of flexibilities in WLTP compared to NEDC The impact of the correction on the repeatability and reproducibility of test results is promising. However Not in all cases the application of this methodology has led to an improvement either r and/or R; The correction of the Road Load coefficients seems to have practical issues (representativeness of coast down checks performed after the tests and time passed between test and checks). SOC Correction on average influenced total NEDC CO2 by ~ 2.1% and total WLTP by ~ 0.4% This correction is already implemented in WLTP
Conclusions Speed + RL Correction had on average ~ 0.5% (NEDC) and ~ 0.6% (WLTP) influence on the total CO2 results Separate effect was not analysed to see which correction has bigger influence; Speed correction is ready to implement (vehicle speed already monitored at 10Hz for DI). If not corrected, effect of the flexibility expected to increase RL correction to be based on the coast down checks performed after the WLTP tests Not ready to implement. Test procedure to be developed (time passed between the tests and coast down checks strongly affects results); Experience from the correlation study is the high variability from different CD checks. More stringent prescription for WLTP dyno setting may not need it. Distance Correction had on average ~ 0.8% (NEDC) and ~ 0.2% (WLTP) influence on the total CO2 results Ready to implement, does not require additional information to implement
Conclusions Temperature Correction had on average ~ 0.5% (NEDC) and ~ 0.1% (WLTP) influence on the total CO2 results Low influence of this correction in WLTP result of more stringent WLTP soak temperature requirements (23±3 C); Correction ready to implement; Possible point of discussion is the correction coefficient (0.18% correction/1 degree Celsius) that is derived based on NEDC tests If derived from WLTP tests it is expected to be even lower that 0.18%, since cold start phase in WLTP has lower impact on the total CO2 result in WLTP compared to the NEDC tests.
Additional elements An additional possible flexibility that has not yet been discussed is the deviation from the prescribed gear-shift strategy Since GS is not monitored and in WLTP varies from vehicle to vehicle and possibly from test to test, this might become the biggest future flexibility of the WLTP Approach proposed by TNO/TUG seems to be sufficiently developed. Formulations should be transferred into legislation. Is the approach valid for all powertrains? To be checked A possible alternative approach to the corrections is the use of CO2MPAS
CO2MPAS to correct test cycle flexibilities In the period 2017-2020 CO2MPAS will be anyhow used for the correlation Once it is calibrated on the WLTP test, CO2MPAS can be used in a very accurate way for the correction of test flexibilities (apart from the RLs) including the gear-shift With the corrections done inside the software, no need to include detailed description in the legislation If other flexibilities are identified, their inclusions in CO2MPAS will possibly be more straightforward CO2MPAS will be maintained and extended towards all powertrains. It will be possible to monitor and improve its performance
Literature Ligterink, N.E., Mensch, P., Cuelenaere, R.F.A., Hausberger, S., Leitner, D., Silberholz, G., 2014. Correction algorithms for WLTP chassis dynamometer and coast-down testing. Pavlovic, J; Marotta, A; Ciuffo, B; Serra, S; Fontaras, G; Anagnostopoulos, K; Tsiakmakis, S; Arcidiacono, V; Hausberger, S; Silberholz, G. Correction of test cycle flexibilities: Evaluating the impact on CO 2 results. Transport Research Procedia, 2016, 14, 3099-3108. UNECE Regulation No. 83 Revision 5. Uniform provisions concerning the approval of vehicles with regard to the emission of pollutants according to engine fuel requirements. UNECE, Geneva, Switzerland, 2015; UNECE Global Technical Regulation No. 15. Worldwide Harmonized Light Vehicles Test Procedure. UNECE, Geneva, Switzerland, 2016;