RDE DEVELOPMENT PROCESS & TOOLS

Transcription

1 Daniel Baumann, IT RDE DEVELOPMENT PROCESS & TOOLS Kieran McAleer

2 SIMULATION LAB ROAD AVL Solutions (A comprehensive approach to RDE) Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 2

3 Road Testing ROAD Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 3

4 RDE TEST EQUIPMENT AVL M.O.V.E is System GAS PEMS is PN PEMS is EFM is System Control Concerto M.O.V.E Postprocessing Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 4

5 AVL M.O.V.E GAS PEMS NOx result as function of altitude ~ m ~15 m Independence of the measured values from ambient conditions, shocks and vibrations, EMV, is a pre-requisite for a successful RDE test! Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 5

6 AVL M.O.V.E PN PEMS Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 6

7 AVL ADVANCED DIFFUSION CHARGER TECHNOLGY INFLUENCE OF PARTICLE DIAMETER Conclusion: The counting efficiency of the AVL Advanced DC is equivalent to PMP systems at typical engine exhaust conditions. Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 7

8 AVL M.O.V.E EXHAUST FLOW METER 6 55 AVL495_ExhaustGasFlowMass OBD_ExhaustFlow 5 AVL495_ExhaustGasFlowMass [kg/h] Time [s] Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 8

9 AVL M.O.V.E IS SYSTEM VERIFICATION (*) AVL M.O.V.E is System CVS System Pollutant [g/km] AVL M.O.V.E is CVS Diff absolute Diff [%] CO2 22,23 24,7-2,47-1,2 (*) CO,8,73,7 1,2 NOx,66,62,4 6,6 WLTC result for 4 cylinder Diesel engine vehicle (*) real vehicle/brand is not shown Note: These results are below the legislative requirments by a factor 2-5. Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 9

10 RDE TESTING HOW TO GET A VALID TEST? TRIP REQUIREMENTS Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 1

11 RDE TEST PROCEDURE Temperature Moderate C to 3 C / Extended (-7 C to 35 C) Altitude Moderate < 7 m / Extended < 135 m Trip / Driver City (34%) / Rural (33%) / Motorway (33%), up to 16 km/h, 9-12 min Load no empty vehicle, heating or air condition on Kieran McAleer 9th AVL Calibration Symposium 4 11 월

12 AVL CONCERTO M.O.V.E POST PROCESSING Highway Rural City Kieran McAleer 9th AVL Calibration Symposium 4 11 월

13 AVL CONCERTO M.O.V.E POST PROCESSING Kieran McAleer 9th AVL Calibration Symposium 4 11 월

14 AVL CONCERTO M.O.V.E POST PROCESSING Kieran McAleer 9th AVL Calibration Symposium 4 11 월

15 Chassis Dynamometer Chassis Dyno Kieran McAleer 9th AVL Calibration Symposium 4 11 월

16 REAL DRIVING EMISSION - DEVELOPMENT Use case 2: Emission development, calibration, verification of development targets, based on reference cycles Type Approval In Service Compliance (Verification, Benchmark) Use case 3: Simulation, virtual development and frontloading Use case 1: Reproduction and analysis of on-road test data Kieran McAleer 9th AVL Calibration Symposium 4 11 월

17 RDE - TEST BED CONSIDERATIONS Are current chassis dyno test bed able to handle vehicle driving with higher loads? RDE engine operation will be far higher than current standard drive cycle requirements. The energy contained in the fuel, which is burned in a combustion engine is converted to : 33% to turn the wheels - Dyno must handle that 33% in heat - Test bed infrastructure (TGA) must handle that 33% heat in the exhaust - CVS and Emission Systems must handle that Test bed infrastructure, like air-conditioning, ventilation, must be able to handle the additional heat produced by the vehicle. Lower dilution ratios + higher temperatures, may require double dilution for PM Higher load requirements due to RDE, especially for high power and/or heavy vehicles, will require higher performance of the chassis dyno and cooling fan. Larger exhaust gas flow and heat may require larger CVS or alternatives (modal diluted or raw, heated CVS, ). i.e. app. 3 m3/min CVS per 1 kw engine operation Temperature??? Street -Slope? -Curves? Altitude??? Which additional environmental conditions do the powertrain development engineers require? That will most likely be different between an OEM (who knows how the vehicle is calibrated) and all others. Kieran McAleer 9th AVL Calibration Symposium 4 11 월

18 REAL DRIVING EMISSION - DEVELOPMENT Use case 2: Emission development, calibration, verification of development targets, based on reference cycles Type Approval In Service Compliance (Verification, Benchmark) Use case 3: Simulation, virtual development and frontloading Use case 1: Reproduction and analysis of on-road test data Kieran McAleer 9th AVL Calibration Symposium 4 11 월

19 RDE REPRODUCTION AND ANALYSIS Reproduction Reproduction and analysis of a RDE road test on a test bed: Target is to reproduce one specific road test of one specific vehicle as close as possible on a test bed. Valid for specific maneuvers or complete road test, engine speed and torque must be directly comparable. With the reproduction of the test it can be repeated as often as needed, which is for analyzing a problem and later to verify a solution, a key element for any development work. Allows to use the full testing and analytical power of an emission laboratory. Implementations: During the PEMS road test all relevant data will be recorded. Data import into the test bed control and automation system, without the need for long set-up or simulation model generation ( drive on the morning and reproduce on the afternoon ) Specific dyno control mode and test execution The road data will be transferred to the test bed and special functionalities (like data import, specific dyno control-modes, ) will allow the reproduction. Kieran McAleer 9th AVL Calibration Symposium 4 11 월

20 RDE REPRODUCTION AND ANALYSIS Reproduction Results of AVL RDE reproduction development work: Euro-5 Diesel passenger car with DOC, SCR and DPF. These data were measured during preliminary tests of a new test methodology under develop. It can not be expected that the here shown good correlation can always be achieved. CO2 mass [g/s] CO2 mass Road 178,1 g/km Lab. 177,9 g/km Deviation -,1 % NOx mass [g/s] NOx mass Road,633 g/km Lab.,692 g/km Deviation +9,3 % Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 2

21 REAL DRIVING EMISSION - DEVELOPMENT Use case 2: Emission development, calibration, verification of development targets, based on reference cycles Type Approval In Service Compliance (Verification, Benchmark) Use case 3: Simulation, virtual development and frontloading Use case 1: Reproduction and analysis of on-road test data Kieran McAleer 9th AVL Calibration Symposium 4 11 월

22 RDE CONVENTIONAL REFERENCE DRIVE CYCLES Conventional Conventional drive cycle testing but with RDE Reference cycles: The current development methodology, which is mostly based on standard reference cycles, is extended by already known or newly developed RDE Reference Cycles. These include variations of velocity, road gradient, curve radius, ambient conditions, Reproducible and comparable results, also between different vehicles Drive Cycle based development: Beside the well known emission cycles, like WLTC, NEDC or FTP75 also other known cycles will be used. Like Artimis (also known as CADC), which was used in the past mainly for emission modeling and emission inventory estimations Standardized Random Test Sequence, like RTS95 ("aggressive ) Drive cycles generated by a Random Cycle Generator, like the one developed by TNO on base of the WLTC data base. Specific RDE Reference cycles which are generated by an OEM specificly for its vehicle types. or specific drive cycle maneuver elements combined to a test sequence like a finite element approach. Kieran McAleer 9th AVL Calibration Symposium 4 11 월

23 Altitude, m Velocity, km/h RDE CONVENTIONAL REFERENCE DRIVE CYCLES Conventional Implementation: AVL Graz RDE Reference road and cycle Distance/Time 52,6km 43s City 2% 47% Highway 43% 23% Interstate 37% 3% GRAZ Gleisdorf warmup Vehicle Speed time (s) v vehicle (km/h) Altitude Kieran McAleer 9th AVL Calibration Symposium 4 11 월

24 RDE CONVENTIONAL REFERENCE DRIVE CYCLES Conventional Implementation: Road load simulation requires additional elements for RDE, like: Slope of street (needs also support for the driver via the driver aid) Curve resistance load simulation based on curve radius and vehicle speed AVL patent bending Ambient conditions (temperature, humidity) Altitude (absolute pressure) Kieran McAleer 9th AVL Calibration Symposium 4 11 월

25 REAL DRIVING EMISSION - DEVELOPMENT Use case 2: Emission development, calibration, verification of development targets, based on reference cycles Type Approval In Service Compliance (Verification, Benchmark) Use case 3: Simulation, virtual development and frontloading Use case 1: Reproduction and analysis of on-road test data Kieran McAleer 9th AVL Calibration Symposium 4 11 월

26 REAL DRIVING EMISSION Simulation Simulation technologies to: address the infinite variables involved in real world driving master todays large number of vehicle models and variants. Virtual Integration and Front-Loading Calibration: Virtual simulation of random driving maneuvers with full variability of ambient conditions, driver types, vehicle variants, connected powertrain, from pure simulation, Hardware in the Loop (HIL) to conventional test beds Evaluation of powertrain and vehicle concepts, definition of solutions and engineering targets, calibration in non standard ambient conditions Implementation: Based on InMotion Engine- or Powertrain test beds 4x4 Chassis dyno test beds with individual wheel dynos. up to GPS Emulation on chassis dyno testbed for car to infrastructure integration Innovation Adoption: Simulation approach, especially for classic emission development and testing groups, require still a high willingness and ramp up time to adopted to such innovations. Kieran McAleer 9th AVL Calibration Symposium 4 11 월

27 REAL DRIVING EMISSION Simulation TÜ V Hessen Advanced Chassis DynoTestbed: Focus on alternative powertrain, alternative fuel and Real Driving Emissions Testing Chassis Dyno Emission Test Cell for Certification and R&D with additional Real Driving Emission Simulation capabilities. also Portable Emission Measurement Systems Kieran McAleer 9th AVL Calibration Symposium 4 11 월

28 CLOSING THE GAP Kieran McAleer 9th AVL Calibration Symposium 4 11 월

29 REAL DRIVING EMISSION Simulation Virtual Front-Loading Calibration: xil Station with Hardware in the loop (HIL) and full variability to simulate ambient conditions, driver types, vehicle variants, connected powertrain, Same AVL user interface and functionalities as AVL conventional test beds ( only the engine sound is missing ) Hardware in the loop: ECU, injection nozzles and intake throttle. Kieran McAleer 9th AVL Calibration Symposium 4 11 월

30 LAB Base Calibration in the LAB Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 3

31 THE CONVENTIONAL APPROACH CAMEO Cluster Points Calculation Determination of Speed/Load Area out of driving cycle Local DoE tests to reduce the testing effort Kieran McAleer 9th AVL Calibration Symposium 4 11 월

32 MULTI-VARIANT PROFILES Weight: 159kg Weight: 181kg Weight: 17kg MORE POINTS SPECIFIC ADJUSTMENTS ON CHASSIS DYNO Different speed/load profile Different cluster points amongst variants Kieran McAleer 9th AVL Calibration Symposium 4 11 월

33 DRIVING STYLE/CONDITIONS Differences due to driving style (exemplary): - Fuel Economy: Factor NOx Emissions: Factor 4.7 RDE NEDC Potential RDE range Individual profile which must meet certain legal requirements Changing real road conditions Changing driver/behavior Changing ambient conditions RDE in worst case can require entire operational range of an engine Compulsory velocity profile Chassis Dyno Tight tolerances Defined ambient conditions NEDC cycle can easily be represented by cluster points with local models NEDC range Kieran McAleer 9th AVL Calibration Symposium 4 11 월

34 GLOBAL ENGINE MODELING LESS POINTS LESS IN-VEHICLE ADAPTIONS Comprehensive and Efficient Engine Identification Simple and Fast Variant Calibration Kieran McAleer 9th AVL Calibration Symposium 4 11 월

35 CAMEO RDE OPTIMIZATION IN 4 STEPS Easy Test Generation Online Test Adaption One-Click Modeling Cycle Optimization Smooth Maps Cycle analysis Automatic clustering of cycle data Easy generation of test points Active DoE Auto-adaption of test points for best model quality No additional iterations needed Automatic Modeling Automatic generation of models direct from raw data. No expert knowledge Expert interaction still possible Smooth Optimization & Cycle Prediction Smoothing constraints during optimization Maps ready to be used Kieran McAleer 9th AVL Calibration Symposium 4 11 월

36 SIMULATION Simulation (Virtual Calibration) Kieran McAleer 9th AVL Calibration Symposium 4 11 월

37 AREA OF OPTIMIZATION EVOLUTION OF MODEL-BASED CALIBRATION TOWARDS VIRTUAL CALIBRATION Semi- Physical Models Real Driving Neural Network Global Polynomial Model Best Point Experimental Designs Local FF EU2 / EU3 EU3 / EU4 EU4 / EU5 EU6 / RDE NUMBER OF VARIATION PARAMETERS Kieran McAleer 9th AVL Calibration Symposium 4 11 월

38 VIRTUAL CALIBRATION SOLUTION AVL connects multiple methods and tools to a beneficial one-stop solution. AVL CRUISE M AVL Virtual.Station AVL PUMA AVL InMotion AVL CONCERTO Kieran McAleer 9th AVL Calibration Symposium 4 11 월

39 FLEXIBILITY ENABLES EFFICIENCY Transient Profiles Calibration Environment Testbed Virtual Testbed (closed loop) Actuator s Vehicle Validation Sensors Production Tolerances Aging effects Kieran McAleer 9th AVL Calibration Symposium 4 11 월

40 AVL VIRTUAL TEST FIELD Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 4

41 WORKFLOW THE MODEL PARAMETRIZATION Model Parametrization Validation Speed [rpm] Torque [Nm] NOx [g/h] Temp. Us. TC [ C] CO2 [kg/h] Measurement Simulation UDC 1 UDC time [s] INPUT DATA EUDC Engine Maps NOx [mg/km] CO2 [g/km] & Cold/Hot NEDC model measurement deviation 6.23 % 2.6 % EUDC time [s] Speed [rpm] Torque [Nm] NOx [g/h] Temp. Us. TC [ C] CO2 [kg/h] Measurement Simulation UDC 1 UDC 1 Speed [rpm] Torque [Nm] NOx [g/h] Temp. Us TC [ C] time [s] 3 CO2 [kg/h] Measurement Simulation Vehicle Measurement Simulation EUDC NOx [mg/km] CO2 [g/km] model measurement deviation 6.23 % 2.6 % EUDC time [s] NOx [mg/km] CO2 [g/km] model measurement deviation 4.5 % 2.5 % time [s] Kieran McAleer 9th AVL Calibration Symposium 4 11 월

42 WORKFLOW THE MODEL PARAMETRIZATION Model Parametrization Validation Speed [rpm] Torque [Nm] NOx [g/h] Temp. Us. TC [ C] CO2 [kg/h] Measurement Simulation UDC 1 UDC time [s] AVAILABLE INPUT DATA AND FEW MORE MEASUREMENTS EUDC Engine Maps NOx [mg/km] CO2 [g/km] 4 15 & 3 UDC Cold/Hot NEDC HIGH ACCURACY WITH 6 model measurement deviation 6.23 % 2.6 % EUDC time [s] Speed [rpm] Torque [Nm] Temp. Us. TC [ C] Measurement Simulation UDC 1 DYNAMIC CAPABILITY NOx [g/h] CO2 [kg/h] Speed [rpm] Torque [Nm] NOx [g/h] Temp. Us TC [ C] time [s] 3 CO2 [kg/h] Measurement Simulation Vehicle Measurement Simulation EUDC NOx [mg/km] CO2 [g/km] model measurement deviation 6.23 % 2.6 % EUDC time [s] NOx [mg/km] CO2 [g/km] model measurement deviation 4.5 % 2.5 % time [s] Kieran McAleer 9th AVL Calibration Symposium 4 11 월

43 RESULTS - NEDC Vehicle Measurement and Virtual Test Bed comparison on NEDC: Torque [Nm] Speed [rpm] Measurement Simulation UDC 1 EUDC NOx [mg/km] CO2 [g/km] model measurement deviation 6.23 % 2.6 % 6 5 NOx [g/h] UDC 1 EUDC 1 Temp. Us. TC [ C] CO2 [kg/h] time [s] time [s] Kieran McAleer 9th AVL Calibration Symposium 4 11 월

44 RESULTS - ARTEMIS Vehicle Measurement and Virtual Test Bed comparison on ARTEMIS: Torque [Nm] Speed [rpm] Measurement Simulation NOx [mg/km] CO2 [g/km] model measurement deviation 4.5 % 2.5 % NOx [g/h] Temp. Us TC [ C] CO2 [kg/h] time [s] Kieran McAleer 9th AVL Calibration Symposium 4 11 월

45 NOx EO (%) Deviation (%) CO2 EO (%) Deviation (%) RESULTS - NEDC Vehicle Measurement and Virtual Test Bed comparison on NEDC: Base Calibration Cold Test Base Calibration Hot Test NOx Optimized Calibration Hot Test NEDC NOx EO NEDC CO2 EO Vehicle Virtual TB Vehicle Virtual TB 14% 5.% 14% 3.% 12% 4.% 12% 25.% 1% 8% 3.% 1% 8% 2.% 15.% 6% 4% 2% 1% 1% 9% 114% 9% 14% 65% 6% 62% 2.% 1.% 6% 4% 2% 1% 4% 14% 91% 97% 91% 7% 91% 1.% 5.% % COLD HOT HOT - NOX OPTIMIZED.% % 1%.% COLD HOT HOT - NOX OPTIMIZED Kieran McAleer 9th AVL Calibration Symposium 4 11 월

46 NOx EO (%) Deviation (%) CO2 EO (%) Deviation (%) RESULTS - NEDC Vehicle Measurement and Virtual Test Bed comparison on NEDC: Base Calibration Cold Test Base Calibration Hot Test NOx Optimized Calibration Hot Test NEDC NOx EO NEDC CO2 EO 14% 12% Vehicle Virtual TB MAX 1% 5.% 14% DEVIATION 4.% 12% Vehicle Virtual TB 3.% 25.% 1% 8% 3.% 1% 8% 2.% 15.% 6% 4% 2% 1% 1% 9% 114% 9% 14% 65% 6% 62% 2.% 1.% 6% 4% 2% 1% 4% 14% 91% 97% 91% 7% 91% 1.% 5.% % COLD HOT HOT - NOX OPTIMIZED.% % 1%.% COLD HOT HOT - NOX OPTIMIZED Kieran McAleer 9th AVL Calibration Symposium 4 11 월

47 NOx EO (%) Deviation (%) CO2 EO (%) Deviation (%) WLTC - RESULTS Vehicle Measurement and Virtual Test Bed comparison on WLTC: NOx Optimized Calibration Cold Test Base Calibration Hot Test NOx Optimized Calibration Hot Test NOx Optimized Calibration Step 2 Hot Test WLTC NOx EO WLTC CO2 EO Vehicle Virtual TB Vehicle Virtual TB 14% 3.% 14% 3.% 12% 25.% 12% 25.% 1% 2.% 1% 2.% 8% 6% 4% 2% % 1% 1% % COLD - NOX OPTIMIZED 144% 134% 7% HOT 111% 16% 5% HOT - NOX OPTIMIZED 9% 89% 15.% 1.% 5.% 1%.% HOT - NOX OPTIMIZED STEP 2 8% 6% 4% 2% % 1% 13% 11% 12% 96% 98% 97% 97% 3% COLD - NOX OPTIMIZED 1% HOT 2% HOT - NOX OPTIMIZED 15.% 1.% 5.% %.% HOT - NOX OPTIMIZED STEP 2 Kieran McAleer 9th AVL Calibration Symposium 4 11 월

48 NOx EO (%) Deviation (%) CO2 EO (%) Deviation (%) WLTC - RESULTS Vehicle Measurement and Virtual Test Bed comparison on WLTC: NOx Optimized Calibration Cold Test Base Calibration Hot Test NOx Optimized Calibration Hot Test NOx Optimized Calibration Step 2 Hot Test WLTC NOx EO WLTC CO2 EO 14% 12% Vehicle Virtual TB DEVIATION 3.% < 1% 14% 25.% DYNAMIC!!! 12% Vehicle Virtual TB 3.% 25.% 1% 2.% 1% 2.% 8% 6% 4% 2% % 1% 1% % COLD - NOX OPTIMIZED 144% 134% 7% HOT 111% 16% 5% HOT - NOX OPTIMIZED 9% 89% 15.% 1.% 5.% 1%.% HOT - NOX OPTIMIZED STEP 2 8% 6% 4% 2% % 1% 13% 11% 12% 96% 98% 97% 97% 3% COLD - NOX OPTIMIZED 1% HOT 2% HOT - NOX OPTIMIZED 15.% 1.% 5.% %.% HOT - NOX OPTIMIZED STEP 2 Kieran McAleer 9th AVL Calibration Symposium 4 11 월

49 AVL Diesel-Calibration Quality Gates ENGINE PROTECTION AND AMBIENT CORRECTIONS ON A VIRTUAL TESTBED Conventional Approach 1 Test Bed Time in Hours 15 Calibration Calibration Validation Validation Frontloading 8% Test Bed Time saved! Model Based Approach Cal. Quality G5 Validated G4 Robust G3 Test Bed Target Achievement G2 Emission Targets Achieved G1 Engine Runnable G Engine Startable conventional Calibration process model based Calibration process Quality Frontloading SOP + Additional effort for model set-up and parameterization has to be considered Time Model Based Calibration Approach: Calibration on XiL instead of Test Bed 8% Test Bed Time Saved per Engine Mode Test Bed available for Frontloading Tasks Dataset Quality & Maturity increased in earlier phase of Development Kieran McAleer 9th AVL Calibration Symposium 4 11 월 215 5

50 THE AVL VIRTUAL TESTBED BASED ON MOBEO-CRUISE M MODELING Fully Transient Capable Test Repeatability Open to any kind of test at any environment condition Flexible to be adapted for variant calibration Kieran McAleer 9th AVL Calibration Symposium 4 11 월

51 CONCLUSIONS AND SUMMARY Hardware in the Loop Chassis Dyno Being RDE Ready is truly a Challenge! We take care Kieran McAleer 9th AVL Calibration Symposium 4 11 월