Institute for Internal Combustion Engines and Powertrain Systems, TU Darmstadt

Institute for Internal Combustion Engines and Powertrain Systems, TU Darmstadt Hauke Maschmeyer, M.Sc.; Prof. Dr. techn. Christian Beidl Methodology Approaches for Real Driving Emissions-Development at Engine Testbeds Extended Version will be presented at: 7th International Symposium on Development Methodology 14th -15th of November, 217 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 1

Real-world Road as new Reference System - RDE is Uncertainty and Randomness Darstellung: Auto Bild Weather Driver Behaviour Height Darstellung: Welt.de Vehicle & Powertrain Concept Randomness RDE Route Traffic Darstellung: fotocommunity.de Darstellung: haeuslerautomobil-gmbh.de Darstellung: faces.ch Loss of Comparison Basis Not reproducible Road Signals & Signs 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 2

Conclusion: Cycle-based vs. RDE-Legislation Loss of a Clear Target How can be assured that RDE is passed? How can system-robustness be achieved? When is the job finished? What are suiting RDE KPIs? What is my engineering target? On what basis can I compare different concepts, functions & components in order to get a valid result for RDE? What is RDE representative? What are RDE concept-kpis? How can I consider and prioritize all realworld influences without being inefficient? How can Over-Engineering be avoided? How can new RDE-approaches be integrated into existing environments? 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 3

Module Real-world Statistics Operating Point (OP)-& OP Change-Histogram Goal: Avoidance of Over-Engineering Analysis of existing PEMS real-world testdrive data from Data Management Identification of relevant OPs and OP- Changes as well as their frequencies Programmed with Concerto-Script TÜV Hessen Soft DUC Soft Normalized Laod Normalized Laod TÜV Hessen Aggr. DUC Aggressiv Normalized Speed Normalized Speed 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 5

Module Real-world Statistics Operating Point (OP)-& OP Change-Histogram Quantified Operating Point Histogram: Histogram for time and distance Specific for powertrain concept In this case analysis of: 92 PEMS tests (warm & cold) 4 routes & 2 vehicles (B-& C-Class) 3 drivers Normalized Laod Time Share in % Normalized Speed Distance Share in % Load change[%/s] Normalized Load Speed change[rpm/s] Normalized Speed 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 6

Module Stationary Validation What is the Engineering Target for stationary RDE Validation? Goal: Assurance of stationary emission conformity Stationary map calibration and measurement of all necessary n/alpha-maps Real-world-Statistics Module provides for vehicle / engine-specific OP frequency for calculations OP-specific emission target EG RDE test drive / homologation estimation with stationary emission maps Real-world driving dominated by almost stationary behaviour: ±1 rpm/s & ±1 %/s 72% oftest Load change[%/s] Reaktionsgrößen: Reaction Variables Alpha [%] 1 75 5 25 Abgasmassenstrom Stickoxide.2.4.5.3.1.6.7.2.8.1.1.1 1 2 3 4 5 Drehzahl 7 15 35 175 28 14 315 175 21.9 35.1 21 28.11.12.12 1 2 3 4 5 Drehzahl [rpm] 35 175 35 245 14 35 15 35 7 245 28 + = Nomierte Last in % 1. 9. 8. 7. 6. 5. 4. 3. 2. 1. Erfüllungsgrad EG = Result/ Limit (Goal: EG<=1): 3. 12. 14. 17. 19. 5. 8. 2. 3. 2. 3. 5. 5. 8. 8. 5. 3. 2. 5. 1. 1.1 3. 3. 8. 3. 14. 17. 19. 12. 8. 12. 14. 17. 19. 3. 5. 1. 1.1 2. 3. 5. 8. 3. 2. 8. 2. 2. 2. 5. 2. 3. 3. 2. 3. 1. 1.1 2. 2. 1. 1.1 1. 1.1 1. 1.1.. 1. 2. 3. 4. 5. 6. 7. 8. 9. 1. Normierte Drehzahl in % 12. 8. 1. 1.1 2. 3. 5. 3. 2. 1. 1.1 1 1.1 2 3 5 8 12 14 17 19 3 Speed change[rpm/s] Stationary behaviour as comparison basis for subsequent Module Dynamic Validation 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 8

Module Stationary Validation Stationary Behaviour Compared to RDE Limits in g/km THC CO OP-specific emission target EG: Normalized Load [%] Nomierte Last in % Normalized Load [%] Nomierte Last in % 1. 9. 8. 7. 6. 5. 4. 3. 2. 1. 3. 2. 1. 1.1 12. 5. 1. 1.1 2. 3. 3. 17. 19. 19. 14. 3. 17. 12. 14. 12. 8. 5. 8. 12. 3. 8. 2. 8. 8. 5. 14. 2. 12. 12. 3. 14. 8. 14. 5. 17. 8. 17. 19. 19. 8. 12. 14. 12. 12. 17. 3. 19. 3. 14. 17. 14... 1. 2. 3. 4. 5. 6. 7. 8. 9. 1. Normierte Drehzahl in % 1. 9. 8. 7. 6. 5. 4. 3. 2. 1. Normalized Speed [%] 19. 17. 12. 14. 12. 8. 5. 3. 5. 2. 3. 2. 1.1 1. 1.1 1. 12. 8. 1. 8. 8. 1.1 14. 1.1 1. 12. 12. 14. 14. 1.1 1. 17. 17. 19. 2. 2. 19. 2. 3. 3. 3. 3. 3. 17. 5... 1. 2. 3. 4. 5. 6. 7. 8. 9. 1. Normierte Drehzahl in % 1 1.1 2 3 5 8 12 14 17 19 3 1 1.1 2 3 5 8 12 14 17 19 3 NO x CO 2 Normalized Load [%] Nomierte Last in % Normalized Load [%] Nomierte Last in % 1. 9. 8. 7. 6. 5. 4. 3. 2. 1. 12. 5. 12. 8. 8. 14. 8. 12. 12. 14. 14. 17. 17. 19. 19. 3. 3. 17... 1. 2. 3. 4. 5. 6. 7. 8. 9. 1. Normierte Drehzahl in % 1. 9. 8. 7. 6. 5. 4. 3. 2. 1. Normalized Speed [%] 12. 12. 1. 12. 1.1 14. 2. 3. 17. 5. 17. 14. 12. 8. 5. 3. 2. 3. 2. 12. 8. 1. 1.1 2. 8. 8. 14. 1.1 1. 3. 2. 12. 12. 3. 14. 14. 5. 17. 17. 19. 19. 8. 12. 5. 14. 17. 3. 19. 3. 17... 1. 2. 3. 4. 5. 6. 7. 8. 9. 1. Normierte Drehzahl in % Normalized Speed [%] Normalized Speed [%] 1 1.1 2 3 5 8 12 14 17 19 3 1 1.1 2 3 5 8 12 14 17 19 3 EG: 1 1.1 2 3 5 8 12 14 17 19 3 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 9

Module Stationary Validation Stationary Behaviour Compared to RDE Limits in g/km Stationary RDE-Prognosis: Critical EG-operating points might be compensated with uncritical ones Prognosis needed EG_stat_prog = RDE histogram based weighted mean value Goal: EG_stat_prog <= 1 basically traditional cycle estimation + = EG_stat_ prog: CO=23,4 + = EG_stat_ prog: CO 2 =6, + = EG_stat_ prog: THC=1,8 + = EG_stat_ prog: NO x =1,1 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 1

1.95 Module Legislative Cycle Validation / Estimation - Generation of Quasi-stationary Signals Transient OPs used for calculation of stationary map signals: Emissions and State Signals Also usedforco 2 -Estimation Hypothesis: Usually in transient operation the engine is not cleaner than during stationary behaviour n, alpha 1 Lambda [-].9 8 6 1 75 1-2.5 5 75 5 5 5 Alpha [%] 25 75 25 25 125 15 15 125 5 175 175 175 275 2 225 2 75 25 1 2 3 4 5 Drehzahl 4 2.5 3 1 7.5 5 Alpha [%] 4 12.5 15 17.5 2 25 3 3 Alpha [%] 2 32.5 32.5 2 3 1.5 1.35.8.75.85 1 1.2.9 1.5 1.9 iga_igc [ KW] 2 1 THC vkat [ppm].85.9 1 1 1 1 1 1.95.95 1.6 1.55 1.5 1.5 1.11.151.21.251.3 1.6 1.6 1.6 1.6 1.6 1.6 1.65 1.351.41.451.5 1.55 25 2 32.5 27.5 27.5 22.5 17.5 35 12.5 7.5 1 27.5 37.5 3 15 1 5 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 12

Module Legislative Cycle Validation / Estimation Quasi-stationary Simulation for KPI Determination Goal: Deduction of RDE Post-processing Parameters & Transient RDE KPIs Measurement of RDE-Reference Test (e.g. Most-Relevant-Test ofpriorproject) & WLTC for concept comparison Idea: high emission OPs might be neglectable due to few occurences Additional map based deduction of their quasi-stationary time dependent signals Difference between stationary emission estimation vs. measured data describes transient engine robustness Outcome: RDE KPIs for Comparisons EG stat = EM stat / Limit important RDE KPI Summary of stationary engine behaviour EG dyn =EM dyn /EM stat RDE KPI for transient robustness Summary of transient engine behavior 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 13

Module Transient Validation Transient Map Analysis in early Development Phase Goal: Detailed influence assessment of transients and states on engine emission behaviour Robust RDE engine concept independent of vehicle basis Concept: DoE-based analysis of OP-ramps for different engine states If ICE is clean in transient behavior: no RDE Problems independent of vehicle type, road, driver behaviour Advantages: Systematic / holistic Concept-specific / Individual Applicable on standard engine testbeds Automatable & Reproducible Speed in rpm Load in % Disadvantages: No direct link to RDE Legislation Risk of Over-Engineering Time in s DoE-based variation of: Starting point & gradient of speed Starting point & gradient of load Coolant & oil temperature 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 15

Module Transient Validation - Difference to Quasi-stationary Signals as Transient Identification Conditioning systems needed measurement of average difference between transient and quasi-stationary behaviour DoE-based measurement of average difference of transient to quasi-stationary emission behaviour [g/s] Ramp time-signal converted to ramp KPI in Concerto post-processing script DoE-based modeling of differences dependent on ramp parameters 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 16

Choice of RDE critical Transients Assessment of Concept-Specifc RDE Relevance Statistic Relevance: Identification of remarkable ramps Difference in mg/s transient vs. stationary > Concept-specific assessmentof statistic relevance Histogram-classification Calculation of ramp weight Gusing individual ramp and OP frequencies Ramp prioritization (for further development) Pollutant specific ramp weight Global worst case transient CO Ramp Weight G CO Most-Relevant-Ramp: high EM difference & high statistic relevance Most-Relevant-Ramp time signal used for following optimization and technical cause identification 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 17

Choice of RDE critical Transients Identification of Technical Causes in Time Domain Ramp 138: CO_dynamic=16 g/km, CO_stationary=35 g/km,86,79 7,5 4 Visual comparisonof quasi-stationary and transient state signals allowanalysison cause of differences 16 11 85 53 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 18

Overview Tasks / Use-Cases for RDE 6 Methodology Approaches for RDE Development on ETBs Real-world Statistics Goal: Avoidance of Over- Engineering Concept:Concept-specific statistcs of realworld data (virt. & real) Result: RDE representative operating points & transients Stationary Validation Goal: Assurance of stationary emission conformity Concept:Assessment of stationary emission behaviourconsidering concept-specific OP frequencies Result: OP-specific RDE result and Homologation estimation Legislative Cycle Validation / Estimation Goal: Deduction of RDE Postprocessing parameters & Transient RDE KPI Concept:Comparison of stationary emission estimation vs. measured data Result: RDE KPI for Transient Robustness & Post-processing parameters Office Stationary Standard ETB Active Standard ETB RDE Sensitivity Analysis Goal: Detection of vehicle individual challenges Concept:Efficient assessment of all possible RDE influences on powertrain; Use of virt. RDE test drives on EiL-ETB Result: RDE Robustness KPI; all concept-specific critical maneuvers Transient Validation Goal: Influence Assessment of transients and states on engine emission behaviour Concept: DoE-based analysis of OPramps for different engine states Result: Identified critical ramps for engine concept and corresponding technical cause Engine-in-the-Loop ETB Active Standard ETB 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 19

Module RDE Sensitivity Analysis X-in-the-Loop RDE Tests on Engine-Testbeds with RDE-Module Goal: Identification of vehicle-specific challenges & engineering targets Testbed Interface Driver Model Desired Cruising Speed Testbed Automation System Set Engine Load Set ICE Throttle Meas. Torque Set ICE Speed Speed limitations Shifting, Pedalry & Steering Forces Vehicle Speed Vehicle Model Resistances & Information 3d Environment Model Signals Traffic Signs & Objects Road & Environment ICE Dyno Realtime System: Real Driving Simulation Module Automatable, reproducible & emission measurement Virtual real-world tests, flexible, holistic 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 2

Module RDE Sensitivity Analysis Definition of RDE relevant Variation Parameters Powertrain (real, virtual) remains unchanged Driving scenario parameters varied for model-based validation of powertrain: Driver Vehicle Environment Tested parameter intervals are concept-specific based on Module Real-world Statistics DoE-Design for efficient testing and optimized modeling Measurement of pollutant emissions for each segment and whole test Exemplary RDE relevant variation parameters: Max. longitudinal acceleration Max. lateral acceleration Desired driver speed Road inclination & decline Curve radius and angle Stoptime cw-value Additional vehicle load (Source: MTZ 1/216) 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 21

Module RDE Sensitivity Analysis From Time Domain to Interaction-Plot a pos_mean? a lat_mean CO 2 v mean Add. Load = kg Small vehicle car load EachPoint in Plot representswholetestin time domain Add. Load =35 kg Large vehicle Time based signals(e.g. emissions) is summed up by mean value Torque [Nm] 45 4 35 45 3 4 25 35 2 45 3 15 4 25 1 35 2 5 45 3 15 4 25 1-5 35 2 5-1 45 15 3-15 -5 VEHICLE 4 1 25 Comparison: -2 5-1 -25 35 2-15 -3 3 15-5 -2-35 25 1-1 -25-4 5 1 15 2 25 3 35 4 45 5 55 2 5-3 -15-35 15-2 -5-4 1-25 5 1 15 2 25 3 35 4 45 5 55-1 -3 5-15 -35-2 -4-5 5 1 15 2 25 3 35 4 45 5 55-25 -1-3 -15-35 -2-4 5 1 15 2 25 3 35 4 45 5 55-25 -3-35 -4 5 1 15 2 25 3 35 4 45 5 55 Most critical tests identified 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 22

Module RDE Sensitivity Analysis From Time Domain to Interaction-Plot a pos_mean If driver doesn t accelerate, additional load is not so critical a lat_mean CO 2 v mean Add. Load = kg car load EachPoint in Plot representswholetestin time domain Add. Load =35 kg Time based signals(e.g. emissions) is summed up by mean value Torque [Nm] 45 4 35 45 3 4 25 35 2 45 3 15 4 25 1 35 2 5 45 3 15 4 25 1-5 35 2 5-1 45 15 3-15 -5 VEHICLE 4 1 25 Comparison: -2 5-1 -25 35 2-15 -3 3 15-5 -2-35 25 1-1 -25-4 5 1 15 2 25 3 35 4 45 5 55 2 5-3 -15-35 15-2 -5-4 1-25 5 1 15 2 25 3 35 4 45 5 55-1 -3 5-15 -35-2 -4-5 5 1 15 2 25 3 35 4 45 5 55-25 -1-3 -15-35 -2-4 5 1 15 2 25 3 35 4 45 5 55-25 -3-35 -4 5 1 15 2 25 3 35 4 45 5 55 Most critical tests identified 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 23

Module RDE Sensitivity Analysis From Time Domain to Interaction-Plot a pos_mean a lat_mean CO 2 a pos_max =1 a pos_mean =,4 a pos_max =7 a pos_mean =,64 v desired =3 v mean =27 v desired =13 v mean =52,5 Powertrain is evaluated for all vehicles, drivers and roads Most critical tests identified v mean Interaction-Plot summarizes all tests car load Vehicle speed [m/s] a pos [m/s 2 ] 45 4 35 45 3 4 25 35 2 45 3 15 4 25 1 35 2 5 45 3 15 4 25 1-5 35 2 5-1 45 15 3-15 1-5 -2 DRIVER 4 25 Comparison: 5-1 -25 35 2-15 -3 3 15-5 -2-35 25 1-1 -25-4 5 1 15 2 25 3 35 4 45 5 55 2 5-3 -15-35 15-2 -5-4 1-25 5 1 15 2 25 3 35 4 45 5 55-1 -3 5-15 -35-2 -4-5 5 1 15 2 25 3 35 4 45 5 55-25 -1-3 -15-35 -2-4 5 1 15 2 25 3 35 4 45 5 55-25 -3-35 -4 5 1 15 2 25 3 35 4 45 5 55 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 24

Overview Tasks / Use-Cases for RDE 6 Methodology Approaches for RDE Development on ETBs Real-world Statistics Goal: Avoidance of Over- Engineering Concept:Concept-specific statistcs of realworld data (virt. & real) Result: RDE representative operating points & transients Stationary Validation Goal: Assurance of stationary emission conformity Concept:Assessment of stationary emission behaviourconsidering concept-specific OP frequencies Result: OP-specific RDE result and Homologation estimation Legislative Cycle Validation / Estimation Goal: Deduction of RDE Postprocessing parameters & Transient RDE KPI Concept:Comparison of stationary emission estimation vs. measured data Result: RDE KPI for Transient Robustness & Post-processing parameters Office Stationary Standard ETB Active Standard ETB Most-Relevant-Test Goal: Condensation of gained concept-specific challenges to one RDE representative test Concept:Merging critical test segments or maneuvers to RDE valid test & elimination of uncritical parts Result: (Short) RDE representative test as new comparison basis for existing engine testbeds RDE Sensitivity Analysis Goal: Detection of vehicle individual challenges Concept:Efficient assessment of all possible RDE influences on powertrain; Use of virt. RDE test drives on EiL-ETB Result: RDE Robustness KPI; all concept-specific critical maneuvers Transient Validation Goal: Influence Assessment of transients and states on engine emission behaviour Concept: DoE-based analysis of OPramps for different engine states Result: Identified critical ramps for engine concept and corresponding technical cause Active Standard ETB Engine-in-the-Loop ETB Active Standard ETB 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 25

Module Most-Relevant-Test - Generation of New RDE-Evaluation Basis / RDE Reference Test Goal: Condensation of gained concept-specific challenges to one RDE representative test What street do you want to drive on and how? How can be assured that RDE certification is passed? What is the comparison basis for concepts, functions and components? Concept: Development basis should be a worst-case test If passed, most likely every other certification test will be passed Test must be individual due to concept-specific challenges Test must be fully relevant for RDE Procedure Most-Relevant Test Scenario Test ismergedof critical test segments or maneuvers from prior modules elimination of uncritical parts 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 26

Module Most-Relevant-Test - Process Structure for Generation of RDE Reference Test Crit. Maneuvers & States from Sensitivity Analysis Literature Statistics Module Maneuver Database Consistent Data Management Crit. Virtual Maneuvers Trigger- Mechanisms Emission critical maneuvers Simulation Environment CO 2 Prognosis Measurement Scenario Generation Test Scenario Definition Simulative Pre-testing & Legislative Relevance Test Definition ready for utilization? CO 2 representative Engine Characterisation on ETB Initial states reenacted? Data Evaluation Raw data Maneuvers identified as crititcal Generation of Most-Relevant-Test Set-Point Signal Deduction Most-Relevant Test Scenario Most-Relevant Test Cycle 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 27

Summary RDE is completely different from NEDC legislation RDE means many more influences & randomness more work and uncertaintyfor the development These challenges are ideally addressed primarily on existing engine testbeds Therefore, 6 RDE-Methodology Modules are defined to enable as early and as much RDE development as possible DoE-approaches allow holistic analysis, yet are efficient Concept-specific statistics avoid over-engineering 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 28

Thank you for your Attention! Contact: Hauke Maschmeyer, M.Sc. Website: www.verbrennungskraftmaschinen.de Tel.: +49 6151 16-21263 Fax: +49 6151 47-63 maschmeyer@vkm.tu-darmstadt.de 18. Mai 217 Institute for Internal Combustion Engines and Powertrain Systems H. Maschmeyer 29