Simulation of Collective Load Data for Integrated Design and Testing of Vehicle Transmissions Andreas Schmidt, Audi AG, May 22, 2014
Content Introduction Usage of collective load data in the development process of transmissions Improvements to the collective load simulation model Engine model Driver model Track model Model validation Integration of the collective load simulation in the development process Outlook Summary 2
Challenges in the Development of Vehicle Transmissions enhanced comfort lower losses more functions higher engine power electrification contradictory requirements lower costs lower weight higher damage to components decreasing safety factor Reliable collective load data is required to meet increasing specifications Integration of collective load data into the design process is essential to save development time and costs 3
Relevant Test Cases and Duty Cycles Ehra durability test (EWP) Defined drive cycle on Ehra test track Volkswagen Group approval test Approx. 2/3 on rural/city roads Various road conditions Various slopes 5-30 % Sharp accelerations Corrosion tests Engine start-stop tests Approx. 1/3 high speed track Special test cases Off-road test track Misuse cases Engine stall Curb scuffing Sand pit Race tracks (e.g. Nürburgring) Test cases have proved to be representative for customer load Basis of collective loads in transmission design for more than 20 years Focus for the following presentation is set on the Ehra durability test (EWP) 4
Integration of Collective Load Data in the Design Process Main steps in the design process EWP collective load data design and calculation of gearbox components component test (gears, bearings, shafts, ) EWP durability test on powertrain test rig time Collective load data is used for: Design and calculation of transmission components Definition of component tests (usually in cooperation with the supplier) Definition of durability tests performed on powertrain test rigs 5
Comparison between Measurement and Simulation Vehicle Measurement Vehicle Simulation + Established process + Vehicle interdependencies are included - Available only for existing powertrain layouts - Environmental influence - Engine power or gears often do not correspond to the actual layout - Expensive and time-consuming - Highly accurate collective load data not available until the end of the design process + Available early in the design process + Variation of vehicle-/gearbox parameters possible with low effort + Rapid generation of collective load data + Less prototype cars required - Application data not available in the early design process assumptions have to be made - Dynamic effects can currently not be simulated (no multi-body-simulation model) In the early design phase, simulation is the preferred method to generate load data 6
Accuracy of Collective Load Data accuracy of load collective load collective simulation Goal load collective simulation measurement of prototype or preseries durability test of gearbox SOP measurement of predecessor development progress Usage of simulation in the early design phase to obtain representative results of high accuracy Collective load simulation has its major benefit in the early design phase Simulation model must be sufficiently detailed to provide accurate results Simulation model must be enhanced to meet the increased requirements 7
Overview of the Simulation Model Longitudinal simulation in MATLAB/Simulink R2010 (32-bit) Differential equations for engine and vehicle inertia Map-based transmission and engine losses Engine Detailed Transmission Model Vehicle Engine power Torque build-up Engine inertia Dual Clutch PI-Controller Application data Gearbox Gear ratios Efficiency map Shift logic Inertia Road resistance Axle load shift Tyre slip Tyre inertia Aerodynamic resistance Driver 30 Environment Acceleration pedal Brake pedal Gear (Manual Transmission) PI-Controller + statistics 25 20 15 10 5 EWP measurement 0 10 20 30 40 50 60 70 80 90 100 Target velocity Road friction and slope Curve radius ac.pedal position [%] Very fast simulation environment: Approx. 5-10 min. per EWP simulation run 8
Improvements to the Simulation Model Improvements were implemented for: Engine model Driver model Environment model Engine Detailed Transmission Model Vehicle Engine power Torque build-up Engine inertia Dual Clutch PI-Controller Application data Gearbox Gear ratios Efficiency map Shift logic Inertia Road resistance Axle load shift Tyre slip Tyre inertia Aerodynamic resistance Driver 30 Environment Acceleration pedal Brake pedal Gear (Manual Transmission) PI-Controller + statistics 25 20 15 10 5 EWP measurement 0 10 20 30 40 50 60 70 80 90 100 Target velocity Road friction and slope Curve radius ac.pedal position [%] 9
1. Engine Model: Dynamic Torque Build-Up Load Step @ 1500 rpm Highly supercharged turbo engines show dynamic, speed dependent torque buildup especially at low engine speeds Heavy influence on collective loads of transmissions with turbo engines T engine [Nm] T 1500 = 0.80 s time [s] Measurement Simulation Load Step @ 1750 rpm Engine model enhanced with a parameter based model for torque build-up Parameter identification to fit the model parameters to engine measurements No physical modelling of the engine necessary increased simulation speed Improved engine model: T engine [Nm] Measurement T 1750 = 0.40 s Simulation time [s] Load Step @ 2000 rpm load signal Engine torque T base T turbo PT 1/2 PT 1/2 + T engine T engine [Nm] T 2000 = 0.35 s time [s] Measurement Simulation 10
1. Engine Model: Verification with Measurements Good match of acceleration measurement and simulation Vehicle Speed and Acc. Pedal [km/h] / [% acc. pedal] 50 0 time [s] Engine and Gearbox Input Speed Acc. Pedal Veh. speed measurement Veh. speed simulation [1/min] 3000 0 time [s] Sum of Wheel Torque (4WD) Gearbox input speed measurement Engine speed measurement Gearbox input speed simulation Engine speed simulation [Nm] 3000 0 time [s] Measurement Simulation 11
2. Driver Model: Analysis of Test Drivers According to the specification of the EWP durability test, drivers are advised to accelerate strongly at all times. This is not always possible due to: Weather conditions (snow, rain, ) Time of day (continuous testing during 3 shifts) Traffic on the test track Engine power (especially in sports cars) rel. frequency [%] 50 40 30 20 10 MIN MEAN MAX Evaluation of 6 Drivers Comparison of rel. damage Mean 100 % Min 48 % Max 147 % 0 20 30 40 50 60 70 80 90 100/KD acc. pedal [%] during accelerations > 1 m/s² Driver model needs to be enhanced to represent the average test driver behavior 12
2. Driver Model: Verification with Measurements Driver modelling based on statistical analysis of the real driver s behavior Evaluation of the Audi durability test database to create driver models for relevant vehicle classes (passenger cars, SUV, sports cars, ) and engine power Distribution of Acc. Pedal Positions During Accelerations 30 EWP measurement 25 EWP simulation relative frequency [%] 20 15 10 5 0 10 20 30 40 50 60 70 80 90 100 acc. pedal position [%] Simulated acc. pedal distribution fits the measurement very well 13
3. Improvements to the Track Model Accurate modelling of the Ehra test track Velocity profile Driving specifications (brake or accel. phases) Track angle Road type Curvature Simulated test track can be assembled in any order Comparison to measurements Changes in the track definition can be regarded 14
Model Verification (1/2) Distribution of engine operating points at gearbox input corresponds very well Ehra durability test operates the engine at high power (1) Constant velocity on high speed track leads to identical operating point (2) Real driver has a more mixed collective at low loads and speeds (3) Minor impact on damage 350 300 Engine Operating Points (Measurement) Engine Operating Points (Simulation) 350 1 1 300 Engine Torque [Nm] 250 200 150 100 2 Engine Torque [Nm] 250 200 3 3 150 100 2 50 50 0 1000 2000 3000 4000 5000 6000 Engine Speed [rpm] 0 1000 2000 3000 4000 5000 6000 Engine Speed [rpm] 15
Model Verification (2/2) Very good correlation between simulated and measured damage 6000 5000 4000 Collective Load (Sum of Front + Rear Axle) 3000 torque (sum front + rear axle) [Nm] 2000 1000 Simulation Measurement Relative damage from simulation is 5 % higher than from measurements of average test drivers EWP measurements from test database EWP sim. ("average test driver") Hypothetical fatigue limit for bearings Hypothetical fatigue limit for gears 10 3 10 4 10 5 10 6 10 7 10 8 number of load cycles Results from improved simulation model can be used in the early design phase 16
Integration of Simulation into the Design Process EWP load collective measurement of predecessor powertrain EWP measurements of pre-series powertrain design and calculation of gearbox components component test (gears, bearings, shafts, ) EWP durability test on powertrain test rig Validation EWP load collective simulation of all relevant powertrain configurations time Collective load data from simulations is used as a basis for design and testing of transmissions throughout the entire development process Validation of simulation results with measurements (if available) is essential Simulation has become an integral part of the development process 17
Outlook EWP load collective measurement of predecessor powertrain EWP measurements of pre-series powertrain design and calculation of gearbox components component test (gears, bearings, shafts, ) EWP durability test on powertrain test rig Validation EWP load collective simulation of all relevant powertrain configurations time Increasing number of transmission variants (conventional, HEV, PHEV) Development of alternative powertrains (more/fewer gears, ) Simulation can reduce the costs of increasingly complex developments Less vehicle measurements needed High informative value in the early stages of development Faster update of collective loads in case of changes to the requirements or specifications 18
Summary Collective load data is an essential part of transmission development Basis for design and approval of Audi transmissions Preferred method to generate load collectives in the early design phase Improvements of the simulation model are essential to generate valid collective load data Engine model including dynamic torque build-up Statistical driver model based on real driver behavior Detailed track and environment model of the test track Model validation shows a good match between measurement and simulation Collective load simulation has become an integral part of the transmission development process at Audi Continuous validation of the simulation model with measurements is essential Integrated design and testing using collective load simulations in the entire process With increasing complexity in transmission development, usage of simulation will become even more important to generate collective load data 19
Thank you very much! Dipl.-Ing. Andreas Schmidt with kind support by Martin Arbesmeier, M.Sc. Dr.-Ing. Alexander Schmidt Audi AG, I/EA-442 Transmission Simulation D-85045 Ingolstadt GERMANY 20