OPTIMIZATION STUDIES OF ENGINE FRICTION EUROPEAN GT CONFERENCE FRANKFURT/MAIN, OCTOBER 8TH, 2018 M.Sc. Oleg Krecker, PhD candidate, BMW B.Eng. Christoph Hiltner, Master s student, Affiliation BMW
AGENDA 1 2 3 4 Motivation and objective Improvements on preliminary results Friction reduction studies with GT-SUITE s Integrated Design Optimizer Conclusion and further developments Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 2
MOTIVATION other losses mech. power Real boundary conditions High Transient cycles Total engine fired Test 1-cylinder fired (floating liner) Total engine / strip-down Friction Friction Measurement data (motored and fired) Speed Simulation data Thermal engine model 1D Simulation warm up & fuel consumption within driving cycles Piston assembly Crankshaft Oil + vacuum pump Temperature TH_ZST_1_2_05 [ C] Low 150 100 50 Low Warm up Cylinder Liner Temperature, NEDC m.th_zst_1_2_05 0 0 200 400 600 800 1000 1200 Zeit [s] Time [s] Single components motored High Data resolution & reproducibility Predicted Correlation & validation Measured 0D/1D Friction Simulation Total engine GT-SUITE Final result: Impact on CO 2 - emissions Cylinder head + chain drive Belt drive Predictive evaluation of concepts and trends in engine friction reduction As simple as possible and as complex as necessary Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 3
OBJECTIVE 0D/1D Friction Simulation Closeness of basic conditions to reality High Transient cycles Requirements Fast prediction Agile transferability Relative comparison Systems Total engine fired 1-cylinder fired (floating liner) Total engine / strip-down Physical evaluation Advanced parameter analysis using GT-SUITE Integrated Design Optimizer (IDO) Validation Calibration Optimization Friction Low Low Crucial for model calibration Single components motored High Possible measuring resolution & reproducibility Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 4
AGENDA 1 2 3 4 Motivation and objective Improvements on preliminary results Friction reduction studies with GT-SUITE s Integrated Design Optimizer Conclusion and further developments Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 5
PRELIMINARY RESULTS TOTAL ENGINE FRICTION, STRIP-DOWN TEST Friction Torque [Nm] Strip-Down Test, 90 C 1000 2000 3000 4000 Gap due to: Friction distribution: previous model assumption Missing simulation of chain friction. Non-validated belt drive friction model. Inaccuracies between single state strip-down measurements, total engine friction behavior and its equivalent simulation. Total Engine Test Rig Transmission Oil- & Vacuum Pump Water pump Acc. Belt Drive Inlet Camshaft + Valvetrain Exhaust Camshaft + Valvetrain Balancer Shaft Piston Skirt Piston Rings Small End Bearing Big End Bearing Main Bearings Seals Experiment Simulation Buildup identification Validation Necessary model improvements: Further investigation on model parameters Validation of single camshaft friction identification Extend measurements for validation Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 6
PRELIMINARY RESULTS TOTAL ENGINE FRICTION, STRIP-DOWN TEST Friction Torque [Nm] Strip-Down Test, 90 C 1000 2000 3000 4000 Gap due to: Friction distribution: previous model assumption Missing simulation of chain friction. Non-validated belt drive friction model. Inaccuracies between single state strip-down measurements, total engine friction behavior and its equivalent simulation. Total Engine Test Rig Transmission Oil- & Vacuum Pump Water pump Acc. Belt Drive Inlet Camshaft + Valvetrain Exhaust Camshaft + Valvetrain Balancer Shaft Piston Skirt Piston Rings Small End Bearing Big End Bearing Main Bearings Seals Experiment Simulation Buildup identification Validation Necessary model improvements: Further investigation on model parameters Validation of single camshaft friction identification Extend measurements for validation Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 7
CHAIN DRIVE FRICTION MODEL Modelling Validation Method Strip-Down states Single component motored subtract Torque transducer subtract Torque transducer Camshaft + valvetrain friction Main friction/ power loss parameters In chain-guide/sprocket contacts: - Friction coefficient guides - Friction coefficient sprockets In chain links: - Longitudinal damping - Torsional damping IDO Calibration Power loss chain drive [W] 0 1000 2000 3000 4000 5000 Feasible magnitude and slope of chain drive power loss. Inaccuracies due to missing system interdependency (e.g. chain tensioner dynamics). Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 8
FEAD FRICTION MODEL Modelling Validation Method Single component motored Friction torque of each accessory Torque transducer Main friction/ power loss parameters In belt-pulley contacts: - Friction coefficient - Contact damping In belt properties: - Bending stiffness - Shearing stiffness - Longitudinal damping IDO Calibration Feasible magnitude FEAD Power loss FEAD [W] 0 1000 2000 3000 4000 5000 6000 power loss. Inaccuracies due to missing system interdependency (e.g. accessory roller bearing). Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 9
CURRENT RESULTS TOTAL ENGINE FRICTION, STRIP-DOWN TEST Buildup identification Validation Necessary model improvements: identification Extend measurements for validation Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 10
Friction Torque [Nm] CURRENT RESULTS TOTAL ENGINE FRICTION, STRIP-DOWN TEST Strip-Down Test, 90 C 1000 2000 3000 4000 Total engine, measured Remaining components, measured Main components, simulated Test Rig Transmission Oil- & Vacuum Pump Water pump Acc. Belt Drive Chain Drive Cylinder Head Unit Piston Assembly Balancer Shaft Crankshaft - Correlation of measurement and simulation has been improved. - Single component friction simulation shows feasible agreement to measured data. - Note: magnitude and trend of single component friction has to be questioned critically if compared to total engine friction losses. Cylinder Head + Chain Balancer Shaft Crankshaft Friction Torque [Nm] Friction Torque [Nm] Friction Torque [Nm] 1000 2000 3000 4000 1000 2000 3000 4000 1000 2000 3000 4000 Buildup identification Validation Necessary model improvements: identification Extend measurements for validation Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 11
AGENDA 1 2 3 4 Motivation and objective Improvements on preliminary results Friction reduction studies with GT-SUITE s Integrated Design Optimizer Conclusion and further developments Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 12
GT-SUITE INTEGRATED DESIGN OPTIMIZER SETTINGS FOR FEAD MODEL 1 st loop 2 nd loop Variables for calibration i = 15 Variables for power loss reduction i = 7 Variables for calibration i = 7 Example parameters: LuGre friction coefficient, belt damping coefficient, belt-pulley connection damping ratio, Search algorithm Population size 50 Number of generations Genetic, NSGA-III 10 Example parameters: LuGre friction coefficient, belt damping coefficient, belt shear stiffness, belt axial stiffness, Objective function [W]*10 5 Objective function [W] 4 3 2 1 0 1000 800 600 400 200 0 300 600 Design [-] 0 0 300 600 Design [-] Power loss [W] Calibrated Best design Base design Target design Power loss [W] (*hypothetical, no-constraint study) 0 2000 4000 6000 Calibrated design No-constraint design 0 2000 4000 6000 Constraint design (*) Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 13
VARIATION STUDIES FEAD FRICTION MODEL Belt pre-tension Belt layout Higher pre-tension leads to more power losses. Good correlation in mid to high speed ranges. Further investigation on low speed power losses necessary. Less power losses by removing a pulley and changing the belt layout (e.g. remove water pump in case it is electrically driven). Good correlation in mid to high speed ranges. Further investigation on low speed power losses necessary. meas., 222N meas., 316N sim., 222N sim., 316N meas., 222N meas., 217N, 2 nd Layout sim., 222N sim., 217N, 2 nd Layout Power loss FEAD [W] Power loss FEAD [W] 0 1000 2000 3000 4000 5000 6000 0 1000 2000 3000 4000 5000 6000 Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 14
OPTIMIZATION STUDIES FEAD FRICTION MODEL FEAD power loss minimization Calibrated design No-constraint design Constraint design Idea Results Critical conclusion Inner belt power dissipation is determined by the belt material properties. Soft belt (less axial stiffness, less shear stiffness) might decrease power losses. comparison base design vs. best design: 1. Axial stiffness & LuGre friction coeff. comparable magnitude. 2. Shear stiffness & belt damping significantly lower in best design. More than 80% power loss reduction. Torque transmission of soft belt still sufficient? Belt slip rate? Functionality of accessories assured? Fatigue strength in long terms? Wear? Power loss FEAD [W] 0 1000 2000 3000 4000 5000 6000-80% power loss Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 15
METHODOLOGY WORKFLOW IMPROVEMENTS Enhanced simulation engineering Validation identification Calibration Optimization Friction f(x1, x2 ) Simulation model categorizing Assumptions Tuning factors Known parameters Sensitivity analysis Interdependency Sensitivity Variables for calibration Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 16
AGENDA 1 2 3 4 Motivation and objective Improvements on preliminary results Friction reduction studies with GT-SUITE s Integrated Design Optimizer Conclusion and further developments Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 17
CONCLUSION AND FURTHER DEVELOPMENTS Conclusion All major friction components of a modern petrol engine have been modelled within GT-SUITE. GT-SUITE s Integrated Design Optimizer is a powerful tool for extensive parameter studies of each friction sub-model. But comprehensive definition of parameter range and magnitude is challenging (usually due to lack of data). Current workflow of model parameter studies will be enhanced by extended sensitivity analyses. Further developments Final parameter freeze of friction sub-models Model calibration and proof of friction prediction Validation of secondary model outputs (besides friction) Development of a friction optimized engine design concept Further investigation on design regarding feasibility Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 18
THANK YOU! Optimization Studies of Engine Friction Oleg Krecker October 8th 2018 Slide 19