VVT optimization with GT-POWER and Genetic Algorithms
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
VVT optimization with GT-POWER and Genetic Algorithms Dipl.-Ing. R. Kuberc...
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
VVT optimization with GT-POWER and Genetic Algorithms Dipl.-Ing. R. Kuberczyk Prof. Dr.-Ing. M. Bargende
Slide 1 ;
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
Overview 1. Introduction 2. GT-Power Model and Combustion Model 3. Model Calibration 4. GA Optimization 5. Conclusion
Slide 2 ;
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
Introduction 11 10 9 8
Variable valve train (VVT):
H u b [m m ]
7 6
• Advantages in efficiency by lower throttling losses 5 4 3
• Complexity raises exponentially 2 1
• Great challenge for simulation
0 0
30
60
90
120
150
180
210
240
270
300
Kurbelw inkel [°KW]
• 3D-CFD is still time consuming 1.2
in testing a high number of possible valve1.1
timing combinations 1
0.9
• 1-D engine simulation tools best solution for these tasks
Druck [bar]
0.8 0.7 0.6 0.5 0.4
How to optimize a VVT in a 1-D simulation ? 0.3 0.2 0
0.1
0.2
0.3
0.4
0.5
Volumen [l]
Slide 3 ;
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
Overview 1. Introduction 2. GT-Power Model and Combustion Model 3. Model Calibration 4. GA Optimization 5. Conclusion
Slide 4 ;
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
GT-Power Model 4-Cylinder SDI-Engine (2198 cm³)
External Cylinder Model used !
How does it work ? Slide 5 ;
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
Combustion Model External Cylinder Model/ Entrainment Model
flame burnt zone unburnt zone
phenomenological model is taking into account: - residual gas fraction (laminar flame speed) - shape of the combustion chamber (flame surface) important for VVT simulation ! Slide 6 ;
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
Overview 1. Introduction 2. GT-Power Model and Combustion Model 3. Model Calibration 4. GA Optimization 5. Conclusion
Slide 7 ;
Model Calibration
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
Entrainment Model Interval of Calculations: Inlet Valve max. lift 4.5mm C
355 350
10 Ventilhub [mm]
345 340 EVO
335
late EVO
330 325 320 310 305 490
8
500
505
510 IVO
515
520
525
2
360
450
540
630
720
810
900
810
900
Kurbelwinkel [°KW]
B
530
8
B
6
…the calibartion 4results ?
4 2
2 0
0
Slide 8
4
10
A
6
270
6
270
Ventilhub [mm]
Ventilhub [mm]
10
495
C
0
late IVO
315 A
8
360
450
540
630
720
Kurbelwinkel [°KW]
810
900
270
360
450
540
630
720
Kurbelwinkel [°KW] ;
Model Calibration/ Results
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
Entrainment Model 10
355
n=3000 1/min pmi=7.01 bar
rpm=3000 1/min imep=7bar
350
8
345
7
340
6
335 EVO
pmi[bar] [bar] imep
9
5 4
325
3 2 1 0
330
0
320 there is only one set of parameters for the 315 310 combustion model for all operating points and 305 500 1000 1500 2000 2500 3000 3500 4000 490 495 500 505 510 515 520 valve timings Drehzahl [1/min] ! IVO IVO
Optimizer has to deal with more than one minimum ! ;
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
GA Optimization Optimizer is coupled with the GT-Power Model for automatic optimization
Optimizer
GT-Model
Aim: Optimizing IVO, EVO, Inlet valve lift to find the global minimum of the ISFC [g/kWh] How does the GA Optimizer work ? Slide 13 ;
Start Population
IVO EVO IV_lift
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
1 0 1 0 0 1
0 1 1 1 1 1
1
1 0 1 1 0 0
1
2 7
3
6
9
Selection 4
5
8
Ind. efficiency
Calculation of the efficiency
next population
Ind. efficiency
completion 9
2
4
8
Crossover
1 0 1 0 0 1
1
2
3
6 4
5
0 1 1 1 1 1
1 0 1 1 0 0
...
1
7 8
9
Slide 14 ;
GA Optimization
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
Result
270
ISFC Distribution: 2000 rpm, 3bar imep
bi [g/kWh] ISFC
260
250 Best Point after 450 calculations
240
………
230 Slide 15
~42h calculation time (3GHz)
0
200
400
600 800 number of calculations Aufruf [-]
1000
1200
1400 ;
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
Overview 1. Introduction 2. GT-Power Model and Combustion Model 3. Model Calibration 4. GA Optimization 5. Conclusion
Slide 16 ;
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
Conclusion VVT optimization: • 1-D simulation in connection with an phenomenological combustion Model (Entrainment Model) is necessary to optimize VVT • Optimization of IVC, EVO, IV lift with Genetic Algorithms to increase the Ind. Efficiency is useful because local and global minima exists • Powerful combination of Entrainment Model and automatic optimizer based on Genetic Algorithms • results are time consuming, but if only “good” results are necessary: these are received quickly ! (result could be used for beginning optimization
Slide 17
by hand then) ;
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart
END
Slide 18 ;
Research Institute of Automotive Engineering and Vehicle Engines Stuttgart