Integrated System Design Optimisation: Combining Powertrain and Control Design

Transcription

1 Integrated System Design Optimisation: Combining Powertrain and Control Design Dr. Ir. Theo Hofman MSc Emilia Silvas. Size Control Technology Topology Wednesday,, 14:15-14:35

2 Are we harming the planet in the name of progress? PAGE 1/16

3 Motivation for Hybrid Powertrains > 200 % increase Today PAGE 2/16

4 What is a hybrid powertrain? Vehicles Boats or Yachts Conventional Vehicle Combustion Engine Transmission Conventional Boat Combustion Engine Transmission Battery Pack Electric Machine Combustion Engine Hybrid Electric Vehicle Transmission Battery Pack Electric Machine Combustion Engine Hybrid Electric Boat Transmission PAGE 3/16

5 Wide Variety of Hybrid Powertrains E. Silvas et al., Review of Optimal Design Strategies for Hybrid Electric Vehicles. IFAC Workshop on Engine and Powertrain Control, Simulation and Modelling, 3(1):57 74, PAGE 4/16

6 Optimal Design of Powertrains PAGE 5/16

7 Optimal Design of Powertrains (example) Which topology and sizes will find the best combination of cost, risk and mission effectiveness for different sea scenarios? 4 sea mission scenarios Multiobjective Genetic Algorithms Five objective functions o Max. propeller efficiency o Max. electric motor efficiency o Min. electric motor size o Min. total energy consumption o Max. steam turbine efficiency 8% improvement in energy consumption for the hybrid solution B.A. Skinner, G.T. Parks and P.R. Palmer Comparison of submarine drivetrain topologies using multiobjective genetic algorithms. IEEE Transactions on Vehicular Technology, 2009 PAGE 6/16

8 Optimal Design of Powertrains (benefits) A hybrid drive train enables: Maximizing the performance; Maximizing the fuel efficiency (minimize emissions); Improving the trade-off between 1 and 2; Performance Fuel efficiency Usage of new technologies; e.g., advanced engines, electrical auxiliaries, and transmissions. PAGE 7/16

9 Optimal Design of Powertrains (problem) General Optimization Problem Find the design variables x by solving min f(x) x s. t. g x 0 h x = 0 x = [x p, x c ] ( ) p denote a plant related variable ( ) c denote a control related variable w n = s(n) d(n) v(n),with n = [1, t f ] Plant and Control Optimization Problem min x s,x c (t) s. t. 0 t f φ x p, x c (t), w dt g p x p 0 h p x p = 0 g c x c 0 h c x c = 0 PAGE 8/16

10 Optimal Design of Powertrains (methods) Optimize the plant Optimize the controller Improve plant without compromising the controller Optimize the controller Plant and Control Optimization Problem min x s,x c (t) s. t. Bi-level / Sequential Nested Simultaneous 0 t f φ x p, x c (t), w dt g p x p 0 h p x p = 0 Optimize the combined system by varying both plant and controller g c x c 0 h c x c = 0 PAGE 9/16

11 Optimal Design of Powertrains (Study Case) Bi-level / Nested Optimal Sizing and Control Design of a Hybrid Electric Vehicle Improve plant without compromising the controller Application Sizing Optimization Genetic Algorithms, Sequential Quadratic Programming, Particle Swarm Optimization or Pattern Search (DIRECT) Optimize the controller Optimal Control Dynamic Programming Scope of the study case: Find optimal engine, motor and battery sizes for minimum fuel and costs Find optimal control inputs (power split signal and gear number) for a given driving profile Compare nested optimization methods PAGE 10/16

12 Optimal Design of Powertrains (Study Case) Full parallel hybrid topology: Backwards modeling Scalable quasi-static models Linear cost-models φ p = max (profit) φ c = x p = t f i=0 P e P m Cb P f, x c = u ps γ P e = engine power P m = motor power C b = battery capacity u ps = power-split signal γ = gear ratio PAGE 11/16

13 Optimal Design of Powertrains (Study Case) Full parallel hybrid topology: Backwards modeling Scalable quasi-static models Linear cost-models φ p = max (profit) φ c = x p = t f i=0 P e P m Cb P f, x c = u ps γ yearly mileage highway typical driving Heavy duty, 40 ton, vehicle Plant and Control Optimization Problem min x s,x c (t) s. t. 0 t f φ x p, x c (t), w dt g p x p 0 h p x p = 0 g c x c 0 h c x c = 0 PAGE 12/16

14 Optimal Design of Powertrains (Study Case) Optimal Sizing Results The choice of optimization target (fuel, hybridization costs, profit) strongly influences the optimal design PAGE 13/16

15 Optimal Design of Powertrains (Study Case) Optimal Sizing Results Pareto Analysis PAGE 14/16

16 Optimal Design of Powertrains (Study Case) Optimal Sizing Results Optimization Algorithms Comparison E. Silvas et al., Comparison of Bi-level Optimization Frameworks for Sizing and Control of a Hybrid Electric Vehicle, (submitted to) IEEE VPPC PAGE 15/16

17 Conclusions Nested optimal design achieves improved fuel efficiency (proven to particular cases), eliminates costly re-design steps, and enables the hybrid powertrains chance to comply with future exhaust emissions legislations. Using brute force search, to find the optimal sizing values becomes too computationally expensive and insufficiently accurate. Optimization algorithms as SQP or DIRECT should be used instead. Current & Future Work Automatic topology generator for hybrid topologies. Extend the design framework to include switchable topologies, automatically generated. PAGE 16/16

18 Thank you! Questions?