Modeling and control strategy for series hydraulic hybrid vehicles

Transcription

1 The University of Toledo The University of Toledo Digital Repository Theses and Dissertations 29 Modeling and control strategy for series hydraulic hybrid vehicles Mingwei Shan The University of Toledo Follow this and additional works at: Recommended Citation Shan, Mingwei, "Modeling and control strategy for series hydraulic hybrid vehicles" (29). Theses and Dissertations This Dissertation is brought to you for free and open access by The University of Toledo Digital Repository. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of The University of Toledo Digital Repository. For more information, please see the repository's About page.

2 A Dissertation Entitled Modeling and Control Strategy for Series Hydraulic Hybrid Vehicles By Mingwei Shan Submitted as partial fulfillment of the requirements for the Doctor of Philosophy Degree in Engineering Advisor: Dr. Roger J. King Co-advisor: Dr. Walter W. Olson College of Graduate Studies The University of Toledo December 29

3 The University of Toledo College of engineering I HEREBY RECOMMEND THAT THE DISSERTATION PREPARED UNDER MY SUPERVISION BY Mingwei Shan ENTITLED Modeling and Control Strategy for Series Hydraulic Hybrid Vehicles BE ACCEPTED IN PARTIAL FULLFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF Doctor of Philosophy in Engineering Dissertation Advisor: Dr. Roger J. King Recommendations Concurred by Dr. Walter W. Olson, Co-advisor Dr. Thomas Stuart Dr. Richard G. Molyet Dr. Gursel Serpen Committee On Final Examination Dean, College of Engineering

4 Copyright 29, Mingwei Shan This document is copyrighted material. Under copyright law, no parts of this document may be reproduced without the expressed permission of the author.

5 An Abstract of Modeling and Control Strategy for Series Hydraulic Hybrid Vehicles By Mingwei Shan Submitted as partial fulfillment of the requirements for the Doctor of Philosophy Degree in Engineering The University of Toledo December 29 Series hydraulic hybrid technology has the potential to significantly improve fuel economy and reduce emission. The series hydraulic hybrid is very different from electric and parallel hydraulic configuration and requires a unique power management control strategy to realize its optimal potential. In this dissertation, three approaches to achieve optimality are proposed and analyzed. These are rule-based, intelligent, and mixed power management control strategy. For evaluating the performance of control strategies, a forward-facing closed-loop simulation model based on physical features is first established in the MATLAB/SIMULINK environment. We then introduce a simple, valid and easily implementable rule-based power management control strategy. To derive the control signals, a PID-based multi-stage controller is presented. A thorough analysis on a class VI medium truck is elucidated. The simulation results demonstrate that a series hydraulic iv

6 hybrid medium truck with the proposed rule-based power management control strategy results in fuel economy increases of 117% and 44% over the conventional baseline respectively over Federal Urban Driving Schedule (FUDS) and Federal Highway Driving Schedule (FHDS). Then, an intelligent power management control strategy incorporating artificial neural networks (ANNs) and dynamic programming (DP) algorithm applied to series hydraulic hybrid propulsion systems is presented. ANNs are used to forecast vehicle speed and DP is utilized to find the optimal control actions for gear shifting and dual power source splitting. A thorough analysis of effect on fuel economy with different prediction window size on the class VI medium truck over FUDS and FHDS is presented. Compared with conventional baseline, the simulation results demonstrate that series hydraulic hybrid medium truck with 2 seconds short-term prediction window enables fuel economy increase of 135% and 48% respectively over FUDS and FHDS. Although the intelligent power management control strategy has obvious advantages over rule-based control strategy in improving fuel economy, this approach is somewhat limited in a realistic application due to prediction error. Finally, we proposed a mixed power management control strategy incorporating intelligent and rule-based approach to obtain a practicable near-optimal control strategy. Validations of these three power management control strategies are performed by Vehicle Propulsion Systems Evaluation Tool (VPSET) developed at Southwest Research Institute. v

7 To my wife Xiaoyan Dou, who is always supporting and encouraging me, and my parents for their spiritual support and backing throughout my years of education. To my son Colin Huize Shan, who is my pride and joy. vi

8 Acknowledgements I would like to express my earnest gratitude to both of my advisors, Professor Roger J. King and Professor Walter W. Olson. Their guidance, assistance, patience, and encouragement have been of enormous importance to my research and the completion of the dissertation. I greatly appreciate the opportunity Professor Olson offered me to conduct research and expand my knowledge in hydraulic and automotive fields. I would also like to thank my other committee members, Dr. Thomas Stuart, Dr. Richard G. Molyet and Dr. Gursel Serpen, for their helpful advices and comments. I particularly convey my most sincere thanks to Dr. Matthew Witte, for the support and guidance he showed to me during the duration of this research. A top specialist in his research field, and an enthusiastic and kind person, Dr. Witte has been always ready to share his knowledge, and show benevolent understanding to simulation and control. His ideas are of inestimable value for this research. I am highly grateful to the U.S. army and Southwest Research Institute (SwRI) of San Antonio, Texas for financial and VPSET software technical support. Last but not the least; I would like to thank all of my friends and teammates for the support they have given me over the years. Thank you! vii

9 Table of Contents Abstract... iv Acknowledgements... vii Table of Contents... viii List of Figures... xi List of Tables...xv Chapter 1 Introduction Background of the Hydraulic Hybrid Vehicle (HHV) Parallel HHV Series HHV Power Management Control Strategy for Hybrid Vehicles Motivation of This Dissertation Contributions... 8 Chapter 2 Simulation Model for a Series Hydraulic Hybrid Vehicle Sub-System Modeling Diesel Engine Model Hydraulic Pump/Motor Model Hydraulic Accumulator Model... 2 viii

10 2.1.4 Vehicle Dynamics Model Sizing of Key Components Sizing of Diesel Engine Sizing of Hydraulic Pump/Motors Sizing of Hydraulic Accumulators Chapter 3 Rule-based Power Management Control Strategy Power Management Control Strategy Discrete Operation Modes Accumulator Propulsion (Mode 1) SOC Regulation (Mode 2) Energy Capture (Mode 3) Frictional Braking (Mode 4) Engine Starter (Mode 5) Implementation of Rule-based Control Strategy Design of Controllers to Emulate the SHHV Controls Simulation Results and Discussion Acceleration Time from to 6 mph FUDS Cycle FHDS Cycle Conclusion Chapter 4 Intelligent Power Management Control Strategy Artificial Neural Networks for Time Series Forecasting Optimization of Power management control strategy through Dynamic ix

11 Programming Dynamic Programming Simulation Results and Discussion Chapter 5 Mixed Power Management Control Strategy Implementation of Mixed Power Management Control Strategy Transition Condition Block Diagram of SHHV with Mixed Control Strategy An Approach of Hardware Implementation Validation of Power Management Control Strategy by VPSET Introduction of VPSET Simulation Model of SHHV with Mixed Power Management Control Strategy in VPSET Environment VPSET Simulation Results and Discussion Mixed Power Management Control Strategy Rule-based Power Management Control Strategy Intelligent Power Management Control Strategy Discussion Chapter 6 Conclusion and Future Work Conclusion Future Work References Appendix x

12 List of Figures Fig. 1.1 Conceptual sketch of the parallel hydraulic hybrid vehicle Fig. 1.2 Conceptual sketch of the series hydraulic hybrid vehicle Fig. 2.1 Schematic diagram of SHHV drivetrain and corresponding control signal path Fig. 2.2 The forward-facing closed-loop simulation model of SHHV in SIMULINK Fig. 2.3 The block diagram of central controller simulation model in SIMULINK Fig. 2.4 The block diagram of engine simulation model in SIMULINK Fig. 2.5 V8 6L DI diesel engine BSFC map Fig. 2.6 The motor total efficiency map: (a) 2 psi; (b) 4 psi... 2 Fig. 2.7 The basic dynamic accumulator model in SIMULINK Fig. 2.8 Fluid volume and pressure as a function of SOC Fig. 3.1 Schematic diagram of switching operation modes in MATLAB / SIMULINK / STATEFLOW Fig. 3.2 PIDMS controller (top level) in SIMULINK Fig. 3.3 Block diagram of Mode 2 control Fig. 3.4 V6 4.5L DI diesel engine BSFC map with minimum BSFC trajectory Fig. 3.5 SHHV(@ SOC=1) vs. CONV in comparison of -6 mph acceleration time Fig. 3.6 Vehicle speed (demanded vs. actual) over FUDS xi

13 Fig. 3.7 SOC, throttle position and engine speed over FUDS Fig. 3.8 Vehicle speed, SOC, throttle position and engine speed during 1-35 sec period Fig. 3.9 Displacement factors of engine pump/motor, primary drive pump/motor and secondary drive motor during 1-35 sec period Fig. 3.1 Frictional brake torque over FUDS: (a) CONV; (b) SHHV Fig Engine visitation points over FUDS: (a) CONV (V8, 6L engine); (b) SHHV (V6, 4.5L engine) Fig Vehicle speed (demanded vs. actual) over FHDS Fig SOC, throttle position and engine speed over FHDS Fig Frictional brake torque over FHDS: (a) CONV; (b) SHHV Fig Engine visitation points over FHDS: (a) CONV (V8, 6L engine); (b) SHHV (V6, 4.5L engine) Fig. 4.1 Schematic of the k th independent neural network... 6 Fig. 4.2 The ideal relationship of hydraulic fluid flow rates in SHHV Fig. 4.3 Simulation results with 2 seconds prediction window size: (a) FUDS; (b) FHDS Fig. 4.4 Some optimal control variables trajectories with 2 seconds prediction window size: (a) FUDS; (b) FHDS Fig. 5.1 The flowchart of mixed power management control strategy Fig. 5.2 Block diagram of SHHV with mixed power management Fig. 5.3 Functional block diagram for hardware implementation Fig. 5.4 Fuel economy validation on level road (PTA) xii

14 Fig. 5.5 FedEx W7 truck acceleration from to 6 mph Fig. 5.6 The forward-facing simulation model of SHHV in VPSET environment Fig. 5.7 The block diagram of mixed power management in VPSET Fig. 5.8 The block diagram of intelligent controller in VPSET Fig. 5.9 The block diagram of rule-based controller in VPSET Fig. 5.1 Desired and actual vehicle speed with mixed power management control strategy: (a) FUDS; (b) FHDS Fig. 5.1 (continue) Desired and actual vehicle speed with mixed power management control strategy: (c) City UDDS; (d) Highway UDDS Fig Engine speed with mixed power management control strategy: (a) FUDS; (b) FHDS Fig (continue) Engine speed with mixed power management control strategy: (c) City UDDS; (d) Highway UDDS Fig State of charge (SOC) with mixed power management control strategy: (a) FUDS; (b) FHDS Fig (continue) State of charge (SOC) with mixed power management control strategy: (c) City UDDS; (d) Highway UDDS Fig Desired and actual vehicle speed with rule-based power management control strategy: (a) FUDS; (b) FHDS Fig (continue) Desired and actual vehicle speed with rule-based power management control strategy: (c) City UDDS; (d) Highway UDDS Fig Engine speed with rule-based power management control strategy: (a) FUDS; (b) FHDS xiii

15 Fig (continue) Engine speed with rule-based power management control strategy: (c) City UDDS; (d) Highway UDDS Fig State of charge (SOC) with rule-based power management control strategy: (a) FUDS; (b) FHDS Fig (continue) State of charge (SOC) with rule-based power management control strategy: (c) City UDDS; (d) Highway UDDS Fig Desired and actual vehicle speed with intelligent power management control strategy: (a) FUDS; (b) FHDS Fig (continue) Desired and actual vehicle speed with intelligent power management control strategy: (c) City UDDS; (d) Highway UDDS Fig Engine speed with intelligent management control strategy: (a) FUDS; (b) FHDS Fig (continue) Engine speed with intelligent management control strategy: (c) City UDDS; (d) Highway UDDS Fig State of charge (SOC) with intelligent power management control strategy: (a) FUDS; (b) FHDS Fig (continue) State of charge (SOC) with intelligent power management control strategy: (c) City UDDS; (d) Highway UDDS xiv

16 List of Tables Table 2.1 Constants and calibrated coefficients for pump/motor model Table 3.1 Controlled variables in each operation mode Table 3.2 The -6 mph acceleration time at different initial SOC... 4 Table 3.3 The effects of different combination of SOCstart and SOCstop over FUDS.. 41 Table 3.4 The effect of initial SOC on the fuel economy and fuel economy improvement over FUDS Table 3.5 The effects of different combination of SOCstart and SOCstop over FHDS.. 51 Table 3.6 The effect of initial SOC on the fuel economy and fuel economy improvement over FHDS Table 4.1 Normalized mean square error of learning and forecasting process over FUDS Table 4.2 Normalized mean square error of learning and forecasting process over FHDS Table 4.3 Summary of SHHV fuel economies with different prediction window size over the convention baseline Table 5.1 Summary of fuel economy validation results for HMMWV Table 5.2 Summary of acceleration comparison for FedEx W7 truck Table 5.3 Intelligent duty for FUDS, FHDS, City UDDS, and Highway UDDS cycles. 86 Table 5.4 Summary of fuel economy results Table 5.5 Summary of fuel economy improvement results xv

17 Chapter 1 Introduction Improving fuel economy and reducing polluting emissions have become more important with recent concerns about global crude oil supplies and green-house effects. Hybridization is a viable approach offering significant breakthroughs for near and midterm application. Over 9% of commercialization effort has focused on hybrid electric vehicles (HEV) and virtually all commercially available passenger cars and light truck hybrids are electric hybrids today. However, it is believed that further advantage can be realized with hydraulic hybrid vehicles (HHV) compared to HEV. While hydraulic accumulators store only a fraction of the total energy that a battery can store, hydraulic pump/motors and accumulators are able to capture and reuse a higher percentage of kinetic energy than their electric counterparts. U.S. Environmental Protection Agency (EPA) estimates that HHV is able to capture and reuse 7-8% of braking energy [1]. The percentage of regenerative braking for HEV, however, is around 15% [2] [3]. This feature is especially beneficial for truck applications in significant stop-and-go activity [4] [5]. Secondly, HHV components are relatively simple and made out of common materials using well established manufacturing processes and are therefore relatively low cost, compared to 1

18 HEV components. Thirdly, the efficiency of bent-axis variable displacement pump/motor and accumulator used in HHV is significantly higher than that of electric generator/motor and battery. 1.1 Background of the Hydraulic Hybrid Vehicle (HHV) In recent years, rising gasoline prices, increased fuel consumption, and overdependence on fossil fuel have driven us to seek more efficient transportation. Electric hybrid vehicle (HEV) has gained acceptance in the light-duty market, but is not appropriate for heavy vehicles that do a lot of stop-and-go driving. Electric hybrid vehicle only has the potential for several percent increases in fuel economy due to conversion efficiency and the low power density limits of electric motors and batteries. Hybrid Hydraulic Vehicle (HHV) technology overcomes these limitations. Unlike electric hybrid which stores regenerated energy in a battery, the hydraulic hybrid recovers the energy in the form of pressurized hydraulic fluid which is stored in a compressed nitrogen gas accumulator. The result is a highly efficient and extremely power-dense hybrid system capable of capturing and reusing a substantially higher portion of braking energy than is possible with an HEV. HHV is generally classified into two categories, parallel or series, depending on whether there is a solid shaft between the engine and the wheels or not. The configuration of parallel hydraulic vehicles (PHHV) is easier to implement, but efficiency gains are limited by the solid link between the wheels and engine. In contrast, series hydraulic hybrid vehicles (SHHV) allow engine speed to be decoupled from vehicle speed. This permits a control strategy where the engine and other hydraulic components operate in 2

19 concert at maximum efficiency. In addition, eliminating idling by shutting off the engine during braking or stopped periods increases the fuel economy to its maximum potential with the series hydraulic hybrid concept Parallel HHV The parallel hydraulic hybrid vehicle (Fig. 1.1) uses a reversible hydraulic pump/motor coupled to the drive shaft with two accumulators. When a driver steps on the brake, the pump/motor forces hydraulic fluid out of a low-pressure accumulator into a high-pressure accumulator, increasing the pressure of nitrogen gas stored there up to 5, psi. During acceleration, the system switches from pump mode to motor mode. The nitrogen gas forces the hydraulic fluid back into the low-pressure accumulator, and the pump/motor applies torque to the driveshaft. Fig. 1.1 Conceptual sketch of the parallel hydraulic hybrid vehicle. The hydraulic hybrid uses the hydraulic power for the acceleration boost, and then blends in the engine. This results in a significant reduction in fuel consumption and 3

20 improved acceleration due to the high power density of hydraulics. In addition, the hydraulic hybrid system increases brake life and reduces engine and transmission wear, potentially extending component life and lowering service costs. It is also more environmentally friendly by decreasing exhaust emissions Series HHV The series hybrid represents the second generation of hydraulic hybrid vehicles (see Fig. 1.2). The engine is connected to the engine pump/motor which, as a pump, transmits pressurized hydraulic fluid to the drive pump/motor and to the high pressure accumulator. The engine pump/motor is used as a motor only when starting the engine, thus the electric starter is eliminated in the series hydraulic hybrid architecture. The conventional driveline is completely removed; instead, soft pipes and drive pump/motor are used to transfer power and propel the vehicle. The drive pump/motor converts high pressure hydraulic fluid into rotating power in the motor mode and transmits regenerative braking energy back to the high pressure accumulator in the pump mode. Since the direct link between the engine and the driveline components is removed, the engine is separated from the road and higher efficiencies are anticipated. 4

21 Fig. 1.2 Conceptual sketch of the series hydraulic hybrid vehicle. 1.2 Power Management Control Strategy for Hybrid Vehicles Hybridization raises the question of how to distribute, split, and coordinate the primary power source and the assistant power source. There are plentiful publications about the power management of hybrid electric propulsion systems. Nevertheless, the relatively low energy density of the hydraulic accumulator requires careful consideration on optimal control. Most of the existing power management control strategies for HHV are developed for parallel configuration. Rule-based control strategy is based on the engineering intuition to devise control rules that would consider load level and move the engine out of the inefficient operating range [4, 6-8]. There has been much other research on power management control strategy by using fuzzy logic techniques [9-13] which are essentially rule-based by heuristic knowledge on the efficient operation region of an engine. Combining GPS information and a fuzzy logic controller is further studied in [14] to enhance vehicle 5

22 performance. Static optimization methods that develop an optimal engine management strategy considering the efficiency maps of the engine and hydraulic components are presented in [15] and [16]. The advantage of this method is that the resulting optimization function can be directly applied in real-time on a vehicle even though the resulting optimal power management control strategy may be suboptimal when compared to results obtained from dynamic programming. However uncertainty of the vehicle system, especially under transient condition, limits its application, i.e. the reference engine speed derived from a predefined objective function may not be the optimal one. Intelligent energy management which incorporates a driving situation identifier, the driving style of the driver and current driving mode by using neural networks and fuzzy logic is proposed in [17] and [18]. The learning vector quantization neural network after training can recognize the current roadway type and congestion level which are predefined. Then the driving trend, driving mode and driving style are determined by means of fuzzy logic, resulting in intelligent decisions for the power splitting. Likewise, this power management can not get optimal fuel economy due to fuzzy rule-based torque distribution. Dynamic programming (DP) algorithm is an approach developed to solve sequential or multi-stage decision problem. DP searches for optimal decisions from a given driving cycle, rather than current operation point, which leads to absolute optimum decisions. DP as a good design tool and benchmark has been successfully applied on hybrid electric vehicles [19-21] and parallel hydraulic hybrid propulsion system [22-24]. However, the real application of DP algorithm to solve the optimal power management 6

23 problem on hybrid vehicles is limited due to the prerequisite of previewed vehicle speed profile. Advantages of SHHV over PHHV have been mentioned before. However, papers devoted to series hydraulic configuration are relatively scarce due to its sophisticated control strategy compared to parallel structure. A simulation-based series hydraulic hybrid propulsion system which relied on thermostatic (on/off) engine power management according to the state of charge (SOC) in the high pressure accumulator was proposed in [25], and first presented the concept of threshold charging power which combined other optimized control parameters, such as the threshold SOC and dead band, to obtain higher fuel efficiency. With a downsized V6 diesel engine, fuel economy improvement is up to 68.2%. 1.3 Motivation of This Dissertation Two power management approaches are proposed in this dissertation: rule-based and intelligent control strategy. All engines have a unique optimum efficiency line on the brake specific fuel consumption (BSFC) map and a supreme optimum point exists on this line, called the eye of BSFC. Theoretically with a good design of power management control strategy, SHHV can keep the engine operating near the eye of BSFC at all times. The objective of this dissertation is to explore a simple, valid and easily implementable rule-based control strategy so that the engine operates near the eye of BSFC a majority of time and along the optimum efficiency line at all other times. In addition, a robust controller which requires no previewed information but only current measurement data from the vehicle is 7

24 presented to implement proposed power management control strategy and produce realistic control signals to drive the simulation models. Besides the rule-based control strategy, an intelligent power management control strategy incorporating artificial neural networks (ANNs) and DP algorithm applied on series hydraulic hybrid propulsion system is proposed. We create a time series approach based vehicle speed prediction model by applying multilayered feed forward neural network and back propagation learning algorithm, and investigate its ability on short-term prediction. A control design procedure tailored for a series hydraulic hybrid medium truck based on dynamic programming is also proposed. Furthermore, the analysis of effect of different prediction window size on fuel economy over Federal Urban Driving Schedule (FUDS) and Federal Highway Driving Schedule (FHDS) is given. For each driving cycle, two cases are solved: engine idling, and engine shut-down. Summarily, the first approach is easier to implement than the second one. However, the intelligent control strategy allows optimum control actions and maximum fuel economy. 1.4 Contributions The major contributions of this dissertation are listed as below: A physics-based forward-facing simulation model of series hydraulic hybrid vehicle is created in the MATLAB/SIMULINK environment for developing the optimal power management strategies. Here, forward-facing manner means everything starts with the driver action rather than wheels. The power management system, cyberdriver model, and sub-system models (including engine, accumulators, hydraulic 8

25 pump/motors, and vehicle dynamics) are integrated to perform a closed-loop simulation. This model can be used to analyze the effects of control strategy on fuel economy. Also, it is very effective for component sizing studies and determining vehicle performance limitations. A control design procedure based on rules is proposed. Operation modes of SHHV are decided by the predefined rules. Our strategy is to run the engine near the eye of BSFC a majority of time or shut it off when not needed. When engine is started, it will run along the optimum fuel efficiency line by controlling engine pump/motor displacement and throttle position. This approach by using engineering intuition provides engineers with qualitative understanding of power splitting. To evaluate the rule-based control strategy, a PID-based multi-stage controller is created which is robust and insensitive to external disturbances. The simulation results are completed over FUDS and FHDS cycle. A time series approach based vehicle speed prediction model by applying multilayered feed forward neural network and back propagation learning algorithm is created. Its ability of short-term prediction is tested on FUDS and FHDS. This prediction model works satisfactorily to provide updated short-term forecasting vehicle speed to dynamic programming (DP) algorithm as inputs. The formulation of DP to find optimal control policy for SHHV is developed. A thorough analysis of effect on fuel economy with different prediction window size on the series hydraulic hybrid medium truck over FUDS and FHDS is presented. The simulation results demonstrate that a twenty second prediction window will be a crossover point at which the prediction window is short enough to forecast accurately 9

26 but long enough to have optimality roughly equivalent to that of the entire driving cycle. A mixed power management control strategy incorporating artificial neural network, dynamic programming optimization, and rule-based approach to obtain a practicable near-optimal control strategy is developed, which inherits the merits of rule-based and intelligent power management control strategy. Models of hydraulic components are incorporated into VPSET. Validations of three power management control strategies are performed by VPSET. 1

27 Chapter 2 Simulation Model for a Series Hydraulic Hybrid Vehicle A physics-based simulation model of SHHV is necessary for developing an optimal power management control strategy. Here, a conventional International series [31], Class VI medium truck with a mass of 734 kg when fully loaded is selected as the baseline and comparable counterpart to SHHV. The conventional truck parameters used during simulation are listed in the Appendix. The main specifications of this truck remain, but the standard engine is replaced with a higher output V8 6L DI diesel engine [25]. For the conventional truck, the engine shaft is connected to the torque converter whose output shaft is coupled to the four-speed automatic transmission. The propulsion shaft transfers torque from the transmission to the differential. The differential delivers torque to the rear wheels through two drive shafts. The schematic diagram of the SHHV drivetrain and corresponding control signal path is shown in Fig The engine is connected to the engine pump/motor which, as a pump, transmits pressurized hydraulic fluid to the drive pump/motor and to the high pressure accumulator. The engine pump/motor is used as a motor only when starting the engine, thus the electric starter is eliminated in the series hydraulic hybrid architecture. The conventional driveline is completely removed; instead, soft pipes and drive 11

28 pump/motor are used to transfer power and propel the vehicle. The drive pump/motor converts high pressure hydraulic fluid into rotating power in the motor mode and transmits regenerative braking energy back to the high pressure accumulator in the pump mode. The two-speed computer shifted transmission keeps the drive pump/motor operating within its efficient area and prevents overspeed. The central controller monitors the driver s acceleration and braking events, and collects the current measured data, the engine angular speed (i.e. the engine pump/motor angular speed due to coupled together), the wheel angular speed and the high pressure accumulator state of charge as inputs from the actual vehicle. In accordance with inputs, the central controller outputs commands based on the optimal control strategy. ω e ω w SOC Central Controller Brake Pedal Position Fuel Pedal Position Mechanical Link Hydraulic Link Control Signal Path Displacement SOC On/ Off Throttle Position Valve High Pressure Accumulator Wheel ω w Engine ω e Engine P/M Drive P/M ω d Integrated Differential & Two-speed Transmission Valve Low Pressure Accumulator Gear Shift Wheel Fig. 2.1 Schematic diagram of SHHV drivetrain and corresponding control signal path. Fig. 2.2 shows a complete series hydraulic hybrid system structure as created in the MATLAB/SIMULINK environment. This architecture maintains a high degree of flexibility. This vehicle simulation model is a forward-facing closed-loop model which presents physical interaction of each sub-model in the real vehicle. 12

29 Scopes Scopes Throttle D1 Q_eng_p/m D2 Q_drive_p/m w _eng w _eng w_out T_eng_p/m DeltaP T_eng_p/m DeltaP T_drive_p/m T_drive _p/m T_wheel Diesel Engine Engine_pump/motor Drive_pump/motors Gear_ratio w_wheel w_out Integrated Transmission and Differential SOC D1 D2 w_eng Throttle T_brake w_wheel Gear_ratio Central Controller Q_eng_p/m DeltaP Q_drive_p/m SOC Accumulators T_wheel w _wheel T_brake Vehicle Dynamics Fig. 2.2 The forward-facing closed-loop simulation model of SHHV in SIMULINK. In this study, a cyber-driver that controls the acceleration or deceleration pedal is integrated inside of the central controller (Fig. 2.3) and used to simulate the driver behavior over the selected driving cycle. The cyber-drive takes the demanded speed prestored in the computer database as an input to calculate the pedal status (fuel pedal or brake pedal). The control signal generated by the PID-based multi-stage controller (PIDMS) is used for acceleration or deceleration along with the control strategy realized by logic control unit in the central controller. The blocks of logic control and PIDMS will be addressed in Chapter 3. 13

30 V_dem Drive Schedule Pedal V_dem V_desired Pedal 1 SOC Pedal Mode SOC V_wheel V_error w_eng V_desired Gear_ratio Logic Control rw 5 Gear _ratio 3 w_wheel 2 w_eng Mode D1 V_error D2 w_eng Throttle w_wheel T_brake Power Management 1 D1 2 D2 3 Throttle 4 T_brake Fig. 2.3 The block diagram of central controller simulation model in SIMULINK. The other models for the simulation include: 1) engine model; 2) hydraulic pump/motor model; 3) accumulator model; 4) vehicle dynamics model. They are addressed in detail in the remainder of this chapter. 2.1 Sub-System Modeling Diesel Engine Model In this work, the engine simulation model (Fig. 2.4) is simplified to accommodate the need of top-level control strategy design. A lookup table of wide-open-throttle (maximum) engine brake torque versus engine speed is directly derived from the engine brake specific fuel consumption map (BSFC) (Fig. 2.5) which was generated from experimental data obtained in the W. E. Lay Automotive Laboratory at the University of Michigan, more details about this BSFC map data see [25] and [32]. 14

31 Lookup Table T_eng T_eng > speed _max *2*pi/6 1 w_eng T_eng _brake 1 Throttle 2 T_eng_p/m 1/(Je+Jepm ) Mode ~= 2 ~= 5 AND w_eng _dot 1 s Fig. 2.4 The block diagram of engine simulation model in SIMULINK. 8 BSFC map [g/kwh] Torque [N-m] Engine speed [rpm] Fig. 2.5 V8 6L DI diesel engine BSFC map. The instantaneous engine torque is approximately decided by the following 15

32 equation: T e ( e e, max e ω ) = ut ( ω ) (2.1) Here, u is a nominal control signal adjusting the engine throttle position and therefore determines the engine torque output. The values of u are in the domain [,1], with and 1 indicating closed-throttle and wide-open-throttle position, respectively. T ( ω ) is the e, max e maximum available engine brake torque which is the function of the engine speed and described by the lookup table of wide-open-throttle engine brake toque. The rate of engine speed change is expressed by dω dt e ( T ( ω ) + T ( ω ) ( J + J ) = ) e e epm e e epm (2.2) Here, Tepm ( ω e ), a signed vector, is load torque determined by the engine pump/motor. J e and J epm are engine and engine pump/motor inertia, respectively. If the engine speed rises above the maximum speed, the engine torque drops to zero. Additionally, a logic control module is added into engine model in order to shut down the engine when not needed in series hydraulic hybrid propulsion configuration Hydraulic Pump/Motor Model The model of bent-axis piston pump/motor is used in this work due to its variable displacement and high efficiency characteristics. The ideal volumetric flow rate and ideal torque expressions are given by Q i T i = xωd (2.3) = xδpd (2.4) where x, a signed nominal number, is displacement factor defined by the ratio of current 16

33 displacement to maximum displacement with positive indicating motor mode and negative denoting pump mode; ω is angular speed; D is maximum displacement in unit of volume per radian; Δ p is pressure difference across the pump/motor. It is obvious that the actual values must be different from ideal ones due to losses in real world. The next four equations ( ) come from updated Wilson s pump/motor theory [33]. The pump volumetric and torque efficiencies, Qa Cs Δp Cst ηv _ pump = = 1 (2.5) Q x S β x σ i η t _ pump Ti 1 = = (2.6) T C a CvS f Ch x σ x x The motor volumetric and torque efficiencies, Qi 1 ηv _ motor = = (2.7) Q Cs Δp C a st xs β xσ T C S C η _ 1 σ t a v f 2 2 motor = = Ch x (2.8) Ti x x where Q a actual pump/motor volumetric flow rate; T a actual pump/motor torque; β fluid bulk modulus of elasticity; C s laminar leakage coefficient; C st turbulent leakage coefficient; C v viscous loss coefficient; 17

34 C f frictional loss coefficient; C h hydrodynamic loss coefficient; S dimensionless number ( S = μω Δp, μ is fluid dynamic viscosity); σ 1/ 3 1/ 2 dimensionless number ( σ = ωd ( 2Δp / ρ), ρ is fluid density). The pump/motor total efficiency is the product of volumetric and torque efficiency. From volumetric and torque efficiency equations of pump/motor, we can see that the efficiency is determined by three variables (angular speed, displacement factor and pressure difference across the pump/motor). In essence, the coefficients used for calculating losses need to be calibrated to match particular pump/motor design. In this study, one pump/motor simulation model is calibrated to roughly match the model used at EPA [34]. The constant numbers and calibrated coefficients are given in Table 2.1. Table 2.1 Constants and calibrated coefficients for pump/motor model. Constants β 166 MPa μ.34 Ns/m 2 ρ 85 kg/m 3 C s C st C v C f C h Calibrated coefficients 2.627e e-7 1.2e-4 1.5e Fig. 2.6(a) and (b) shows the motor total efficiency as a function of displacement (the vertical axis) and speed (the horizontal axis) at 2 and 4 psi, respectively. The pump efficiency trend is similar to that of the motor. It is assumed that the value of each dimensionless loss coefficient is of uniform for the same machine regardless of pump or 18

35 motor operation mode. From Fig. 2.6, we can see that displacement factor plays an important role on the pump/motor efficiency. Generally speaking, the larger the displacement factor, the higher the pump/motor efficiency is. Hence, working the pump/motor at high load is very desirable. For the purpose of simplifying the simulation model, it is reasonable to set a constant pump/motor efficiency because the pump/motor can operate within certain efficiency area (e.g. above 94%) most of the time when a good control strategy associating with proper components sizing is exerted. This article will use the model with dynamic efficiency to exactly simulate the behavior of pump/motor. 1 Efficiency at 2 psi Displacement factor Speed [rpm] (a) 19

36 1.95 Efficiency at 4 psi Displacement factor Speed [rpm] (b) Fig. 2.6 The motor total efficiency map: (a) 2 psi; (b) 4 psi Hydraulic Accumulator Model Hydraulic accumulators store energy as a secondary power source for hydraulic hybrid vehicles. In this way, accumulators are similar in function to the electric batteries of electric hybrid vehicles. The hydraulic hybrid configuration requires both a high pressure accumulator and a reservoir (low pressure accumulator). The low pressure accumulator precharge is required to avoid cavitation in the pump. Hence the term of low pressure accumulator is more exact than that of reservoir. The difference in pressure between the accumulators and the associated pump/motor displacement determines the pump/motor shaft torque that is used to accelerate or decelerate the 2

37 vehicle, or absorb engine power. The accumulator model is a hydro-pneumatic accumulator which consists of a precharged inert gas chamber and a fluid chamber connected to a hydraulic system. The chambers are separated by a bladder with elastomeric foams in the gas-side to reduce heat loss. When the fluid chamber is empty, a valve is activated to shut off the accumulator fluid port to prevent the bag from being herniated. The dynamic accumulator model is based on the following equation: PV k = C (2.9) Here, P is supply pressure; V is gas volume; k is specific heat ratio within a range between 1 (isothermal process) and 1.4 (adiabatic process); C is constant and depends on precharge. For modern bladder accumulators, the higher cycle rates and the application of elastomeric foams on the inner bladder wall lead to little or no heat transferred into or out of the accumulators, so we should assume that the compression and expansion of gas is adiabatic. In this dissertation, we select k to be 1.4. One basic dynamic accumulator model in SIMULINK is given in Fig Its governing equation is derived from Eqn. (2.9). P t V 1.4P = t V (2.1) V where, > means that accumulator is discharging. t 21

38 1 flow rate 1 s V_gas 1 s pressure Fig. 2.7 The basic dynamic accumulator model in SIMULINK. The internal friction resulting in the loss of pressure is also included in this accumulator model. The loss of pressure between gas ( P gas ) and fluid at the accumulator port ( P ) is proportional to the ratio of overall frictional loss energy ( E ) to input energy ( E ) [33]. f P gas P E f P = ± (2.11) 2 E where ± implies discharging and charging, respectively Vehicle Dynamics Model In the modeling of vehicle dynamics, many approaches may be used to mimic the dynamic behavior of vehicles. Overall simulation objectives determine which is best. The studies in this dissertation consider only the effect of power management control strategy on the fuel economy from the top level. That is we just explore acceleration and deceleration on a smooth road in longitudinal direction without analysis of ride and pitch motion. So the single mass model is sufficient for this purpose. In the model of single 22

39 mass, the entire vehicle is regarded as a unit, and it is lumped at the center of gravity with appropriate mass and inertia properties. Many acceleration and deceleration events would occur in the heavy duty driving schedule. Therefore, the impact of equivalent mass of rotating components cannot be neglected. The equations for conventional vehicle [35], do not apply due to series hydraulic hybrid architecture. The new modified equations here are derived from Newton s Second Law. T = T ( J + J ) a (2.12) t dpm dpm t dpm T = ( T J a ) N (2.13) df t t dpm t T = ( T J a ) N (2.14) w df df df df F r = T nj a (2.15) tr w w w where T dpm drive pump/motor torque; T t input torque to transmission; T df input torque to differential; T w input torque to wheel axles; J dpm drive pump/motor inertia; J t transmission inertia (as seen from dive pump/motor side); J df differential inertia (as seen from transmission side); J w wheel and axle shaft inertia; a dpm angular acceleration of drive pump/motor; 23

40 a df angular acceleration of differential (as seen from transmission side); a w angular acceleration of wheel; N t gear ratio of transmission; N df gear ratio of differential; F tr tractive force at ground; r n wheel radius; the number of driving wheels. Now, the angular accelerations of drive pump/motor and differential are related to that of the wheels by the gear ratios. a dpm = N t N df aw and df N df aw a = (2.16) Combining Eqns. ( ) and (2.16), the tractive force available at the ground can be obtained. And with the assumption of no wheel slips ( a w = a / r ; here, a is vehicle acceleration), yields: F tr T N r N T N N a dpm t df [ J + 2J ) N N + J N + nj ] = M a dpm t df = 2 ( dpm t t df df df w r (2.17) r r Here, M ( J + 2J ) N N + J N nj ) r = is defined as equivalent rotating r dpm t t df df df + mass. Hence, the final governing equation for vehicle dynamics is shown below: w ( M z dv Tdpm N t N df ( ηtη df ) 2 + M r ) =.5ρCD Av Mgf r cosθ Mg sinθ (2.18) dt r where M v vehicle mass; vehicle velocity 24

41 ρ,c d, A air density, aerodynamic drag coefficient and frontal area of vehicle, respectively; g gravity acceleration f r θ rolling resistance coefficient grade angle; η t, η df efficiencies of transmission and differential, respectively; z + 1 motor mode z = sgn( T dpm ) =. 1 pump mode In Eqn. (2.18), M r doesn t contain the equivalent rotating mass of the engine and engine pump/motor, because they have been included in the engine model. The rolling resistance is the primary resistance force at low speed. This implies that the rolling resistance coefficient f r would be very important for the simulation study over the city driving cycle, such as FUDS with 19.6 mph average speed. Two equations for estimating f r of heavy truck were developed at the University of Michigan Transportation Research Institute [36]. f = ( v) for radial tires (2.19) r C h f = ( v) for bias-ply tires (2.2) r C h Here, v is vehicle speed in mph; C h is road surface coefficient (usually in the range of ). 2.2 Sizing of Key Components To describe the process of fuel economy in this study, the size of components 25

42 must be tailored for special cases. The design progress begins with a scaled down diesel engine and hydraulic parts. The selection is based on the operation requirements, such as acceleration ability and constrains of continuous speed at 2% and 3% grade. Because many variables interact with each other, a multi-start methodology associating design-ofexperiments (DOE) approach [23] is applied to get optimal parameters. The optimized components sizes are directly presented in each sub-section without showing the optimization progress. Finally, an estimated mass increase of 25 kg is accounted for in this simulation. This value is based on the data of series hydraulic hybrid SUV from EPA report (see [34]) by means of reasonable modification for the medium truck Sizing of Diesel Engine For the SHHV configuration, the accumulator provides an auxiliary power source with sufficient power density to ensure the vehicle s acceleration performance, even if the engine is downsized. The engine used in conventional medium truck is V8 6L DI diesel engine. When changing the size of engine, it is hypothetic that the downsized engine keeps the same design characteristics, i.e. the engine of SHHV scales proportionally to the V8 6L diesel engine. Thus the overall trends of original engine, such as engine BSFC map, are preserved during scaling progress. Since -6 mph acceleration time is the common ground for all configurations, a downsized V6 4.5L DI diesel engine for SHHV configuration is selected to provide favorably comparable -6 mph acceleration time to that of the baseline vehicle. More 26

43 details can be found in Chapter Sizing of Hydraulic Pump/Motors In order to maximize the hydraulic machine efficiency, two drive motors are coupled to the integrated two-speed transmission and differential. One of them operates as a primary motor and provides most of the propulsion. The other one, with a gear ratio of 1.5, runs only as an assistant motor during hard acceleration and/or grade requirement. Once the primary motor is able to fulfill the requirement, the secondary one is set to idle immediately. The result is that the primary pump/motor will operate at a much higher displacement range with improved efficiency compared to a single large drive pump/motor. Thus the in-use efficiency is improved. The engine pump/motor is initially sized to have the ability of completely absorbing engine power, as a pump, at minimum pressure (about 1 psi). It is not necessary to consider the efficiency as a motor (its use as a starter is extremely short). However, the system pressure will remain above 1 psi most of the time, which requires resizing the engine pump/motor. The final optimization results for hydraulic machine are shown below: The maximum displacement of primary drive pump/motor The maximum displacement of secondary drive motor The maximum displacement of engine pump/motor 175 cc/rev 15 cc/rev 265 cc/rev Sizing of Hydraulic Accumulators In this study, the low and high pressure accumulators are designed to have the 27

44 identical size. The pressure in the former one is between 5 and 25 psi, while between 1 and 5 psi in the latter one. The size of accumulators is configured to absorb sufficient required regenerative braking energy and act as a buffer to engine ON/OFF operation, if engine shut-down strategy is applied. Theoretically, the larger the accumulators, the better it is. A smaller accumulator will lead to more frequent engine start-and-stop events and much shorter each engine run time. However, it is impossible to design infinite accumulators due to the weight and space penalty. For this SHHV, the capacity of each accumulator is designed to 1 liters. The maximum fluid capacity is 68.3 liters. The high pressure accumulator can store up to 17 kj energy at maximum operation pressure, 5 psi. And the state of charge (SOC) is defined here as: SOC P P P P min = (2.21) max min where, P is instantaneous pressure of the high pressure accumulator; P min and Pmax are allowable minimum and maximum pressures, respectively, at high pressure side. SOC is a continuous variable. is corresponding to the status of no fluid, while 1 means the high pressure accumulator is full. The relationships of SOC to fluid volume and pressure are shown in Fig

45 8 Fluid volume Pressure Fluid volume [liter] 4 3 Pressure [psi] SOC Fig. 2.8 Fluid volume and pressure as a function of SOC. 29

46 Chapter 3 Rule-based Power Management Control Strategy 3.1 Power Management Control Strategy The strategy for SHHV is straightforward, run the engine at its most fuel efficiency point (the eye of BSFC) or shut it off when not needed. When the engine is started, it will run along the optimum fuel efficiency line by controlling engine pump displacement and throttle position. Most of the time, the engine will operate at the eye of BSFC and the engine power will be dynamically divided along two paths: to the drive motor or to the accumulator. For the case of low power demand, engine power is used to satisfy the requirement of the drive motor, with the surplus engine power being absorbed by the accumulator. The accumulator will discharge to act as the secondary power source if the power request is greater than the engine can deliver at optimal efficiency. In the event that the driver demands even more power, the engine operation point will move toward higher power range along the optimum trajectory. In the extreme case, such as hard acceleration, the system pressure may drop quickly and seriously. If so, priority shifts to track the speed profile and the engine deviates from the optimum line to produce maximum power. 3

47 In order to maximize the hydraulic machine efficiency, two drive motors are coupled to the integrated two-speed transmission and differential. One of them operates as a primary motor and provides most of the propulsion. The other one, with a gear ratio of 1.5, runs only as an assistant motor during hard acceleration and/or grade requirement. Once the primary motor is able to fulfill the requirement, the secondary one is set to idle immediately. The result is that the primary pump/motor will operate at a much higher displacement range with improved efficiency compared to a single large drive pump/motor. 3.2 Discrete Operation Modes Five discrete operation modes occur during SHHV rule-based power management Accumulator Propulsion (Mode 1) In accumulator propulsion, the power propelling the vehicle comes only from the accumulator until EITHER SOC drops to the lower threshold (namely SOCstart) OR the displacement of the drive motor reaches the maximum value, namely D2max. (For the two-drive-motors configuration, D2max is decided by the sum of the primary and secondary motor displacements.) In the former event, the engine must be engaged before the fluid is depleted. A higher value for SOCstart can improve the vehicle acceleration performance, but a lower SOCstart will reduce the number of engine restarts required. Therefore, a proper SOCstart should be selected over the driving cycle, especially during aggressive sections. In the latter event where D2max is reached, the system pressure is not sufficient to drive the vehicle, hence the engine kicks in to increase system pressure. 31

48 3.2.2 SOC Regulation (Mode 2) The engine is the primary power source in this mode. The objective is to increase SOC while maintaining the optimum fuel economy. During SOC regulation, the SOC will increase steadily (as long as no rapid acceleration is demanded) until reaching the predetermined upper SOC threshold, denoted by SOCstop. This threshold is one of two factors used to determine engine shut-down. The other factor is D2max. The condition for turning the engine off is that pressure reaches SOCstop AND the sum of two drive motor displacements is less than D2max. Always, SOCstop is greater than SOCstart. The difference between them determines the buffer capacity of the accumulator. Therefore, SOCstop deserves particular attention. If this value is set too high, there will not be enough room to capture the energy during braking events. Keeping a low value inevitably results in more frequent, shorter engine runs. This not only deteriorates the vehicle drivability but decreases the fuel economy Energy Capture (Mode 3) The drive pump/motor operates as a pump to absorb the braking energy until SOC rises to 1. The resistant torque coming from pump can satisfy the brake requirement most of the time. In the limiting case, when the pump displacement reaches its maximum, the frictional brake is engaged too Frictional Braking (Mode 4) If SOC is equal to or greater than 1, for a safety precaution, the traditional frictional brake system will be applied and the accumulator will be disconnected from the 32

49 hydraulic line. When the vehicle speed is close to zero, the frictional brake system will be engaged also and energy storage is terminated. In this mode, the displacement of the drive pump is forced to zero Engine Starter (Mode 5) As the accumulator is an auxiliary power source, the engine can be started by this source, eliminating the need for an electric starter motor. In this mode, the engine pump/motor works as a motor to start the engine. Care must be taken to set SOCstart above zero to assure sufficient restart energy. 3.3 Implementation of Rule-based Control Strategy In this simulation study, the logic control model for switching operation modes is implemented by MATLAB/SIMULINK/STATEFLOW (see Fig. 3.1). Therefore, the model is compatible with other models established in the MATLAB environment. Moreover, STATEFLOW makes the strategy easily downloadable to the electric control unit (ECU) of an actual vehicle. One of objectives in this paper is to emulate physical vehicle characteristics. MATLAB and SIMULINK provide the platform to perform this simulation in a realistic manner. 33

50 Fig. 3.1 Schematic diagram of switching operation modes in MATLAB/SIMULINK/STATEFLOW. The logic control model monitors the pedal position every sample time (here 1 ms) during simulation. When pedal position status is changed by the driver, e.g. from brake to fuel, the corresponding state (Positive-Power or Negative-Power) will be activated. Then the model decides which operation mode should be executed according to the current transition conditions. One or two of five input variables fully decide the transition conditions. They are SOC, vehicle speed, engine speed, displacements of primary drive pump/motor and secondary drive motor. 34