Development of Series Mode Control of a Parallel-Series Plug-In Hybrid Electric Vehicle THESIS

Transcription

1 Development of Series Mode Control of a Parallel-Series Plug-In Hybrid Electric Vehicle THESIS Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in the Graduate School of The Ohio State University By Eric Michael Gallo Graduate Program in Mechanical Engineering The Ohio State University 214 Thesis Committee: Dr. Shawn Midlam-Mohler, Advisor Dr. Giorgio Rizzoni

2 Copyright by Eric Michael Gallo 214

3 ABSTRACT Due to the environmental and political landscape of the past few decades, the automotive industry has been rapidly evolving to meet demanding fuel economy and emissions standards. The increasingly stringent restrictions put forth by governments across the world have opened the door for hybrid vehicle development. To help develop these technologies, automotive companies are training their workforce to become well versed in the area of hybrid electric vehicles (HEVs). Companies are also focusing on developing future engineers for this task. One example of this is the EcoCAR 2 competition. This is a three year competition sponsored by General Motors (GM), the Department of Energy (DOE), and many others that challenges students from universities across North America to redesign a Chevrolet Malibu to be more fuel efficient and reduce emissions while maintaining consumer acceptability. This thesis deals with the parallel-series plug-in hybrid electric vehicle (PHEV) developed by students at The Ohio State University for this competition. Specifically, this document details the development of the series mode control in Software-in-the-Loop (SIL). This mode involves using the engine and front electric machine (FEM) to generate electricity while the rear electric machine (REM) drives the vehicle. Three separate control strategies were developed: thermostatic, load following, and Equivalent Consumption Minimization Strategy (ECMS) control. Each method went ii

4 through multiple iterations and were optimized to produce the best results to minimize a common objective function. The optimal solutions for each of the three strategies were then compared to see which would be the best solution to implement on the vehicle. The final solution points to thermostatic control towards being the best solution of the three strategies for series mode control. It is shown that this is mainly due to the fact that thermostatic control limits losses from the engine, which dominate the total losses across the vehicle. It is recognized that this solution is highly dependent on the objective function selected and that in the future, much more work needs to be done to develop a more comprehensive objective function that includes detailed emissions information. iii

5 DEDICATION I dedicate this dedication to my parents, my brother, my sister, and my girlfriend. Without them I would have no one to dedicate this dedication to. iv

6 ACKNOWLEDGEMENTS I would first like to thank Dr. Shawn Midlam-Mohler for his guidance throughout my undergraduate and graduate career. He has a true passion for teaching and I would not have pursued a Master's degree if it had not been for him. I would also like to thank Dr. Giorgio Rizzoni for his support and guidance. I am proud to have been a part of the Center for Automotive Research and he is a large part of that. I would also like to thank my EcoCAR 2 comrades, including (but not limited to), Matt Yard, Matt Organiscak, Katherine Bovee, Jason Ward, Amanda Hyde, Bharatkumar Hegde, MJ Yatsko, Andy Garcia, Dave Walters, and Curtiss Stewart. v

7 VITA June 28...Blackstone-Millville Regional High School May B.S. Mechanical Engineering, The Ohio State University August 213 to Present...Graduate Research Associate, Department of Mechanical Engineering, The Ohio State University PUBLICATIONS Bovee, K., Hyde, A., Yard, M., Gallo, E. et al., "Fabrication of a Parallel-Series PHEV for the EcoCAR 2 Competition," SAE Technical Paper , 213 K. Bovee, A. Hyde, M. Yard, T. Trippel, M. Organiscak, A. Garcia, E. Gallo, M. Hornak, A. Palmer, J. Hendricks, S. Midlam-Mohler and G. Rizzoni, "Design of a Parallel- Series PHEV for the EcoCAR 2 Competition," SAE, FIELDS OF STUDY Major Field: Mechanical Engineering vi

8 TABLE OF CONTENTS ABSTRACT... ii DEDICATION... iv ACKNOWLEDGEMENTS... v VITA... vi TABLE OF CONTENTS... vii LIST OF TABLES... x LIST OF FIGURES... xii CHAPTER 1: INTRODUCTION Motivation EcoCAR Vehicle Technical Specifications Thesis Overview... 4 CHAPTER 2: LITERATURE REVIEW Series Hybrid Electric Vehicles Control Strategies for Series HEVs Thermostatic Load Following Equivalent Consumption Minimization Strategies... 9 CHAPTER 3: Tools and Methods Vehicle Overview Vehicle Architecture Operating Modes EcoSIM vii

9 3.2.1 Top-Level Structure Driver Subsystem Supervisory Controller Front Powertrain Engine Front Electric Machine Automated Manual Transmission Rear Powertrain Rear Electric Machine Single Speed Gearbox High Voltage Battery Pack Brakes Tires Vehicle Dynamics Objective Function Simulation Setup Drive Cycles Cumulative Fuel Economy CHAPTER 4: Thermostatic Control Introduction Strategy Overview Baseline Thermostatic Optimized Thermostatic Thermostatic Control Strategies Comparison CHAPTER 5: Load Following Control Introduction Strategy Overview Baseline Load Following Optimized Load Following Load Following Strategy Comparisons CHAPTER 6: Equivalent consumption minimization strategy Introduction Strategy Overview Adaptive ECMS Allowing Engine Shutoffs Adaptive ECMS Not Allowing Engine to Shutoff ECMS Strategy Comparisons CHAPTER 7: Comparison of Strategies viii

10 7.1 Component Losses CHAPTER 8: CONCLUSIONS AND FUTURE WORK Conclusions Future Work BIBLIOGRAPHY APPENDIX: LIST OF SYMBOLS AND ABBREVIATIONS ix

11 LIST OF TABLES Table 1: Vehicle Technical Specifications (VTS) Targets... 3 Table 2: Engine Speed Determination Table 3: FEM Torque or Speed Control Determination Table 4: Transmission State Determination Table 5: Battery Information Table 6: ICE and FEM Thermostatic Operating Points Table 7: Baseline Thermostatic Results Table 8: Optimal SOC Bounds for Thermostatic Control Table 9 Optimized Thermostatic REsults Table 1: Comparsion of Thermostatic Control Strategies Table 11: Baseline Load Following Results Table 12: Optimized PI Gains for Load Following x

12 Table 13: Optimized Load Following Results Table 14: Load Following Strategy Results Comparisons... 7 Table 15: Optimized Tunable ECMS with Engine Shutoffs Parameters Table 16: Adaptive ECMS with Engine Shutoffs Results Table 17: Optimized Tunable ECMS with Engine Idle Parameters Table 18: Adaptive ECMS with Engine Idle Results Table 19: ECMS Strategies Comparison Table 2: Comparison for All Control Strategies xi

13 LIST OF FIGURES Figure 1: Series HEV Archietecture Overview [2]... 7 Figure 2: Vehicle Architecture Overview Figure 3: Front Powertrain (CAD Image) Figure 4: Rear Powertrain (CAD Image) Figure 5: Energy Storage System (ESS) Enclosure Figure 6: Charge Depleting Mode Power Flow Figure 7: Charge-Sustaining Series Mode Power Flow Figure 8: Charge-Sustaining Parallel Mode Power Flow Figure 9: Regenerative Braking Power Flow... 2 Figure 1: EcoSIM Overall Structure Figure 11: Driver Overview Figure 12: Supervisory Control Overview xii

14 Figure 13: Engine Model Overview Figure 14:ICE Efficiency Map Figure 15: ICE Fuel Consumption Map Figure 16: FEM Model Overview Figure 17: FEM Efficiency Map Figure 18: Automated Manual Transmission Model Overview... 3 Figure 19: REM Model Overview Figure 2: REM Efficiency Map Figure 21: Single Speed Model Overview Figure 22: High Voltage Battery Model Overview Figure 23 Brake Model Overview Figure 24: Tire Model Overview Figure 25 Vehicle Dynamics Model Overview Figure 26: Modified Drive Cycles Figure 27: Cumulative Fuel Economy Example Figure 28: Thermostatic Control Example xiii

15 Figure 29: ICE and FEM Combined Efficiency Map Figure 3: FEM Thermostatic Operating Point Figure 31: ICE Thermostatic Operating Point Figure 32: Thermostatic Baseline Performance (FUDS Modified)... 5 Figure 33: Thermostatic Baseline Performance (US6 City Modified) Figure 34: Optimized Thermostatic SOC Bounds Performance (FUDS Modified) Figure 35: Optimized Thermostatic SOC Bounds Performance (US6 City Modified).. 55 Figure 36: FEM Load-Following Operating Points... 6 Figure 37: ICE Load-Following Operating Point... 6 Figure 38: Load-Following Operating Points on the Combined Efficiency Map Figure 39: Baseline Load Following Performance (FUDS Modified) Figure 4: Baseline Load Following Performance (US6 Modified) Figure 41: fmincon Results for Load Following Optimization (FUDS Modified) Figure 42: fmincon Results for Load Following Optimization (US6 City Modified) Figure 43: Optimized Load Following (Performance Performance (FUDS Modified) Figure 44: OptimizedLoad Following Performance (US6 Modified) xiv

16 Figure 45: Genetic Algorithm Results for Adaptive ECMS with Engine Shutoffs (FUDS Modified) Figure 46: Genetic Algorithm Results for Adaptive ECMS with Engine Shutoffs (US6 City Cycle Modified) Figure 47: Adaptive ECMS with Engine Shutoffs Performance (FUDS Modified) Figure 48: Adaptive ECMS with Engine Shutoffs Performance (US6 City Cycle Modified) Figure 49: Genetic Algorithm Results for Adaptive ECMS with Engine idle (FUDS Modified) Figure 5: Genetic Algorithm Results for Adaptive ECMS with Engine idle (US6 City Cycle Modified)... 8 Figure 51: Adaptive ECMS with Engine Idle Performance (FUDS Modified) Figure 52: Adaptive ECMS with Engine Idle Performance (US6 City Cycle Modified) Figure 53: Component Losses for the FUDS Modified Drive Cycle Figure 54: Component Losses for the US6 Modified Drive Cycle xv

17 CHAPTER 1: INTRODUCTION 1.1 Motivation Recently, vehicle fuel consumption and emissions have become major focuses of the transportation industry. Due to major factors such as new legislation, consumer demand, and rising oil prices, automotive manufacturers are turning to new, innovative solutions to improve their fleet of vehicles. These solutions include aerodynamic improvements, light-weighting of vehicles, alternative fuels, and hybridization and electrification. If current trends provide for an indication of where the automotive industry is headed, hybrid electric vehicles (HEVs) and electric vehicles (EVs) will become very prevalent in the market in the coming years. HEVs consist of powertrains that combine more than one form of energy. For example, the Chevrolet Volt is an HEV that combines a conventional gasoline powered engine with electrical power generation provided from a battery pack. EVs consist of a powertrain powered solely by electricity, usually coming from a battery pack. These technologies are all relatively new and still need much development before they can hold a significant place in the automotive market. It is for this reason that companies are investing heavily in training the next generation of engineers to work on this problem. 1

18 EcoCAR 2, an Advanced Vehicle Technology Competition (AVTC), is one way companies are pursuing this. 1.2 EcoCAR 2 EcoCAR 2 is a three year competition headline-sponsored by the U.S. Department of Energy (DOE) and General Motors (GM). There are a total of fifteen university teams spread across the U.S. and Canada that participate. The goal of this competition is for students to completely redesign and build a 213 Chevrolet Malibu to make it more fuel efficient while maintaining consumer acceptability. The competition is broken up into three years, each with a final competition at the end. In year 1, the teams first create vehicle technical specifications (VTS) to design their vehicle to. After these specifications are set, the team decides on a vehicle architecture to pursue. Following this, the team works on modeling and simulation to further design the vehicle. This includes computer-aided design (CAD) to package the components in the vehicle and hardware-in-the-loop (HIL) testing to begin testing control strategies. Year 2 is when the teams receive the vehicle and proceed to modify it to build what they designed the previous year. The end goal of this year is to built a functioning mule vehicle that they can test. Year 3 is a refinement year; Teams make any changes to the vehicle and attempt to create a showroom ready car that a customer would want to purchase. The teams are judged on things ranging from static consumer acceptability, fuel economy, emissions, vehicle performance, and ride quality. 2

19 1.3 Vehicle Technical Specifications One of the most important first steps in designing a vehicle in the EcoCAR 2 competition is developing a set of Vehicle Technical Specifications (VTS). These VTS targets motivate just about every design decision that goes into the vehicle. The Ohio State EcoCAR 2 team's VTS table is shown in Table 1. Table 1: Vehicle Technical Specifications (VTS) Targets Specification Production 213 Malibu Competition Design Target Competition Requirement OSU Proposed Design Acceleration -6 mph 8.2 sec 9.5 sec 11.5 sec 1 sec Acceleration 5-7 mph (Passing) 8. sec 8. sec 1 sec 4.6 sec Braking 6- mph ft (43.7 m) ft (43.7 m) 18 ft (54.8 m) ft (43.7 m) Highway 1+% 3.5% 3.5% mph 2 min Cargo Capacity 16.3 ft ft 3 7 ft 3 1 ft 3 Passenger Capacity 5 >=4 2 5 Mass 1589 kg <225 kg <225 kg 275 kg Starting Time <2 sec <2 sec <15 sec <1 sec Ground Clearance Vehicle Range mm 155 mm 127 mm >127 mm 736 km [457 mi] (CAFE) 322 km [2 mi]* 322 km [2 mi]* >398 km [247 mi]* (Continued) 3

20 Table 1: Continued Charge-Depleting Range* Charge-Depleting Fuel Consumption* Charge- Sustaining Fuel Consumption* UF-Weighted Fuel Energy Consumption* UF-Weighted AC Electric Energy Consumption* UF-Weighted Total Energy Consumption* UF-Weighted WTW Petroleum Energy (PE) Use* UF-Weighted WTW GHG Emissions* Criteria Emissions N/A ** N/A N/A ** N/A N/A ** N/A 8.83 (lge/1 km) [787 Wh/km] 7.12 (lge/1 km) [634 Wh/km] N/A N/A ** N/A 787 (Wh/km) 634 (Wh/km) N/A 774 (Wh PE/km) 253 (g GHG/km) 624 (Wh PE/km) 24 (g GHG/km) N/A N/A 97.4km [6.5 mi] (Wh/km) 5.35 (lge/1km) [476.4 Wh/km] 1.36 (lge/1km) [121.1 Wh/km] (Wh/km) (Wh/km) 43.1 (Wh PE/km) (g GHG/km) Tier 2 Bin 5 Tier 2 Bin 5 N/A <Tier 2 Bin Thesis Overview This thesis sets to find an ideal control strategy for the series mode of the Ohio State EcoCAR 2 vehicle that helps achieve the VTS targets mentioned previously. It goes through three separate strategies and optimizes each with respect to a specific objective 4

21 function. Once the best candidate from each type of strategy is selected, these three separate strategies are compared and the ideal strategy is selected. The organization of this thesis is as follows: Chapter 2: Literature Review This chapter provides background information on control strategies that exist for series hybrid electric vehicles. Chapter 3: Tools and Methods This chapter discusses the vehicle architecture and the different operating modes of the vehicle. It also describes the EcoSIM vehicle simulator that was used in this analysis. This includes a brief description of every major subsystem and its components. It then goes on to discuss the formulation of the objective function and how the analysis was conducted. Chapter 4: Thermostatic Development This chapter describes the development of the thermostatic control strategy for the series mode of the vehicle. It begins with a baseline model and then optimizes it in attempts to improve performance. Chapter 5: Load Following Development This chapter discusses the load following control strategy and its development. It begins developing a strategy with no tunable parameters and then introduces modifications to enable it to be implementable on the vehicle. Chapter 6: ECMS Development 5

22 This chapter goes through the ECMS strategy development. It begins with a baseline strategy which is then optimized to improve performance. Chapter 7: Comparison of Strategies This chapter compares all three strategies developed in Chapter 4 through 6 and determines which is the ideal control candidate. It then goes in depth to see why the ideal strategy is such. Chapter 8: Conclusions and Future Work This chapter summarizes the results found and suggests future steps to take to improve the analysis. 6

23 CHAPTER 2: LITERATURE REVIEW 2.1 Series Hybrid Electric Vehicles In series Hybrid Electric Vehicles (HEVs), all of the traction power is directly from electricity [1]. The internal combustion engine (ICE) is decoupled from the wheels and connected to a electric machine that operates as a generator. This generator converts the power produced by the ICE to provide electrical power that either powers the drive motor directly or charges the battery pack. An overview of the series HEV configuration is shown below in Figure 1. Figure 1: Series HEV Archietecture Overview [2] 7

24 Some advantages of series mode vehicles include allowing the ICE to operate at optimal operating points to allow for better efficiency and simpler control strategies. A major disadvantage associated with this architecture is the fact that there are inefficiencies associated with converting energy from mechanical energy produced by the engine to electric energy converted by the generator and then back to mechanical energy to propel the vehicle. There are added inefficiencies from the charging and discharging losses from the batteries. Also, this mode often requires a large traction motor since it is the sole source of tractive force [3], [4]. 2.2 Control Strategies for Series HEVs Thermostatic Thermostatic, or on/off, control is one of the simplest forms of control. It involves turning a controlled system on to a specified and constant set-point to achieve a desired state of the system. When the desired state of the system is achieved, the controlled component will turn off until the state of the system no longer meets certain criteria. In the case of the series hybrid vehicle, the engine will turn on and power the generator when the SOC reaches its lower threshold and remain on at a fixed point until the SOC reaches its upper threshold, at which point the engine will shut off [5]. Typically there are two ways of determining the engine operating point. First, and more commonly, the engine is selected to operate at the speed and torque that offer the highest brake specific fuel consumption [5]. The desired set-point for the engine is the speed and torque at 8

25 which the highest system efficiency is achieved [6]. This system efficiency includes the components involved in the charging process. In the case of the series hybrid-electric vehicle, this would include the efficiency of the generator and the charging efficiency of the batteries. It is widely acknowledged that thermostatic control, while simple, does not offer close to near-optimal results [5], [7] Load Following Load following control is a common control strategy amongst series HEVs. This strategy calculates the power to be generated by the generator based on the overall power demand for the vehicle to meet torque and accessory power demands. This control strategy was used for the previous Ohio State AVTC vehicle [8]. The advantages of this control strategy are that the battery is used less than other strategies. This helps to reduce charge and discharge losses associated with battery usage. This strategy is also beneficial because of its relative ease to implement on a vehicle with minimum tuning [8]. A negative aspect of this strategy is the fact that the ICE and FEM set (genset) will stray from the single most efficient operating point based on power request Equivalent Consumption Minimization Strategies The issue with global optimization methods is that the entire drive schedule must be know ahead of time to find a solution. The Equivalent Fuel Consumption Strategy 9

26 (ECMS) reduces the global optimization goal to a local optimization carried out at each time step. This allows ECMS to be implemented in real time in practical applications. The foundation of ECMS is based on the fact that in a charge-sustaining vehicle, whatever energy the battery supplies in the form of electrical power must be replenished by the either the engine or regenerative braking at some point in the future. Otherwise, if the battery is charging, it will result in a future fuel savings because the battery power will supplant power from the ICE. Because of this, the energy provided by the battery can be considered as virtual fuel consumption. The total equivalent fuel consumption can represented as shown in Equation (1). The purpose of ECMS is to minimize this cost function at every time step while meeting overall power demand. (1) where represents the total equivalent fuel consumption of the ICE and battery, represents the ICE fuel consumption, and represents the equivalent battery fuel consumption. The lower heating value of the fuel used is represented as, while the battery power is represented as. The equivalency factor, s, is what converts the electrical power to fuel consumption [9]. The difficulty with ECMS lies in selecting the appropriate equivalency factor. The equivalency factor is different for battery discharge and charge conditions, as well as different diving cycles. The makes it unrealistic to implement ECMS with a constant s in 1

27 any real application. To remedy this, a few methods have been introduced. One method is to use a penalty function, p(x) as seen in Equation (2). (2) This penalty function penalizes any deviation from the desired charge sustaining SOC. At this set SOC, the value of the penalty function is 1. If the SOC drops below the set point, the penalty function will take a value greater than 1. This serves to increase the cost of electricity so the control favors using the engine to meet more of the power demand. This will allow the battery to gain SOC. The opposite is true if the SOC is higher than the set point; the penalty function is less than 1, effectively reducing the cost of electrical power, which, in the end, will decrease SOC to the set point. Another method that has been introduced to account for the variation of equivalency factors is adaptive ECMS. Adaptive ECMS uses vehicle information to adjust the equivalency factor. This can either be done through driving pattern recognition, drive cycle prediction, or SOC feedback [1]. For driving pattern recognition, an algorithm examines a window of vehicle speeds and relates them to previously defined driving patterns, each of which has their own specific equivalency factor. To predict drive cycles, information can be analyzed from wheel speed and current SOC, or the controller can utilize information from Intelligent Transportation Systems to analyze what traffic conditions the vehicle will see. Adaptation based solely on SOC feedback usually is 11

28 accomplished using a PI control to adjust s based on SOC. These adaptive strategies are shown to be able to perform close to as well as the optimized ECMS for specific drive cycles [1]. 12

29 CHAPTER 3: TOOLS AND METHODS 3.1 Vehicle Overview The following sections describe to Ohio State EcoCAR 2 vehicle architecture and operating modes Vehicle Architecture The Ohio State EcoCAR 2 vehicle is a Parallel-Series Plug-in Hybrid-Electric Vehicle (PHEV). An overall diagram of the vehicle is shown in Figure 2. Figure 2: Vehicle Architecture Overview 13

30 As can be seen from the image above, the vehicle has separate powertrains on the front and rear axles. The front powertrain consists of a 1.8L high-compression internal combustion engine (ICE). This engine was originally a compressed natural gas engine but the team in a previous competition converted it to run E85. This modification allowed the engine to operate much more efficiently than typical ICE engines. The engine outputs to a clutch that connects the engine to the input shaft of an automated 6-speed manual transmission. Also at the input of the transmission is an 8 kw electric machine that connects via belt drive. The front powertrain is shown in Figure 3. Figure 3: Front Powertrain (CAD Image) 14

31 The rear powertrain consists of an electric machine coupled directly to a single speed transmission. This is shown in Figure 4. Figure 4: Rear Powertrain (CAD Image) A kw-hr 34V A123 battery pack consisting of 7 modules provides the highvoltage electrical power for the vehicle. This module was placed in the trunk of the vehicle, close to the rear powertrain. The pack enclosure can be seen in Figure 5. 15

32 Figure 5: Energy Storage System (ESS) Enclosure This complicated vehicle architecture allows the vehicle to operate in many different operating modes, which are discussed in the next section Operating Modes This vehicle is able to operate in four distinct modes. The first mode is charge depleting (Figure 6). In this mode, the ICE is off and the vehicle is powered by the two electric motors. In this mode, the 6-speed automated manual transmission switches between gears to keep the front electric machine (FEM) at its most efficient operating point. 16

33 Figure 6: Charge Depleting Mode Power Flow Once the state of charge (SOC) of the battery drops below 18%, the vehicle then enters one of two charge sustaining modes. If the vehicle is travelling at a speed less than 3 miles-per-hour (MPH), the vehicle will be operating in charge sustaining series mode (Figure 7). In this mode, the ICE is on and spinning the FEM, which acts as a generator to charge the battery pack. The 6-speed manual transmission is in neutral so the tractive power is coming from the rear electric motor (REM) alone. 17

34 Figure 7: Charge-Sustaining Series Mode Power Flow At speeds greater than 35 MPH, the vehicle enters charge sustaining parallel mode (Figure 8). In this mode, the ICE, FEM, and REM are all driving the vehicle. In this mode the 6-speed manual transmission changes gears to ensure optimal operation of the front powertrain. 18

35 Figure 8: Charge-Sustaining Parallel Mode Power Flow The final mode of the vehicle is regenerative braking (Figure 9). This mode occurs when the driver requests negative torque by applying the brakes. This mode uses the FEM and REM as generators to absorb the torque from the wheels, slowing down the vehicle while charging the battery pack. 19

36 Figure 9: Regenerative Braking Power Flow 3.2 EcoSIM 2 The Ohio State EcoCAR 2 team used an energy-based model called EcoSIM 2 to develop control strategies for the vehicle. This model allows the team to see how different strategies affect fuel economy and performance. This chapter contains an overview of this model. A more detailed description can be found in [11] Top-Level Structure EcoSIM2 is broken up into 4 major systems. These are the driver, supervisory control, powertrain, and the vehicle dynamics subsystems. The driver subsystem acts as the human driver would in real life and changes the brake and accelerator pedal position based on the desired speed based on a predetermined drive cycle. The supervisory control 2

37 subsystem considers the pedal positions along with vehicle speed to determine an appropriate torque and speed request for the powertrain components. These requests are sent to the powertrain subsystem which contains the component level controllers and plant models. This subsystem then outputs a total tractive force that the tires exert on the road which is fed into a vehicle dynamics subsystem which determines vehicle speed. This speed is then fed back into the driver subsystem. A diagram of the EcoSIM 2 overview is shown in Figure 1. Supervisory Controller Torque/Speed Request Powertrain Traction Force Vehicle Dynamics Speed and Accelerator/Brake Pedal Positions Vehicle Speed Driver Figure 1: EcoSIM Overall Structure 21

38 3.2.2 Driver Subsystem The vehicle power request originates from the brake and accelerator pedal positions that come from the driver model. This model is a simple PID controller that adjusts pedal position based on the difference between the current vehicle speed to the desired vehicle speed and adjusts. The desired vehicle speed comes from a set drive cycle. An overview of this model can be seen in Figure 11. Desired Speed Current Speed Driver (PID Controller) Brake Pedal Position Accelerator Pedal Position Current Speed Figure 11: Driver Overview Supervisory Controller The Supervisory Controller subsystem is responsible for determining what operating mode the vehicle is in as well as torque, speed, and gear requests. An overview of the supervisory controller model is shown in Figure

39 Gear Request ICE Torque Request Accelerator/Brake Position Vehicle Speed Vehicle States Mode Selector (StateFlow) Mode Mode Operation FEM Speed/ Torque Request REM Torque Request Figure 12: Supervisory Control Overview First, the controller must decide what mode the vehicle is to be in. To do this, the mode selector StateFlow considers the pedal position, vehicle speed as well as the current state of the vehicle. This vehicle state includes a variety of parameters such as battery SOC and component states. Based on these inputs, the mode selector selects an operating mode at which point this command is passed to the mode operation block. This block contains how the each of the modes determine the torque and speed requests for each component as well as the gear command Front Powertrain The front powertrain model consists of the engine, front electric machine (FEM) and the automated manual transmission 23

40 Engine The engine model contains a soft ECU and the engine plant model. The soft ECU models the engine controller that is present in the vehicle. The soft ECU receives the ICE Torque request, the FEM speed, vehicle speed, and vehicle state and from this is commands a torque. An overview of this model is seen in Figure 13. ICE Torque Request ICE Torque Command ICE Torque Vehicle Speed FEM Speed Vehicle State Engine Soft ECU ICE Speed Engine Plant Model ICE Speed Figure 13: Engine Model Overview The engine speed is not commanded by the engine ECU. The engine speed is determined by either the FEM or wheel speed based on what mode it is in. This is summarized in Table 2. 24

41 Table 2: Engine Speed Determination Engine State Conditions Applicable Modes Speed Determination Engine off, decoupled from wheels Clutch disengaged Charge Depleting RPM Engine on, coupled to wheels Transmission in gear, clutch engaged Charge-Sustaining Parallel Wheel speed dependent Engine on, decoupled from wheels Transmission in neutral, clutch disengaged Charge-Sustaining Series FEM speed dependent The engine plant model includes simplified dynamics of the engine torque and speed response and output the actual torque and speed. With this torque and speed it is possible to find the engine efficiency and engine fuel consumption rate based on the maps shown in Figure 14 and Figure 15, respectively. 25

42 Engine Brake Torque [Nm] Engine Brake Torque [Nm] Engine Efficiency [%] Engine Speed [RPM] Figure 14:ICE Efficiency Map Engine Fuel Consumtion [kg/s] Engine Speed [RPM].1 Figure 15: ICE Fuel Consumption Map 26

43 These maps were generated by students from the previous Ohio State EcoCAR team using an engine dynamometer Front Electric Machine The FEM contains a soft ECU that considers the torque and/or speed requests from the supervisory control as well as vehicle speed, ICE torque, and the vehicle state to determine FEM torque and speed commands. The overview of the FEM model can be seen in Figure 16. FEM Torque Request FEM Speed Request FEM Torque Command FEM Torque Vehicle Speed ICE Torque Vehicle State FEM Soft ECU FEM Speed Command FEM Plant Model FEM Speed FEM Current Figure 16: FEM Model Overview The FEM can be either commanded a speed or a torque. Which of these is commanded is determined by the state of the vehicle, including clutch engagement, transmission gear, and vehicle operating mode. A summary of this is shown in Table 3. 27

44 Table 3: FEM Torque or Speed Control Determination FEM State Conditions Applicable Modes Torque or Speed Control FEM on, coupled to wheels Clutch disengaged Charge Depleting Torque FEM on, coupled to wheels Transmission in gear, clutch engaged Charge-Sustaining Parallel Torque FEM on, decoupled from wheels Transmission in neutral, clutch disengaged Charge-Sustaining Series Speed Once the FEM torque and speed commands are determined, they get passed to the FEM plant model which contains simplified dynamics. The plant model then outputs a torque and speed. From this, the efficiency of the FEM can be determined using Parker- Hannefin supplied efficiency maps seen in Figure

45 FEM Torque [Nm] FEM Efficiency [%] FEM [RPM] Figure 17: FEM Efficiency Map 29

46 Automated Manual Transmission An overview of the automated manual transmission model can be seen in Figure 18. ICE Torque Engine Clutch FEM Torque Belt Coupling (Connects FEM and ICE) Combined FEM and ICE Torque Torque at Axle Gear Command Transmission Plant Model Gear Ratio Gear Request Transmission Soft ECU Figure 18: Automated Manual Transmission Model Overview This model contains the ICE clutch, which has one of two states: engaged or disengaged. It also contains the belt coupling model that sums the FEM and ICE torque. The transmission soft ECU receives the gear request from the supervisory control and in turn commands the transmission plant model to change gears. The transmission plant model then outputs the torque at the axle based on the gear ratio of the selected gear. The transmission can be either in neutral or in gear depending on the vehicle operating mode. This is summarized in Table 4. 3

47 Table 4: Transmission State Determination Transmission State Neutral In Gear Modes Charge-Sustaining Series Charge-Sustaining Parallel, Charge Depleting Rear Powertrain The rear powertrain model consists of the rear electric machine and single speed gearbox models Rear Electric Machine The rear electric machine (REM) model consists of a soft ECU that receives the REM torque request from the supervisory control and the rear axle speed. This ECU then commands a REM torque to the REM plant model. This plant model contains simplified dynamics that outputs a torque and current for the REM. An overview of this model can be seen in Figure

48 REM Torque Request Rear Axle Speed REM Soft ECU REM Torque Command REM Plant Model REM Torque REM Current Figure 19: REM Model Overview Unlike the FEM, the REM is strictly torque controlled due to it being constantly coupled to the rear axle through the single speed gearbox. Using the REM torque and speed, it is possible to find the efficiencies based on the Parker-Hannefin supplied map seen in Figure 2. 32

49 REM Torque [Nm] REM Efficiency [%] REM [RPM] Figure 2: REM Efficiency Map Single Speed Gearbox The single speed gearbox is part of the rear powertrain, connecting the REM to the rear axle. The gear ratio is fixed at An overview of this model is shown in Figure

50 REM Torque Single Speed Gearbox Plant Model Rear Axle Torque Figure 21: Single Speed Model Overview High Voltage Battery Pack An overview of the high voltage battery pack is shown in Figure 22. The high voltage battery model determines battery SOC, voltage, and current based on a first-order equivalent circuit model. 34

51 Battery Current FEM Current Battery Voltage Battery Voltage REM Current High Voltage Battery Model Battery SOC Battery Soft ECU Battery Current Accessory Current Battery SOC Heat Generation Figure 22: High Voltage Battery Model Overview The battery model also calculates heat generation based on current and the internal resistance of the battery. The battery soft ECU is meant to represent the A123 battery controller and outputs the battery voltage, current, and SOC. More information on the battery pack can be found in Table 5. 35

52 Table 5: Battery Information Pack Configuration 7x15s3p Cell Min Capacity 19.6 Amp-hr Cell Nominal Voltage 3.24 V Cells in Pack 315 Pack Voltage 34 V Continuous Charge Limit -21 kw Continuous Discharge Limits 6 kw Brakes The brake model overview can be seen in Figure 23. This model considers the brake pedal position from the driver and converts that to friction brake torque. This friction brake torque is then subtracted from the axle torque, leaving the torque at the wheel. Brake Pedal Position Axle Torque Brake Plant Model Torque to Wheel Figure 23 Brake Model Overview 36

53 3.2.8 Tires The tire model overview can be seen in Figure 24. This model considers the wheel torque and converts it to tractive force based on the wheel diameter. Wheel Torque Tire Plant Model Tractive Force Figure 24: Tire Model Overview Vehicle Dynamics To determine vehicle speed, a simplified vehicle dynamics model was created by the team. This model considers the front and rear wheel tractive force, road load, and vehicle mass, to determine the vehicle speed. An overview of this model can be seen in Figure

54 Front Wheels Tractive Force Rear Wheels Tractive Force Road Load Vehicle Dynamics Model Vehicle Speed Vehicle Mass Figure 25 Vehicle Dynamics Model Overview The road load is determined by considering the resistive forces from gravity based on the grade of the road, the aerodynamic forces, and the rolling resistance from the road surface. 3.3 Objective Function In order to compare the different control strategies that were developed for series mode, an objective function was created. The goal of this objective function is to create an objective method of comparison of the strategies based on fuel economy, engine operation, and charge sustainability. The objective function is shown in Equation (3). 38

55 (3) Where:, The variables,, and are all normalized and scaled by the scaling factors. The normalized fuel variable considers the fuel economy achieved for a given drive cycle and normalizes it with respect to the best fuel economy feasible ( kg/km) and worst fuel economy expected (2 kg/km). The normalized engine variable considers the amount of times the engine shuts off during a drive cycle and normalizes it against the best performance ( shutoffs/km) and the worst (5 shutoffs/km). This engine term was included because it is attempting to discourage strategies that turn the engine on and off too frequently. This type of operation is bad for the engine, bad for noise, vibrations, and harshness (NVH), and also can lead to increased emissions. The 39

56 normalized SOC variable considers the deviation from the set charge sustaining SOC (18%). The SOC variance is given by Equation (4). (4) This equation penalizes the control strategy for having an SOC trace that tends to be either charge gaining or charge depleting. It is divided by the total distance traveled to normalize and allow for comparisons across all drive cycles. The scaling factors,, and, are using to weight each factor with regards to how important they are to developing a desirable control strategy. For this particular analysis, was set to.75, to.2, and to.5. These scaling factors demonstrate the overall importance of fuel economy. The engine operation is also important and is weighted such that it penalizes strategies that attempt to shut the engine off too many times. The SOC variance term is weighted lightly in comparison to the other two variables because it is seen as not as important and is weighted just enough to penalize and eliminate any charge gaining or charge depleting strategies. 4

57 3.4 Simulation Setup In order to compare the results of different control strategies, it was necessary to setup the simulations such that the vehicle was constantly in series mode. This involved having the initial SOC be 18% to ensure that it started in series mode and modifying the drive cycles. It was also necessary to determine cumulative fuel economy that was not highly dependent on final SOC. The simulation setup is discussed below Drive Cycles The control strategies were developed using two different modified drive cycles. These drive cycles are based off of the US6 city cycle and the FUDS cycle. The cycles were modified to have a maximum vehicle speed of 35 MPH. This speed was selected because this is the threshold at which the vehicle transitions from charge sustaining series mode to charge sustaining parallel mode. Series mode was isolated to make it easier to observe how the control strategies performed without diluting the results with operation in other modes. In Figure 26 below, the original drive cycle is shown on the left followed by the modified cycle on the right. 41

58 Figure 26: Modified Drive Cycles Cumulative Fuel Economy It is common when comparing control strategies to force the final SOC to be the same for every simulation. This is to ensure that the fuel economy results can fairly be compared to each other because a lower final SOC will most likely have a smaller fuel consumption. This is not, however, necessarily representative of the most fuel efficient control strategy because the simulations do not have the same end point. 42

59 Fuel Economy (kg/km) To get rid of this necessity each simulation was run for multiple drive cycles until the cumulative fuel economy per cycle remained the same. By running multiple cycles, the fuel economy dependence on final SOC goes away. This can be seen in Figure 27. The red lines represent the end of one FUDS drive cycle. After three consecutive cycles, the cumulative fuel economy trace looks the same. The average fuel economy of the final cycle is then taken to be the overall fuel economy for that given strategy for that drive cycle..15 Cumulative Fuel Economy Time(s) Figure 27: Cumulative Fuel Economy Example 43

60 CHAPTER 4: THERMOSTATIC CONTROL 4.1 Introduction Thermostatic, or on-off control is one of the simplest ways to control a series hybridelectric vehicle. This mode is advantageous because it is simple to implement as it does not require a lot of tuning and the engine tends to operate smoothly at steady state since it stays at a fixed operating point. This is beneficial for NVH. Also, because the engine is decoupled from the wheels, the speed and torque can be set to the most efficient operating point of the engine. One common disadvantage associated with this mode is that it uses the battery as a buffer too much. Since the genset is supplying constant power, the battery pack will either absorb excess power produced or provide the extra power necessary to meet the vehicle power demand. This is undesirable because there are losses associated with charging and discharging the batteries. This increased loss can be mostly avoided in different strategies as will be discussed in later chapters. Another disadvantage with this method of control is the lack of flexibility. Because of the lack of tunable parameters, this mode has little room for improvement and optimization. This chapter discusses the development of thermostatic control for the vehicle. 44

61 4.2 Strategy Overview Thermostatic control has a fixed operating point for the genset and SOC bounds that decide whether the genset should be on or off. When the SOC hits the upper bound, the genset is switched off and the vehicle depletes charge until it reaches the lower SOC bound at which point the genset turns on and generates power until the SOC reaches the upper SOC bound again. A simple depiction of this is shown in Figure 28. Figure 28: Thermostatic Control Example The first step in developing this method of control is to select the operating point of the FEM and ICE. Because these components are not connected to the wheels in series mode, the operating point can be at whatever speed and torque is desired. As discussed in 45

62 Chapter 3, when the vehicle is in series mode, the speed of the engine is determined by the FEM controller and torque is determined by the engine controller. To find the best operating point of the genset it is necessary to consider the combined efficiency of both the ICE and the FEM. To do this, the efficiency maps for each component shown in Chapter 3 were combined. To do this, the efficiency of both components were multiplied together at each possible operating point. The ICE and FEM speeds and torques are related by the 1.75 belt ratio that couples both components together. Figure 29 below shows the combined efficiency map in terms of engine torque and engine speed. 46

63 Engine Brake Torque [Nm] Combined Efficiency [%] Engine Speed [RPM] Figure 29: ICE and FEM Combined Efficiency Map With the combined efficiency of both components, the next step is to find the most efficient operating point for the genset to operate at. Since thermostatic control leaves the engine running at a constant power for extended periods of time, it is important to consider the battery charging limits when selection an operation point. As discussed in Chapter 3 in Table 5, the maximum continuous power rating for charging the battery is 21 kw. This means that the genset cannot have a set point that generates more than 21 kw. To find this set point, every possible combination of torque and speed was examined and the combination that had the highest efficiency while being under 21 kw was selected. The specific speed and torque combination with the corresponding efficiencies of each component are shown in Table 6. 47

64 Table 6: ICE and FEM Thermostatic Operating Points Component Operating Point Efficiency ICE Nm 41% FEM Nm 9% This points can be seen on the efficiency maps of both the FEM and ICE in Figure 3 and Figure 31, respectively. Figure 3: FEM Thermostatic Operating Point 48

65 Figure 31: ICE Thermostatic Operating Point The selection of the thermostatic operating point allows the rest of the strategy to be developed, as discussed in the next sections. 4.3 Baseline Thermostatic As discussed in Section 4.1, thermostatic control is very limited in the amount of optimization that can be done. The only tunable parameters are the SOC bounds where the control decides to turn the genset on and off. To create a baseline strategy, the entire SOC range for series mode (13-23%) was first considered. This served as a baseline and the SOC bounds were optimized based on these results. 49

66 speed [kph] Battery SOC (%) Torque [Nm] speed [rpm] To test this strategy, the vehicle was simulated over both the FUDS and US6 modified drive cycles as discussed in Chapter 3 and component operation and fuel economy results were recorded. The component speeds and torques, vehicle speed, and battery SOC for both drive cycles are shown in Figure 32 and Figure FEM REM ICE 4 3 FEM REM ICE time [s] time [s] Vehicle velocity Cycle Velocity time [s] time [s] Figure 32: Thermostatic Baseline Performance (FUDS Modified) 5

67 speed [kph] Battery SOC (%) Torque [Nm] speed [rpm] 2 1 FEM REM ICE 4 3 FEM REM ICE time [s] time [s] 6 5 Vehicle velocity Cycle Velocity time [s] time [s] Figure 33: Thermostatic Baseline Performance (US6 City Modified) As can be seen by these results, the ICE and FEM perform as expected. The genset operates smoothly and maintains the SOC with the desired bounds. In order to compare these results objectively to future iterations, the objective function value discussed in Chapter 3 was calculated along with fuel economy and the number of engine shutoffs. These results are summarized in Table 7. 51

68 Table 7: Baseline Thermostatic Results Drive Cycle US6 Modified FUDS Modified Fuel Economy (kg/km) Engine Shutoffs Objective Function These results were then compared to an optimized thermostatic strategy discussed in the next section. 4.4 Optimized Thermostatic The next step was to optimize the SOC bounds of the thermostatic control strategy. To do this an SOC bound sweep was performed. The upper SOC was limited between 2% and 23% in increments of 1% while the lower bound was limited between 13% and 16% in increments of 1%. To do the SOC bound sweep, every possible combination of upper and lower SOC bounds were tested for both drive cycles. The SOC bounds combination that produced the lowest value of the objective value was selected as the optimal SOC bounds. The optimal SOC bounds for both the US6 city and FUDS modified drive cycles are shown below in Table 8. 52