Get Complete Project Material File(s) Now! »

LITERATURE REVIEW

Mahapatra et al. Paper on Model-Based Design

A paper by Mahapatra et al. [5] offers several insights into the motivation for and advantages of using model-based design as an integral tool throughout the design and implementation process. Mahapatra says that while model based design is traditionally only used for controller development, one of the goals of the paper is to demonstrate how model-based techniques can be used throughout the design process. The paper goes on to discuss general challenges in HEV system design. According to Mahapatra the key aspect of model-based design is the ability to continually verify that design requirements are being met at each step of the way.
The paper follows the development of a hybrid vehicle that uses two electric machines and a planetary power splitting device. While this architecture is distinctly different from the architectures that will be considered for this research, there is still much insight to be gained from this paper. Mahapatra describes how different steps of design require models with varying levels of abstraction and fidelity. The model-based design process is described as a process of continually elaborating simulation models from a concept to a detailed system design that can verify performance. The overall goal is to ensure first pass success when a prototype is built.
The paper suggests starting with a high level system model. The subsystems in the model are then each passed on to various specialists who refine the models while also refining and updating requirements. Detailed models are then integrated back into the system level model and verified by simulation. This is described as an iterative process that converges to an optimal design. Code generation is described as the next step to facilitate testing on a target processor as well as accelerating simulation and doing HIL testing.
The paper also gives some description about the models used. The models of the mechanical powertrain components use the Mathworks SimDriveLine toolkit for calculations. Mathworks SimPowerSystems is used for some electrical components. Mahapatra discusses the tradeoffs of creating highly detailed models and suggests against models with excessive detail. He describes some of the simplified subsystem models used for simulation. The control strategy developed for the vehicle is a state machine type architecture programmed using Mathworks Stateflow.
This research will address some of the short comings of Mahapatra’s paper by describing in more detail system design of the vehicle control system. For instance interaction and communication between vehicle ECUs will be examined. This paper will also present in detail a control strategy aimed at minimizing fuel consumption. This paper will also describe in detail HIL testing using dSPACE Automotive Simulation Models (ASM).

Marco and Cacciatori Paper on Model-Based Design Techniques

A paper by Marco and Cacciatori [6] aims to show how use of model based design techniques can be used to reduce complexity and development time of HEV systems. The paper focuses on two main areas, architecture and component selection and control algorithm design. The paper discusses differences between online and offline simulation, as well as forward looking and backward looking simulation models during initial simulation and component selection phase. Two case studies are presented as examples for architecture selection using simulation. For the first, a backward looking model is used in the development of a fuel cell hybrid sports car to examine tradeoffs between parameters such as vehicle weight, powertrain efficiency and aerodynamics. The model requires a generic control strategy to run drive cycles for measuring fuel economy. The results provide a sensitivity analysis for varying the parameters to fuel economy. The second case study looks to compare actual powertrain component selection choices. In particular, simulation is used to choose between different gearing configurations for the traction motor and selection of the motor itself.
While the methods presented can be effective they have some shortcomings. While the paper looks at component selection, it does not address choosing between distinctly different architectures. Another important issue is that the simulations used by Marco and Cacciatori require control strategies to operate the powertrain. At the architecture selection point in design however, a detailed control strategy has usually not been developed. This means a simple strategy must be created for each architecture just for the purpose of determining if that architecture is a desirable option. The goal should be to determine the potential that a given vehicle architecture has of improving fuel economy without looking at the actual control strategy.
The paper then goes on to discuss control architecture and strategy development. Universal modeling languages SysML and UML are presented as tools for this process. A visual model is developed to show interaction of the requirements of the control system between the driver and the vehicle. Mathworks Simulink is then used to implement the actual control strategy. The Wren Project is introduced as a tool to facilitate interchangeability of control functions between different powertrain concepts but is not described in detail. The paper does not actually present a detailed control architecture or strategy. The current work aims to directly demonstrate how model-based design tools have actually been used to develop and refine a control system.

Katrašnik et al. Article on Energy Conversion Efficiency

An IEEE article by Katrašnik et al. [7] presents analysis on conversion efficiency for both series and parallel hybrid powertrains. This is important to this research because choosing between these powertrain options is an important step during architecture selection. The article mentions an important limitation in lookup table based models traditionally used for simulations. Lookup table based engine models, for instance, do not account for dynamics and transients. Turbo charged engines cannot be effectively modeled in this way. The article therefore points to forward looking dynamic models as the most effective for simulation. The article presents two approaches of analyzing powertrain efficiency, analytical and through simulation. The analytical approach uses energy balance equations to present a solution for calculating the powertrain efficiency. The equations however use somewhat arbitrary efficiency constants and the article offers no explanation of how to derive the constants.
The simulation software used is a thermodynamic and fluid mechanics code developed by Katrašnik. The article goes into detail about the parameters for each of the component models for the series and parallel architecture. Rather than running vehicle drive cycle simulations, simulations are run for the European Transient Cycle engine dynamometer test, using the engine load as a powertrain load. This was done to compare the hybrid powertrain to the baseline conventional powertrain without the influence of gear shift strategies and vehicle parameters. For the simulation a thermostatic type control strategy was used for the series hybrid. The control strategy for the parallel was not described in detail but offered several modes of operation. The article does not address how the control strategy affects the powertrain efficiency or support the control strategies used. This is a limitation that this research aims to address. Comparing different architectures independent of a specific control strategy is important to identifying whether one architecture is superior to another. The findings of the paper are that overall the parallel powertrain offered better fuel economy. The series powertrain offered an advantage in cycles featuring frequent rapid decelerations and transients. Another finding was that the hybrid powertrain efficiency was very sensitive to the efficiencies of the electric machines and storage devices, thus emerging electric machines with higher efficiency offer promise in increasing fuel economy of hybrid powertrains.

Alley Master’s Thesis on Energy Flow and Losses in Hybrid Powertrain

This master’s thesis by Alley introduces a new tool and method for analyzing losses throughout a hybrid electric powertrain system. The purpose of the tool developed, called VTool, is to facilitate better understanding of a powertrain architecture and provide a means of comparing the potential of two different architectures for fuel economy. Like the method that will be presented in the current work, the VTool method allows the comparison of different architectures without the need to develop a real-time control strategy. Alley uses “Sovran” coefficients to characterize a drive cycle and vehicle glider properties to characterize the drive cycle demands on the powertrain over a cycle. The tool can than calculate how much fuel energy it would take to meet the demands. The effects of the control strategy are summarized into a single parameter called the Power Split Fraction. This parameter indicates how much energy from the engine goes directly to meeting driver demand versus the fraction of energy that gets stored in the battery pack then later discharged. The VTool method provides some advantages and disadvantages compared to the method proposed in the current research. One advantage is that it requires less work and can rapidly be used to study the energy flows and losses in the system. An advantage of the method introduced in the current research is that it provides more insight into what control techniques might be effective when the selected architecture moves into control development.
Another interesting component of VTool is the use of drive cycle average efficiencies. VTool makes use of multiple average component efficiencies in order to calculate energy consumption. Alley uses test data collected from a vehicle for these parameters and tunes them to validate the model against the collected data. The current research uses drive cycle average efficiencies to determine the value of stored battery energy. These parameters are estimated during architecture selection through iteratively running the model. Later they will be used as tunable parameters to tweak behavior of the control strategy.

Pisu and Rizzoni Article on Supervisory Control Strategies

Pisu and Rizzoni authored an article [8] that compares some supervisory control techniques for hybrid electric vehicles. The paper uses a pre-transmission belt coupled parallel hybrid electric vehicle to analyze the impact of different control strategies. The electric machine is an 18 kW continuous, 42 kW peak induction machine. While this powertrain has some similarities to the powertrain that will be examined for the current work, larger electric machines and thus a higher degree of hybridization will give the control strategy in the current work much more impact in the fuel economy. This article still offers some useful insight. The vehicle model used for simulating the control strategies is outlined first. The mathematical models for all the vehicle components are presented. A curve fit model is used for engine fuel consumption. A similar method is used for the electric motor. While initial modeling for the current research will use simple lookup table based models, further development will be done using dynamic simulation models.
The article then goes into four different proposed control strategies. Pisu and Rizzoni state that since the parallel powertrain is a pre-transmission architecture the gear shifting strategy can be separated from the torque split strategy. This may be a reasonable assumption on a vehicle with a low degree of hybridization, but for a vehicle with more powerful electric machines the gear ratio is important to efficiency and is a degree of freedom that should be examined. The first control strategy described is a finite-state machine (FSM) that is rule and event driven and operated the powertrain in discrete states such as acceleration, deceleration, cruise, and recharge. The advantage of this method is simplicity and low computational complexity. The disadvantage however is that this method is not directly based on fuel economy. The strategy is mainly effective for maintaining battery SOC within limits, but does not directly aim to reduce fuel consumption.
The second strategy is called the equivalent consumption minimization strategy (ECMS). This strategy forms a cost function with the goal of minimizing fuel consumption at each point in time. It works by equating energy used from the battery to fuel that was consumed to charge the battery. Energy being stored into the battery can be deducted from fuel being consumed at the present time since that stored energy will offset fuel use. This method is sensitive to how electric energy is related to fuel. An equivalence factor is derived for this purpose that changes based on driving conditions. This method is very similar to the control strategy that will be proposed in this paper. This paper will however use the strategy to explore all modes of operation including transmission gear, rather than having a separate gear shift strategy. The article then goes on to describe an adaptive version of the ECMS strategy.
The last two strategies described are called ^H_∞ control and dynamic programming. The ^H_∞ control method is based on a state space closed loop control model. Dynamic programming analyzes all possible torque split combinations at each point in time to find the optimal trajectory of battery SOC that gives the highest fuel economy. These methods are not practical for in-vehicle use because it requires a high level of offline computational complexity and prior knowledge of the drive cycle to develop the optimal solution. The article summarizes by discussing the pros and cons of the different control methods. ECMS is found to offer the best compromise delivering fuel economy close to the optimal solution.

Schacht Master’s Thesis on Development of EcoCAR Vehicle Controls

The Master’s thesis by Schacht [9] covers development of the Ohio State University vehicle for the EcoCAR competition. The architecture has some similarities with the vehicle used for this research and some key differences. Both have a high power electric machine on the rear axle. On the front both architectures have an engine and electric motor that can be used in series mode or deliver torque to the front axle. The Ohio State vehicle can decouple the front electric motor from the engine and provide electric assist in electric-only mode. The vehicle used for this research cannot disconnect the electric motor from the engine, but has multiple gear ratios for parallel operation, where the Ohio State vehicle only has one gear. The electric motor on the front for the Ohio State vehicle is also much bigger, offering 82 kW peak, versus 27 kW for the vehicle in this research.
A detailed description of the vehicle supervisory control system is included. There are many similarities between the control system of the Ohio State vehicle and the control system designed in this research. The control system architecture is very similar with distributed component controllers and a supervisory controller interpreting driver demand and commanding powertrain components, communicating via a multiple CAN bus structure. The control hardware selected for the supervisory controller was a MicroAutoBox just like the controller in this research. While the CAN bus structure is presented, the process of designing this structure and verification of reliability is not addressed. This is one of the goals of this research.
Schacht then goes into the actual design of the control software that runs on the supervisory controller. At the highest level the control software utilizes a rule based control system to control flow between discreet modes. Within the operation modes an optimization strategy is run to choose an operating point. This is also similar to the strategy employed for this research but there is major difference. While the strategy used by Schacht uses rule based control to choose operating mode, the strategy developed for this research only uses rule based control for choosing high level operating state (e.g. charge depleting vs. charge sustaining). Selection of operating mode and of an operating point within a mode is done using an optimization strategy. Schacht used a rule based system, primarily a vehicle speed criteria, to choose between series and parallel modes. In series mode, a load following strategy is used to choose an engine generator operating point while in parallel mode the ECMS strategy described in the Pisu and Rizzoni article [8] is used to choose torque split. The strategy in this research uses an algorithm similar to ECMS to choose between series and parallel mode and the operating point within the mode. The parallel mode for the vehicle architecture used in this research also has multiple gear ratios to choose from for parallel operation. This is another variable for the strategy to optimize.

1. Introduction 
1.1 Introduction to Hybrid Electric Vehicles
1.2 Motivation for Development of Vehicle Control Systems
1.3 Model-Based Design
1.4 Objectives
2. Literature Review
2.1 Mahapatra et al. Paper on Model-Based Design
2.2 Marco and Cacciatori Paper on Model-Based Design Techniques
2.3 Katrašnik et al. Article on Energy Conversion Efficiency
2.4 Alley Master’s Thesis on Energy Flow and Losses in Hybrid Powertrains
2.5 Pisu and Rizzoni Article on Supervisory Control Strategies
2.6 Schacht Master’s Thesis on Development of EcoCAR Vehicle Controls
2.7 Ramaswamy et al. Case Study in Hardware-In-the-Loop Testing
2.8 Deng et al. Paper on Controller Hardware-In-the-Loop Simulation
2.9 Summary of Literature Review
3. Hybrid Vehicle Architecture Selection 
3.1 EcoCAR Competition and Team Goals
3.2 Primary Design Concepts
3.4 Energy Storage System Sizing
3.4 Architecture Layout Selection
4. Control System Architecture Development .
4.1 Expansion of Base Vehicle Control System
4.2 Controller Interaction
4.3 Vehicle CAN Bus Design
5. Hybrid Vehicle Supervisory Controller development 
5.1 Hybrid Vehicle Supervisory Controller Requirements
5.2 Control Strategy Development
6. Conclusions
References
GET THE COMPLETE PROJECT
Model-Based Design of a Plug-In Hybrid Electric Vehicle Control Strategy