1
Electric Vehicle Simulator to Determine motor and battery specifications J.W.K.K. Jayasundara∗ , Rohan Munasinghe† ∗ Department
of Electronic and Telecommunication University of Moratuwa, Katubedda, Moratuwa Email:
[email protected] † Department of Electronic and Telecommunication,University of Moratuwa, Email:
[email protected]
Abstract— The determination of the motor and battery specification is one of the major challenges in the electric vehicle designs, specially considering the different kinds of requirements and specifications in different designs. This paper discusses about a method of simulating a vehicle in given terrains and driver models for extracting the torque, power and battery usage; and hence get the most optimized set of parameters for the given design requirement. The simulation takes the driver as a fuzzy model and the overall vehicle as a non-linear model for calculating the motor and battery requirements according to the terrain, driver commands and traffic condition.
I. I NTRODUCTION One of the major challenges that electric vehicle designers face is the selection of the set of most optimum motor and battery parameters. With the wide variety of the designs and requirements, it is difficult to obtain a fixed set of equations for choosing the optimum values. For an example, a vehicle intended to operate on rocky roads for cab services should have different performance than a vehicle designed for VIP personals which normally runs on flat and smooth terrains. Electric vehicle simulators are available in the research and the software market [1],[2],[3],[4]. The paper [1] discusses about a total hybrid vehicle simulator package mainly focusing on the cost function based on the fuel economy running on a rural terrain. It also considers the system power management and regeneration, with the output plots on power, torque and speed. But the driver signal is implemented only as a ’prescribed vehicle speed schedule’. Therefore it is a simulation of a vehicle running on a pre defined speed curve with no feedback from the road conditions and the vehicle parameters. The paper [2] discusses the torque, power and energy requirements of the vehicle in different terrains (urban, federal highway etc.) in the time domain which can be very useful in determining the power requirements. But the paper lacks proper driver modeling, which is implemented only as a variable torque command input. The researches with a focus on the driver characteristics and the terrain conditions to simulate the maximum torque, power and energy needed for a given test run are hard to find. These test runs with different types of driver domains and terrains are crucial for designing a vehicle for a specific target group. This paper discusses about a method of generating the torque, power, efficiency and battery consumption according
Fig. 1.
Functional block diagram of the vehicle simulator
to the given terrain and driver model. By this way, the designer can input different types of road conditions that the vehicle is expected to operate, with the driver model of the target group. The driver model depends on the application which describes the driver expectations and behavior according to the terrain, road traffic and the speed / acceleration parameters. II. E LECTRIC V EHICLE S IMULATOR The electric vehicle simulator is designed as an integration among the driver, vehicle and the terrain. These three parameters are simulated in a single control loop to obtain the power, torque and energy requirements as in the Fig. 1. In the system; the driver gets all the inputs from the vehicle and the road conditions, makes the decisions about the breaking, gearing and accelerating, and execute the commands. The other parameters such as overcurrent indicators, over temperature indicators, battery meters etc. also affect the driver behavior. The power controller gets the driver accelerator pedal and breaking commands and decide on the increasing/decreasing the power output or regeneration from the motor. The vehicle gets the power output from the motor and change its output parameters accordingly (speed, acceleration etc..) The entire system is taken into consideration in the simulation for obtaining the desired results. A. System and Driver Modeling The MATLAB GUI used is given in Fig. 2. The GUIDE is used to implement the user interface to enable any unfamiliar
2
user can easily use the simulation. In the GUI, all the necessary parameters of the designed vehicle should be entered by the user and start the simulation. But the driver modeling fuzzy functions and road condition should be changed within the Matlab fuzzy editor. All the units of the control system except the driver block can be implemented in conventional control logic while the driver model is a fuzzy function. Therefore the driver model is implemented as a separate unit for easy altering of the driver in the simulation and the accuracy of the design. The rules of the driver model fuzzy function are given by, Rule1: If the speed high, pedal is low Rule2: If the speed low, pedal is high Rule3: If the slope high, expected speed is low Rule4: If the slope low, expected speed is high Rule5: If the road condition low, expected speed is low Rule6: If the road slope high, expected speed is low Rule7: If the traffic density high, expected speed is low Rule8: If the acceleration is high, decrease pedal Rule9: If the over temp or over current indicator on, pedal is low Figure 4 shows the rules combination (rule viewer) for getting the final pedal output. This implementation was derived according to the general three wheeler taxi driver expectations, which can be altered to obtain the accelerator pedal output for any kind of driver group. According to the fuzzy model of the driver, the actual driver pedal command according to the road angle and vehicle speed can be simulated. Fig. 3 shows the variation of driver accelerator command with vehicle speed and road angle. The fuzzy parameters are changed with the driver model and passed to the system to simulate with the different types of drivers. The system is used to calculate the output and hence the power, torque and the battery usage.
This matrix can be used as the full road data set for a given simulation and used to simulate the drive.
B. Terrain Modeling
C. The Vehicle Parameters
One of the key parameters in motor and battery selection is the terrain. As an example; for terrains with high road slopes, the motor must come with a higher torque and farely high
Fig. 4.
Accelerator command with Speed and Road Angle
power output and lower RPM. But for flat terrains, the motor can be low torque one but with a high RPM. The road is modeled in terms of the slope, rolling loss coefficient, road quality factor (the inequalities and damages on the road) Q, traffic density on the road and the speed limit on the road. These parameters are arranged into two dimensional matrix with the distance from the beguning to the end of the road in fixed gaps and passed to the system for simulation. The generated terrain matrix would be, Slope1 Slope2 ... Slopen RLoss1 RLoss2 ... RLossn Qf act1 Qf act2 ... Qf actn T den1 T den2 ... T denn Slimit1 Slimit2 ... Slimitn
The set of input parameters includes all the necessary factors that the designer can change in the vehicle design. These include the total weight of the vehicle, motor characteristics, battery type and capacity, gear ratios, wheel size, controller efficiency, mechanical transmission efficiency, wheel windage, drag coefficient, rolling loss coefficient etc.. These parameters can be altered according to the vehicle design within some limitations. Even for the same design, the power, torque and energy requirement would depend on the user preferences and the usage of the vehicle. Once all these are taken in to consideration, the designer should choose the motor, battery pack and the controller accordingly. III. I MPLEMENTATION AND R ESULTS
Fig. 3.
Rules set and Overall Fuzzy model
The system is modeled to obtain the torque and power needed to travel at any given road angle and speed. This is plotted with the maximum available power to obtain the limiting values of the road angle and speed for the given vehicle in Fig. 5. In the simulation presented, the peak power of the motor is 16kW and the figure shows clearly the
3
Fig. 2.
MATLAB GUI
maximum road angle and relevant peak speed that the vehicle can move with that power. Maximum RPM allowed is used to obtain the possible gear combinations for the vehicle. This kind of calculation can be used to optimize the gear ratios and the number of gears used in the gear box for the cost and efficiency optimization, as given in the fig. 6. A five gear system is simulated and it can be clearly seen that the last two gears are not required to obtain the 70KmPH peak speed. This is highly useful for deciding on the gearbox, and to eliminate the gearbox if possible. In the vehicle simulation, the true run in a given terrain is simulated and the system parameters such as speed, acceleration, vibrations, efficiency etc. and the output parameters such as power, torque and battery usage are plotted to obtain the system requirements in the real world. The important parameters obtained within the simulation are shown. Figure 7 shows the angle of road in the test simulation. The
simulated terrains is a stiff angle road of 100m high ascending and descending within a range of 1km. This kind of terrain is expected to be the highest possible angles in the general road conditions. Therefore if the vehicle can perform well in this terrain, then it can be considered as having enough power and torque for the general usage. In Fig. 7 through 11, the mark A denotes the start of the ascend, B denotes the peak point and C denotes the end of the descend. The output and input parameter changes on these points are clearly seen in each of the graphs. Figure 8 shows the driver accelerator command. A general driver would be satisfied with a fair speed at the flat run and a slower speed with the climbing angle, which is taken from the fuzzy model of the driver. Therefore the throttle will be full at the start of the run and the climbing angle (mark A), but will come to zero throttle when descending at mark B and continue to be so until the speed is slown down to desired
4
Fig. 5.
Power Requirement for given speed and angle Fig. 7.
Fig. 6.
Gear Requirement and RPM Vs speed
level even in the flat road as show in the last part of the figure after mark C. With a very complex fuzzy model that implements the exact behavior of a driver, this acceleration would be having some fluctuations than shown in the Fig. 8. But these models would highly alter from driver to driver, therefore such a simulation can’t be taken as a generalized model for a target group of drivers. On the other hand, a basic fuzzy model is expected to reflect the collective and average model of a given driver set better. The vehicle behavior (speed and acceleration) can be seen in the Fig. 9. These parameters are calculated from the given accelerator command, current speed and the road conditions. These calculations clearly define the maximum torque and power needed for the run, and hence the requirement of the vehicle power system. Also it can be clearly checked if the speed and the acceleration are within the acceptable limits for the given driver mind set (whether the vehicle designer
Fig. 8.
Road altitude along the test run
The driver Accelerator command (Throttle percentage)
expectations are achieved with the test run). At mark A, the start of the road ascend would slow down the vehicle even with an increased accelerator command. But the driver would be satisfied with a low speed on that road angle, so the speed would be settled down to a low value even with a high throttle. At mark B, there would be a good acceleration because the driver would try to accelerate on top of the hill to a higher speed. Even though there would not be having any accelerator command after the descending, there will be an increase in the speed and the driver would control the speed and acceleration through the breaking. After the point C, there would a little deceleration; but still the driver might be satisfied with that slightly reduced speed at the flat road. Finally, the overall torque requirement of the test run is shown in the Fig. 10. This clearly shows that the peak torque required for this run is around 160Nm. Also the torque reflects the accelerator command shown in Fig. 8 as the accelerator
5
Fig. 10.
Fig. 9.
Torque Requirement In the Total Run
Acceleration and Speed Plots Fig. 11.
input is nearly proportional to the torque requirement of the motor before the gear box. As this torque requirement depends on the road angle, and very highly on the driver mind set; it is recommended that the two input parameters are well defined. For the given simulation, a more economical mind set of a general threewheeler taxi driver was selected. But for the people who require a Sporty type vehicle, the torque requirement will highly increase. Figure 11 shows the total percentage of the battery usage within this simulation. It is highly important to select the battery pack. For example; the designer can set 100km run with a general terrain selected, simulate and check if the battery pack can afford that run. Also the total energy requirement can be obtained, which became around 1.4MJ for the given test run. The test simulation is for a controller with no regeneration; but if the proper regeneration parameters are used (which highly depend on the motor controller), there will be a little increase in the remaining battery capacity at the descending end of the simulation after the point B.
Battery Usage of the Travel
A. Practical Implementation The vehicle simulator was used for simulating the existing three-wheelers commonly found in the local market in an attempt to convert them to electric three wheelers. This has been a crying need of the country for the existing two stroke high emmision three wheeler vehicles are to be removed from the roads within a few years. The existing three wheeler is having a maximum torque of 350Nm at the wheel and a maximum engine power output of 7.5kW. But the simulations suggest the actual specifications for the given three wheelers according to the drive pattern of their main users to be as given in the table I. Figure 12 shows the three wheeler motor at the test run within the laboratory running at 0.5 accelerator input. It also shows the controller, controller power supply, variable resistor used for accelerator command, power controller circuit and the turn on spike of the switching IGBTs.
6
1 2 3 4 5 6 7
Parameter Average Power Output Peak Power Output Rated Torque (before GB) Peak Torque (before GB) Rated Speed Max Allowed Acceleration Battery for 100km run
Pav Ppeak Trated Tpeak ωrated amax Pbat
Value 5kW 10kW 30Nm 70Nm 2000RPM 3 ms−2 120Ah x 4
TABLE I R ESULTS OF THE T HREE W HEELER S IMULATION
Fig. 12.
motor At Test Run
IV. C ONCLUSION AND F URTHER W ORKS Using the generated outputs, it is possible to determine the exact requirements for the given specifications. As one of the most important part of the electric vehicle design is to clearly determine the specifications for the given target group, this is intended to optimize the vehicle and drive system design, and ultimately uplift the quality of the electric vehicles. The practical implementation has proved its performances in the actual electric vehicle design field. Finally this simulator is believed to be further developed and used in the electric vehicle design industry as an easy tool for determining specifications for the motor and the battery. According to the simulation results, a motor was selected out of the commercial market and a controller was designed accordingly. The motor-controller assembly at a laboratory test is shown in the Fig. 12. The selection of the motor, controller and the battery pack were well justified with the simulator even if the overall torque and power output of the converted vehicle are a little lower than the existing values of the three-wheeler. The simulation of the three-wheeler and the practical implementation has well proved that the simulator works according to the requirements and that it is a very useful tool in any vehicle design or electric conversion in selecting the motor, battery and the controller, which would be very painstaking and time consuming otherwise.
ACKNOWLEDGMENT J. W. K. K. Jayasundara Author thanks the extensive contribution that late Dr. D.A.I. Munindradasa had rended for the research.” R EFERENCES [1] Chan-Chiao Lin, Zoran Filipi, Yongsheng Wang, Loucas Louca, Huei Peng, Dennis Assanis, Jeffrey Stein, Integrated, Feed-Forward Hybrid Electric Vehicle Simulation in SIMULINK and its Use for Power Management Studies, Automotive Research Center, University of Michigan, 2001 [2] Butler, K.L.; Ehsani, M.; Kamath, P., A Matlab-based modeling and simulation package for electric andhybrid electric vehicle design, IEEE Transactions on Vehicular Technology, Volume 48, Issue 6, Nov 1999 Page(s):1770 - 1778 [3] Qingyuan Li, Development and Refinement of a Hybrid Electric Vehicle Simulator and its Application in ’Design Space Exploration’, Master Thesis, The Ohio State University, 1998. [4] Sung Chul Oh, Evaluation of motor characteristics for hybrid electric vehicles using the hardware-in-the-loop concept, IEEE Transactions on Vehicular Technology, Volume 54, Issue 3, May 2005 Page(s) 817- 824 [5] J.W.K.K. Jayasundara, Design of multi phase in-wheel axial flux permanent magnet motor for electric vehicles, ICIIS conference 2006 [6] J.W.K.K. Jayasundara, Novel Sinusoidal PWM Controller stratagy for axial flux permanent magnet motors, ICIIS conference 2006
