Energy Management Strategies for Modern Electric Vehicles Using MATLAB/Simulink

Jinesh Pradip Shah, Prashant Kumar Soori, and Sibi Chacko

Energy Management Strategies for Modern Electric Vehicles Using MATLAB/Simulink

Jinesh Pradip Shah, Prashant Kumar Soori, and Sibi Chacko

—The paper describes the various energy management techniques that can be implemented for a modern electric vehicle by using MATLAB/Simulink. The Renault Twizy vehicle is considered for MATLAB simulation. Regenerative braking technique is discussed, in which the kinetic energy is converted to electricity to charge the battery of the vehicle when the brakes are applied or when the vehicle is moving down the hill. A solar photovoltaic (PV) on the roof-top of the vehicle is implemented to charge the battery used in the vehicle. The simulation results are highlighted and energy management strategies are presented. The results showed that the speed control of direct current (DC) motor during the motoring mode and regenerative braking mode was successfully achieved by using a bi-directional DC-DC converter and a proportional-integral (PI) controller at various reference speeds set by the user by applying a variable load torques to the motor. The size of solar PV on roof-top of the vehicle was found to be 280 W that charged the 48 V battery of the vehicle by using a bi-directional DC-DC converter, which was evaluated by using MATLAB/Simulink.

Index Terms—Battery vehicle, DC-DC converter, DC motor, MATLAB.

1. Introduction

The major issue relating the depletion of fossil fuels, the increase in the price of oil and gas, greenhouse gas (GHG) emissions, and high environmental pollution are creating a major concern for the society and the world as well. The automotive industry is greatly attached to all the above issues. The major concern with internal combustion engine (ICE) vehicles is the continuous reduction in fossil fuels which directly leads to the increase in the prices of petrol and gas. The other issues with ICE based vehicles are its environmental concerns of producing high percentage of greenhouse gas emissions and thus polluting the environment. The development of electric vehicles in the automotive industry can reduce these major issues to a great extent. In the recent era, most of the automotive industries have taken steps to develop green vehicles, which are clean, safe, highly efficient, and also eco-friendly. With the development of electric vehicles, battery electric vehicles (BEV), hybrid electric vehicles (HEV), and fuel cell based electric vehicles (FCEV) have the potential to replace the present conventional based vehicles (ICE based vehicles)[1]. The current concern for the electric vehicles, which limits itself from entering the market, is due to the battery technology. The battery technology is found to be the weakest and this poor storage capability limits the battery electric vehicle range (BEV) to only specific applications, like airport stations and for small drive range applications[1]. Some major challenges for BEV are the poor mileage, sizing and capacity constraints, dependency on power, and longer charging hours. In addition, the improvements in battery technology and public recharging installations are required[2]. This paper describes the various charging techniques that can be implemented in an electric car in order to overcome the above constraints.

2. Methodology

The block schematics of the proposed system are shown in Fig. 1.

Fig. 1. Design of solar hydrogen-based fuel-cell battery-electric vehicle.

The design consists of a solar photovoltaic (PV) on the roof-top of the vehicle which is connected to a charge controller. This charge controller has a control strategy of producing hydrogen via electrolyzer and storing in the hydrogen storage tank. Unless it detects that the state ofcharge (SOC) of the battery is full, it directly starts to charge the battery until its SOC is reached to the maximum limit. The lithium ion battery and a proton exchange membrane (PEM) based electrolyzer is considered. The primary power source used here to propel the wheels of the vehicle is the battery. The design also has an additional feature of regenerative braking to charge the battery effectively during braking and down-hill motion of the vehicle. The vehicle is powered by a separately excited DC motor. Fuel cell is designed to power the auxiliary power unit. In this case a small application is considered, such as power windows which use the permanent magnet DC motor (PMDC) for its operation.

3. Vehicle Specifications

The vehicle specifications are tabulated in Table 1. Fig. 2 shows the Renault Twizy vehicle considered for simulation.

Fig. 2. Renault Twizy[3].

Table 1: Renault Twizy specifications[4]

The vehicle also has an extendable cable for battery charging. It is capable of charging by plugging in the voltage of 220 V at 10 A (domestic electrical supply). The battery can be fully charged in three to four hours of continuous charging. The maximum torque during the vehicle’s initial start provided by the motor is around 57 Nm, almost four times greater than a three wheeled scooter of 125 cm3[4].

4. Case 1: Closed Loop Speed Control of DC Motor Using PI Controller

The closed loop speed control of the DC motor used in this paper is carried out by using a proportional-integral (PI) controller. Fig. 3 shows the control strategy used for the closed loop speed control of DC motor by using a half bridge non-isolated bi-directional DC-DC converter and a PI controller[5].

Fig. 3. Closed-loop speed control of DC motor using PI controller[5].

In order to control the speed of the DC motor, the output voltage of the bi-directional DC-DC converter must be controlled. In order to control the output voltage of the bi-directional DC-DC converter and to bring the speed of the vehicle running at a different speed to the desired speed, a PI controller is used to provide a quick response to sudden and quick speed changes during the driving cycles. In this control strategy, the motor speed or the actual speed ωmotoris sensed and compared with a reference speed or the desired speed ωrefas shown in Fig. 4. The error signal (ωref-ωmotor) is calculated and fed into the PI controller, which minimizes the error by using the proportional and integral gain values and sends the signal to the pulse width modulation (PWM) generator. The signal which is fed into the PWM generator is then compared with the high frequency saw-tooth wave equivalent to the switching frequency of the converter in order to generate pulse width modulated (PWM) control signals for the switches used in the bi-directional DC-DC converter and make the vehicle to run at the desired speed[5].

Fig. 4. Generation of gate pulses using PI controller[5].

5. MATLAB/Simulink Model for Motoring and Regenerative Braking Mode

In this case, both motoring and regenerative braking take place in the single simulation which runs for a time period of two seconds. The speed control of the DC motor for motoring and regenerative braking is done at various reference speeds (desired speed) by applying various load torques to the motor. The concept of complementary gate switching during motoring and regenerative braking modes is used. The bi-directional DC-DC converter circuit is shown in Fig. 5.

Fig. 5. Bidirectional DC-DC converter[5].

5.1 Parameters

The below parameters are considered in MATLAB/Simulink model for running the vehicle in motoring mode:

· Motor: 5 HP, 240 V, and 1750 rpm (183 rad/s).

· Bidirectional DC-DC converter specifications:

LC=1600 μH,

CH=470 μF, CL=470 μF

· Battery:

Voltage=48 V, battery capacity=140 Ah

·

PI controller:

Kp=0.001, KI=0.02

5.2 Motoring Mode

The Simulink model of motoring mode is shown in Fig. 6. During the motoring mode, the desired speed of 70 rad/s isachieved,when a positive load torque of 10 Nm is applied to the motor. It can be observed that the armature current is positive and proportional to the electrical load torque.

5.3 Regenerative Braking Mode

Fig. 7 shows that during the regenerative braking mode, the desired speed of 70 rad/s is achieved, when a negative load torque of -10 Nm is applied to the motor after a period of one second during the simulation. It can be observed that the armature current is negative and proportional to the electrical load torque.

Fig. 6. MATLAB/Simulink model during motoring mode.

Fig. 7. MATLAB/Simulink model during regenerative braking mode.

6. Case 2: Design of Solar PV on the Roof-Top of the Vehicle

The design of solar photovoltaic (PV) for the vehicle is calculated by using the length, width, and height of the car Renault Twizy. From the dimensions of the vehicle, it is possible to calculate the area that can be covered by solar PV on the roof-top of the vehicle. The total roof-top area of Renault Twizy that can be covered is 2.76 m2. However, as per the vehicle dynamics, the total area cannot be utilized and hence an approximate area of 1.96 m2is considered for installing the solar PV on the roof-top of the car. The module selected here is a mono-crystalline silicon cell based solar PV. The module efficiency ranges between 13% and 19%[6]. The rating of solar PV to be installed on the roof-top of the vehicle is calculated by using this efficiency and area. Assuming in mono-crystalline solar PV system, the area/Kilowatt (kW) is considered to be 7 m2. Sizing of solar PV is given as

Hence, a 280 Wp solar PV module is considered on the roof-top of Renault Twizy for charging the 48 V, 7 kWh lithium-ion battery of the vehicle.

6.1 Boost Converter Design Parameters

In the boost converter, the output voltage is

The input voltage Vinis taken as 30.6 V (the maximum power point voltage of 280 Wp solar PV module exposed to 1000 W/m2). The output voltage V0is taken as 54 V which is the maximum charging voltage limit of the battery. So, a duty cycle (D) of 0.433 has been obtained by using (2). The switching frequency fs=20 kHz[7].

6.2 Inductor Design

The inductor for the boost converter is designed by where ΔIL is the inductor ripple current, IOUT(max)is the maximum output current necessary in the application[8]. A factor of 0.3 is chosen to calculate the value of ΔIL andis taken as 70 A since the maximum charging limit of the battery is 70 A. The output voltage V is

outconsidered to be 48 V, since the battery used in the vehicle is 48 V. By substituting these values in (3), the inductance (L) is found to be 1.68×10-5H.

6.3 Capacitor Design

The capacitance value is calculated using (5):

whereoutVΔ is typically 0 to 10% of output voltage.

In this design, ΔVoutis assumed to be 5% of the output voltage and its value is found to be 2.4 V. The capacitance is found to be C=Cout=6.31×10-4F .

The above inductor, capacitor, and duty cycle values obtained are used for the boost converter (charge controller) for safe battery charging by using solar PV[9].

7. MATLAB/Simulink Model of Solar PV—Battery Charging Using Boost Converter

Fig. 8 describes the Simulink model of solar PV to charge the battery by using the boost converter at a constant voltage of 48 V and also within the battery charging current of 70 A for the safe battery charging when a solar irradiance of 1000 W/m2is applied on the 60 cells module rated 280 Wp.

Fig. 8. MATLAB/Simulink model of solar PV to charge the battery by using boost converter.

Fig. 9. MATLAB/Simulink model of fuel cell powered PMDC motors used in windows.

A controlled voltage source is used in the above Simulink model in order to feed the output voltage from solar PV to the boost converter circuit. The parameters used for the boost converter are the inductance L=1.68×10-5H , capacitance C=6.31×10-4F, resistor R=200 ?, duty cycle D=0.43. The model is simulated for a simulation time of one second and the results are obtained.

8. Proton Exchange Membrane (PEM) Based Fuel Cell for Powering Auxiliary Loads

The fuel cell considered here is of smaller rating compared with the one usually used in the fuel cell based hybrid electric vehicle. The main use of the fuel cell designed for this vehicle is to provide power to the auxiliary applications used in a vehicle. The overall power consumption of auxiliary loads is around 4.3 kW. The MATLAB/Simulink model of the fuel cell powered permanent magnet direct current (PMDC) motors for the power windows application is shown in Fig. 9.

The above Simulink model presented in Fig. 9 shows how the fuel cell stacks is connected to the four PMDC motors of the power windows. It is clear that the fuel flow rate of the fuel cell stack is controlled by using a fuel flow regulator. The above block also shows that the four PMDC motors of the power windows are run by applying a load torque of 2 Nm at a 12 V DC system voltage.

9. Results and Discussions

9.1 Motoring Mode

In the motoring mode, the duty cycle of the switches for a bi-directional DC-DC converter (as shown in Fig. 5) fed separately excited DC motor is calculated as

where DQ1and DQ2are the duty cycles of switches Q1 and Q2, respectively; Vbattis the battery voltage.

In Case 1, a positive load torque of 10 Nm is applied to the motor till a time period of one second and then the negative load torque is applied for another one second. The armature voltage V2is around 98 V which can be observed in Fig. 10. The battery used here is 48 V, 140 Ah. The duty cycle of gates Q1 and Q2 are calculated by using (6) and (7), respectively. Its value is found to be DQ2=0.49 and DQ1=0.51.

It is seen that the complementary gate switching takes place as in the motoring mode DQ1＞0.4 and DQ2＜0.6. Hence the bi-directional DC-DC converter is acting in a boost mode, which propels the DC motor of the vehicle to run at the desired speed of 70 rad/s.

From Fig. 10, it is clear that an armature voltage around 98 V is achieved for the motor running at a speed of 70 rad/s with a load torque of 10 Nm. The inductor current is around 21 A which is positive in the case of the motoring mode.

Fig. 11 shows the battery state of charge (SOC), battery current, and battery voltage. The state of charge of the battery is seen to decrease, as in the motoring mode the battery starts discharging. It is clear from Fig. 11 that the battery current is positive, since the current is drawn from the battery to power the motor. The discharging current is in the range of 20 A to 30 A for running the vehicle at 70 rad/s. The battery voltage is around 48 V as the maximum battery voltage limit is 54 V.

Fi g. 10. Armature voltage and inductor current .

Fii g. 11. Battery SOC, voltage, and current.

9.2 Regenerative Mode

In the regenerative mode, the duty cycle of the switches for a bi-directional DC-DC converter fed separately excited DC motor is obtained by using (8) and (9), respectively.

Fi g. 12. Armature voltage and inductor current .

In the second case, a negatiive load torquue of -10 Nm is applied to the motor after a time period of one second. The battery rating is 48 V, 140 Ah. The armature voltage V2is around 50.9 V which can be observed in Fig. 12. Duty cycle of Q2 (DQ2) is found to be 0.94 and that of Q1 (DQ1)is 0.06.

From Fig. 12 it is clear that the armaturre voltage of 50 V is achieved for the motor ruunning at a speed of 70 rad/s with a load torque of -10 Nm. Thus the voltage from 98V comes down to 50V (nearing the batery voltage) to charge the batery during this mode.The inductor cuurent is around -10A which isnegative inthe case of the regenerative braking mode. Thus the motor voltage is steped down by thebi-directional DC-DC converter to thebatery voltage and the curent flow is reversed during this mode.

Fig. 13 represents the batery state of charge (SOC), batery curent, and batery voltage. The state of chargeof thebatery is sen to be increasing, as inthe regenerative mode the batery starts charging. It is clear from Fig.13 thathe batery curent is negative, since the curent flows in the reverse direction. The charging curent is in the range of 20 A to 30 A for runing the vehicle at 70 rad/s, whichis wel within the charging limitof the batery. The batery voltage is found to be around48 V as the maximum batery voltage limit is 54 V.

Fig.13. Battery SOC, voltage, and current.

Fig.14. Maximum power point current, voltag2e, and power ffor 280 Wp module at solar irradiance of 1000 W/m.

9.3 Solar PVto Battery Chharging

Fig. 14 shows that the solar PV output for a 280 Wp solar PV exposed to a solar irradiance of 1000 W/m2 produces the maximum power Pmax of 268 W with a maximum power point voltage Vmpp of 30.6 Vand maximum powerr point curent Impp of 8.7A.

The maximum power point voltage Vmpp of 30.6V is fed to the boost converter too step up thevoltage from30.6 Vto a constannt voltage of48 V at a chharging current of --60 A, which is within the charging limmit of the batery. From Fig. 15,it is clear that the batery state of charge (SOC) increases as the batery starts charging using the soolar PV.

Fii g. 15. Battery SOC, voltage, and current.

9.4 PEM Fueel Cells to Power Windows

Fig. 16 shows the fuel ceell voltage and fuel cell current, which is around 12 V and 8.3 A. So it produces around 100 W of power at standard operating conditions and powers the 4 PMDC motors of the power window used in the electric vehicle.

Fi g. 16. Fuel cell voltage and current.

10. Conclusions

This paper presents various energy managemeent techhniques that can be implemented for thelectric vehicle apllication. Theperformancee of the system has been verified by simmulating with the MATLAB/Simulink environment. It isfound that the use of regenerative braking technique leads to energy saving in vehicles and the idea of using solar PV on roof-top of the vehicleleads to further energy saving making the pure electricvehicle more energy-eficient and improving the range of the vehicle.

[1]M. Ehansi, Y. Gao, S. E.Gay, and A.Emadi, Modern Electric, Hybrid Electric and Fuel CelVehicles, 1st ed. Florida: CRCpress, 2005.

[2]Battery, hybrid and fuel celelectric vehicles are the keysto a sustainable mobility. AVERE European Association for Battery, Hybrid and Fuel Cell Electric Vehicles, Brussels. [Online]. Available: http://www.avere.org/www/Images/ files/about_ev/Brochure.pdf

[3]OVMS: Renault Twizy UserGuide, Renault, 2014.

[4]The 2010 Paris Motor Show, Renault, 2010.

[5]P. Pany, R. Singh, and R. Tripathi, “Bidirectional DC-DC converter fed drive for electric vehicle system,” Intl. J.of Enginering,Science andTechnology, vol. 3, no. 3, pp. 101-110, 2011.

[6]IEA-ETSAPand IRENA, U.A.E.(2003). Solar Photovoltaics TechnologyBrief. [Onlline]. Available: https://www.irena.org/IRENA-ETSAP%20Tech%20Brief% 20E10%20

[7]N. Mohan, T. M. Undeland, and W. P.Robins, Power Electronics: Converters, Aplications and Design, 2nd eed. Vancouver: John Wiley, 1999.

[8]Basic Calculation of a Bost Converter’s Power Stage, Texas Instruments, Dalas, 2014.

[9]A. P. Vyshakh and M. R. Unii, “BLIL PFC bost converter for plug in hybrid electric vehicle batery charger,” Int. J.of Scientific Eniginering and Research, voll. 2, no. 1, pp. 22-26, 2014.

Jinesh Pradip Shahwas born in Coimbatore, India in 1991. He received the B.Tech. degre in electrical& electronics engineringfrom Amrita VishwaVidyapetham University, Coimbatore in 2013 and the M.S. degre in ren ewab le energy engine ring from HeriotWatUniversity((DubaiCampus),

Dubal in 2014.Now,he works with Heriot Watt Unlvsity Dubal Campus,Dubai.His research intersts include wind energy,solar energy,electric vehicles,hybrid energy based vehicles,and fuelcell techology.

Prashant Kumar Soori’s and Sibi Chacko’sphotographs and biographies are notavailable at the time of publication.

Appendix

Supporting Information Document

Abbreviation List

AMOLED: Active-matrix organic light-emitting diode

CO2e: Carbon dioxide equivalent

CPU: Central processing unit

GHG: Greenhouse gas

ICT: Information and communications technology

OS: Operating system

RSE: Relative standard error

1. Proposed Methodology Scope

Table S1: Summary of the devices tested

Table S2: Summary of the activities tested

2. Device Preparation

To begin, each device’s hard disk was formatted and the OS reinstalled in the default manner. Each OS was updated using the OS update software so that it was running the most up to date version available. In the case of the laptop, the manufacturer’s hardware specific drivers were also installed. The phone had specific drivers bundled with the OS installation package. Where possible, both the OS and driver versions were captured in case of future re-testing (Table S3 and Table S4). Each device was then configured to not enter a standby mode and was left on for at least 48 hours so that any OS initial installation tasks could occur. For example, after a new installation, Microsoft Windows 7 and Windows 8 require up to 3 days to let idle tasks run in the background[1](It was noted that this time could be lowered by calling the ‘ProcessIdleTask’ API from Advapi32.dll to force idle tasks to run). The power management settings of each device were also configured to not enter any variable mode. For the laptop, power management settings were set to the maximum performance pre-set. The brightness level of each display was set to the brightest level possible and the audio device volume set to 50%. In the case of the laptop, the adaptive light sensor and related service were disabled on all OS versions. Any extra software that could interfere with tests such as external virus checkers and device specific add-on software were removed.

The software required for each test was installed where relevant and run at least three times to allow for any initial setup tasks. The versions of the software were also noted, in case a retest was required (Table S3 and Table S4). Eachtest was performed at least twice before final assessment to let any OS specific performance adjustment processes to take place. For example, Microsoft Windows 7 and 8 adjusts performance over time based on observed usage patterns. Additionally, the process allowed any testing difficulties to be identified and rectified before final assessment.

3. Device Preparation

3.1 OS and Software Energy Measurement and Apportionment

The energy consumption of each device and OS combination was measured, and the software activity energy apportioned. The energy consumption (Wh), environmental impact (grams of CO2e), and energy percentage increase over idle of each test are presented in Tables S5, S6, S7, S8, S9, and S10.

3.2 Measurement Variation

The relative standard error (RSE) of each tests energy consumption on each device and OS combination was calculated (Table S11). This was used to determine how variable each of the three repeat tests were, indicating how reliable the results would be. The largest error calculated was for idle testing on the phone (Windows 7.5) at 5.34%, and the smallest of 0.00% for test 7 (Game) on the Laptop (Windows 7 OS). Across all tests the average difference in measurement variability was 0.96%; however, when segmented, the average increased for the phone to 1.57%, and dropped to 0.65% for the laptop. Likewise, when segmented across the different connectivity options, the average dropped slightly for no connectivity (0.87%) and wireless (0.79%); however, the variability increased to 2.58% for mobile connectivity. The phone (windows 7.8), across all tests, had the highest average variability of 1.58% and the laptop (Windows Vista) had the lowest variability of 0.20%. By individual test average and across all devices, the largest variability was calculated for test 4 (streaming video) using a mobile data connection at 3.19%, and the lowest for test 6 (email) using wireless data connection at 0.40%.

4. Discussion

4.1 Result Verification

Measurement results were difficult to verify as there was little data published by the device manufacturers to compare results. Some device manufacturers participate in the Energy Star scheme[2], whereby typical energy consumption of a device is estimated from unpublished measurements; however, none of the tested devices were listed. Attempting to verify the results also highlighted the lack of detailed consumer information available. An approximate method to verify results was created by calculating the average power consumption of the device using the advertised battery life of each device, and comparing it to the measured average idle power consumption of each device at different brightness levels. Results of this method were unclear because of the lack of data, but illustrated that power consumption measurements were within valid ranges. The results of the method were attained by dividing the battery life (hours) by the batteries capacity (Wh), whereby the approximate average power consumption (W) of each device could be calculated (Table S14). This method assumed that each device manufacturer calculated battery life using the same process, which was not known. Nonetheless, the results provided an approximate power consumption for each device which was comparable to the measured idle power consumption of each device (Table S15).

4.2 Result Discussion

In order to inform these determinations, firstly, the graphics, chipset, wireless connection, HD audio and management interface drivers on the 32 bit version of Windows XP were updated to the latest Intel versions using the Intel driver website. The idle power consumption was measured (using the same techniques as in the method) and an average rise of 11% was observed over both connectivity states. Secondly, the 64 bit version of Windows XP was installed on the laptop using a mixture of standard OS drivers and some manufacturer drivers intended for different OS, as it was not supported. The idle power consumption was again measured in both disabled and enabled wireless connection states, and was on average 75% higher than the 32 bit version. It was also, on average, 31% higher than Windows Vista 64 bit’s idle power consumption, its successor OS. It was noted that the 64 bit version of XP does not have the same core instructions as the 32bit version, and thus is difficult to perform a true comparison. Furthermore, to gauge whether 64 bit computing in general consumed more power, the 32 bit version of Windows 7 was installed on the laptop and tested. Across all tests the power consumption of the 32 bit version was between 4% and 25% higher than the 64 bit version. This test highlighted that 64 bit computing is not necessarily a higher consumer of power and that the power consumption of the OS is highly variable by OS type.

Table S3: OS and activity software versions for Nokia phone

Table S4: OS and activity software versions for the laptop

Table S5: Energy consumption and CO2e emissions for the Laptop using Windows 8

Table S6: Energy consumption and CO2e emissions for the laptop using Windows 7

Table S7: Energy consumption and CO2e emissions for the Laptop using Windows Vista

Table S8: Energy consumption and CO2e emissions for the laptop using Windows XP (note that Windows XP could not enter different brightness modes)

Table S9: Energy consumption and CO2e emissions for phone using Windows 7.5.

Table S10: Energy consumption and CO2e emissions for the phone using Windows 7.8.

Table S11: Relative standard error of measurement for each test and device (higher is worse)

Table S12: ‘User’ use model for laptop

Table S13: ‘User’ use model for the phone

Table S14: Data and results of calculating the average power consumption of each device using the battery life and capacity statistics

Table S15: Measured idle power consumption and inferred battery life of each device. *Wireless ON

Manuscript received May 24, 2015; revised July 24, 2015.

J. P. Shah and S. Chacko are with the Department of Electrical Engineering, School of Engineering & Physical Sciences, Heriot Watt University Dubai Campus, Dubai, U.A.E. (e-mail: jineshshaheee@ gmail.com; c.sibi@hw.ac.uk).

P. K. Soori is with the Department of Electrical Engineering, School of Engineering & Physical Sciences, Heriot Watt University Dubai Campus, Dubai, U.A.E. (Corresponding author e-mail: p.k.soori@hw.ac.uk).

Color versions of one or more of the figures in this paper are available onlineat http://www.journal.uestc.edu.cn.

Digital Object Identifier: 10.11989/JEST.1674-862X.505241