BATTERY MANAGEMENT SYSTEMS FOR ELECTRIC VEHICLE
APPLICATIONS
AYMAN S. ELWER (1) SAMY M. GHANIA (2) NAGAT M. K. A. GAWAD (3)
(1) Electronics Research Institute – Cairo- Egypt (2) Shoubra Faculty of Engineering – Cairo- Egypt
(3) Shoubra Faculty of Engineering – Cairo- Egypt
Abstract:
Battery Management Systems (BMS) are used in many embedded with distributed BMS in the battery cell to the cell
industrial and commercial systems to make the battery interconnections.
operation more efficient and for the estimation to keep the
battery state, as long as possible, away from destructive state, In practice, the BMS may also be coupled to other vehicle
to increase battery life time. For this purpose, many systems which communicate with the BMS via the CAN bus
monitoring techniques are used to monitor the battery state such as the Thermal Management System or to anti theft
of charge, temperature and current. In the current paper, the devices which disable the battery. Also, there are monitoring
monitoring system for battery powered Electric Vehicles system and programming data logging using an RS232 /
(EV) has been implemented and tested. This system RS485 serial buses.
evaluates and displays the battery temperature,
charging/discharging current and State Of Charge (SOC) for
the considered model battery. For monitoring purpose,
digital and analog sensors with microcontrollers are used.
The battery information and the obtained results explaining
the main characteristics of the system are presented by
photographs and some experimental results are given by the
LCD screen.
Keywords: Battery management systems, Electric
vehicles, Monitoring techniques, microcontroller and State
of charge.
1- INTRODUCTION
A Battery management system (BMS) consists of software
and hardware, designed to increase the discharge cycle of the
battery to maximize the battery lifetime [1]. To explain the
battery management systems (BMS), there are two variables
that should be considered. The first variable is the battery
State Of Charge (SOC) which refers to the amount of charge
presented in a battery in a charge or discharge cycle. The
second variable is the Battery State Of Health (SOH) which
represents the performance of the battery compared to its
past and expected future.
Figure 1: Vehicle Energy management
As shown in Figure 1, the basic BMS consists of three main
building blocks which are, the Battery Monitoring Unit
(BMU), the Battery Control Unit (BCU) and the CAN bus
2-SOC DETERMINATION
vehicle communication network. Also, it is clear that, BMS
building blocks interface with the rest of the vehicle energy A good SOC calculation provides many advantages for EV
management systems. While, the other configurations are such as; longer battery life, better battery performance and
failure warning of the battery pack. The residual battery
capacity can be determined by measurement of the density battery voltage ensures that only the correct type of battery is
of chemical components of the battery; however it is not a used.
practical solution [2]. Accordingly, various methods have
been proposed based on battery voltage and current Reporting refine/calculate to the vehicle applications
measurement [3]. Most of these methods ignore the include the voltage and charging/discharging. Battery
temperature effect in calculation of the battery SOC. remaining life is based on the battery parameters such as
The SOC should be determined accurately, especially for voltage, discharge current and the battery
electrical vehicles applications to predict the travel / charging/discharging characteristic which are stored in the
remaining distance of the vehicle. There are three basic MCU data memory. Figure 2 shows the associated MCU
interrogation methods existing for determining the SOC of a requirements.
battery. The first method is the Coulomb counting, the
second one is the voltage delay and the third one is the Table 1: Summarize the different methods for SOC
impedance methods. Other methods, such as measuring determination.
electrolyte specific gravity are not practical for
implementation in an electric vehicle fuel gauge/battery Application
Technique Advantages Drawbacks
management system. field
Used for capacity Easy and Offline, time
The coulomb counting method is the most widely used Determination at accurate; intensive, modifies
Discharge Test
the beginning of Independent of the battery state, loss
technique for battery fuel gauging in EVs. This method is life SOH of energy
reasonably accurate when SOC estimates are compensated Accurate if
for temperature and discharge rate variations. However, All battery
enough re- Sensitive to parasite
coulomb counting provides no diagnostic capability, which Coulomb calibration points reactions; needs
systems, most
Counting are available and regular re-
can be used to determine the state of health (SOH) of applications
with good current calibrations points
batteries. measurements
Lead,Lithium,Zn/ Online,cheap,OC Needs long rest time
OCV
The voltage delay method is commonly used to perform Br V prediction (current=0)
Online,cheap,EM Needs long rest time
battery tests outside a vehicle. In this case, the battery is EMF Lead, Lithium
F prediction (current=0)
subjected to a transient load discharge and the voltage Lead Needs reference data
response of the battery monitored. The voltage recovery Linear model Online, easy
photovoltaic for fitting parameters
transient is then used to characterize the SOC of the battery. Impedance
Gives Temperature
This technique is again limited in its ability to be All systems information about sensitive, cost
spectroscopy
SOH and quality intensive
implemented as an in-vehicle battery SOC/SOH instrument. Gives
information about Good accuracy, but
DC internal
The impedance method uses a current/voltage excitation resistance
Lead,Ni/Cd SOH, possibility only for a short time
waveform to a battery and the monitoring of the battery’s of online interval
measurements
voltage/current response. In Electrochemical Impedance Needs training data
spectroscopy (EIS), the applied signal is small amplitude, ac Artifical
All battery of a similar battery,
neural Online
waveform, so that the battery system is perturbed about its networks
systems expensive to
equilibrium condition. implement
All battery Ask memory in real-
Fuzzy logic Online
systems word application
The frequency of the excitation waveform may be swept Difficult to
over a wide frequency range and the resultant battery implement the
response can be used to determine an equivalent circuit All battery filtering algorithm
model of the battery and correlated with battery SOC. Table Kalman filters systems,PV,dyna Online, dynamics that considers all
mic application features as, e.g
1 summarizes the different techniques used for SOC nonnormalities and
determination and its advantages and drawbacks. nonlinearities
Due to the modern advances in microcontroller, a battery
having a degree of logic via microcontroller unit (MCU) is The proposed approach that has been developed, is the
called Smart Battery [4]. The Smart Battery has many determination of SOC, its temperature and
advantages such as; acquisition signal recognition, reporting charging/discharging current by directly modeling the
battery voltage and reporting charging/discharging current to impedance response at a few discrete frequencies (using a
refine and calculate data to the main applications. Also, the fuzzy logic methodology), without the intermediate step of
dangerous states such as; too high charge/discharge currents extracting equivalent circuit models for the batteries.
can be reported to predict, automatically, the dangerous Impedance measurements have the distinct advantage of
situations and to make the cell balancing. The task reporting
I n * Ti = const
being rich with information related to the state-of-health of
the battery and can therefore, be useful in battery (1)
management systems for battery diagnostic purposes.
Where:
I: The discharge current (Amp.).
n: The battery constant (n=1.35 for typical lead-acid
batteries).
Ti: The time to discharge at current I (Sec.).
The Peukert relationship can be written to relate the
discharge current at one discharge rate to another
combination of current and discharge rate as:
( n −1)
⎛I ⎞
C1 = C 2 * ⎜⎜ 2 ⎟⎟ (2)
⎝ I1 ⎠
Where :
C1, C2: The discharge rates at different discharge rate states.
Figure 2: The associated MCU requirements. I1, I2: The currents at the two different discharge rate states.
The Battery state of charge (SOC) at a constant discharge
3-BATTERY SOC MODELING rate is given as:
The SOC normally refers to the rated capacity of a new cell. ⎛ I * Time ⎞
It is not the fully charged capacity of the cell when it was SOC = 1 − ⎜ ⎟ (3)
last charged (i.e. the current charge-discharge cycle). This is ⎝ C ⎠
because the cell capacity gradually reduces as the cell ages
and it is also affected by temperature and discharge rate. At For non-constant discharge rates, the current and discharge
the end of the cell's life its actual capacity will be rate (C1 and I1) should be previously known. Given the
approaching only 80% of its rated capacity and in this case, current at the present time step I2, the corresponding
even if the cell were fully charged, its SOC would only be discharge rate is calculated using equation for C1 and
80%. This difference is important if the user is depending on plugged into a incremental form of equation for SOC. The
the SOC estimation as he would in a real gas gauge incremental change in the battery ΔSOC can be given as:
application in a car. These ageing and environmental factors
must therefore be taken into account if an accurate estimate ⎡ ⎤
is required. If the SOC reference was defined as; the current ⎢ ⎥
⎢ Δt / 3600 ⎥ (4)
fully charged capacity of the cell, then the adjustment factors ΔSOC = I 2 ⎢ (n −1) ⎥
would have to be applied to the rated capacity to determine ⎢ C * ⎛⎜ I 2 ⎞⎟ ⎥
the new reference capacity. ⎢ 1 ⎜I ⎟ ⎥
⎣ ⎝ 1⎠ ⎦
In this case, a fully charged cell would have an SOC of
100%, but it would only have a capacity of 80% of a new For the battery operating range used in Hybrid Electric
cell. For cell balancing applications, it is only necessary to Vehicle (HEV), determining the SOC of the battery is the
know the SOC of any cell relative to the other cells in the second major function of the BMS. The SOC is needed not
battery chain. Since all the cells will have been subject to the only for providing the fuel gauge indication, but also to
same influences during their lifetime, the ageing and check for uniform charge in all of the battery cells in order to
environmental adjustments, which apply equally to all cells, verify that individual cells do not become overstressed. The
can be ignored for this purpose. SOC indication is also used to determine the end of the
charging and discharging cycles. Over-charging and over-
Based on the Peukert equation, the discharge current of a discharging are two of the prime causes of battery failure
battery decreases with increasing the discharge time [5]. and the BMS must maintain the cells within the desired
operating limits. Hybrid vehicle batteries require both high
power charge capabilities for regenerative braking and high 4-EXPERIMENT SYSTEM
power discharge capabilities for launch assist or boost. For
this reason, their batteries must be maintained at a SOC that The battery pack used in this study is considered the main
can discharge the required power but still have enough power source for a prototype model, which is based on re-
headroom to accept the necessary regenerative power engineering of ICE SUZUKI car. The battery pack consists
without risking overcharging the cells. of 6 modules of 12 V traction lead acid batteries connecting
in series to supply the main traction DC motors of 75 V and
To fully charge the HEV battery for cell balancing, as shown 9 hp. Figure 5 shows the combination of the battery pack for
in Figure 3, would diminish charge acceptance capability for the prototype model of EV. Firstly, to perform the SOC
regenerative braking and hence braking efficiency. The experimentation, two batteries of 12 V are used. Secondly,
lower limit is set to optimize fuel economy and also to the developed system is applied for whole battery pack in the
prevent over discharge which could shorten the life of the EV prototype.
battery. Accurate SOC information is therefore needed for
HEV to keep the battery operating within the required safe
limits.
Figure 5: Battery pack above the experimental prototype
model.
4.1 DESIGNED SYSTEM AND RESULTS
For monitoring the battery SOC, a hardware system is
designed using implementation of a microcontroller built
with a flash memory. Figure 6 shows the processor unit
utilized for SOC estimation. The used CPU is a microchip
Figure 3: HEV battery operating ranges Atmel 89C51. The battery charging/discharging current is
determined by using a LA255-S Hall Effect current sensor
A battery model applying the above equations and with manufactured by LEM Corporation as shown in Figure 7.
using Matlab software is achieved to estimate the Battery The battery terminal voltage is simply measured by a
SOC, and the result of SOC during both charging and resistive voltage divider. According to the manufacture
discharging is as shown in Figure 4. recommendation, a K-type thermocouple along with AD595
thermocouple signal conditioner is employed.
Figure 6: Picture of the SOC indicator and its wiring
diagram.
Figure 7: The used current sensor
A microcontroller, in which the SOC algorithm is
stored, determines the SOC of the battery system
Figure (4): SOC for typical lead acid battery
based on the measured signals by the current
sensor. Two types of memories are needed. First, reporting data at each half hour. A captured photo results of
the read-only memory (ROM) from which the SOC and temperature are shown in Figure 10 and these
results are not drawn later for the purpose of comparison
amount of self discharge as a function of T and the symmetrically between the no-loading and loading
discharging efficiency as a function of I and T are conditions.
read and when the SOC algorithm is based on
EMF measurements, the EMF–SOC relationship
can be stored in ROM together with other battery
specific data. Second, the random access memory
(RAM) is used to store the history of use, such as
the number of charge/discharge cycles, which can
be used to update the maximum battery capacity.
The simple flowchart of the developed algorithm
is shown in Figure 8.
Start Figure 9: Practical Setup for measuring Temp. And SOC
Read current
and Temp.
Estimate SOC using eq.(1) to eq.(4) (a) (b)
Figure 10: Captures of SOC and temperature reading by
LCD at no-loading conditions.
yes yes Figure 11 presents the variation of the SOC at different times
T < max Max>SOC>min for a period of three hours. It can be noticed that the SOC
varies from about 100% to about 70 % for a period of three
hours. Figure 12 presents the variation of the battery
No temperature at different times for a period of three hours. It
can be noticed that the battery temperature increased by
Give an alarm and disconnect DC link about 20 % from its initial temperature during the
discharging cycle.
Figure 8: The flow chart for the developed Algorithm.
120 SOC (%)
Each part of this system (software algorithm or hardware 100
device) will influence the final accuracy in the SOC SOC (%)
indication, as if there is inaccuracy in the V, T and I 80
measurements this will give an inaccuracy in the final SOC.
SOC %
60
An important role must be recorded also to the calibration of
the SOC as if the SOC algorithm is based on current 40
measurement and integration the obtained error caused by
the current measurement inaccuracy will accumulate over 20
time.
0
0 0.5H 1H 1.5 H 2H 2.5H 3H
The system is put together, as shown in Figure 9, with Discharging Time
loading the motor by light load that draws a constant current Figure 11: Variation of the battery SOC during the
of about 5 Ampere (by lifting the rear axel of the car from discharging time
the ground)) and using one battery pack to test our model.
And the data of the drawn current from the battery and its
temperature are stored for a period of three hours at
40 Temperature ( C ) 5. CONCLUSIONS
35
30 A simulated model for the applied battery using Matlab
Temperature ( C )
25
Software, is achieved to determine the SOC which is the
major function of Battery Management System and the
20
Temperature ( C ) model gives a good result during both charging and
15 discharging cycles. Also, a very simple and cheap
10 experimental circuit applying MCU is designed to estimate
5 the SOC and measure temperature of the battery the
designed circuit gives a good estimation of SOC and an
0
0 0.5H 1H 1.5 H 2H 2.5H 3H alarm is obtained when SOC exceeds the minimum or the
Discharging time maximum limits, also, an alarm is obtained if the
Figure 12: Variation of the battery temperature during the temperature exceeds the maximum set point, and these
discharging time alarms are protecting the battery to increase its life time.
The rear axel wheels are braked to make the motor draw its
rated current and with connecting the whole 6 battery packs 6. REFERENCES
and the SOC and temperature are reported for a period of
3 hours. Figure 13 presents the variation of the SOC at [1] John Chatzakis, Kostas Kalaitzakis, Nicholas C.
different times for a period of three hours. It can be noticed Voulgaris and Stefanos N. Manias, "Designing a New
that the SOC varies from about 100% to about 40% for Generalized Battery Management System", IEEE Trans.
a period of three hours. On I.E., VOL. 50, NO. 5, Oct. 2003.
120 % SOC with rated load [2] R. Weiss et.al., "Battery SOC determination in
photovoltaic system", J. Electrochemical society:
100
electrochemical science & technology, Vol-129 No. 9,
80 September 1982, pp. 1928-1933.
SOC %
SOC (%)
60
[3] Caumont O., Le Moigne P., Muneret, X. Lenain P.,
40
Rombaut C., "An optimized State of Charge Algorithm for
20 Lead Acid battery in Electric Vehicle", pp. 93-96, EVS
0 1999.
0 0.5H 1H 1.5 H 2H 2.5H 3H
Time [4] Z. M. Salameh M. A. Casacca, W. A. Lynch, "A
Figure 13: Variation of the battery SOC during the Mathematical Model for Lead Acid Battery", IEEE Trans.
On Energy Conversion Vol-7, pp.93-96, March, 1992.
discharging time with rated current.
[5] Kiessling, R., “A battery model for monitoring of and
Figure 14 presents the variation of the battery temperature at
corrective action on lead-acid EV batteries”, Battery
different times for a period of three hours. It can be noticed
Conference on Applications and Advances, Proceedings of
that the battery temperature increased by about 110 % from
the Ninth Annual Meeting, Long Beach, CA, USA
its initial temperature during the discharging cycle.
January 11-13, 1994.
Temperature ( C ) with rated load
70
60
Temperature ( C )
50
40 Temperature ( C )
30
20
10
0
0 0.5H 1H 1.5 H 2H 2.5H 3H
Time
Figure 14: Variation of the battery Temperature during the
discharging time with rated current
