2016 IEEE 2nd International Forum on Research and Technologies for Society and Industry Leveraging a better tomorrow

(RTSI)

Energy-Efficient Battery Charging in

Electric Vehicles with Solar Panels
Michelangelo Grosso 1 , Davide Lena 1, Alberto Bocca 2, Alberto Macii 2 , Salvatore Rinaudo 3
1
ST-Polito S.c.ar.l., 2 Politecnico di Torino, 3 STMicroelectronics
1,2
Corso Duca degli Abruzzi 24, 10129 Torino, Italy
3
Str. Primosole 50, 95121, Catania, Italy
{michelangelo.grosso, davide.lena}@st-polito.com, {alberto.bocca, alberto.macii}@polito.it, salvatore.rinaudo@st.com

Abstract—The problem of pollution due to transportation is connection of multiple cells, in series and/or in parallel, so as
extremely serious. Hybrid and fully electric vehicles are increas- to obtain the voltage and the capacity required by a specific
ingly popular, to reduce fuel consumption and gas emissions, and application. The management of a battery pack is performed
car makers are also starting to equip vehicles with solar panels
to extend driving range. These photovoltaic installations pose through a Battery Management System (BMS), which can be
many challenges, e.g., how to increase their efficiency and decide implemented in many ways [4]. This system is mainly devoted
where and how the solar energy is stored. This paper analyzes to (i) constantly monitoring the state-of-charge (SOC) of the
the problem of transferring the energy generated by an on-board various battery cells, (ii) equalizing the SOC of all the cells,
solar cell system to the main battery pack of an electric vehicle and (iii) regulating the charging current.
and proposes an optimized solution based on the interaction with
the BMS. This work analyzes the problem of transferring the energy
Index Terms— battery charging; cells balancing; electric ve- generated by an on-board solar cell system to the main
hicle; photovoltaic panels. battery pack of an electric vehicle. Among different possible
architectures for solar energy transfer, this paper proposes
I. I NTRODUCTION and describes in additional detail a solution based on the
In modern cities, the problem of pollution due to transporta- interaction with the BMS. For this solution, we propose
tion is becoming more and more important. From this point a dedicated management algorithm and we demonstrate its
of view, solutions are always searched for reducing fuel con- effectiveness by means of simulation in a set of realistic
sumption and costs by exploiting a sustainable and renewable scenarios.
energy. For instance, many vehicle manufacturers have started
II. BACKGROUND AND RELATED WORK
to equip their products with solar cell systems, although used
mainly for powering fans and auxiliary circuitry. Fully solar A. Battery Cell Balancing
vehicles are mostly still prototypes, although some solutions The individual cells in a battery pack typically have some-
have been proposed to charge traction batteries in hybrid solar what different capacities and they may be at different SOC
vehicles [1]. Perhaps, in this nanotechnology era, the final goal levels. This phenomenon may limit the energy that can be
may be the generation of a complete car as a solar cell through taken from and returned to the battery since the discharging
some nanocomposites [2]. However, intermediate steps are phase must stop when the cell with the lowest capacity is
necessary to improve the present technology and reach the empty, even though other cells have an SOC 6= 0. Therefore,
future. without balancing the SOC of the cells belonging to a battery
Firstly, the presence of solar cells on-board the vehicle raises pack, the cell with the smallest capacity is a “weak point”,
the issue of where and how the energy from the solar panels since it can be easily over-charged or discharged, while the
is stored. Hence, it is necessary to make an accurate efficiency cells with a higher capacity only undergo a partial cycle.
analysis of the best electrical energy storage system based on Although charging Li-ion batteries is generally based on
batteries [3]. Furthermore, it becomes essential to increase the a standard protocol, the Constant Current - Constant Voltage
energy efficiency in photovoltaic (PV) installations in such a (CC-CV) mode [5], methods for cell balancing differ depend-
way as to maximize the amount of the solar energy that is ing on the battery management policy. These methods can
converted into electricity. be divided into two main classes: passive and active [6], [7].
Currently, the most typical rechargeable batteries in hybrid Passive balancing relies on the principle that the energy of
and fully electric vehicles are lithium-ion (Li-ion) cells, which the cells with the highest voltage is “wasted” through a power
are characterized by a greater specific energy [Wh/kg] with resistor connected in series with a control MOSFET transistor.
respect to other common battery cells of different chemistry. In active balancing, an energy source supplies or distributes
In addition, a Li-ion cell is not affected by the memory the charge among the various battery cells in order to keep
effect and it has a slow loss of charge when not in use. them equalized in voltage. This balancing method does not
An accumulator (i.e., a battery pack), is realized by the depend on the chemical characteristics of the cells, and can

978-1-5090-1131-5/16/$31.00 ©2016 IEEE

 2016 IEEE 2nd International Forum on Research and Technologies for Society and Industry Leveraging a better tomorrow
(RTSI)

be used for most types of modern batteries. There are several

types of active balancing methods based on the energy transfer
type [7]. It might be from one cell to the whole battery, from
the whole battery to one cell, or from cell to cell.
B. Vehicles with solar panels
In general, the power generated by the solar cells on board
a vehicle is directed to the loads or to low voltage batteries,
often adopted for powering the auxiliary systems. For example,
a retrofit for a typical bus with diesel motor was proposed
in [8]. In this work, flexible PV modules on the roof made
it possible to charge traditional lead acid batteries to avoid
energy deficit due, for instance, to when buses are stationary Fig. 1. Architecture with converters in cascade
for a long time (e.g., terminus) or very frequent stops in urban
mobility (e.g., traffic lights and queues). In these conditions, N in the range 3÷4, thus yielding ηres =0.72÷0.85 when
a bus usually requires continuous auxiliary power for certain considering high-efficiency devices with η=0.92÷0.95). A vi-
utilities like heating, ventilation and air conditioning (HVAC), able solution would use no more than two DC-DC conversion
other than lighting, even when the engine is off. stages, with a maximum voltage multiplying factor of 20x.
A more comprehensive analysis of conversion efficiency, An alternative methodology, aimed at avoiding the afore-
from solar energy to battery charge, is provided in [3]. In the mentioned limits, is based on the addition of switching ma-
patent by Ward [9], a low-voltage solar panel on a hybrid trices in combination with MPPT and boost converters. For
vehicle directly charges battery cells, individually or in series, this case, a general architecture is illustrated in Fig. 2. The
but with no power optimization through Maximum Power switches can be inserted to enable the dynamic customization
Point Tracking (MPPT) techniques [10]. of cells in series, that is, strings according to the working and
This paper proposes an active balancing approach that uses operating conditions (e.g., shadowing).
solar power from PV modules installed on the vehicle. The
battery pack is charged with the PV energy following an
algorithm (described later), which balances the cells voltage
that is, the SOC of the cell strings belonging to the battery
pack.
III. M ETHOD AND MATERIAL
Generally, electric vehicles have a high voltage (HV) battery
pack connected to a HV bus for transferring energy from the
battery to the electric propulsion system. Low power loads can
be supplied through DC-DC converters or, alternatively, using
a different, low voltage (LV) battery [11], [12].
The presence of an auxiliary power supply system based on
solar cells, that can provide on average about 1.5kWh per day,
can be exploited to power the low-voltage system and thus the
Fig. 2. Architecture with switches matrices
low voltage battery may no longer be needed.
In this work, we focus on the transfer of the energy from Let us consider first the switching matrix at the output of
the solar cells to the HV battery. the solar cells. As an example, in clear sky conditions, it is
An individual solar cell provides energy at low voltage. possible to arrange a long series of cells and get a total voltage
For example, a C60 SunPower cell provides up to 6A at a close to that of the battery pack, thus requiring moderate
voltage of about 0.6 V. In order to work at higher voltages boost conversion and improving the resulting efficiency. If
(e.g., 120V), it is therefore necessary to connect more cells some solar cells are shadowed, they can be excluded from the
in series and use boost converters, such as the SPV1020 of series. Similarly, a switching matrix situated after an initial
STMicroelectronics, here adopted. These include the MPPT conversion stage may be used to connect in series or in
technique as well as boosting the voltage with good efficiency parallel different solar cell/MPPT subsystems. This leads to
values within limited multiplicative ratios (<4÷5x). However, a higher voltage without the use of additional boost DC-DC
the conversion efficiency is inversely proportional to the boost converters, and adapts the configuration to the photovoltaic
ratio. Furthermore, the performance is also reduced with a system conditions. However, solutions like these require insu-
non-constant current flow. The use of N stages of cascaded lated converters to enable system reconfiguration, as well as a
converters may simplify the design, as shown in Fig. 1, but centralized smart module (microcontroller unit or MCU) and a
the resulting efficiency would be much lower (ηres =η N , with sensing system that is able to recognize the status of each cell

978-1-5090-1131-5/16/$31.00 ©2016 IEEE

 2016 IEEE 2nd International Forum on Research and Technologies for Society and Industry Leveraging a better tomorrow
(RTSI)

(or cell set), implementing a system management algorithm the other can still provide the power, even though partially, to
for power transfer maximization that may be quite complex. the vehicle. In this case, considering the operating mode of
A final solution, initially proposed in [9], is based on the the converter, there is no need to use insulated converters, and
direct interaction between the battery and the solar cell system. simple Schottky diodes at the output of each conversion stage
This solution takes advantage of the fact that a battery is a can guarantee the full functionality in a simple and robust way.
series of cells, each characterized by a much lower nominal As previously stated, the BMS MCU dynamically selects
voltage than the pack one. In this proposed methodology, the which battery pack module has to be connected to the solar
solar cell output voltage does not need to be increased to source. To this purpose, the BMS MCU periodically queries
charge the battery pack, since the energy is transferred directly the module-level BMSs to get knowledge of the SOC, and
to subsets of the battery cells. then it turns on or off the switches sw xx depicted in Fig. 3.
Except in the unlikely case in which all battery elements are
fully charged and no loads are applied (in this case it would
not be possible to transfer energy from the solar source), in
every instant a single couple of switches is on (i.e., sw na and
sw nb, where n is the identification number of the switches),
while the others are off. A suitable algorithm is employed to
optimize the choice, as will be described in the next section.
For the sake of simplicity, the solar source is disconnected
!
from the system when an external energy charger is connected
to the batteries.
# $

A. Operating mode analysis

The following paragraphs show the capability of the system
# $

to transfer energy efficiently in different operating modes of

!
the car. We analyze three main different scenarios, which are
based on the current requested for the traction (Imotor ) and
# $!

!
auxiliary systems (Iaux ), other than on the current provided
by the PV cells (Isolar ).
"
1) Car Stopped, Auxiliary Systems Turned Off: In this con-
# $"

figuration, typical of long-stay parking, no charge is applied

to the battery pack, but the solar cells can provide energy.
Without solar contribution, the state of charge of each module
stays the same, or decreases moderately due to leakage.
# $ % Imotor = 0; Iaux = 0; Isolar > 0 (1)
The proposed system will charge the modules of the battery
Fig. 3. Proposed architecture
pack, alternating the connection to the solar source in time.
In fact, as can be seen in Fig. 3, various solar cell/MPPT After some time, the whole battery pack will be fully charged.
subsystems are connected in parallel, thus adding currents and 2) Car Stopped, Auxiliary Systems Turned On: This con-
maximizing the available PV power. Each of them raises the figuration is typical, for example, of a parked car with an anti-
voltage to an intermediate common value; the battery pack theft system powered with a converter attached to the main car
consists of the series of four subsets (i.e., strings) of cells battery.
in series, with a nominal voltage of 30V each. The available Imotor = 0; Iaux > 0; Isolar > 0 (2)
solar energy can thus be directly used to charge one of the In this case, two situations are possible, depending on the
subsets at a time, or to reduce its discharge in case the battery proportion between the energy amount requested by the loads
is being used to power the vehicle, by suitably arranging the and the one available from the solar source.
array of switches sw xx. The charge/discharge of the battery is In the former case,
managed by a low-level BMS acting on each subset, and on an
additional BMS MCU that operates the switches. The selection Isolar ≥ Iaux (3)
of which battery element to connect to the solar source has
therefore the solar energy contribution will partly feed the
to be done dynamically, after taking into account the SOC of
auxiliary load, and will partly charge one of the battery
each module.
modules. By periodically switching the connected battery
In the case of partial shading of one of the solar sub-
module, it will be possible to slowly fill up the battery pack.
modules, the related MPPT device is still able to maximize
In the latter case we have
the power transfer and to provide a contribution to the total
current; in the case of complete shading of one sub-module, Isolar < Iaux (4)

978-1-5090-1131-5/16/$31.00 ©2016 IEEE

 2016 IEEE 2nd International Forum on Research and Technologies for Society and Industry Leveraging a better tomorrow
(RTSI)

where the energy from the solar source cannot fully sustain of this algorithm is to transfer the PV energy to the strings
the load, so part of the energy will be taken from the battery. individually, and equalize them if necessary, at the beginning
However, the battery discharge will be slower thanks to the of the charging phase.
solar contribution. Similarly, this algorithm can also be applied during the
3) Car Moving: This last configuration, is characterized by discharge of the battery, allowing an increase in the total
available energy and, therefore, the battery run-time before
Imotor > 0; Iaux ≥ 0; Isolar > 0; Imotor + Iaux > Isolar (5)
re-charging.
This situation is equivalent to 4. The only difference being
IV. M ODEL VALIDATION
that the energy requested from the load is much larger. Most
of the requested energy will therefore come from the battery For this application, we considered a PV system of about
pack, but the solar contribution will still contribute to the 340 Wp , with a power rate per unit area of 177 Wp /m2 . The
extension of the vehicle autonomy. total surface of the mono crystalline solar panels on board is
1.92 m2 , distributed on various parts of the vehicle such as
B. System Management Strategies the roof, trunk and doors.
Although the battery pack has a high nominal voltage, e.g., A variable solar irradiance, that is, the power per unit area
120V, a lower voltage is needed for charging the cells. The [W/m2 ] received by the panels, is considered during all the
boost converter SPV1020 adopted here provides an output simulations. In fact, it is seldom the case that a constant clear
voltage in the range between 0 and 40V and, therefore, it sky condition and absence of side effects (e.g., shadowing)
can be connected to the strings, one at a time. This technique allow to get the maximum power from the PV system. So,
requires special care to avoid creating a high unbalancing of variable weather conditions and presence of, for instance,
the various strings during charging. This can be achieved by surrounding buildings are considered through a parameter.
transferring to each string, alternately, the energy generated by This derating factor for the peak power allows a near real
the solar cells, starting from the string with the lowest SOC. case application during the simulations.
Hence, the charge of each string will increase each time by a The results obtained from applying the algorithm 1 of
certain percentage until the SOC is equal to 100%. Section III-B are shown in Fig. 4.
The algorithm 1 describes the proposed method for charging
the battery pack with four strings Si , each consisting of N 0.74

elementary cells belonging to the set S of the battery pack.

0.72
Algorithm 1 Algorithm for the charge phase, and sustain
during the discharge phase, of the battery pack through the 0.7
available PV power
1: while SOCpack 6= SOCmax do 0.68
SOC

2: for all S ∈ [S1 , S4 ] do

3: Find the string Si : SOCSi = SOCmin (t), i ∈ [1, 4] 0.66

4: end for
5: while SOCSi 6= SOCmax do 0.64 S1

6: Si ← power(P V ) S2

7: if SOCSi 6= SOCmin (t) then 0.62 S3

S4
8: SOCL ← SOCmin (t) + T h
9: if SOCSi ≥ SOCL then 0.6
0 2000 4000 6000 8000 10000 12000
10: break Time [s]
11: end if
Fig. 4. Simulation of the algorithm for charging the strings of the battery
12: end if pack through the solar cells while providing a current of 0.8A for the auxiliary
13: end while loads.
14: end while In this first test simulation, a 0.8A current for the auxiliary
circuits is continuously supplied. At the beginning this current
At the beginning, the algorithm searches the string that, is provided by the battery. Then, after the equalization phase,
at time t, has the lower state of charge. The entire available the energy to the loads is supplied by the PV system, while
photovoltaic power is then transferred to it until it reaches an the surplus PV energy is transferred to the battery to charge
SOC equal to SOCL , given by the sum of the new absolute the cells. The slightly waving form of the plot is due to
minimum state of charge and the threshold value Th defined the changing conditions of the solar irradiance during the
in the BMS. In this way, it keeps a proper equalization. simulation.
After reaching this charge level, the process re-starts for the Fig. 5 reports an example of such application for a discharge
string with the lowest SOC, and so on until every string phase, when a 12A current is demanded by the load. In
reaches the maximum state of charge SOCmax . The effect this case, one can notice that the discharge of the battery

978-1-5090-1131-5/16/$31.00 ©2016 IEEE

 2016 IEEE 2nd International Forum on Research and Technologies for Society and Industry Leveraging a better tomorrow
(RTSI)

takes place more slowly after the effect of the initial string About the variability of the contribution of PV energy along
equalization (see the difference between the actual discharge the time, it is more visible during the charge phase (see the
curve of the string S4 after the equalization, and the dotted non linear increase of the SOC in Fig. 4); on the other hand,
line reporting the discharge in the case of no energy support it is mostly negligible during the discharge phases (see Fig. 5
by the PV system). and 6), since the effect of the load currents overlook the effect
of the variable irradiance.
0.95 V. C ONCLUSION
w/o PV cells
In this paper we addressed the problem of transferring the
S1
0.9 energy generated by an on-board solar cell system to the main
S2
S3 battery pack of an electric vehicle. Three alternatives were
0.85 S4
presented and one was selected due to the improved efficiency
0.8 thanks to the reduced number of conversion steps required. For
this solution, based on the interaction with the BMS devices,
SOC

0.75 we performed additional analyses and presented an algorithm

for balancing the SOC of the battery cells during the charging
0.7 phase.
The analysis and results of the simulations performed in
0.65
realistic conditions demonstrate that the proposed architecture
0.6
is suitable for extending the driving range of electrical or
hybrid vehicles, by maximizing efficiency and by guaranteeing
0.55 a good battery cells balancing. By the latter improvement, it
0 1000 2000 3000 4000 5000 6000 7000
Time [s] is also reasonable to expect an extended battery life.
R EFERENCES
Fig. 5. Simulation of the cell balancing during the discharge phase of 12A
for the battery pack with the PV energy support. [1] G. Rizzo, I. Arsie, M. Sorrentino, “Hybrid Solar Vehicles,”
in Solar Collectors and Panels, Theory and Applications, Dr.
The third simulation refers to a 75A current, which is an ex- Reccab Manyala, Ed., InTech, 2010, ch. 4, pp. 79–96. [On-
line] Available: http://www.intechopen.com/books/solar-collectors-and-
treme test condition. Fig. 6 reports the result, which obviously panels–theory-and-applications/hybrid-solar-vehicles.
shows the limited contribution of the energy provided by the [2] H. Presting, U. König, “Future nanotechnology developments for auto-
PV system in this case, when comparing the discharge phases motive applications,” Materials Science and Engineering, vol. C 23, no.
6, pp. 737–741, 2003.
of the battery in the vehicle with and without the solar panels, [3] T.L. Gibson, N.A. Kelly, “Solar photovoltaic charging of lithium-ion
respectively. However, a slight edge is still evident, even in batteries,” Journal of Power Sources, vol. 195, no. 12, pp. 3928–3932,
this stressful working condition for the battery pack. 15 June 2010.
[4] D. Andrea, “Battery Management Systems for Large Lithium-Ion Bat-
tery Packs”, Artech House, 2010.
0.9 [5] W. Shen, T.T. Vo, and A. Kapoor, “Charging Algorithms of Lithium-Ion
Batteries: an Overview,” Proceedings of the 7th Conference on Industrial
w/o PV cells Electronics and Applications (ICIEA), pp 1567–1572, July 2012.
0.8 with PV cells [6] Y. Xiaolu, E.W. Ma, and M. Pecht, “Cell balancing technology in
battery packs,” in Proceedings of the 13th International Conference on
0.7 Electronic Packaging Technology and High Density Packaging (ICEPT-
HDP), pp. 1038–1041, August 2012.
[7] M. Daowd, N. Omar, P. Van Den Bossche, and J. Van Mierlo, “Passive
0.6 and active battery balancing comparison based on MATLAB simulation,”
in 2011 Vehicle Power and Propulsion Conference (VPPC), September
SOC

0.5 2011.
[8] M. Donno, A. Ferrari, A. Scarpelli, P. Perlo, A. Bocca, “Mechatronic
system for energy efficiency in bus transport,” in Proceedings of the
0.4
2012 Conference on Design, Automation and Test in Europe (DATE),
pp. 342–343, March 2012.
0.3 [9] T.A. Ward, “Hybrid vehicle with a low voltage solar panel charging a
high voltage battery using a series charger to separately charge individual
0.2
cells of the series connected battery,” U.S. Patent No. 7,884,569, 8
February 2011.
[10] T. Esram, P.L. Chapman, “Comparison of Photovoltaic Array Maximum
0.1 Power Point Tracking Techniques,” IEEE Transactions on Energy Con-
0 500 1000 1500 2000 2500
Time [s] version, vol. 22, no. 2, pp. 439–449, June 2007.
[11] A. Emadi, S.S. Williamson, and A. Khaligh, “Power electronics intensive
Fig. 6. Simulation of the discharge phase of 75A of the battery pack. solutions for advanced electric, hybrid electric, and fuel cell vehicular
power systems,” IEEE Transactions on Power Electronics, vol. 21, no.
These results confirm that solar panels on board the ve- 3, pp. 567–577, May 2006.
hicle can offer a valid energy support during the charge [12] D. Costinett, K. Hathaway, M.U. Rehman, Y. Levron, and D. Maksi-
movic, “Active Balancing System for Electric Vehicles with Incorporated
and discharge phases of the battery pack. Additionally, active Low Voltage Bus,” in the Twenty-Ninth Annual IEEE Applied Power
balancing can occur without requiring extra energy. Electronics Conference and Exposition (APEC), 16–20 March 2014.

978-1-5090-1131-5/16/$31.00 ©2016 IEEE