Experimental Analysis of an Electric Minibus
with Small Battery and Fast Charge Policy
R. Di Rienzo
∗
, F. Baronti
∗
, F. Vellucci
§
, F. Cignini
∗∗
, F. Ortenzi
§
, G. Pede
§
, R. Roncella
∗
, and R. Saletti
∗
∗
Dip. di Ingegneria dell’Informazione, Universit
`
a di Pisa, Italy
§
Laboratorio Veicoli a Basso Impatto Ambientale, ENEA, Santa Maria di Galeria (Roma), Italy
∗∗
Universit
`
a La Sapienza, Centro di ricerca per il trasporto e la logistica, Roma, Italy
Abstract—The lead-acid battery of an electric minibus has
been replaced with a smaller size lithium-ion battery system
consisting of standard 12 V modules and a hierarchical battery
management system. The minibus has experimentally been tested
to show that the reduced battery capacity, which also cuts costs,
does not affect the daily operational mission. This is assuming
that the driving phases are alternated with fast charging periods.
Experiments show that fast charging of 8 min guarantees up to
1 h of operation.
I. INTRODUCTION
The Kyoto protocol requires that the greenhouse gas (GHG)
emissions within the European Union in 2050 are about 80–
95 % below the 1990 levels. The protocol suggests halving the
number of conventional vehicles by 2030 and their complete
elimination by 2050. Since a reduction of mobility is not a
viable option, a realistic solution can be the increase of public
transportation based on the sustainable energy paradigm [1],
[2].
A study by the Joint Research Center (JCR) in Ispra (Italy)
[3] shows that the average length covered by a vehicle in a
single trip is about 5–20 km. The study is based on the GPS
data collected from 16 000 vehicles in the Italian province
of Modena in May 2011. According to these driving range
data, electrified vehicles equipped with the currently available
battery technology are a feasible solution and a very good
opportunity to meet the protocol requirements. A large GHG
emission reduction may also be achieved by the electrification
of public transportation, especially in small cities and historical
city centres. This concept has successfully been applied in
Coimbra (Portugal), where a fleet of electric minibusses for
tourists has been introduced in 2003 [4].
Commercial electric minibusses are usually equipped with
lead-acid batteries. This mature battery technology has the
disadvantage of a low gravimetric energy density and short
lifetime, especially when subjected to deep discharges and
high currents [5], as it happens in this application. However,
the replacement of the traction battery with one of higher
performance significantly increases the minibus costs. This
issue is addressed in [6], where the authors conducted a
simulative study to demonstrate the effectiveness of replacing
the original lead-acid battery of a minibus with a lithium one.
The lithium battery technology provides indeed higher energy
and power densities, together with a longer lifetime [7].
TABLE I
CHARACTERISTICS OF THE CASE-STUDY ELECTRIC MINIBUS
Curb weight
(without the traction battery) (2800 ± 100) kg
Length 5.1 m
# Passengers 27 (10 of which are seated)
Traction power 25 kW (peak), 20 kW (nominal)
Average energy
consumption per kilometer 500 W h km
−1
Original lead acid battery 72 V, 585 A h, 1500 kg
Volume available for the battery 2×(1210 × 541 × 375) mm
3
This work aims at describing the implementation and test
of an electric minibus based on a lithium battery and the
fast charging concept. This allows the reduction of the bat-
tery capacity by alternating the driving phases with rapid
charging periods. A preliminary study of the battery design
has been carried out in [8], which suggested that a modular
architecture based on 12 V standard modules may be effective
in reducing costs due to the economies of scale. A first
implementation of the standard module has been described in
[9], [10]. The application of the fast charging concept to local
public transportation is discussed in [11]. The current work
provides an insight into the actual battery implementation
and its integration into an electric minibus, similarly to what
done in [6]. Moreover, this work reports on the experimental
campaign carried out on a dynamometer chassis and discusses
the experimental results in the framework of the fast charging
scenario outlined in [12].
The paper is organized as follows. The target electric
minibus and the battery charger representing the case-study
are described in Section II. Section III provides details of the
battery implementation. The experimental test-bed and results
are explained in Section IV. Finally, Section V draws some
conclusions.
II. CASE-STUDY
This Section presents the electric minibus and the battery
charger, which constitute the starting points of this work.
A. Electric minibus
The electric minibus used in this work is the TECNOBUS
Gulliver U520, shown in Fig. 1. It was originally equipped
Fig. 1. Photograph of the electric minibus used in this work.
TABLE II
CHARACTERISTICS OF THE BATTERY CHARGER
Input 400 V three-phase AC , 50-60 Hz
Output voltage 60 V to 87.6 V
Continuous output current 360 A
Continuous power @ 72.8 V 26 kW
Efficiency >85 %
with a 72 V 585 A h lead-acid battery. The minibus is pro-
vided with a 25 kW electric motor with a maximum torque
of 235 N m at 950 rpm. The energy consumption is about
500 W h km
−1
and the speed is limited at 33 km h
−1
by the
vehicle control unit. The main characteristics are summarized
in Table I.
B. Battery charger
The battery charger was selected from commercially avail-
able devices, in order to reduce the initial cost of the project.
The charger is manufactured by Zivan. It consists of four RG9
modules connected in parallel capable of providing an overall
continuous charging power of 26 kW. It can be controlled
via a Controller Area Network (CAN) bus interface. Its main
characteristics are summarized in Table II.
III. BATTERY IMPLEMENTATION
The lithium battery has been designed to be electrically
compatible with the original lead-acid one. The minibus
electric propulsion system can thus be reused. Therefore,
the nominal voltage of the new battery must be 72 V, a
value achieved by 6 standard 12 V battery modules [9], [10]
connected in series. The battery capacity has been chosen to
be 240 A h. This choice provides a good trade-off between
the battery cost and the driving range (approximately 30 km,
given a 500 W h km
−1
energy consumption). Such a relatively
small sizing of the battery energy is possible thanks to the fast
charging concept, which allows the energy consumed for the
journey to be balanced with the energy recharged at the bus
stops [12].
Fig. 2. Photograph of the standard 12 V 60 A h battery module (the case
lid has been removed to show the module management (MMU) electronic
boards.
TABLE III
CHARACTERISTICS OF THE LFP CELLS (HP-PW-60AH)
Nominal voltage 3.2 V
Nominal capacity 60 A h
Dimensions (L x W x H) (114 × 61 × 203) mm
3
Weight 2.04 kg
Working temperature (discharge) -20 to 65°C
Working temperature (charge) 0 to 45°C
A. Standard 12 V battery module
A photograph of the designed standard 12 V battery module
is shown in Fig. 2. The module incorporates 4 lithium-
iron-phosphate (LFP) cells, whose main characteristics are
summarized in Table III and an advanced module management
unit (MMU). The core of the MMU is a 32-bit ARM Cortex-
M3 microcontroller (LPC1754 from NXP). It manages the
acquisition of the voltage and temperature of the 4 cells (via
a dedicated stack monitor IC, the LTC6803-3 from Linear
Technology), the activation of the module fans (which are
controlled so that the maximum cell temperature is kept
between two configurable thresholds) and the communication
via an isolated CAN bus.
The MMU also includes a section for handling analog inputs
and isolated general purpose (GP) I/Os. Two analog input
channels can be connected to the LEM S/18 sensor, used to
acquire the module current. The LEM S/18 is a bidirectional
Hall Effect sensor with two output channels having different
sensing ranges, namely, ±30 A and ±350 A. The module
current is numerically integrated by the microcontroller as part
of the State-of-Charge (SoC) estimation algorithm [9], which
is based on the Coulomb Counting method combined with
Open Circuit Voltage (OCV) compensation [13].
An innovative feature of the MMU is the circuit for the
active charge equalization of the battery cells. It is based
on an isolated DC/DC converter (Cincon EC6A01) and a
Fig. 3. Battery architecture.
switch matrix, which allows the individual connection of each
module cell to the converter output (the converter is fed by the
module voltage), thus implementing a module to cell active
balancing topology. A novel and interesting feature of the
circuit is the possibility of connecting the converter output also
to a cell in another module, therefore enabling inter-module
active balancing. This is obtained by a circular balancing
bus, controlled in such a way that the battery is never short
circuited, independently of any decision taken by the single
microcontrollers in the MMUs [14].
Compared to a previous implementation of the battery
module [9], [10], particular care has been given to the me-
chanical assembly and the thermal management to ensure good
reliability. In particular, as seen in Fig. 2, the MMU consists of
two boards: the upper one hosts the control circuitry, whereas
the lower one acts as connection board holding the busbars
and routing the cells’ terminals and temperature sensors to
the upper board. This dramatically reduces the complexity of
the wire harness in the module increasing its reliability.
Fig. 4. Photograph of battery pack and its integration into the minibus. The
lower inset shows one battery housing containing 2 battery strings, without the
power and signal wire harness. The upper inset shows the pack management
unit.
Fig. 5. Schematic block diagram of the PMU.
B. Battery pack
The battery pack consists of 24 standard modules to reach
the required 72 V terminal voltage and the 240 A h battery
capacity. The 24 modules are partitioned in 4 strings, each of
them composed of 6 series-connected modules (see Fig. 3).
Compared to other configurations that yield the same voltage
and capacity, this topology simplifies the power connection
between the modules’ terminals, at the expense of a higher
sensitivity to cell inhomogeneities, which reduce the usable
capacity of the battery [15]. Moreover, the chosen topology
facilitates the assembling of the battery pack and its integration
into the minibus, as two complete strings can be allocated in
each of two housings containing the original lead acid battery
(see Fig. 4).
Fig. 3 shows the connections between the modules and the
pack management unit (PMU). Each string has a local routing,
which establishes the series connection of the 6 modules and
the communication among the MMUs, and a global routing
toward the PMU and its ancillary components. Consequently,
there is no direct wiring between the two strings, which
supports the above mentioned objective of simplifying the
integration of the battery into the minibus. In order to reduce
costs, only one current sensor connected to the Module 1 is
used per string, as seen in Fig. 3. The same current flows
indeed in all the modules of a string and its values is shared
by the Module 1 to the other modules via the CAN-bus.
The battery connection to the load/charger is controlled by
the switches (Sw
B+
, Sw
B-
), in series with the positive battery
terminal and the negative one, respectively. The switches are
implemented with a power relay or contactor (Tyco EV200),
which can carry a continuos current of 500 A @ 85
◦
C
and withstand a voltage of 320 V. The same contactors are
also used for the string switches (Sw
1
–Sw
4
) that control the
parallelization of the 4 strings, to reduce the bill of material. A
pre-charge resistor and a relay are connected in parallel to the
Sw
B+
switch. This limits the in-rush current flowing when the
battery is connected to the load due to its input capacitance.
All the switches are driven by the PMU, which controls
their state according to the commands received from the user
interface (UI), the string total voltages, and the internal status
of the modules. In particular, a string switch may be turned
on only if the difference between the string voltage and the
battery voltage is in the range of ±3 V and if all the string
modules are in the safe state. This avoids an extremely high
circulating current between the strings because of their very
low series resistance and keeps the voltage and the temperature
of all the 4 module cells in their safe operating area. In more
detail, the activation signal of a string switch, generated by
the microcontroller inside the PMU, is AND-wired with the
consent signals produced by 6 MMUs (and routed to the
PMU through the isolated GPIOs). In this way, each module
can interrupt the current flowing in the corresponding string
independently of the PMU to achieve a prompt and robust
protection mechanism.
As seen in the schematic block diagram of Fig. 5, the PMU
is based on the NXP LPC1756 microcontroller, which belongs
to the same family of the one used in the MMU, but with an
additional CAN-bus peripheral. One CAN-bus peripheral is
used for the communication with the modules and the other
one for the communication with the battery charger. The PMU
also incorporates 4 isolated DC/DC converters (one per string)
powered by the string with highest voltage. The output of each
converter is connected to the balancing bus of one of the 4
strings. Consequently, energy can be moved from the cells
of the averagely more charged string to a generic cell of the
battery. The MMU equalization method is used to balance the
4 cells in the module and this can be achieved in parallel
for all the modules. After this step, cells are balanced within
modules, but charge imbalance can still be present between
cells belonged to different modules. The latter imbalance is
recovered by using the PMU equalization method.
Fig. 6. Photograph of the minibus during a test session on the dynamometer
chassis.
IV. EXPERIMENTAL VALIDATION
A. Experimental test-bed
The battery pack has been assembled and mounted in the
minibus, as shown in Fig. 4. An extensive test campaign has
been carried out using a dynamometer chassis (see Fig. 6),
which measures and stores mechanical quantities, such as
speed and torque. The minibus performance in terms of
energy consumption and driving range has been assessed using
the Standardized On-Road Test (SORT) driving cycles, as
defined by the international organisation for public transport
authorities and operators (UITP).
During the tests the PMU is connected to a laptop which
implements the UI developed in LabVIEW. Fig. 7 shows
a screenshot of the UI, which allows the user to control,
monitor and log the status of the battery. In particular, voltage,
current, and temperature of each cell of the battery can be
logged and saved for offline processing. Moreover, the UI
makes it possible to individually turn on the 4 strings in the
battery. This feature has been used to test the minibus when
powered by 4 strings in parallel (full battery) and by just 2
strings (half battery). Fig. 8 displays the speed acquired by the
dynamometer chassis and the electrical power provided by the
battery, during two SORT 1 cycles.
B. Experimental results
1) Full battery case: Table IV summarizes the results
of a test consisting of 86 repetitions of the SORT 1 cycle
(corresponding to 31.7 km), executed with the full battery,
starting from the full charged state. The test was completed
with success without fully discharging the battery. This means
that the full battery can guarantee the expected driving range
(see Section III). It is worth noting that the charge and energy
provided by the battery is not uniformly distributed among the
4 strings. This is an undesired behavior, which tends to degrade
the usable capacity of the battery. This is caused by variations
in the characteristics of the strings connected in parallel, as
discussed in [15]. In this case, we observed a significant
variations in the string series resistance, especially between
the first and fourth string. This phenomenon can mainly be
ascribed to some critical aspects in the power connections
Fig. 7. Screenshot of the Main tab of the User Interface developed in LabVIEW.
Time (s)
0 50 100 150 200 250 300
-10
-5
0
5
10
15
20
25
30
35
Speed (km/h)
-10
-5
0
5
10
15
20
25
30
35
Power (kW)
Fig. 8. Speed and electric power during two SORT 1 cycles.
TABLE IV
FULL BATTERY TEST (4 STRINGS)
Time 3.5 h
Length 31.7 km
Number of SORT1 cycles 86
Start / End voltage 80 V / 77 V
string 1 string 2 string 3 string 4
Charge delivered by (A h ) 35 40 41 44
Energy delivered by (kW h) 2.64 3.07 3.09 3.34
between the cells within a module and the external terminals,
leading to a wide dispersion of the module series resistances.
2) Half battery case: Another interesting test is the evalu-
ation of the minibus driving range when powered by only two
strings. This configuration uses a battery with halved capacity
yielding a significant reduction of costs. Table V shows the
relevant results. The minibus powered by two strings runs
TABLE V
HALF BATTERY TEST (2 STRINGS)
Time 2.5 h
Length 22.5 km
Number of SORT 1 cycles 61
Start / End voltage 81 V / 69 V
string 3 string 4
Charge delivered by (A h) 53.8 53.7
Energy delivered by (kW h) 3.92 4.08
about 22.5 km in 2.5 h, using approximately 8 kW h.
A driving range of 22.5 km is not sufficient for an urban
minibus. However, it may be increasd by intermediate fast
recharges in a very effective way. The required daily operation
is achieved provided that the energy consumed during the
travelling time is balanced by the recharged energy during
standstill time [16]. To guarantee the balance, the ratio r of
the charging time (T
chg
) to the travelling time (T
run
) must
fulfil the following equation
r =
T
chg
T
run
=
Cv
P
chg
(1)
where C is the average energy consumption per kilometer,
P
chg
the available charging power and v the average speed.
Thus, it is crucial to assess the maximum charging power that
the battery can withstand.
To this end, the charging procedure shown in Fig. 9 was
applied to a fully discharged battery consisting of 2 strings.
In more detail, the PMU first configures the charger to deliver
a current of 330 A, which corresponds to a 2.75 C-rate. This
current value remains constant until one cell reaches the upper
cut-off voltage. At this point, the charger is set to constant-
voltage with a value slightly lower than the battery voltage at
