A Concept of a High-Energy, Low-Voltage EV
Battery Pack
Matthias Kerler∗ , Peter Burda† , Michael Baumann‡ and Markus Lienkamp§
Technische Universität München
Institute of Automotive Technology
Email: ∗ kerler@ftm.mw.tum.de, † burda@ftm.mw.tum.de, ‡ baumann@ftm.mw.tum.de, § lienkamp@ftm.mw.tum.de
Abstract—Today’s electric vehicles (EV) use a high system
voltage due to the reduction of the appearing current. This results
in a much more complex electrical system (high-voltage side: IT
topology) compared to that of a conventional car with an internal
combustion engine. This can be explained by the need to ensure
the safety of the passengers, the maintenance staff or the rescue
personnel, in the case of an accident.
Current battery systems for hybrid and battery electric vehicles
typically have operating voltages of 200–800 Volts [1], [2].
Hence the idea is to investigate whether lower voltage levels (down
to 24 V) are a possible alternative. The question is divided into
three topics. Firstly, the optimum cell size, secondly, the battery
management system and, thirdly, the system level/battery pack
design. After a detailed investigation of those subjects, three
voltage levels (24 V, 48 V, 300 V) of a possible battery pack
will be examined in terms of costs and technical effects.
I. I NTRODUCTION
The voltage level of electric vehicles’ battery systems lead
to a multitude of safety issues. Voltages above 60 V DC
imply potential danger to the life of persons involved in the
assembly, operation, maintenance and recycling of electric
vehicles. Current battery systems for electric cars typically use
voltage levels between 200 and 800 V. The voltage level is
usually not determined by the battery system, but the battery
system is defined to operate within the voltage range demanded
for by the power electronics and electric engine. With the
ISCAD (Intelligent Stator Cage Drive) drivetrain becoming a
possible option, new approaches to the battery system topology
to work with such a low-voltage, high-power system becomes
an interesting demand.
II. N EW APPROACH FOR AN EV BATTERY PACK
In this paper, the meaningfulness, feasibility and the tech-
nical, as well as the monetary effects of two low-voltage,
high-energy battery packs (24 V and 48 V) for EVs shall be
investigated and compared to a battery pack with the same
energy content and a nowadays typical voltage level (300
V). The study thereby is conducted on three different levels:
Firstly, the cell itself is considered with the question of its
optimum size for the various voltage levels. Subsequently, the
consequences for the design and functionality of the battery
management system are taken into account. Then, the different
approaches are investigated on a system/battery pack level.
Chapter III to V comprise the fundamental thoughts of these
three topics while, in chapter VI, the three different battery
pack configurations are compared according to the topics.
978-1-4799-6075-0/14/$31.00 ©2014 IEEE
III. O PTIMUM CELL SIZE
In the early stage of the concept for a new BEV battery
pack, an important part of the development is the decision
for the correct battery cell. It has a major influence on many
of the following decisions, like the cooling system or the
internal interconnections, etc. However, the performance of
the system also partly depends on the right choice of cells
as parallel interconnections within the battery, for example,
lead to a more efficient energy utilisation (compare chapter
III-A). Looking more closely at today’s BEV car market, a
bright variety (shape and capacity) of cells installed in BEV’s
battery packs can be found. On the one hand, it is mainly
German OEMs which are using very big cells in accordance
with the German VDA standard with approximately 60 - 66
Ah capacity. When such cells are used to reach a voltage level
of 300 - 400 V, no parallel connection is necessary to get a
sufficient battery capacity (e.g. BMW i3). On the other hand,
Tesla Motors is using 18650 battery cells to build up packs
under the use of a massively parallel connection. At this point,
the questions emerge as to whether a rather simple battery
pack layout with big cells is the best solution or where the
benefit is to Tesla Motors greater effort with its utilisation of
a large quantity of small cells (more than 7000). To answer this
question, especially with regard to a high-energy, low-voltage
battery-pack, in the following, the influence of the cell size
on efficiency, safety aspects, weight, costs, thermal behaviour
and ageing will be investigated. As a result, a method will
emerge to define the optimum cell size. The topics mentioned
are picked up again in chapter VI to evaluate the three different
voltage levels.
A. Efficiency
Considering the efficiency of a battery pack, the question
is how much of the stored energy can be utilised. Or, in other
words, how much energy will get lost, or converted, into heat
respectively, by the demand of a specific current. It is obvious
that the internal resistance of the battery pack is a crucial factor
in answering this question.
The total static resistance consists of the resistance of the
battery cells, the contact resistance and the resistance of the
wiring. As the influence of the cell size on the efficiency
shall be investigated, a correlation between the cell size or the
capacity respectively, and the internal resistance is required.
G. Pistoia postulated the following relation [3].
Rcell · Ah = constant = CR (1)

Fig. 1. Thermal loss of an 85kWh / 360kW battery pack
Rcell = internal resistance of a battery cell
Ah = capacity of a battery cell
CR = normalized resistance
In a good approximation, CR is constant for a type of cell
chemistry, this factor can be used to scale a cell size up or
down, for further investigations. Assuming that a battery pack
with more, but smaller, cells will not need significantly more
wiring than one with larger cells, a constant factor Rw can be
added for the calculation of the total resistance of the pack.
1 role in the overall energy consumption of a car. Indeed a great
Rpack = s · 1 + Rw (2)
z· Rcell (Ah,CR )+2·Rc
s = number of serial connected cells
z = number of parallel connected cells
Rcell (Ah, CR ) = internal resistance of a cell, ref. equation 1
Rc = contact resistance
Rw = resistance of the wiring
With formula 2, the internal resistance of a whole battery
pack can be calculated. In figure 1 (Rc = 0.2mΩ [4], CR =
0.058ΩAh, Nominal voltage per cell 3.7 V), the thermal loss of
a possible 85 kWh battery pack with a peak power of 360kW
is plotted, based on formula 2 and the emerging current,
which is a function of the amount of serial interconnections. A
theoretical interconnection of 1s1p means a possible 85kWh
battery pack, realised with only one large battery cell. The
resulting high currents produce a lot of thermal loss in the
battery pack, compared to a battery system with a 100s100p
interconnection.
As a result, with knowledge of all the boundary conditions, like
the cooling system, or the required voltage level, it is possible
to choose the correct interconnection or cell size respectively.
Obviously, a large quantity of serial and/or parallel intercon-
nections, and thus smaller cells, are desirable to increase the
efficiency of a battery pack.
Nevertheless, further investigation of the contact resistance Rc
is necessary, because it is likely that this is not a constant
value but also a function of the cell size or the cell terminals
respectively.
B. Safety aspects
The safety of a BEV battery pack is very important. One
scenario which must be avoided at all costs is the so-called
thermal runaway. According to [5], this term describes an
exothermic reaction within a lithium ion cell which goes out
of control. In the worst case, a thermal runaway can lead to a
fire and explosion of the cell. In the case of fire, the average
calorific value of a lithium ion cell is of importance. According
to [6], the average calorific value of a lithium ion cell is about
ten times higher than the electrical energy stored within it.
This aspect or circumstance must be considered in the design
of a potential battery pack. This can, for instance, be done
by allowing specific distances between the battery cells, or
by utilising particular fireproof films: this is because of the
risk of a cascade failure of the cells. The term describes the
phenomenon of one cell, which sustains a thermal runaway,
producing a critical amount of thermal energy, with the re-
sult that neighbouring cells also sustain a thermal runaway.
Knowledge of this problem means that smaller cells should be
preferred.
C. Weight
The weight of BEV is a crucial parameter. Due to of the
limited storable energy in the battery, the aim is to reduce the
weight of a BEV as much as possible in order to maximise
the electric range - this is because weight plays an important
proportion of the overall weight of a BEV can be allotted to
the battery itself [7]. Hence, the lightweight issue has to be
considered in the design of a new battery pack. The decision
process can be supported by analysing the weight capacity
density of common cells available on the market (Fig. 2; based
on the database [8]). Generally speaking, without considering
detailed use cases, cylindrical cells seem to have a slightly
higher energy capacity compared to that of prismatic-shaped
cells. Moreover, it can be recognised that cells with a capacity
of between 1–10 Ah approximately, have a significantly higher
weight-energy density. It can be assumed that the rising
demand in cells larger than 10 Ah will lead to a progression
of the weight-energy density in the future. Nevertheless, the
use of too smaller cells should be avoided because the ratio of
active material to casing material decreases massively with the
size. This is the reason why only cells with a capacity greater
than 1 Ah are considered.
Pouch-Cells are not considered in this investigation, because
they would manipulate the image due to their missing casing.
At this point, additional weight in the design of the battery
pack casing has to be considered, which would lead, in the
end, to similar results as for the prismatic cells.
D. Costs
Cost is also a very important factor in the field of batteries
for a BEV. An optimum design solution is useless if it is too
expensive for both the manufacturer and the customer. Hence,
it is necessary to examine this topic as well. The costs which
can be influenced by the cell size are the costs of production
and the material costs.
The costs for the cells, as well as for the production of a
battery pack, are influenced, amongst other things, by the
phenomenon of technological learning. According to [9], the
Fig. 2. Weight capacity density of cylindrical and prismatic cells available
today considering capacities from 1 Ah to 100 Ah; V = 3.7 V [8]
technical and economic performance increases significantly as
producers and consumers gain experience with a technology.
Equation 3 describes the correlation between experience and
costs.
m
CCum = C0 Cum
with
CCum = cost per unit
C0 = cost of the first unit produced
Cum = cumulative (unit) production
m = experience index [9]
On the one hand, producing a battery pack with more, but
smaller, cells does lead to increased assembly costs but, on
the other, the experience in the bonding process grows faster
and smaller cells do show a better cost/kWh-ratio. Hence, the
decrease in costs for the production of a battery pack with
smaller cells will be bigger, but it is reasonable to assume that
it will never fall below the costs of a battery pack with larger
cells. Further investigations are necessary to assess this topic
in greater detail.
E. Thermal behaviour of the cells
One of the major concerns about a BEV battery pack
is the process of ageing, and the loss of driving range over
the years of utilisation. The phenomenon of capacity loss of
lithium ion batteries has been under scientific investigation
for many years. Meanwhile, it is commonly accepted that
the influence of the temperature plays an important role in
this context [10]–[12]. Due to this fact, the aim is to keep
the battery pack and the cells within it in a perfect thermal
window. Hence, too low, as well as too higher temperatures
need to be avoided, but the thermal gradient in a battery cell
also has to be kept low.
With regard to the cell size, the question is whether there is
an optimum geometrical size of a cell, which complies with
the given specifications. The following boundary conditions
are identified, to assess the cooling of a cell within a battery
pack.
Fig. 3. Sequence of the thermal optimisation
• Heat capacity, mass flow and Inlet temperature of the
cooling medium
• Heat transfer coefficient
• Internal resistance of the battery cell
(3) • Demanded current
To solve this problem, an optimisation tool has been
developed which uses MatLab R2013b and Ansys15, presented
in figure 3. The goal is to minimise the maximum temperature
and the gradient to a defined value. Beginning with a set of
geometrical data of a battery cell, and the mentioned boundary
conditions, the algorithm sends the data to Ansys. After the
simulation, the results are then sent back to MatLab and
the algorithm generates a new set of geometrical data. The
sequence will be terminated when both the temperature and
the gradient have fallen below the defined values. As a result,
the geometrical data of a thermal optimised cell size will be
generated.
In this context, equation 1 is used again, to enable the calcula-
tion of the new resistance of the altered cell. The FE-Model of
the cell is a pure thermal model. All remaining calculations,
like the thermal loss based on the current, are made by the
MatLab algorithm. Nevertheless, further refinement of the FE-
Model will be necessary. With a thermal-electrical coupling,
it will be possible to map the current distribution and thus the
exact heat distribution in the cell. By this means, the ampli-
fication of thermal inhomogeneities within the cell, induced
by the wrong cooling system, can be avoided. Only then will
the investigation of the ageing process within a battery cell be
useful.
Generally, there are already some effects that can be observed
with the simple model. There are two possible ways to decrease
the maximum temperature within the cell; firstly, the cooling
can be increased or secondly, the current must be decreased.
If these two possibilities are not an option, it is possible to
reduce the cell size which leads to a lower current per cell. Of
course, such a decision is only possible in a very early stage
of the development process.
There are three possible ways to minimise the temperature
gradient. Firstly, the current can be reduced; secondly, the
cooling can be decreased or, thirdly, the cell size can be
reduced.
The thermal dimensioning of a battery pack is a well-balanced
equilibrium between the cell size, the cooling system (active
or passive) and the whole battery pack. An optimised cell is
the first step in achieving such a solution.
IV. BATTERY MANAGEMENT SYSTEM
Lithium-ion-based batteries are very sensitive when oper-
ated outside their specified safe operation areas (SOA) for
voltage, temperature and current, with consequences which
range from irreversible cell damage to thermal runaway and
even cell burst [13], [14]. In order to ensure battery pack safety
and maximise useful cell life and capacity, battery management
systems (BMS) are utilised. In general, the basic functionality
of such systems can be divided into [15]:
• Cell data acquisition (Voltage, Current, Temperature)
• Battery state estimation (e.g. SOC, SOH)
• Electrical management (e.g. charge balancing)
• Thermal and Safety Management
• Communication (with user or external devices)
In the specific use case of battery and hybrid electric vehicles,
the BMS fulfils further tasks like e. g. insulation monitor-
ing. Subsequently, the most important features of a battery
management system are detailed in order to provide the basic
knowledge, before differences of the BMS layout for high and
low-voltage battery packs are discussed in chapter VI-B.
A. Electrical monitoring
In order to operate lithium-ion cells within their electrical
specifications, as well as to provide input data for state
estimation, the voltage of each parallel block of cells and the
overall battery pack current is monitored by the BMS.
For cell voltage measurement integrated circuits like [16]
are typically utilised which can handle a certain number of
serial cell blocks (12 in total for [16]). Dependent on the total
number of serial cell blocks and the layout of the battery pack,
various topologies for the BMS are possible respectively (see
figure 4). In terms of battery pack current measurement, either
shunt or hall-effect based sensors are commonly used.
B. Balancing
Usually identical cells with the same nominal voltage
(same cell chemistry) and capacity (same cell size) are used
for a traction battery. However, due to production tolerances,
inhomogeneous ageing behaviour or temperature distributions,
the individual cells may have slightly varying capacities and
leakage currents. This would cause drifting SOCs of the serial
cell blocks. In order to keep all the cells at equal charge
levels, and to avoid overcharging of individual ones, balanc-
ing mechanisms are deployed. Thereby, passive ones, which
simply dissipate excessive energy via a switched resistor, and
active ones (e.g. bidirectional synchronous flyback converters
like those described in [17]) which transfer energy between
cells and/or the whole battery stack, can be differentiated. In
table I, a brief comparison of both balancing methods is listed.
Further information can be found in [18] and [19], for example.
Fig. 4. Various BMS topologies (top left to bottom right): Centralised,
modular, Master-Slave and distributed topology (CM = cell monitor; µC =
Microcontroller)
TABLE I. PASSIVE AND ACTIVE BALANCING COMPARISON
Passive Balancing Active Balancing (Flyback
(Switched Resistor) converter)
Equalisation speed + ++
Control Complexity Simple Complex
Size/Cost Low - medium/cheap High/expensive
Charge/ Discharge Only charge Bidirectional
Efficiency 0% (energy dissipated in Up to 92% [17]
heat)
C. Thermal management
Generally speaking, the task of the thermal management
is to minimise temperature gradients within the entire battery
pack and to keep cell temperatures inside designated limits.
Therefore, acquiring cell temperatures as input for thermal
management is an essential task of battery management sys-
tems. The sensors used vary depending on the intended purpose
(required accuracy) and range from simple NTC/PTC sensors,
via PT100/1000 elements, to digital bus-based sensors.
One essential question is, where to place the temperature
sensors and which cells to monitor within the battery pack,
since the quantity of sensors is typically limited due to costs
and a limited number of measurement inputs. For safety
purposes, sensors need to be placed at the battery packs hot
spots which can be identified by e. g. thermal simulation.
Besides cells, temperature monitoring of the current col-
lector connected to the pols of the cells, and at some critical
connections within the high-voltage cabling, may be consid-
ered dependent on the particular battery pack design in order
to avoid heat input to cells through damaged connections with
high contact resistance [14].
D. State estimation
Precise state estimation, like the state of charge (SOC)
and state of health (SOH), is essential for maximising the
efficiency of a battery pack and is an important input for user
feedback, such as the remaining range calculation of a BEV. In
general, two different approaches can be differentiated for SOC
estimation of lithium ion cells which can be applied depending
on cell chemistry and battery use case: on the one hand, the
state of charge is directly determined by some gauging or
balancing methods, like Ah counting, or it is identified via
adaptive, model-based algorithms like those described in [20]
or [21]. While the accuracy of the first category is highly
dependent on the precision and drift of the measurement
equipment (especially the current sensor), the second category
requires the not negligible processing power of the battery
management system (e.g. 32-bit micro-controllers).
E. Insulation monitoring
In order to avoid personal damage, and to guarantee oper-
ational reliability, IT systems in accordance with ISO/FDIS
6469-3:2011-5 [22] are chosen for todays electric vehicles
high voltage DC power systems, which provide two crucial
advantages: an initial insulation error doesnt necessarily cause
a shut-down and also does not give rise to a dangerous touch
voltage which tremendously reduces the threat of electric
shock for humans.
Specific hardware devices, like described in [23], can be
utilised for monitoring the insulation resistance of the IT
system. While these devices are able to independently signal
a critical insulation status, it is also reasonable to feedback
the insulation resistance’s value to the battery management
system. This way, the BMS can detect gradual degradations
of the insulation resistance and can react accordingly.
V. S YSTEM LEVEL / BATTERY PACK DESIGN
Current battery systems for hybrid and battery electric
vehicles typically have operating voltages of 200–800 Volts
[1], [2].
Reducing the voltage level below 60 V would lead to a more
simplistic design of the electrical system and the reduction
of the safety risks for all users. However, in order to achieve
this, the battery pack needs to be redesigned. A battery system
for the peak power demand of an electric vehicle of 360 kW,
but a nominal voltage below 60 V (16 lithium ion cells in
a series connection or less), requires a new approach to the
development and dimensioning of the battery system.
The most obvious advantage is the aspect of safety; a low
operating voltage eliminates one of the main dangers of BEVs.
No special precautions to ensure high-voltage safety are neces-
sary. This particularly and greatly improves the assembly of the
battery system and its service. In most current battery module
designs, the connection of cells is limited by its maximum
desired voltage level. Furthermore, with the complete battery
system remaining below 60 V, no special precautions in respect
of the high voltage are necessary, thus dramatically simplifying
the safety precautions necessary, especially in the event of
a crash or similar accident. Maintenance and recycling are
also simplified, with no special trained personnel and safety
precautions or tools for handling the high-voltage system
necessary.
A possible low voltage level massively influences con-
ception and packaging. Depending on the cell size selected,
a huge quantity of cells have to be connected in parallel.
In a common high-voltage battery system in a small vehicle
utilising standard consumer cells of the type 18650, a mature
cell type that allows for low costs and good availability leads to
a reasonable complexity of the system. Larger battery systems
lead to more complex systems, such as in the Tesla Model S
with 85 kWh in a 74s96p connection [24], [25].
Special challenges to packaging and vehicle integration
arise as a result of the changed ratio of parallel and series
interconnections. If the battery is split up into smaller modules
for more easy packaging and handling, every module contains
one or a series connection of the parallel connected cell blocks.
Having huge parallel connections in a low-voltage vehicle
battery would lead to very large modules. In the given example
system, only seven modules would be possible. This makes
splitting the parallel connection necessary. Such segmentation
into modules has not yet been considered for vehicle usage
and demands in-depth analysis.
Another advantage of large parallel connections is the
higher redundancy of the system. A high ohmic failure of
a cell in one parallel interconnection leads to a performance
reduction in the amount of the share of the cell in its parallel
connection [26], [27]. By using a lower voltage level for the
battery system, whereby a larger amount of energy is stored
in every parallel connected cell strain, this advantage can be
utilised, even for large individual cells.
The high currents, resulting from the low voltage at high
power, demand new ideas for electric conductors and their
interconnections. Changes to the cell terminal contacting tech-
nologies are not required, as the current is split between the
cells interconnected in parallel and the current at one single
cell is no larger than in common systems. Only for very large
single cells would high currents occur at the cell terminals.
Special attention, on the other hand, has to be given to the
design of the current collector joining the terminal-contacts,
the interconnecting of the battery modules and the electric
conductor towards the drivetrain.
The voltage level makes it possible to combine the typical
low-voltage power grid of a car with the drivetrain’s power
system in a BEV, further reducing complexity. Here, a strong
focus would have to be put on the power grid’s stability
because of huge variations in the current demand of the electric
drivetrain.
Electric conductors designed for high currents have to be
selected appropriately and integrated into the car. To limit
material costs, short distances are important. The utilisation
of alternative lower cost materials for the conductors, such
as aluminium instead of copper, becomes of great interest.
Magnetic fields, resulting from the high currents, have to be
considered for integration and safety.
The diameter of an electric conductor needed for certain
currents mainly depends on the thermal load. For further
reduction of the diameter, active cooling, not only of the
battery, but also of the conductors, could be considered.
The design of the electric terminal conductor depends
very much on the quantity of cells in parallel. With many
small cells, and the resulting small currents for every terminal
contact, no new contacting principles need to be utilised.
Thought, on the other hand, has to be put into the current
distribution in the electric conductor connection of the parallel
cells, as the current increases with every cell towards its
connection to the vehicle power conductor.
 Special focus has to be put on the system’s power con-
ductors. The massive currents cannot be handled by common
types of power conductors and connectors. Using copper would
increase the costs way too much, so new approaches have to
be found. One possible approach, besides the utilisation of
aluminium instead of copper, could be to reduce the diameter
necessary for the power conductors by using active cooling
technologies.
The vehicle concept should be developed with a focus
on minimal distances for power conduction from battery to
power electronics and electric motor, to decreased weight and
material by design. From the aspect of the battery pack system,
the main question remains of how to produce a cost and
weight-effective system. Future research and prototypes will
be needed to give a reliable answer.
VI. C OMPARISON OF A 24 V, 48 V AND A 300 V
BATTERY PACK
Hereafter, a battery pack with the following specifications
will be investigated. Voltage levels: 25.9 V, 48.1 V, 303.4 V;
capacity 85 kWh; peak power output 360 kW (for 10 seconds
at most); continuous output: 80 kW.
A. Cell size
In the following a possible cell size, or a serial-parallel
interconnection respectively, will be proposed considering the
efficiency, costs and thermal loss. In figure 5, the thermal losses
of all three mentioned voltage levels at the two power stages
are plotted (partly based on figure 1). The horizontal line marks
the limit of air cooling in a BEV (empirical estimation). The
area below the line can be cooled by air. Above the line, it
is more reasonable to cool the battery pack with water. A
cooling system with air should be the preferred choice due
to the lower costs and weight. As mentioned in chapter VI-C
Tesla Motors is currently using a 96p interconnection. But it
must be considered, that the behaviour of a larger amount
of parallel connected cells needs further investigations. This
is why the recommendation is, not to simply exceeded this
amount of parallel connected cells over a 96p interconnection.
Nevertheless it must be mentioned, that in the following
scenarios, the efficiency can be raised substantially by using a
larger amount of parallel connected cells and therefore, it will
be also mentioned as an alternative.
• 25.9 V (7s): At peak performance (360 kW), a lot
of thermal loss is generated. The majority of the
heat would emerge in the main connectors of the
battery because the current density would be the
highest there. At nominal power (80 kW), based on
figure 5, starting from a 27p interconnection, the loss
falls below 1000 W (label 1, figure 5). As it is not
reasonable to use cells larger than 60 Ah capacity
(decreasing weight capacity density; compare figure
2, the suggestion is to use mature VDA standard cells
(BEV2, 60 Ah [28]) which lead to a 55p (equation 4)
interconnection with a water cooling system, to resist
the peak power demand loss (referring to figure 5,
label 2, 9.8 kW). As mentioned before, the cooling
of the conductors becomes of great interest. Due to
the parallel interconnection, and thus the division of
the current, the cooling of the cells is not critical.
For reasons of safety, precautions and tests have to
be carried out to avoid a cascade failure, in the case
of a thermal runaway.
85000W h
25.9V
⇒ 55p interconnection (4)
60Ah
Thinking about efficiency, it is possible to beat down
the thermal loss under the use of the before addressed
massively parallel interconnection. Assuming, that at
a voltage level of 25.9 V approximately 7104 (men-
tioned in chapter VI-C) cell are used, the thermal loss
will come to 533 W. Using the same amount of cells
for the following voltage levels, this result is also
valid.
• 48.1 V (13s): At nominal power, based on figure 5
label 3 and starting from a 15p interconnection, the
loss falls below 1000 W. Despite the now doubled
voltage level, a 17p interconnection would lead to a
cell size of approximately 100 Ah (equation 5).
85000W h
48.1V
= 103.9Ah (5)
17
Using VDA standard cells (BEV2) as well would
lead to a 30p interconnection but, again, a too greater
thermal loss at peak power of almost 9.7 kW (label
4). To reduce the thermal loss, by increasing to a 55p
interconnection, as previously suggested in the 25.9 V
version, a 32 Ah cell can be used and the thermal loss
at peak power halved (5.3 kW, figure 5, label 5). Using
an air cooling system, this would still be too much, but
utilising the thermal capacity of the battery pack, such
an overload for 10 seconds could be possible without
water cooling (further investigations are necessary).
Despite the smaller battery cells, precautions have to
be taken against a cascade failure. ⇒ 13s30p (60Ah),
water cooling system; or 13s55p (32Ah) air cooling
system
• 303.4 V (82s): Even at peak performance starting
from a 47p interconnection, the thermal loss falls
below 1000 W (figure 5, label 6). Such a battery pack
would be very powerful because it can perform at
peak power for more than 10 seconds. A lightweight
solution with high-energy cells (figure 2) is also
possible, because using standard consumer cells (type
18650, around 3 Ah) makes sense at this voltage
level and therefore the efficiency can be significantly
increased (figure 5, label 7). Safety issues can be
retained very easily with such small cells. This will
lead, according to equation 6, to a 94p interconnection.
85000W h
303.4V
⇒ 94p interconnection (6)
3Ah
Of course, a low-cost version with fewer, but larger,
cells is conceivable. Utilising VDA standard cells
(BEV2) with 60 Ah capacity again will lead to a
5p interconnection. Such a version will not be as
efficient as the one with small cells (figure 5, compare
label 6 and 7 with 8), but the effort in building such
a battery pack will be less. ⇒ 82s5p (60Ah), air

Fig. 5. Thermal loss of a 25.9 V, 48.1 V and 303.4 V battery pack from a
2P to 100P interconnection
cooling system (low cost version); or 82s94p (3Ah),
air cooling system (high power version)
The assessment of the costs for the three different packs
will be made in chapter VI-C, table II.
B. Battery management system
• Electrical Monitoring: As lithium ion battery packs
with 24 V nominal voltage can be realised with only 7
serial cell blocks, one cell monitor is sufficient which
eases the BMS topology tremendously. In fact, only
a single controller is needed to acquire and process
the measurement data from the cell monitor. For the
48 V battery pack concept with a 13s configuration, a
modular BMS topology, as illustrated in figure 4, can
be utilised by just adding another cell monitor without
an own logic unit (micro-controller).
In contrast to the two low voltage battery pack con-
cepts, a more complex BMS topology is necessary
for the 300V pack. Because of its 82s configuration 7
cell monitors are needed. Since this high number of
monitors is not suited for processing with small micro-
controllers, a Master-Slave or distributed architecture
with a more powerful controller needs to be employed.
In terms of current measurement, hall-effect-based
sensors should be used for the two low-voltage battery
pack concepts in order to minimise thermal losses,
since power dissipation at a shunt-based sensor grows
quadratically with current and can go up to 900 W (at
60 mV, 15000 A for the 24 V pack), for example.
• Balancing: As illustrated in table I, active balancing
circuits are bigger/heavier and more expensive than
passive ones. Nevertheless, for the low-voltage bat-
tery packs, the additional weight and costs are much
smaller than for the 300V pack due to the low number
of serial cell blocks. However, before utilising active
balancing, the capacity spreading of the serial cell
blocks would need to be investigated which might be
compensated for by the high number of parallel cells.
For the 300 V pack, and its 82s configuration, passive
balancing is the favoured choice with respect to weight
and costs.
• Thermal Management: The differences, in terms
of the cooling system for the various battery pack
concepts, are discussed in VI-A. However, these dif-
ference only affect the BMS on the software side
(cooling control algorithm) if a universal interface is
chosen for controlling the cooling systems. To define
the extent of cell temperature measurement (only hot-
spot based or individual cell monitoring) and therefore
the number of temperature sensors, detailed data of the
battery pack design would be necessary.
• State Estimation: The question of the best-suited
SOC algorithm for a given battery pack is not an easy
one since it is highly dependent on a multitude of
factors, from cell chemistry, via battery pack design,
to BMS hardware. Nevertheless, the extraordinarily
high currents which accompany a low-voltage battery
pack do limit the current sensor choice to hall-effect
based ones and therefore the accuracy of current
measurement compared to shunt sensors. The SOC
algorithm has to take this disadvantage into account
which restricts the suitability of balancing SOC algo-
rithms and favours model-based ones.
• Insulation Monitoring: As detailed in chapter IV-E,
the requirement of insulation monitoring only arises
with DC systems over 60 V. Therefore, the 24 V and
the 48 V battery packs would not necessarily need to
have this feature and the corresponding hardware, this
reduces BMS complexity and cuts costs dramatically.
C. System design
The voltage level for a battery system massively influences
system design.
• 25.9 V (7s): At peak performance (360 kW), a current
of 13900 Ampere has to be transferred through the
main power conductors of the battery pack. Huge
magnetic fields occurring inside the conductors have
to be considered, when designing the battery system.
At nominal power (80 kW), 3089 Ampere still have
to be conducted generating a great deal of thermal
loss. Adding to the thermal losses in the conductors is
the heat transferred to the connectors from the cells.
Just using massive copper bars for power conduction
is not feasible for an electric car, as weight and costs
would increase too much. Alternative approaches for
connectors, such as active cooling, will have to be
investigated to predict the feasibility of such a system
for car usage.
• 48.1 V (13s): Despite having much less dramatic
currents and thermal losses, in principle the same ideas
and questions as for the 25.9 voltage level have to
be faced. The pack design typically used in todays
electric battery systems is not feasible for such a
system. At exactly what voltage level feasibility can be
guaranteed can only be ascertained by the conducting
of in-depth design, simulation and testing.
• 303.4 V (82s): State-of-the-art technology battery
packs for current electric cars usually have this,
 TABLE II. C OST COMPARISON
[3] G. Pistoia, Lithium-Ion Batteries: Advances and Applications. Elsevier,
Voltage level Costs based Costs Annotation 2014.
on (20 years
[4] M. Brand, “Technische Universität München, Lehrstuhl für elektrische
303.4 V projection)
system Energiespeichertechnik (EES).”
25.9 V 250 % 150 % - very complex water cooling system [5] Q. Wang, P. Ping, X. Zhao, G. Chu, J. Sun, and C. Chen, “Thermal
- water cooled conductors runaway caused fire and explosion of lithium ion battery,” in Journal
- unpredictable problems: no suitable of Power Sources, 2012.
connectors and electric contactor and
other secondary components available, [6] A. J. Rene Groiss, Sicherheitsaspekte beim Testen von Lithium-Ionen
magnetic fields disturbing other con- Batterien.
trollers in the car, weight problems, [7] S. Dr.-Ing. Fuchs, “Verfahren zur parameterbasierten Gewichtsab-
- bad efficiency or untested massive schtzung neuer Fahrzeugkonzepte,” Ph.D. dissertation, TU München,
parallel interconnection
2014.
- new connectors to the battery are
necessary due to the high current [8] L. Shmuel De-Leon Energy. (2014, October) Power sources database.
+ simple BMS topology www.batteriesdatabase.com.
48.1 V 150 % 125 % - more complex cooling system [9] A. F. Martin Junginger, Wilfried van Sark, Technological Learning in
- overload must be tested
- new connectors design to the battery
the Energy Sector: Lessons for Policy, Industry and Science, 2010.
is necessary due to the high current [10] M. Fleckenstein, O. Bohlen, and B. Bäker, “Aging effect of temperature
◦ efficiency must be evaluated with gradients in li-ion cells experimental and simulative investigations and
driving profiles the consequences on thermal battery management,” in Electric Vehicle
+ relative simple BMS topology Symposium 26, 2012.
303.4 V 100 % 100 % - complex BMS topology
+ Well known system [11] N. Omar, M. A. Monem, Y. Firouz, J. Salminen, J. Smekens, O. Hegazy,
+ Well known parts H. Gaulous, G. Mulder, P. Van den Bossche, T. Coosemans, and J. van
Mierlo, “Lithium iron phosphate based battery Assessment of the aging
parameters and development of cycle life model,” in Applied Energy,
2013.
or even higher ranges of voltage. Standardised and
[12] M. Broussely, P. Biensan, F. Bonhomme, P. Blanchard, S. Herreyre,
approved technological solutions and battery system K. Nechev, and R. J. Staniewicz, “Main aging mechanisms in li ion
designs are available. batteries,” in Journal of Power Sources, 2005.
[13] C. Mikolajczak, M. Kahn, K. White, and R. T. Long, Lithium-Ion
Batteries Hazard and Use Assessment - Final Report. The Fire
Protection Research Foundation, 2011.
VII. D ISCUSSION AND C ONCLUSION [14] G. P. Beauregard, “Report of investigation: Hybrids plus plug in hybrid
electric vehicles,” ETEC, Tech. Rep., 2008.
It can therefore be concluded that a high-energy, low- [15] A. Davide, Battery Management Systems. Artech House, 2010.
voltage battery pack has many still unsolved or not considered [16] LTC6804-1/LTC6804-2 Multicell Battery Monitors, Linear Technology,
problems (for example, cooled conductors for over 13 kA, 6 2014.
or a massively parallel connection) in the field of mobile [17] LTC3300-2 Addressable High Efficiency Bidirectional Multicell Battery
applications. Of course, components which can withstand Balancer, Linear Technology, 2013.
such high currents do exist in the field of control cabinets, [18] J. Cao, N. Schofield, and A. Emadi, “Battery balancing methods:
etc., however, these are usually heavy parts which are not A comprehensive review,” in IEEE Vehicle Power and Propulsion
tested against influences which emerge in the field of mobile Conference (VPPC), 2008.
applications (vibration, moisture, acceleration, through to [19] M. Daowd, N. Omar, P. V. D. Bossche, and J. V. Mierlo, “A Review of
shock, active cooling systems, etc.). The emerging thermal Passive and Active Battery Balancing based on MATLAB/Simulink,”
in International Review of Electrical Engineering (I.R.E.E.), 2011.
loss that might occur, especially at 25.9 V (without a massive
[20] K. S. Ng, C.-S. Moo, Y.-P. Chen, and Y.-C. Hsieh, “Enhanced coulomb
parallel connection), are a challenge. More promising are counting method for estimating state-of-charge and state-of-health of
voltages around 48–60 V, but it is very clear that further lithium-ion batteries,” in Applied Energy, 2009.
investigation and research are necessary to assess the full [21] G. Walder, C. Campestrini, M. Lienkamp, and A. Jossen, “Adaptive
potential of such a low-voltage approach. State and Parameter Estimation of Lithium-Ion Batteries based on a
Dual Linear Kalman Filter,” in The Second International Conference
on Technological Advances in Electrical, Electronics and Computer
Engineering, 2014.
ACKNOWLEDGEMENTS [22] “Iso/fdis 6469-3:2011-5 elecelectric propelled road vehicles - safety
This article contains results of term papers, Bachelors specification - part 3: Protection of persons against electric shock.”
and Masters theses from Ralf Köhler. The authors would [23] W. Hofheinz and H. Sellner, “Elektrische Sicherheit bei der Ladung von
Elektrofahrzeugen.”
like to thank them for their contributions. Again, the authors
[24] Tesla Motors, “Tesla Model S - Features and Specs,” 2013. [Online].
would like to thank them for the support. The project Available: http://tinyurl.com/66c4ms9
was independently funded by the Institute of Automotive [25] IAV, “Test-drive,” 11th Symposium: Hybrid and Electric Vehicles,
Technology at the Technische Universität München. Braunschweig, 02 2014.
[26] A. Davide, “Parallel cells vs. strings in parallel,” 07 2011. [Online].
Available: http://tinyurl.com/qab99hw
R EFERENCES [27] D. H. Doughty and A. A. Pesaran, “Vehicle battery safety roadmap
guidance,” NREL Subcontract Report, p. 109, 10 2012.
[1] A. Freialdenhoven, “Stärkung der Wettbewerbsfähigkeit der Automo-
bilindustrie durch Vernetzung von Wissenschaft und Industrie,” Ph.D. [28] D.-I. A. Lamm, “Entwicklungen auf dem Batteriesektor aus Sicht eines
dissertation, RWTH Aachen University, 2009. Automobilherstellers,” 2012.
[2] H. Wallentowitz, A. Freialdenhoven, and I. Olschewski, Strategien zur
Elektrifizierung des Antriebsstranges. Vieweg Verlag Teubner, 2010.