1
2
Li-ion Battery Aging Conscious Intelligent Energy
Management Strategy for Hybrid Electric Buses
Estrategia inteligente basada en envejecimiento de la bater
´
ıa
de litio-ion para la gesti
´
on energ
´
etica de autobuses h
´
ıbridos
el
´
ectricos
Abstract— This paper aims to propose a battery aging
conscious energy management strategy. The initial design
of an energy management strategy is a significant point
to fulfill the efficiency goals in the short term. However,
with aging, the initial conditions may vary. The new
trend of digitalization allows monitoring the operation,
having the possibility to improve the performance of the
initially proposed strategy in the long term. Therefore, a
methodology for updating the energy management strategy
along the bus lifetime is intended to improve the operating
costs and extend the battery lifetime. This methodology is
based on a dynamic programming optimization, tuning
the membership functions in a fuzzy logic control. The
simulation results show a reduction of the operation costs
up to 47% as long as it stands for battery (BT) lifetime
extension of around 2.94%.
Keywords— Energy management, dynamic program-
ming, fuzzy logic, hybrid electric bus, state of health,
energy storage systems.
Resumen— El objetivo de este trabajo es proponer
una estrategia inteligente basada en el envejecimiento
de la bater
´
ıa de litio-ion instalada abordo de veh
´
ıculo
como aplicaci
´
on de gesti
´
on energ
´
etica. El dise
˜
no inicial
de una estrategia de gesti
´
on energ
´
etica (EMS) es un
paso significativo para cumplir los objetivos de eficiencia
en la operaci
´
on a corto plazo. Sin embargo, debido al
envejecimiento de la bater
´
ıa las condiciones iniciales de
la EMS pueden variar. La nueva tendencia hacia la
digitalizaci
´
on permite monitorizar la operaci
´
on, brindando
la posiblidad de mejorar el desempe
˜
no de la estrategia
inicialmente propuesta en el largo plazo. Por lo tanto, se
propone una metodolog
´
ıa para actualizar la EMS con el
objetivo de mejorar los costos de operaci
´
on y extender
la vida
´
util de la bater
´
ıa. La metodolog
´
ıa se basa en una
optimizaci
´
on mediante programaci
´
on din
´
amica para para-
metrizar las funciones de pertenencia de un control difuso.
Los resultados de simulaci
´
on muestran una reducci
´
on en
los costos de operaci
´
on entorno al 47 % junto con una
extensi
´
on de la vida
´
util de la bater
´
ıa de alrededor de
2.94 %.
Palabras Clave— Gesti
´
on energ
´
etica, programaci
´
on
din
´
amica, l
´
ogica difusa, autob
´
us h
´
ıbrido el
´
ectrico, estado
de salud, sistema de almacenamiento de energ
´
ıa.
I. INTRODUCTION
Nowadays, urban transport is undergoing a funda-
mental shift to greener and more sustainable solutions.
This evolution has been driven by two main factors, a
growing awareness of the emitted pollution, especially
in urban areas and the lithium-based batteries (BTs)
price decrease (nearly 79% since 2010 [1]). However,
the implementation of hybrid and electric buses is a
challenging process due to the high investment cost
besides conventional buses.
Some studies pointed out the influence of the BT price
on the buses total cost, reaching values of the 39% [2].
Adding the fact that BTs have a shorter lifetime than
power electronic systems, the BT system is identified as
a bottleneck in the lifetime of the bus. Moreover, that BT
pack replacements are required, this affects significantly
to the operation costs during the vehicle lifetime [3], [4],
increasing the total cost of ownership (TCO). Therefore,
it can be said that the BT lifetime is closely related to
vehicle operation.
Jon Ander López-Ibarra
1,3,♦
, Mattin Lucu
1,3
, Nerea Goitia-Zabaleta
1
, Haizea Gaztañaga
1
, Victor Isaac
Herrera
2,♣
, Haritza Camblong
3,4,♠
1
IKERLAN Technology Research Centre Energy Storage and Management Area Gipuzkoa, Spain
2
Escuela Superior Politécnica de Chimborazo, Facultad de Informática y Electrónica, Riobamba, Ecuador
Email:
♦
jonander.lopez@ikerlan.es,
♣
isaac.herrera@espoch.edu.ec,
♠
aritza.camblong@ehu.eus
3
University of the Basque Country Gipuzkoa, Spain
4
ESTIA Research, France
Fecha de Recepción: 16 – May – 2019 Fecha de Aceptación: 13 – Jun – 2019
Revista Técnico - Cientíca PERSPECTIVAS
Volumen 1, Número 2. (Julio - Dicimbre 2019)
e -ISSN: 2661-6688
3
For bus manufacturers, the developed initial energy
management strategy (EMS) for fulfilling the efficiency
operation goals is a significant point. However, the
conditions used for the initial EMS vary throughout
the bus lifetime. Therefore, an update of the initial
EMS will adjust the EMS to the new situation. For the
correct update of the EMS, the continuous operation
monitoring is needed. This need is fulfilled with the latest
vehicles digitalization trend, which allows the constant
monitoring of the operation. Consequently, having the
possibility to analyze the current operation and take
action to correct it.
The operation information with the needed BT ad-
vance knowledge will allow managing the BT lifetime,
going a step further on the EMS. In this regard, new
techniques for managing the BT aging are needed, since
BT replacements are directly related to the TCO. Con-
sequently, the operation management conscious of the
BT aging will allow improving the TCO further. As a
result, they are making urban mobility electrification a
more attractive and viable solution for investors.
Dealing with hybrid propulsion systems, several EMSs
have been proposed in the literature with a multi-
objective strategy to minimize fuel consumption and at
the same time minimize the capacity loss and extend the
BT lifetime [5], [6]. It has been underlined that these
strategies do not consider the state of health (SOH) of
the BT as an input. The SOH is a piece of valuable
information, which enables the maximization of the BT
lifetime in a long-term scope.
The SOH enables to act to manage the BT lifetime in
a long term scope. In [4], Du J. et. al propose an EMS
for an hybrid electric bus (HEB) with a hybrid energy
storage system (HESS), to minimize the BT degradation
and total costs. However, this minimization process is
based on the power management of the HESS, without
any updates of the EMS within the bus lifetime.
This paper aims to develop a BT aging conscious
intelligent EMS focused on imporving the BT lifetime.
In the first part, a fuzzy logic (FL) algorithm is devel-
oped based on the short term EMS initially designed.
This strategy is tuned by using the optimal operation
obtained by dynamic programming (DP) optimization.
This strategy is tuned based on the optimal operation
obtained by dynamic programming (DP) optimization.
As the initial conditions vary and the BT capacity fades,
a methodology for updating the EMS is proposed, to
maximize the BT aging and improve the operating costs.
Operation costs improvements up to 47% are obtained
beside a rule-based strategy, and a BT lifetime extension
of 2.94% is reached.
II. S
CENARIO OVERVIEW
The scenario analyzed in this paper is based on a
series hybrid electric bus (SHEB) that performs in an
urban route. As shown in Fig. 1, this type of bus
power-train is pulled by an electric motor (EM). In this
configuration, the required power is provided by a genset
(GS, composed by an internal combustion engine (ICE)
coupled to an electric generator) and a BT pack.
The parameters of this scenario are listed in Table I
[7].
Simulation in MATLAB has been performed. For this
simulation, line 28 from Donostia-San Sebastian (Spain)
urban bus route has been used. It is worth to highlight
that the BT used in this SHEB is fast charged after every
round-trip, except for the last round-trip, which is depot
charged. The driving profile is depicted in Fig 2.
III. B
ATTERY AGING CONSCIOUS INTELLIGENT
ENERGY MANAGEMENT STRATEGY
The suitable and efficient power split among the GS
and BT became a complex problem. There are several
factors to take into account to minimize fuel consump-
tion. Furthermore, if BT aging management is included
in the problem, it becomes even more complex. To an-
swer this problem, advanced techniques and knowledge
of the BT and application are needed, as a fair trade
between the fuel consumption minimization and the BT
utilization management has to be determined.
In this context, FL rule-based EMS has been consid-
ered as the technique to deal with the complex problem.
The main reason for using FL has been the capacity of
this technique to adapt to the uneven events during the
real operation, due to the non-linear variable response
depending on the operation conditions [7].
Therefore, a FL based BT conscious EMS has been
proposed. In the following subsections the proposed
EMS FL membership functions, optimization used for
Electric
Generator
Gen-Set
Electric
Motor
DC Bus
DC
AC
Battery
Mechanical Link
DC Link
AC Link
T
Clutch
ICE
DC
DC
AC
DC
Figure 1: Series HEB configuration.
Table I: Scenario approach.
Elements Units Parameters
Electric Motor Power kW 196.5
Genset Power kW 160
Battery Pack
Voltage V 650
Energy kWh 24
Revista Técnico - Cientíca PERSPECTIVAS
Volumen 1, Número 2. (Julio - Dicimbre 2019)
e -ISSN: 2661-6688
4
the tuning and the output are described. In Fig. 3 an
overview of the proposed FL EMS is shown.
A. Fuzzy Logic Membership functions
The proposed FL controller have been based on a
Sugeno Fuzzy Inference System [8], due to the less pro-
cessing needed time. The utilized membership functions
are detailed in Fig. 4. The names of the membership
functions have been defined as follows:
- GS power (
P
GS
(k − 1) [W ]): the input data is the
previous GS power value, to avoid sudden GS power
demands.
- Regeneration capability (
P
Regen
[W ]): the input data
is the difference between the maximum BT charging
power and the regenerated power in the continuous bus,
to maximize the BT regenerative charging.
- DC link power balance (
P
DCLink
[W ]): the input stands
for the power balance of the continuous bus, to evaluate
the power demand
- BT State of Charge (
SOC): the input is the current
SOC of the BT, to evaluate if the BT needs to be charged
or discharged.
B. Fuzzy Logic Tuning
The tuning of the proposed EMS has been based
on both advance application knowledge and an off-line
optimization of the driving profile. The DC link power
balance [W], GS Power [W], and Regeneration capability
[W] membership functions have been tuned based on
the advanced application knowledge. On the contrary,
the SOC [%] has been adjusted based on the optimal
operation obtained from DP optimization.
The predefined driving profile has been optimized
based on DP optimization. The optimization problem is
based on the following cost function (
J), for the fuel
consumption minimization [9]:
J =
N−1
i=0
m
f
(U(i)) · T
s
(1)
where
m
f
· T
s
is the fuel mass consumption at each
time step (
T
s
=1 s), determined by the split factor (U),
within the urban route length (N) [10]. Therefore the
optimized parameter is the split factor.
In the SHEB configuration, the split factor stands for
the division of the power demand (
P
dem
[W ]) among the
GS (
P
GS
[W ]) and the ESS (P
ESS
[W ]).
P
dem
(i)=
P
GS
(i)=P
dem
(i) · (1 − U(i))
P
BT
(i)=P
dem
(i) · U(i)
(2)
To maximize the BT use obtaining the lowest fuel
consumption, the optimization has been designed, to
reach a predefined minimum SOC. This minimum SOC
has been calculated based on the amount of energy that
can be recharged with the available fast charger power
in 2.5 minutes.
From the obtained optimal SOC operation, the min-
imum and maximum SOC values are extracted. These
points are used to set the A (minimum SOC) and B
(maximum SOC) points, as shown in Fig. 4.
C. Fuzzy Logic Output
The output variable of the proposed EMS is the GS
power [W]. In Sugeno Fuzzy Inference System, the
output values are defuzzified by statistic functions [11].
The output statistics function used in this design is the
weighted arithmetic mean value. Each data point has a
degree of contribution, also denominated as weight, on
the weighted arithmetic mean calculus. Consequently,
for calculating the output, weighted values are required.
These weighted values are computed by another mathe-
matical function denominated as the propagation of error,
Eq. (3). In this case, the weight refers to the contribution
of the FL controller rules (
w
r ule
n
) for each GS power
value,
w
GSvalue
.
w
GSvalue
=
w
2
rule
1
+ w
2
rule
2
+ ... + w
2
rule
n
(3)
0
10
20
30
40
50
Go
. . .
Operation [hours]
Back
Fast Charging Zone
Speed Profile
Depot Charging Zone
Go
Back
Go
Back
1
16
+
. . .
24
Speed [km/h]
Figure 2: Driving profile.
Revista Técnico - Cientíca PERSPECTIVAS
Volumen 1, Número 2. (Julio - Dicimbre 2019)
e -ISSN: 2661-6688
5
After obtaining the weighted value, weighted arith-
metic means statistics function can be applied to obtain
GS (
GS
output
) output value. The following formula is
used for the output definition:
GS
output
=
n
i=1
(w
GSi
· GS
i
)
n
i=1
w
GSi
(4)
where,
GS
i
[W ] are the GS constants, defined by n the
number of constants and
w
GSi
represents the weighted
value of each constant.
GS
i
is tuned based on the obtained DP optimal GS
operation values. This tuning is based on extracting the
minimum, quantiles 25, 50 and 75, and the maximum
values.
IV. EMS U
PDATING METHODOLOGY
The complexity of the energy management problem
increases as the conditions used for the initial EMS
design conditions vary. This initial EMS is a significant
point for fulfilling the operation and efficiency goals.
However, the state of the bus with the aging differs,
identifying the BT as a bottleneck in the lifetime of
the bus. For this reason, BT conscious EMS has been
proposed.
The BT aging analysis and estimation is also a
complex task as it has been aforementioned in Sec. I.
This is the main reason for the need of updating the
initial EMS. For the correct update of the EMS, the
continuous operation monitoring is needed. This will
allow to analyze the current operation and take action
to correct it.
Therefore, a methodology for updating the EMS to
correct the SOH of the BT, fulfilling the operation
conditions is presented. The proposed methodology is
shown in Fig. 5. For the initial and subsequent EMS
Fuzzification
μP_GS [-]
SOC [%]
DC_link
P [W]
P [W]
Regen
Rules
Inference
Defuzzification
μP_DC_link [-]
μP_Regen [-]
μSOC [-]
GS
P (k-1) [W]
Tuning-Update
GS
P [W]
GSvalue
w [-]
GS
i
[W]
DP based optimisation
Figure 3: Fuzzy Logic control block diagram.
1
0
-1.5
-1
-0.5 0 0.5
1
x10
5
1.5
DC Link Power Balance [W]
SOC [%]
1
0
0
100
0.5
0.5
Deg. of Membership
Negative
Null
Low
Medium
High
Very Low
Low
Medium
High
1
0
0.5
0
15
x10
4
10
5
Genset Power [W]
Very Low
Low
Medium
High
Very High
1
0
0.5
x10
4
Very Low
Low
Medium
1
0
2
34
5
67
Regeneration capability [W]
B
A
μP_GS [-]
Deg. of MembershipDeg. of MembershipDeg. of Membership
μP_Regen [-]
μSOC [-]
μP_DC_link [-]
Figure 4: Fuzzy Logic membership functions.
designs, two periods have been set, the short and the
long terms subsequently. In the following lines, the
methodology stages are presented in detail.
Stage 1: Driving Profile Operation Optimization
In the first stage, the Short Term EMS design is
developed. Thes initial EMS has been designed for
fulfilling the operation efficiency goals.
Stage 2: Urban bus operation
In this stage, a simulation in MATLAB of the urban
bus real operation driving behavior (described in Sec. II)
is performed. For the real driving operation conditions,
driving disruptions have been considered, such as passen-
ger, auxiliary and traffic disruptions [10]. Additionally,
in this simulation, continuous operations monitoring is
done during 15 days.
Stage 3: Analysis and Decision Maker
This stage consists of analyzing the current aging of
the BT and updating the EMS, to improve the operation
from the initial EMS design. The continuous operation
data gathered in the urban bus operation simulation is
processed, and the BT aging is obtained and analyzed.
Subsequently, to adapt the new EMS to the new condi-
Revista Técnico - Cientíca PERSPECTIVAS
Volumen 1, Número 2. (Julio - Dicimbre 2019)
e -ISSN: 2661-6688
6
Figure 5: EMS updating methodology.
tions, decisions are made. For that purpose, the following
three stages are required:
Stage 3.1: Battery Aging Analysis
For a correct BT aging management, apart from ad-
vanced BT knowledge, an appropriate BT aging model
is needed. BT lifetime estimation is a complex task due
to the dependence on multiple stress factors [12]. The
literature proposes different aging models, which can
be categorized as physiochemical, empirical or semi-
empirical with uneven levels of complexity and accuracy
[12], [13]. The semi-empirical model is the most widely
used method [12]. For this work, a W
¨
ohler curve based
fatigue method has been used.
Firstly, both estimated and real aging curves are com-
pared in a determined evaluation period. This evaluation
period has been set of steps of 5% of the SOH decrease
(
SOH) until the end-of-life (EOL) of the BT is
reached. In other words, BT aging is evaluated in
P
1
95%, P
2
90%, P
3
85% and P
4
80% of SOH. The real
aging estimation (blue curve), reaches a lifetime of
Ψ
.
Consequently, the aging curve varies from the initially
estimated one. In this evaluation points, to extend the BT
lifetime, there is a need for updating the EMS, to reach
Ψ
.
Stage 3.2: Long Term Design
To improve the operation obtained from the initial
EMS, to determine the most convenient operation. This
stage has been divided into two, the driving profile re-
optimization and the techno-economic evaluation.
Driving Profile Re-optimization
These re-optimizations consist of modifying the re-
quested charging power from 50 kW to 150 kW by a step
of 10 kW. In this way, as the EMS is designed based on
the available energy to be recharged (explained in Sec.
III-B), the BT utilization will be reduced as the charging
power decreases. Consequently, the strategy will increase
the GS, improving the BT lifetime. On the contrary, if
the BT aging is considered to be as designed, the BT
utilization is increased, re-tuning the SOC membership
function.
Techno-Economic Evaluation
At this point, the obtained optimization results are
techno-economically analyzed. In this outline, the fol-
lowing operating costs are taken into account: fuel con-
sumption cost, BT cost, and BT charging energy cost,
Eqs. (5), (6) and (6), respectively.
F
cost
=
p
k=1
(mf
ICE
(k) · k
cs
· C
L,fuel
)
ρ
fuel
[e/day] (5)
where
ρ
fuel
is the fuel volumetric density [kg/l], k
cs
is
the global factor to cold starts [-] and
C
L,fuel
[e/l] is
fuel, in this case diesel, cost per liter.
BT
cost
=
C
BT
· E
BT
· n
L · O
·
Y
deg
Y
deg,current
[e/day] (6)
where
C
BT
is the BT cost in [e/kWh], E
BT
is the BT
size [kWh], n are the number of replacements,
L is the
lifetime operation of the BT [years],
O is the yearly
operation in [day/years],
Y
deg
are the predicted years and
Y
deg,current
are the current years.
C
el
= C
e,fix
+ C
e,var
[e/day]
(7)
where
C
el,f ix
is the fix power cost in [e/day] and C
e,var
is the consumed energy cost in [e/day]. Both costs are
calculated as follows:
C
el,f ix
= C
kW
· P
cha
[e/day]
(8)
C
el,var
= C
kW h
· E
cha
[e/day]
(9)
Revista Técnico - Cientíca PERSPECTIVAS
Volumen 1, Número 2. (Julio - Dicimbre 2019)
e -ISSN: 2661-6688
7
where, C
kW
represents the cost of the charging power,
P
cha
the charging infrastructure power, C
kW h
the energy
cost and
E
cha
the consumed energy from the grid.
The objective of this analysis is to reduce fuel con-
sumption and extend BT lifetime. According to the devi-
ation of the SOH and the techno-economical boundaries,
an optimization is selected.
Stage 3.3: Decision Maker
In this sub-stage, based on the obtained results form
the re-optimization and techno-economic analysis, the
most suitable operation will be determined for the long
term EMS design. For this, a fuel consumption limit
has been established, which limits the obtained opti-
mizations results. Therefore, based on the performed
techno-economical analysis, the most suitable operation,
according to the BT aging, is chosen.
Stage 4 EMS update
In this Long Term EMS, the selected SOH optimization
via DP is used for tuning off-line FL strategy, until
the next evaluation period. In the same way as short
term EMS, SOC input, and GS output are the adjusted
variables. In this EMS operation targets are modified, to
correct the BT degradation, obtaining the re-evaluated
BT aging curve (green curve, Fig 5). The re-evaluated
BT aging maximizes the BT lifetime until point A”.
After this design, the EMS is updated to Long Term
EMS, and the operation is simulated in MATLAB until
the next evaluation period.
V. R
ESULTS AND ANALYSIS
To validate the proposed BT conscious EMS and EMS
updating methodology, a simulation of the presented
SHEB in Section II has been carried out.
In this section, the obtained results are presented. First,
the proposed EMS evaluation has been performed. Then,
the obtained results from the short and long terms are
presented.
A. Proposed EMS evaluation
In Fig. 6 the proposed EMS comparison is shown. The
proposed EMS has been first compared against a DP
off-line optimization and with the proposed rule-based
non-adaptive EMS named LUT (Look up table based
EMS) [10]. The operation costs of the proposed approach
increase compared to the DP global optimization which
has a value of 31.5%. Another comparison is done
against the LUT strategy obtaining an increase of 63.7%.
Additionally, the obtained operation costs improvement
of the proposed EMS compared to the LUT EMS is up to
47%. Indeed, to ensure the adaptability of the proposed
approach, a comparison with the maximum demand of
auxiliaries and number of passengers on the bus, against
High Price
Scenario
Low Price
Scenario
- 63.7%
- 47.0%
- 44.3%
- 31.5%
+ 5.1%
- 21.2%
Figure 6: EMS comparison.
the LUT EMS has been done resulting on a decrease of
44.3% of operating costs for the proposed EMS.
For further evaluation and to ensure the stability of
the proposed EMS, the worst and best cases have been
assessed. The obtained result in the best case, i.e. the bus
without passengers and with the minimum of auxiliary
consumption, has been a decrease of 21.2%. On the
contrary, the worst case has been the evaluation with
the maximum of passengers and maximum auxiliary
consumption, obtaining an increase of 5.1%.
B. Short and long term evaluations
In this subsection, the proposed methodology for up-
dating the EMS has been analyzed. For this comparison,
the non-updated EMS has been defined as the short term
ST, and the updated one as the long term LT.
In Fig. 7 the obtained results regarding the BT aging
are shown. The BT aging extension for the updated EMS
2.94% against the non-updated EMS. It is noteworthy the
BT aging lifetime of the ST, as it does not fulfill the vehi-
cle end-of-life (EOL) planned years. On the contrary, the
updated LT EMS overcomes the expected vehicle EOL
point. Concerning the BT aging deviation concerning the
vehicle lifespan, two scenarios are analyzed. In a first
analysis, a unique BT replacement is considered in the
whole vehicle life, and the vehicle not reaching EOL is
removed before the planned date. The second case study,
80
85
90
95
100
SOH [%]
0 0.2 0.4 0.6 0.8
1
Battery aging [pu]
Updated strategy
Non-updated
Evaluation points
Vehicle end-of-
life
strategy
P
1
P
2
P
3
P
4
0.97
Figure 7: Corrected BT aging.
Revista Técnico - Cientíca PERSPECTIVAS
Volumen 1, Número 2. (Julio - Dicimbre 2019)
e -ISSN: 2661-6688
8
Operation Costs [€/day]
4 eval
th
3 eval
rd
2 eval
nd
1 eval
st
Evaluation Periods
2.5% 1.3% 0.5%
Figure 8: ST and LT comparison with a single replace-
ment.
Operation Costs [€/day]
4 eval
th
3 eval
rd
2 eval
nd
1 eval
st
Evaluation Periods
3.0% 4.8% 4.1%
Figure 9: ST and LT comparison with two replacements
for the bus not reaching the vehicle EOL point, a second
BT replacement has been considered.
Fig. 8 shows the operation costs obtained on each
evaluation point for each EMS, the ST, and the LT. The
ST EMS shows lower operation costs than LT EMS,
reaching values up to 2.5%.
By contrast, as shown in Fig. 9, if the vehicle requires
two BT replacements, the opposite happens. A decrease
up to 4.8% in the LT EMS is obtained in the operation
costs.
VI. C
ONCLUSIONS
In this paper, a BT aging conscious intelligent energy
management strategy was presented focused on BT life-
time maximization. For the validation of the proposed
BT conscious EMS and EMS update methodology, a
simulation as described in Section II is carried out.
The obtained results against a non-adaptive EMS has
been up to 47% of operation costs decrease. For the
stability evaluation, the worst and best cases have been
evaluated, obtaining an increase of the operation costs
up to 5.1% and a decrease up to 21.2% respectively.
The BT aging extension has been of 2.94%, compared
to the non-updated EMS, reaching the planned bus
EOL. However, in the case of the non-updated EMS,
the scheduled EOL date is not reached. In this regard,
the two analyzed possibilities are to remove the bus
before the planned time or to replace the BT. In the
case of removing the bus, the obtained results for the
ST operating cost have a decrease up to 2.5%. On the
contrary, considering to replace the BT, the operation
costs of the LT EMS decrease up to a 4.8%, since the
BT lifetime overcomes the planned EOL date and an
only BT replacement is needed. It is worth to highlight
that the penalization for not reaching the bus EOL was
not taken into account for the scenario of removing the
bus.
On the ongoing research, an improved and self-
adaptive [14] BT aging estimation model will be im-
plemented to the BT aging conscious EMS.
A
CKNOWLEDGMENT
This work was partially supported by the Gipuzkoa
Provincial Council under Project On-Mobility (Regional
Program Red Guipuzcoana de Ciencia, Technology and
Innovation).
R
EFERENCES
[1] The latest bull case for electric cars: the cheapest batteries ever.
Available:https://www.bloomberg.com/news /articles/2017-12-
05/latest-bull-case-forelectric- cars-the-cheapest-batteries-ever,
05-12-2017 [Accessed: 26-04-2018].
[2] Z. Bi, L. Song, R. De Kleine, C. C. Mi, and G. A.
Keoleian, “Plug-in vs. wireless charging: Life cycle energy and
greenhouse gas emissions for an electric bus system,” Applied
Energy, vol. 146, no. February, pp. 11–19, 2015. [Online].
Available: http://dx.doi.org/10.1016/j.apenergy.2015.02.031
[3] R. Dufo-L
´
opez, “Optimisation of size and control of grid-
connected storage under real time electricity pricing condi-
tions,” Applied Energy, vol. 140, pp. 395 – 408, 2015.
[4] J. Du, X. Zhang, T. Wang, Z. Song, X. Yang,
H. Wang, M. Ouyang, and X. Wu, “Battery degradation
minimization oriented energy management strategy for plug-
in hybrid electric bus with multi-energy storage system,”
Energy, vol. 165, pp. 153–163, 2018. [Online]. Available:
https://doi.org/10.1016/j.energy.2018.09.084
[5] L. Tang, G. Rizzoni, and S. Onori, “Energy management
strategy for HEVs including battery life optimization,” IEEE
Transactions on Transportation Electrification, vol. 1, no. 3,
pp. 211–222, 2015.
[6] L. Tang and G. Rizzoni, “Energy management strategy in-
cluding battery life optimization for a HEV with a CVT,”
2016 IEEE Transportation Electrification Conference and Expo,
Asia-Pacific, ITEC Asia-Pacific 2016, pp. 549–554, 2016.
[7] V. I. Herrera, A. Milo, H. Gazta
˜
naga, J. Ramos, and H. Camb-
long, “Adaptive and non-adaptive strategies for optimal energy
management and sizing of a dual storage system in a hybrid
electric bus,” IEEE Transactions on Intelligent Transportation
Systems, pp. 1–13, 2018.
Revista Técnico - Cientíca PERSPECTIVAS
Volumen 1, Número 2. (Julio - Dicimbre 2019)
e -ISSN: 2661-6688
9
[8] T. Takagi and M. Sugeno, “Fuzzy Identification of Systems and
Its Applications to Modeling and Control,” IEEE Transactions
on Systems, Man, and Cybernetics, vol. SMC-15, no. 1, pp.
116–132, 1985.
[9] O. Sundstr
¨
om and L. Guzzella, “A Generic Dynamic Program-
ming Matlab Function,” 18th IEEE International Conference
on Control Applications, Saint Petersburg, Russia, no. 7, pp.
1625–1630, 2009.
[10] J. A. Lopez, V. I. Herrera, A. Milo, and H. Gazta
˜
naga,
“Energy Management improvement based on Fleet
Learning for Hybrid Electric Buses,” in IEEE Vehicle
Power and Propulsion Conference (VPPC). Chicago, IL,
USA, USA: IEEE, 2018, pp. 2–7. [Online]. Available:
https://ieeexplore.ieee.org/document/8605025
[11] Mathworks. Sugeno fuzzy interference system. Available:
https://es.mathworks.com/help/fuzzy/sugfis.html [Accessed: 05-
12-2018].
[12] M. Dubarry, N. Qin, and P. Brooker, “Calendar aging
of commercial Li-ion cells of different chemistries
– A review,” Current Opinion in Electrochemistry,
vol. 9, pp. 106–113, 2018. [Online]. Available:
https://doi.org/10.1016/j.coelec.2018.05.023
[13] D. U. Sauer and H. Wenzl, “Comparison of different approaches
for lifetime prediction of electrochemical systems-Using lead-
acid batteries as example,” Journal of Power Sources, vol. 176,
no. 2, pp. 534–546, 2008.
[14] M. Lucu, E. Martinez-Laserna, I. Gandiaga, and
H. Camblong, “A critical review on self-adaptive li-ion
battery ageing models,” Journal of Power Sources,
vol. 401, pp. 85 – 101, 2018. [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S0378775318
309297
Revista Técnico - Cientíca PERSPECTIVAS
Volumen 1, Número 2. (Julio - Dicimbre 2019)
e -ISSN: 2661-6688
