Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 15, NO.

3, JUNE 2014 1193

Comparison of Three Electrochemical Energy


Buffers Applied to a Hybrid Bus Powertrain
With Simultaneous Optimal Sizing
and Energy Management
Xiaosong Hu, Member, IEEE, Nikolce Murgovski, Lars Mårdh Johannesson, Member, IEEE, and
Bo Egardt, Fellow, IEEE

Abstract—This paper comparatively examines three different solutions [1]–[3]. It has been projected that electrifying the
electrochemical energy storage systems (ESSs), i.e., a Li-ion bat- whole transportation system could reduce energy dissipation
tery pack, a supercapacitor pack, and a dual buffer, for a hy- by 80% of the current level [4]. The use of electricity is able
brid bus powertrain operated in Gothenburg, Sweden. Existing
studies focus on comparing these ESSs, in terms of either general to diversify the power sources of vehicles, thus downsizing or
attributes (e.g., energy density and power density) or their impli- even getting rid of low-efficiency internal combustion engines
cations to the fuel economy of hybrid vehicles with a heuristic/ (ICEs), while facilitating the growth of renewable energy in the
nonoptimal ESS size and power management strategy. This paper power sector, e.g., wind and solar energy sources and hydro-
adds four original contributions to the related literature. First, the electric power [5]. A good interaction between the transporta-
three ESSs are compared in a framework of simultaneous optimal
ESS sizing and energy management, where the ESSs can serve tion and power sectors is therefore anticipated to accomplish a
the powertrain in the most cost-effective manner. Second, convex sustainable energy future [6].
optimization is used to implement the framework, which allows the Two key technologies accelerating an evolution toward elec-
hybrid powertrain designers/integrators to rapidly and optimally trified transportation are hybrid electric vehicles (HEVs) and
perform integrated ESS selection, sizing, and power manage- plug-in HEVs (PHEVs). HEVs/PHEVs are attracting increas-
ment. Third, both hybrid electric vehicle (HEV) and plug-in HEV
(PHEV) scenarios for the powertrain are considered, in order to ing attention from automotive industry and academia, owing to
systematically examine how different the ESS requirements are better fuel economy and lower exhaust emissions in comparison
for HEV and PHEV applications. Finally, a sensitivity analysis is to conventional ICE vehicles [7]–[12]. The performance of
carried out to evaluate how price variations of the onboard energy HEVs/PHEVs is affected by a multitude of factors, in which the
carriers affect the results and conclusions. characteristics, sizing, and control of the energy storage system
Index Terms—Convex optimization, electrified vehicle, energy (ESS, onboard electricity carrier) are instrumental. For exam-
management strategy, energy storage, optimal sizing. ple, the fuel consumption, recuperation efficiency, and drivabil-
ity of HEVs/PHEVs highly depend on the specific power and
I. I NTRODUCTION energy of their ESSs [13]–[15]. Additional characteristics of
ESS are lifetime, safety, cost, etc. [16], [17].
T HE current transportation system is heavily dependent
on fossil fuels, resulting in serious concerns on energy
and economic sustainability, as well as environmental impact
Several survey papers have compared the characteristics of
prevalent vehicular ESSs, including batteries, supercapacitors,
in densely populated areas. Endeavors are being actively un- flywheels, and fuel cells [18]–[21]. It has been demonstrated
dertaken to alleviate this dependence, in which transportation that the electrochemical power sources, i.e., battery and su-
electrification has been recognized as one of the most promising percapacitor technologies, possess better overall performance
than flywheels, thus being the most common options for current
electrified vehicles. Although the fuel cell, another electro-
Manuscript received July 22, 2013; revised October 16, 2013; accepted chemical system, is a good choice of vehicle prime mover, the
December 4, 2013. Date of publication January 9, 2014; date of current version
May 30, 2014. This work was supported in part by the Swedish Energy Agency
long time constant restricts its usage as an ESS [21]. More
and in part by the Swedish Hybrid Vehicle Center. The Associate Editor for this specifically, Li-ion batteries have been perceived as one of the
paper was L. Li. most promising options in the battery group because of its
X. Hu, N. Murgovski, and B. Egardt are with the Department of Sig-
nals and Systems, Chalmers University of Technology, 412 96 Gothenburg, significant advantage in energy density. Compared with Li-ion
Sweden (e-mail: xiaosong@chalmers.se; nikolce.murgovski@chalmers.se; batteries, supercapacitors typically have higher power density
bo.egardt@chalmers.se). and longer lifespan, enabling a faster response to vehicle power
L. M. Johannesson is with the Department of Signals and Systems, Chalmers
University of Technology, 412 96 Gothenburg, Sweden and also with the Vik- demand, whereas their energy density is much lower [22], [23].
toria Swedish ICT, 417 56 Gothenburg, Sweden (e-mail: larsjo@chalmers.se; A dual energy buffer combining two electrochemical power
lars.johannesson@viktoria.se). sources with complementary characteristics, such as a Li-ion
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. battery pack and a supercapacitor pack, has been also proposed
Digital Object Identifier 10.1109/TITS.2013.2294675 and analyzed [19], [21], [24]–[26]. Superior performance could

1524-9050 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1194 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 15, NO. 3, JUNE 2014

be expected for this type of hybrid power source, compensating


the individual shortcomings of the pure battery and the pure
supercapacitor ESSs.
The aforementioned literature [18]–[26] presented a general
description of primary attributes of vehicular ESSs. However,
it did not quantify and compare ESSs’ impacts on the per-
formance (e.g., fuel consumption and economic viability) of
HEVs/PHEVs through a system-level powertrain simulation
and assessment. A vehicle-level energy management strategy,
coordinating the power allocation between the prime mover
(e.g., ICE) and the ESS, also plays an important role in the
efficacy and potential of ESS utilization. A Li-ion battery and a
supercapacitor stack were compared in a simulation of HEV
powertrains in the commercial electrified-vehicle-simulation Fig. 1. Bus powertrain configuration.
platform, Advanced Vehicle Simulator (ADVISOR) [27]. The
results showed that both the battery and the supercapacitor conducted in a framework of simultaneous optimal ESS sizing
could be used in HEVs to gain fuel economy improvement and vehicle energy management, in which the sum of operating
relative to ICE cars, while meeting the requirement of perfor- cost (i.e., fuel/electricity consumption) and ESS cost for the bus
mance and drivability. A comparison among a Li-ion battery, is minimized. The ESS is, hence, able to serve the hybrid pow-
a supercapacitor stack, and a dual energy buffer combining ertrain in the most cost-effective manner, which is of greatest
the former two has been also conducted for both HEVs and concern to the bus operators. Second, to rapidly and efficiently
PHEVs in ADVISOR environment [28]. The study indicated implement the framework, convex optimization is employed.
that the dual energy buffer could lead to the best fuel savings in Its effectiveness and computational advantages over DP and
both scenarios, while reducing the battery stress with the load- evolutionary algorithms for optimal sizing and control of hybrid
leveling assistance of the supercapacitor stack. Additionally, powertrains have been demonstrated in our prior work [17],
the fuel economy and ESS cost of a fuel-cell-based HEV were [38]–[41]. The proposed convex-programming-based frame-
compared in a MATLAB-based vehicle simulator for concepts work can provide the hybrid powertrain developers/integrators
consisting of a high-speed flywheel, a Li-ion battery, and a with a new efficient tool for integrated ESS selection, sizing,
supercapacitor system [29]. These studies, however, considered and power management at the initial exploratory stage of
rule-based vehicle energy controls that cannot realize the opti- vehicle design. Third, both HEV and PHEV configurations
mal power split between multiple energy sources for minimum for the bus powertrain are considered while comparing the
fuel consumption. The consequent nonoptimal ESS power man- three different ESSs. This makes it possible to systematically
agement may not sufficiently exploit the ESS potential [15], examine how different the requirements on the ESS are for
[30]–[33]. This has also been concluded in [8], [9], [11], [17], HEV and PHEV applications. Finally, a sensitivity analysis of
and [30]–[34], by showing the advantages of optimization- the result is performed with respect to the prices of fuel, battery,
based energy management strategies over the conventional and supercapacitor. Although the aforementioned contributions
rule-based strategies. For example, in [34], the optimal energy are made specifically for a series hybrid bus powertrain, the
management strategy has been obtained by dynamic program- proposed framework and implementation approach are readily
ming (DP), in order to compare fuel consumption, battery extendible to other topologies.
lifetime, and hybridization cost of a parallel HEV with either The remainder of the paper is organized as follows. Section II
a single Li-ion battery or a dual energy buffer. The dual buffer details the bus line and the powertrain model; the convex-
was found advantageous, and it was argued that the method- programming-based framework of simultaneous optimal ESS
ology considering the optimal energy control could lead to a sizing and energy management is introduced in Section III;
relatively fair comparison. the comparative results for both HEV and PHEV scenarios are
An additional critical factor that most of the foregoing papers discussed in Section IV; a sensitivity analysis of the comparison
did not consider is the influence of ESSs’ size on the per- result is described in Section V; limitations and future work
formance and cost of HEVs/PHEVs, when comparing various are briefly discussed in Section VI followed by conclusions
ESSs. First insights have been given in [36]–[39], where it presented in Section VII.
has been demonstrated that the optimal ESS sizing is closely
coupled with the energy management strategy. These studies
II. B US L INE AND P OWERTRAIN M ODEL
implied that a framework of simultaneous optimal sizing and
energy management is needed to pursue the “truly” optimal A series hybrid bus powertrain is considered [17], [38], as
ESS size and vehicle energy control. shown in Fig. 1. The bus traction is provided by an electric
This paper presents a comparative study of three electro- motor (EM) with a power rating of 220 kW. The two power
chemical ESSs (i.e., a Li-ion battery pack, a supercapacitor sources are an ESS and a 180-kW engine-generator unit (EGU).
pack, and a dual buffer) for a hybrid bus powertrain operated in An HEV or PHEV operation of the series hybrid drivetrain
Gothenburg, Sweden. This study adds four important original architecture can be implemented, depending on the ESS size
contributions to the related literature. First, the comparison is and the accessibility of grid electricity. The demanded velocity
HU et al.: THREE ELECTROCHEMICAL ENERGY BUFFERS APPLIED TO A HYBRID BUS POWERTRAIN 1195

Fig. 3. EGU and EM efficiencies. (a) EGU efficiency. (b) EM efficiency.

Fig. 2. Bus line in Gothenburg, Sweden. (a) Speed. (b) Road slope.

and road gradient for the bus, i.e., a realistic bus line in
Gothenburg, Sweden, are shown in Fig. 2. In order to ensure a
quick simulation of the complete driving cycle with reasonable
Fig. 4. Pack configuration.
accuracy, the quasi-static modeling methodology is employed
to model the powertrain [31]. The power demand at the wheels
is interpreted as an angular velocity ω(t) and a torque Tv (t) on with ai ≥ 0, i ∈ {0, 2}. The underlying assumption is that
the shaft between the EM and the final drive, which is based on the EGU operates along the optimal operating line, which can
the longitudinal vehicle dynamics (see [40] for more details). be realized by a local EGU controller [33]. The quadratic
The EM with a delivered torque T (t) is able to both sat- loss functions, i.e., (3) and (4), are deliberately employed to
isfy high torque demand for propelling and allow regenerative preserve the optimization problem convexity (see Section III).
braking. However, when either its torque threshold Tmin (ω(t)) The effectiveness of convexifying these energy conversion units
or the ESS charge limit is violated, conventional dissipative was elaborated in [17], [38], and [39]. The binary term e(t)
braking Tbrk (t) will act as a complement. The torque and power is used to eliminate the idling loss a2 , while shutting off the
balance equations are represented as engine, which is herein decided by
T (t) = Tv (t) − Tbrk (t)  ∗
1, Tv (t)ω(t) ≥ Pon
Pm (t) + Ploss,m (t) + Pau (t) = Pess (t) + P egu (t) (1) e(t) = (5)
0, otherwise.
where Pm is the EM power applied to the final drive computed by
It is meant to switch on the engine if the bus power demand ex-
Pm (t) = T (t)ω(t) (2) ∗ ∗
ceeds a threshold Pon . The best Pon is determined by iteratively
and the EM loss Ploss,m is delineated by a quadratic function of solving the convex optimization problem, given different Pon
torque values in an outer loop. The detailed search routine is presented
in [38] and [40], in which it has been demonstrated that the
Ploss,m = b0 (ω)T 2 + b1 (ω)T + b2 (ω) (3) heuristic technique guarantees a solution close to the global op-
timum. The efficiencies of the EM and EGU are given in Fig. 3.
where bi (i = 0, 1, 2) are nonnegative speed-dependent coeffi-
We consider an ESS consisting of one or two electrochem-
cients. Pau , Pess , and P egu denote the auxiliary, ESS, and EGU
ical buffers, where the buffer is a lithium-ion battery or a
powers, respectively. Without loss of a fair comparison among
supercapacitor pack. Each pack consists of strings connected in
ESSs, as in [17] and [38], and for simplicity, the losses of power
parallel, with each string containing the same number of cells
electronics are neglected because they are considerably more
connected in series, as illustrated in Fig. 4. By modeling the cell
efficient than the key onboard power sources. Pau is assumed
as an open-circuit voltage, i.e., uj (t), in series with a resistor
to be constant.
Rj , the power at the pack terminals can be calculated by
The fossil fuel (diesel) power is characterized by
 
Pf (t) = a0 P 2egu (t) + (a1 + 1)P egu (t) + a2 e(t) (4) Ptj (t) = uj (t)ij (t) − Rj i2j (t) nj (6)
1196 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 15, NO. 3, JUNE 2014

which is a concave function of P j (t), E j (t), and nj , provided


that E j (t) and nj are nonnegative real values. When sizing
the ESS, this study investigates the optimal power ratings
and energy capacity of the individual packs as nonquantized
values, which is achieved by relaxing nj to a real value. This
makes it possible to lower the dependence on premanufactured
cells. Instead, the focus is kept on the battery/supercapacitor
technology, assuming that cells can be fabricated and assembled
in accordance with the desired (optimized) pack power and
energy content.
Several constraints are imposed on the buffer pack, including
limits on the cell current (iminj and imaxj ), limits on the
allowed SOC range (socminj and socmaxj ), ESS energy conser-
vation for HEV, and ESS energy depletion for PHEV. Including
the cell dynamics, all the constraints can be written as convex
Fig. 5. Battery open-circuit voltage versus SOC. with respect to P j (t), E j (t), and nj below

where ij (t) is the cell current, nj is the total number of cells, ⎪
⎪ Ė j (t) = −P j (t)

⎪ 
and the subscript j ∈ {b, c} denotes a battery or a superca- ⎪


⎪ i n
2E j (t)
+ u 2 n
pacitor, respectively. Then, the power of a dual-buffer system ⎪

minj j Cj 0j j

⎪ 
can be readily represented as Pess (t) = Ptb (t) + Ptc (t). The ⎪


⎨ ≤ P j (t) ≤ imaxj nj
2E j (t)
+ u20j nj
discharge power is assumed to be positive, by convention. Cj
The battery cell’s open-circuit voltage is given as a nonlinear C j nj  2  (10)

⎪ u (soc ) − u 2
function of state of charge (SOC), as illustrated in Fig. 5. ⎪
⎪ 2 j minj
0j


⎪ C n
≤ E j (t) ≤ j2 j u2j (socmaxj ) − u20j
However, throughout this paper, we will use an approximated ⎪



affine voltage–SOC model ⎪

E j (t0 ) = E j (tf ), (HEV case)


⎪ E
⎩ j f
(t ) = E j (socminj ), (PHEV case)
Qb j ∈ {b, c}, ∀ t ∈ [t0 , tf ]
ub (t) = soc(t) + u0b (7)
Cb
where t0 and tf are the initial and final times of the driving
which gives a good fit within the allowed SOC range. In (7), cycle, respectively. For further details on the convex modeling
Qb is the cell capacity in ampere-seconds, and Cb and u0b steps, interested readers are referred to [42].
are coefficients in farads and volts, respectively. It has been
certified that the affine approximation will lead to a convex bat-
III. S IMULTANEOUS O PTIMAL ESS S IZING AND E NERGY
tery model, in terms of pack energy E b (t) and terminal power
M ANAGEMENT VIA C ONVEX P ROGRAMMING
Ptb (t) [42]. Here, we briefly repeat the convex modeling steps,
but the internal pack power (without considering loss) P b (t) = A. Convex Modeling
nb ub (t)ib (t) is used as an optimization variable, instead of
A convex programming problem is formulated as follows:
Ptb (t). Hence, the battery pack energy can be expressed as
minimize F (x)

soc(t)
subject to fi (x) ≤ 0, i = 1, . . . , n
E b (t) = nb Qb ub (t) d soc(t)
hl (x) = 0, l = 1, . . . , q
0
x∈ℵ (11)
 
soc(t) 
Qb
= nb Qb soc(t) + u0b d soc(t) where ℵ ∈ Rn is a convex set, F (x) and fi (x) are convex
Cb
functions, and hl (x) are affine functions of the optimization
0 
Qb 2
vector x [43]. Here, we take the dual buffer as an example
= n b Qb soc (t) + u0b soc(t) to show how a convex optimization problem for ESS sizing
2Cb
nb Cb  2  and energy management can be formulated, since those for the
= ub (t) − u20b . (8) battery-only ESS and the supercapacitor-only ESS can be easily
2
obtained as special cases. The optimization variables are P b (t),
The supercapacitor pack energy can be obtained in a similar E b (t), nb , P c (t), E c (t), nc , P egu (t), and T (t); the con-
way by replacing the subscript b with c in (8), and using straints are the bus power balance (1), the ESS constraints (10),
u0c = 0. Then, by expressing the open-circuit voltage from (8) the EM torque bounds, the EGU power bounds, nb ≥ 0, and
as a function of energy, the terminal pack power (6) can be nc ≥ 0. Relaxing the equalities (1) to inequalities preserves
written as convexity while not qualitatively changing the problem, i.e.,
Rj Cj P 2j (t) T (t) ≥ Tv (t)
Ptj (t) = P j (t)− , j ∈ {b, c}, u0c = 0 (9)
2E j (t)+u20j Cj nj Pm (t) + Ploss,m (t) + Pau (t) ≤ Pess (t) + P egu (t) (12)
HU et al.: THREE ELECTROCHEMICAL ENERGY BUFFERS APPLIED TO A HYBRID BUS POWERTRAIN 1197

where TABLE I
C ONVEX -O PTIMIZATION -BASED ESS S IZING
AND E NERGY M ANAGEMENT
Pess (t) = P b (t) + P c (t) − Rb i2b (t)nb − Rc i2c (t)nc . (13)

As in [17] and [38]–[40], the convex objective function F (x),


which is of greatest interest to bus operators, is formulated to
minimize a summation of the diesel/electricity cost and the ESS
cost, i.e.,
tf
cf ce
F = Pf (t) dt + cbe nb + cce nc +
Cf 3.6 × 106 ηgrid
t0

× (E b (t0 ) + E c (t0 ) − E b (tf ) − E c (tf )) (14)

where cf is the diesel price per liter, Cf is the lower heating


value of diesel in joules per liter, ce is the electricity price
per kilowatthour, and ηgrid is the charging efficiency. The
recharging time and power levels are not explicitly considered TABLE II
in the study. As in [44] and [45], a constant charging efficiency L I - ION C ELL S PECIFICATION
is used, and all the three ESSs are assumed to have the same
efficiency. The equivalent cost of the ESS cje is the cycle-
normalized ESS cost (i.e., the purchasing cost times the per-
centage of the driving-cycle length with respect to the total bus
mileage in service). The ESS is paid with 5% yearly interest
rate in the period of bus operation, thus yielding
Ld
cje = 1.075cj Enj mj (15)
S
where cj is the ESS price per kilowatthour (including packaging
and circuitry), Enj is the specific energy in kilowatthours per
kilogram, and mj is the cell mass in kilograms. The driving-
cycle length and the total bus mileage (in kilometers) are
denoted by Ld and S, respectively. Here, the underlying as-
sumption is that there is no ESS replacement during the two-
year operation (100 000 km, the bus operation period), as in
our prior work [41]. Typically, there is no concern over the
supercapacitor lifetime in vehicular applications. With contin- IV. C OMPARATIVE R ESULTS
ually ameliorated chemistry and control, the battery durability
A. ESS Specifications and Main Vehicle Parameters
is also probably sufficient for automotive use [46], [47]. Battery
replacement was often neglected in similar studies, such as [17], The Li-ion battery (Saft VL 45 E) and the supercapacitor
[29], and [36]–[42], for component sizing and energy controls (Maxwell BCAP 2000) for hybrid and electric vehicles are
of hybrid propulsion systems at the initial exploratory stage. considered for this study. Since Saft and Maxwell are well-
known Li-ion battery and supercapacitor manufacturers for
transportation applications, respectively, their products can be
B. Framework of Simultaneous Optimal ESS Sizing and
considered as typical representatives of electrochemical cell
Energy Management
technologies. The main specifications of the Li-ion battery
The convex programming algorithm for simultaneous op- and the supercapacitor are summarized in Tables II and III,
timal ESS sizing and energy management of the hybrid bus respectively. It is clear that the Li-ion battery has much higher
with the dual buffer is summarized in Table I. The so-called specific energy, whereas the supercapacitor has considerably
expressions in the table are not equality constraints, but are better capability of sourcing and sinking power. The ESS price
only used for better readability. Their values are derived from includes costs for materials, manufacturing, packaging, and
the associated mathematical equations. A tool, CVX [43], [48], circuitry. Cell packaging and circuitry are assumed to account
is applied to parse the problem in Table I, leading to a general for 12.3% of the total mass of the ESS pack [54]. Note that the
semidefinite program that can be efficiently solved by Self Dual power-type battery technology is not considered, since it typi-
Minimization (SeDuMi) [17], [49]. Owing to the convexity, a cally has substantially higher cost per kilowatt and lower power,
globally optimal solution is guaranteed with arbitrary initial- compared with the supercapacitor technology [50]. Both the
ization. The theoretical and algorithmic properties of convex HEV and PHEV scenarios are discussed. In the HEV scenario,
optimization are elaborated in [43]. the ESS is operated in a charge-sustaining mode, and the energy
1198 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 15, NO. 3, JUNE 2014

TABLE III
S UPERCAPACITOR S PECIFICATION

TABLE IV
M AIN V EHICLE PARAMETERS

Fig. 6. Optimal power split of the bus in the HEV scenario. (a) EGU output
power. (b) Battery terminal power. (c) Supercapacitor terminal power.

power in the driving cycle. In addition, the supercapacitor in the


dual buffer has an obvious load-leveling effect on the battery.
Furthermore, it can be observed that the dual buffer makes it
possible to turn off the engine more frequently and for longer
durations. The SOC trajectory and operating efficiency of the
ESS are shown in Fig. 7. In contrast to the supercapacitor-
only ESS, the battery-only ESS has a larger efficiency span and
even sometimes reaches its power bounds where the efficiency
consumed for traction is from the engine; whereas in the PHEV is relatively low. With the assistance of the supercapacitor,
scenario, the ESS charge is used (depleted to a specified value the efficiency of the battery in the dual buffer is improved.
and then recharged through electricity grid), and the engine The cycle-averaged efficiencies of the power sources of the
also provides traction energy. A general control strategy that bus with the three different ESSs are summarized in Table V
optimally blends the ESS and EGU powers throughout the (the idling points are excluded). The supercapacitor has higher
driving cycle is adopted in the PHEV scenario (see Table I). average efficiency than the battery, despite the substantially
The detailed specifications of ESS SOC constraints are listed larger current and relatively low SOC (e.g., 20%), where the
in Tables II and III. Owing to high power and long lifetime, a efficiency is relatively low. Nevertheless, the average efficiency
large SOC fluctuation of the supercapacitor is often allowed in of the EGU of the bus with the supercapacitor-only ESS is
charge-sustaining HEVs [27], [53]. lowest (slightly worse than the other two) because of the lowest
The main vehicle parameters are shown in Table IV. For EGU power (see Fig. 3). The EGU of the bus with the dual
simplicity, the number of passengers onboard is assumed to be buffer has the highest efficiency, since the optimization is more
constant. likely to yield the best EGU operation, given additional degrees
of freedom (e.g., power split between the battery and the
supercapacitor). The optimal ESS size and costs are compared
B. HEV Scenario
in Table VI. It can be observed that the optimized dual buffer
The convex optimization can rapidly and efficiently solve leads to the smallest total cost, whereas the bus with the battery-
the optimal ESS size and power split between the engine and only ESS is most costly. Compared with the battery-only ESS,
the ESS. The optimal power distribution of the bus powertrain the supercapacitor-only ESS and the dual buffer can reduce the
in the HEV scenario is shown in Fig. 6. Compared with the total cost by 9.9% and 13.7% in the driving cycle, respectively.
battery, the supercapacitor can provide or sink substantially This gives a considerable economic benefit in the service
larger power, leading to a smaller demand on the EGU power period of the bus. Additionally, we can see that, although the
and better recuperation capability. Because of the relatively supercapacitor-only ESS induces lower EGU efficiency, the
high cost per kilowatt, it is not optimal to size the battery-only consequent diesel consumption is less than that of the battery-
ESS big enough to be able to recuperate the maximum brake only ESS, owing to the much less energy requested by the EGU.
HU et al.: THREE ELECTROCHEMICAL ENERGY BUFFERS APPLIED TO A HYBRID BUS POWERTRAIN 1199

Fig. 8. Optimal power split of the bus in the PHEV scenario. (a) EGU output
power. (b) Battery terminal power. (c) Supercapacitor terminal power.

C. PHEV Scenario
The ESS charge is depleted to a lower bound in the PHEV
scenario. The blending of the ESS and engine powers in the
driving cycle is optimized by the convex programming intro-
duced in Section III. The optimal power distribution in the
PHEV scenario is shown in Fig. 8. As opposed to the HEV
scenario, the supercapacitor-only ESS leads to the most fre-
quent EGU utilization, since the low specific energy and quite
high price per kilowatthour (see Table III) give the result that a
small supercapacitor pack will minimize the total cost (the large
majority of energy comes from the EGU). It is apparent that the
Fig. 7. ESS SOC and efficiency in the HEV scenario. (a) Battery-only ESS. optimized battery-only ESS and the dual buffer can effectively
(b) Supercapacitor-only ESS. (c) Dual-buffer ESS. reduce the EGU energy output. In the dual buffer, the domi-
nant element is the battery, while the supercapacitor provides
TABLE V complementary power. The SOC trajectory and efficiency of the
AVERAGE E FFICIENCIES OF P OWER S OURCES IN THE HEV S CENARIO
ESS are presented in Fig. 9. The supercapacitor-only ESS still
shows better efficiency, due to the much smaller resistance. The
average efficiencies of the bus power sources in the PHEV sce-
nario are listed in Table VII. As in the HEV scenario, the dual
buffer results in the best EGU efficiency. The optimal ESS size
and costs are presented in Table VIII. The results show that the
dual buffer is most cost effective, whereas the supercapacitor-
only ESS is worst. Compared with the battery-only ESS, the
supercapacitor-only ESS increases the total cost by 25.2%,
TABLE VI implying that it may not be suitable for PHEV applications. On
O PTIMAL ESS S IZE AND C OSTS IN THE HEV S CENARIO the other hand, it can be seen that the battery in the dual buffer is
larger than the battery-only ESS, since the high power of the su-
percapacitor helps to use more electricity through reducing the
duration of the EGU operation (see Fig. 8). Further enlarging
the battery-only ESS to reduce the diesel consumption would
be more costly, owing to the relatively high battery cost per
kilowatt. The dual buffer outperforms the battery-only ESS by
1.7%, in terms of the total cost, which is, however, slight.
1200 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 15, NO. 3, JUNE 2014

TABLE IX
P RICES OF E NERGY C ARRIERS FOR S ENSITIVITY A NALYSIS

V. S ENSITIVITY A NALYSIS
The optimization results discussed in Section IV are based
on the baseline prices of diesel, electricity, Li-ion battery, and
supercapacitor shown in Tables II–IV, which can be considered
as the current market level in Europe (Sweden). From (14) and
Table I, we can see that, with the weights being these prices,
the simultaneous optimal ESS sizing and power management
can be interpreted as a multiobjective optimal control problem.
Therefore, it is important to investigate how changes in these
prices influence the optimization results and the comparison
between the three different ESSs. Since the focus of the paper is
not to accurately project the prices of different onboard energy
carriers in the future (which belongs to economics area), and it
is not feasible to consider all the possibilities, a simple but use-
ful sensitivity analysis is carried out by taking four likely values
of each price into account. The prices of the diesel, Li-ion
battery, and supercapacitor for the sensitivity analysis are listed
in Table IX. Variation in electricity price is not included because
electricity is substantially cheaper than others and often steadily
regulated for recharging purposes in public transportation [17],
[55]. Because fossil fuel has been extensively analyzed to have
an overall rising price in future years [15], [55], [56], 50%,
100%, and 150% increases of the baseline diesel price are
considered. On the other hand, owing to economies of scale,
the costs of the Li-ion battery and supercapacitor will gradually
diminish [19], [50], [55]. Here, we take 25%, 50%, and 75%
decreases of the baseline price as an example. An analogous
Fig. 9. ESS SOC and efficiency in the PHEV scenario. (a) Battery-only ESS. price treatment for sensitivity analysis can be found in [57].
(b) Supercapacitor-only ESS. (c) Dual-buffer ESS. Instead of considering complete three-factor combinations (i.e.,
43 ), for simplicity, we examine the sensitivity of the comparison
TABLE VII outcome in the following cases.
AVERAGE E FFICIENCIES OF P OWER S OURCES IN THE PHEV S CENARIO 1) Only one price (factor) is varying, while the others are
assigned to baseline values.
2) All the three prices are varying according to Table IX (i.e.,
Baseline–Change 1–Change 2–Change 3).

A. HEV Scenario
The total cost and ESS size in the HEV scenario, with respect
to the varying diesel price, are shown in Fig. 10. It is clear that
TABLE VIII
the increase in diesel price causes a rising total cost for all the
O PTIMAL ESS S IZE AND C OSTS IN THE PHEV S CENARIO ESS options. More importantly, as the diesel price increases,
the cost-saving advantage of the supercapacitor-only ESS over
the battery-only ESS becomes increasingly obvious, as well
as that of the dual buffer over the supercapacitor-only ESS.
The optimal size of the battery-only ESS is enlarged with the
increased diesel price; that of the supercapacitor-only ESS first
increases and then kept the same after the diesel price reaches
2.01 C/l, since the increase in the averaged EGU efficiency is
HU et al.: THREE ELECTROCHEMICAL ENERGY BUFFERS APPLIED TO A HYBRID BUS POWERTRAIN 1201

Fig. 10. Sensitivity of ESS comparison, with respect to the diesel price, in the Fig. 12. Sensitivity of ESS comparison, with respect to the supercapacitor
HEV scenario. (a) Total cost. (b) Single buffer (battery or supercapacitor) size. price, in the HEV scenario. (a) Total cost. (b) Single buffer (battery or super-
(c) Dual buffer size. capacitor) size. (c) Dual buffer size.

Fig. 11. Sensitivity of ESS comparison, with respect to the battery price, in Fig. 13. Sensitivity of ESS comparison when all the three prices alter in the
the HEV scenario. (a) Total cost. (b) Single buffer (battery or supercapacitor) HEV scenario. (a) Total cost. (b) Single buffer (battery or supercapacitor) size.
size. (c) Dual buffer size. (c) Dual buffer size.

too small to compensate for the ESS size/cost increase in the


HEV scenario. The battery and supercapacitor sizes in the dual
buffer have a nonmonotonic alteration: when the diesel price
is not larger than 2.01 C/l, the supercapacitor size increases,
and the battery size reduces; when the diesel price is larger than
2.01 C/l, an opposite size change is noticed.
The result with respect to the varying battery price is illus-
trated in Fig. 11. It is shown that, as the battery price decreases,
the battery-only ESS has an increasingly small deviation with
the supercapacitor-only ESS and the dual buffer. When the price
is less than a certain threshold, the battery-only ESS is even
better than the supercapacitor-only ESS, but still worse than Fig. 14. Sensitivity of ESS comparison, with respect to the diesel price, in the
the dual buffer. The battery size (the battery-only ESS and the PHEV scenario. (a) Total cost. (b) Single buffer (battery or supercapacitor) size.
(c) Dual buffer size.
dual buffer) increases, while the supercapacitor size in the dual
buffer becomes smaller.
B. PHEV Scenario
The sensitivity to the supercapacitor price is given in Fig. 12.
As opposed to the case of varying battery price, the battery-only The total cost and ESS size in the PHEV scenario, with
ESS has a larger deviation with the supercapacitor-only ESS respect to the diesel price and the battery price, are shown
and the dual buffer. Furthermore, the reduced supercapacitor in Figs. 14 and 15, respectively. It is clear that the gain of
price benefits the supercapacitor-only ESS more than the dual choosing the battery-only ESS rather than the supercapacitor-
buffer. The supercapacitor size (the supercapacitor-only ESS only ESS becomes increasingly large. The improvement in the
and the dual buffer) increases, while the battery size in the dual dual buffer relative to the battery-only ESS becomes larger with
buffer becomes smaller. the increased diesel price, whereas the outcome is opposite
Fig. 13 shows the comparison result when all the three prices when the battery price decreases (see Fig. 15). After the diesel
alter in a manner shown in Table IX. Although the prices of the price reaches 2.01 C/l, further increasing the diesel price will
supercapacitor and the battery decrease, the minimal total cost not change the battery size in the dual buffer, since the increased
increases for all the three ESSs, since only the diesel energy is supercapacitor size ensures the minimum diesel consumption.
finally consumed in the HEV scenario subject to ESS charge The comparison with respect to the supercapacitor price in
sustenance, and the increased diesel price is the dominating the PHEV scenario is shown in Fig. 16. Although the reduced
factor. In this case, the sizes of all the buffers increase. supercapacitor price is favorable to the supercapacitor-only
1202 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 15, NO. 3, JUNE 2014

VI. D ISCUSSION
Based on the assumption that the battery will last for at least
two years of operation and 100 000 km of driving, battery aging
is not considered in detail in this paper. It is, however, relevant to
briefly discuss whether battery aging would clearly point toward
either of the ESS designs by increasing the value of the bus after
the end of the studied two-year period. Known causes of battery
aging, in addition to calendar time, are high C-rates, ampere-
hour throughput, and cycling with high depth of discharge.
Focusing on the HEV, the results in Fig. 7 show that the
battery in the dual-buffer ESS uses the entire allowed SOC-
Fig. 15. Sensitivity of ESS comparison, with respect to the battery price, in
the PHEV scenario. (a) Total cost. (b) Single buffer (battery or supercapacitor) window from 40% to 60%, whereas the battery-only ESS has a
size. (c) Dual buffer size. slightly smaller window from 55% to 60%. Furthermore, Fig. 7
shows that the maximum used charging and discharging battery
power is about 10% lower in the dual-buffer ESS compared
with the single-buffer ESS. Total ampere-hour throughput is
71 Ah for the battery-only ESS and 63 Ah for the dual buffer.
Since the difference in how the cells are used in the dual
buffer and in the battery-only ESS is relatively small, it is
a reasonable assumption that the battery aging and, thus, the
traveled distance between battery replacements will be similar
for both designs. If so, the battery replacement cost would
be significantly smaller for the dual buffer compared with the
battery-only ESS, which has approximately eight times larger
battery. Hence, it can be concluded that, when considering
Fig. 16. Sensitivity of ESS comparison, with respect to the supercapacitor battery aging, the competitiveness of the dual-buffer HEV is
price, in the PHEV scenario. (a) Total cost. (b) Single buffer (battery or increased, compared with the battery-only ESS.
supercapacitor) size. (c) Dual buffer size.
In the PHEV, the optimal battery sizes are almost equal for
the dual buffer and the battery-only ESS. Both ESSs use the
entire SOC-window with the difference that the double-buffer
ESS relies more on the supercapacitor during high charging
and discharging power, which could reduce the battery wear.
However, the difference in the battery cell utilization is not large
enough to claim that considering battery aging would clearly
benefit the competitiveness of the dual-buffer PHEV.
It is not straightforward to use the same quantitative rea-
soning as aforementioned when considering whether battery
aging would significantly increase the competitiveness of the
supercapacitor-only ESS with respect to the dual buffer. The
Fig. 17. Sensitivity of ESS comparison when all the three prices alter in the supercapacitor will definitely last longer than the battery and
PHEV scenario. (a) Total cost. (b) Single buffer (battery or supercapacitor) size. possibly require no replacements before the bus is scrapped.
(c) Dual buffer size. To weigh this against the 4% cost advantage of the dual-buffer
HEV (shown in Table VI), we need a detailed battery wear
ESS, it still leads to substantially higher cost than the battery- model that can predict the number of replacements over the en-
only ESS and the dual buffer. The growing superiority of the tire bus lifetime. However, since battery wear characterization
dual buffer to the battery-only ESS is also observed. and prediction is clearly out of scope in this paper, the effects of
The result of changing all the three prices in the PHEV battery wear on the competitiveness of the supercapacitor-only
scenario is shown in Fig. 17. The supercapacitor-only ESS ESS versus the dual buffer is left to be studied in future papers.
becomes even worse, since the consequent dominant energy Since public transit buses often have a fixed route, which is
source is still diesel. On the contrary, the total cost of the very different from passenger cars, the convex optimization and
battery-only ESS and the dual buffer is reduced, despite the the associated ESS comparison are for a hybrid bus powertrain
increased diesel price, because the vast majority of energy with a known driving trip operated in Gothenburg, Sweden. A
consumed is electricity provided by the ESS. It is not hard to natural question arises: what would happen for very different
envision that the optimization will approach a purely electric trips, in terms of the applicability of convex optimization and
bus, given the fact that electricity is considerably cheaper than the ESS comparison results? First, the convex-optimization-
hydrocarbon fuel, as the ESS price continually decreases. After based ESS sizing and energy control methodology also applies
Change 1, the size of the dual buffer remains fixed, since the to different driving cycles, which has been demonstrated in
bus has already evolved as a purely electric bus. our prior work [39]. The ESS comparison results may be only
HU et al.: THREE ELECTROCHEMICAL ENERGY BUFFERS APPLIED TO A HYBRID BUS POWERTRAIN 1203

quantitatively different, in terms of the total cost for different 3) When only the supercapacitor price decreases, in the
trips. However, it will not lead to qualitative change on the best HEV scenario, the battery-only ESS has an enlarged
ESS option for both HEV and PHEV cases, which has been difference from the supercapacitor-only ESS and the
argued and supported by several existing survey papers in the dual buffer; the superiority of the dual buffer to the
literature [19]–[21], [27]. Our focus is on the ESS comparison supercapacitor-only ESS becomes smaller (still visible).
of a hybrid bus with the optimal ESS sizing and power manage- The supercapacitor-only ESS, in the PHEV scenario, be-
ment in a fixed route, so that we did not consider varying routes. comes better, but still substantially worse than the battery-
The convex-optimization-based approach can be efficiently and only ESS and the dual buffer; a growing advantage of the
rapidly used, if needed, to evaluate the quantitative deviation of dual buffer over the battery-only ESS is found.
the ESS comparison results for discrepant bus lines (trips). 4) When 1) to 3) happen simultaneously, the dual buffer
outperforms the other two ESSs more clearly in the HEV
VII. C ONCLUSION scenario. In the PHEV scenario, however, the dual buffer
has a negligible advantage over the battery-only ESS.
This paper has presented a comparative analysis of three
different electrochemical energy buffers (a Li-ion battery pack, It can be concluded that the dual buffer with the optimal size
a supercapacitor pack, and a dual buffer combining the former and power management is the preferable choice for the hybrid
two) applied to a hybrid bus powertrain in a framework of bus powertrain. In contrast, its superiority in the HEV scenario
simultaneous optimal ESS sizing and vehicle energy man- is more prominent; the difference between the dual buffer
agement. In the framework, the sum of operating cost (i.e., and the battery-only ESS in the PHEV scenario is relatively
fuel/electricity consumption) and ESS cost for the bus is min- small. The proposed convex-programming-based comparison
imized, such that the ESS can serve the hybrid powertrain in framework and the achieved comparison results are based on a
the most cost-effective manner. Convex modeling/optimization system-level hybrid powertrain optimization, which is valuable
is used to implement the framework, which is a new efficient to the initial exploration and assessment of vehicle design (the
tool for the hybrid powertrain developers/integrators to rapidly implications of some local components, such as power elec-
achieve the optimal ESS selection, dimensioning, and power tronics and battery aging, are often modeled using simplifying
management. assumptions at that stage).
Given the baseline prices of the diesel, Li-ion battery, and
supercapacitor that reflect the current market level, the com- R EFERENCES
parison result in the HEV scenario shows that the optimized [1] A. G. Boulanger, A. C. Chu, S. Maxx, and D. L. Waltz, “Vehicle electri-
dual buffer induces the smallest diesel consumption and the fication: Status and issues,” Proc. IEEE, vol. 99, no. 6, pp. 1116–1138,
Jun. 2011.
best economy; the supercapacitor-only ESS is second best; the [2] A. Emadi, “Transportation 2.0,” IEEE Power Energy Mag., vol. 9, no. 4,
Li-ion battery-only ESS is worst. Compared with the battery- pp. 18–29, Jul./Aug. 2011.
only ESS, the cost savings of the supercapacitor-only ESS [3] W. Su, H. Rahimi-Eichi, W. Zeng, and M.-Y. Chow, “A survey on the
electrification of transportation in a smart grid environment,” IEEE Trans
and the dual buffer are 9.9% and 13.7%, respectively. In the Ind. Informat., vol. 8, no. 1, pp. 1–10, Feb. 2012.
PHEV scenario, the dual buffer is still most cost-effective; the [4] D. MacKay, Sustainable Energy: Without the Hot Air. Cambridge, U.K.:
battery-only ESS is second best; the supercapacitor-only ESS UIT Cambridge, 2009.
[5] R. A. Waraich, M. D. Galus, C. Dobler, M. Balmer, G. Andersson, and
is worst. A significant cost increase relative to the battery-only K. W. Axhausen, “Plug-in hybrid electric vehicles and smart grid: Investi-
ESS implies that the supercapacitor-only ESS may not be an gations based on a microsimulation,” Transp. Res. C, Emerging Technol.,
appropriate option for PHEV applications. The advantage of the vol. 28, pp. 74–86, Mar. 2013.
[6] S. Chu and A. Majumdar, “Opportunities and challenges for a sustainable
dual buffer over the battery-only ESS is slight (even negligible) energy future,” Nature, vol. 488, no. 7411, pp. 294–303, Aug. 2012.
in the PHEV scenario. [7] X. He and J. W. Hodgson, “Modeling and simulation for hybrid electric
The impact of price variations of the diesel, Li-ion battery, vehicles. II. Simulation,” IEEE Trans. Intell. Transp. Syst., vol. 3, no. 4,
pp. 244–251, Dec. 2002.
and supercapacitor on the comparison is assessed by a simple [8] C.-C. Lin, H. Peng, J. W. Grizzle, and J.-M. Kang, “Power management
but useful sensitivity analysis considering four possible cases. strategy for a parallel hybrid electric truck,” IEEE Trans. Control Syst.
The following outcomes are acquired. Technol., vol. 11, no. 6, pp. 839–849, Nov. 2003.
[9] L. Johannesson, M. Åsbogård, and B. Egardt, “Assessing the potential of
1) When only the diesel price increases, in both the HEV predictive control for hybrid vehicle powertrains using stochastic dynamic
and PHEV scenarios, the advantage of the dual buffer programming,” IEEE Trans. Intell. Transp. Syst., vol. 8, no. 1, pp. 71–83,
Mar. 2007.
becomes increasingly obvious. [10] K. Li, T. Chen, Y. Luo, and J. Wang, “Intelligent environment-friendly
2) When only the battery price decreases, in the HEV sce- vehicles: Concept and case studies,” IEEE Trans. Intell. Transp. Syst.,
nario, the battery-only ESS has a smaller deviation from vol. 13, no. 1, pp. 318–328, Mar. 2012.
[11] S. Stockar, V. Marano, M. Canova, G. Rizzoni, and L. Guzzella, “Energy-
the supercapacitor-only ESS and the dual buffer. When optimal control of plug-in hybrid electric vehicles for real-world driv-
the price reduces to a certain threshold, the battery-only ing cycles,” IEEE Trans. Veh. Technol., vol. 60, no. 7, pp. 2949–2962,
ESS becomes even better than the supercapacitor-only Sep. 2011.
[12] C. Zhang and A. Vahidi, “Route preview in energy management of plug-
ESS, but still worse than the dual buffer. In the PHEV sce- in hybrid vehicles,” IEEE Trans. Control Syst. Technol., vol. 20, no. 2,
nario, the gain of using the battery-only ESS rather than pp. 546–553, Mar. 2012.
the supercapacitor-only ESS, however, becomes larger. [13] I. L. Sarıoğlu, O. P. Klein, H. Schröder, and F. Küçükay, “Energy man-
agement for fuel-cell hybrid vehicles based on specific fuel consumption
The advantage of the dual buffer over the battery-only due to load shifting,” IEEE Trans. Intell. Transp. Syst., vol. 13, no. 4,
ESS becomes indiscernible. pp. 1772–1781, Dec. 2012.
1204 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 15, NO. 3, JUNE 2014

[14] M. S. Whittingham, “History, evolution, and future status of energy stor- [39] M. Pourabdollah, N. Murgovski, A. Grauers, and B. Egardt, “Optimal
age,” Proc. IEEE, vol. 100, Special Centennial, pp. 1518–1534, May 2012. sizing of a parallel PHEV powertrain,” IEEE Trans. Veh. Technol., vol. 62,
[15] X. Zhang and C. Mi, Vehicle Power Management-Modeling, Control and no. 6, pp. 2469–2480, Jul. 2013.
Optimization. New York, NY, USA: Springer-Verlag, 2011. [40] N. Murgovski, L. Johannesson, and J. Sjöberg, “Engine on/off control for
[16] C.-S. Ernst, A. Hackbarth, R. Madlener, B. Lunz, D. U. Sauer, and dimensioning hybrid electric powertrains via convex optimization,” IEEE
L. Eckstein, “Battery sizing for serial plug-in hybrid electric vehicles: Trans. Veh. Technol., vol. 62, no. 7, pp. 2949–2962, Sep. 2013.
A model-based economic analysis for Germany,” Energy Policy, vol. 39, [41] X. Hu, N. Murgovski, L. Johannesson, and B. Egardt, “Energy efficiency
no. 10, pp. 5871–5882, Oct. 2011. analysis of a series plug-in hybrid electric bus with different energy man-
[17] N. Murgovski, “Optimal powertrain dimensioning and potential assess- agement strategies and battery sizes,” Appl. Energy, vol. 111, pp. 1001–
ment of hybrid electric vehicles,” Ph.D. dissertation, Chalmers Univ. 1009, Nov. 2013.
Technol., Gothenburg, Sweden, 2012. [42] N. Murgovski, L. Johannesson, and J. Sjöberg, “Convex modeling of
[18] E. Karden, S. Ploumen, B. Fricke, T. Miller, and K. Snyder, “Energy energy buffers in power control applications,” in Proc. IFAC Workshop
storage devices for future hybrid electric vehicles,” J. Power Sources, E-CoSM Powertrain, Rueil-Malmaison, France, Oct. 23–25, 2012,
vol. 168, no. 1, pp. 2–11, May 2007. pp. 92–99.
[19] S. M. Lukic, J. Cao, R. C. Bansal, F. Rodriguez, and A. Emadi, “Energy [43] S. Boyd and L. Vandenberghe, Convex Optimization. Cambridge, U.K.:
storage systems for automotive applications,” IEEE Trans. Ind. Electron., Cambridge Univ. Press, 2004.
vol. 55, no. 6, pp. 2258–2267, Jun. 2008. [44] S. Moura, D. S. Callaway, H. K. Fathy, and J. L. Stein, “Tradeoffs be-
[20] S. Vazquez, S. M. Lukic, E. Galvan, L. G. Franquelo, and J. M. Carrasco, tween battery energy capacity and stochastic optimal power management
“Energy storage systems for transport and grid applications,” IEEE Trans. in plug-in hybrid electric vehicles,” J. Power Sources, vol. 195, no. 9,
Ind. Electron., vol. 57, no. 12, pp. 3881–3895, Dec. 2010. pp. 2979–2988, May 2010.
[21] A. Khaligh and Z. Li, “Battery, ultracapacitor, fuel cell, and hybrid energy [45] S. Moura, J. L. Stein, and H. K. Fathy, “Battery-health conscious power
storage systems for electric, hybrid electric, fuel cell, and plug-in hybrid management in plug-in hybrid electric vehicles via electrochemical mod-
electric vehicles: State of the art,” IEEE Trans. Veh. Technol., vol. 59, eling and stochastic control,” IEEE Trans. Control Syst. Technol., vol. 21,
no. 6, pp. 2806–2814, Jul. 2010. no. 3, pp. 679–694, May 2013.
[22] A. Burke, “Ultracapacitors: Why, how, and where is the technology,” [46] B. G. Pollet, I. Staffell, and J. Shang, “Current status of hybrid, battery and
J. Power Sources, vol. 91, no. 1, pp. 37–50, Nov. 2000. fuel cell electric vehicles: From electrochemistry to market prospects,”
[23] A. Burke and M. Miller, “The power capability of ultracapacitors and Electrochim. Acta, vol. 84, pp. 235–249, Dec. 2012.
lithium batteries for electric and hybrid vehicle applications,” J. Power [47] T. J. Knipe, L. Gaillac, and J. Aargueta, “100 000-mile evaluation of the
Sources, vol. 196, no. 1, pp. 514–522, Jan. 2011. Toyota RAV4 EV,” Southern California Edison, Elect. Veh. Tech. Center,
[24] E. Schaltz, A. Khaligh, and P. O. Rasmussen, “Influence of battery/ Rosemead, CA, USA, 2003.
ultracapacitor energy-storage sizing on battery lifetime in a fuel cell hy- [48] Accessed by May 2010 M. Grant and S. Boyd, CVX: Matlab software for
brid electric vehicle,” IEEE Trans. Veh. Technol., vol. 58, no. 8, pp. 3882– disciplined convex programming, Accessed by May 2010.
3891, Oct. 2009. [49] Y. Labit, D. Peaucelle, and D. Henrion, “SeDuMi interface 1.02: A tool for
[25] M. Ehsani, Y. Gao, and A. Emadi, Modern Electric, Hybrid Electric, and solving LMI problems with SeDuMi,” in Proc. IEEE Int. Symp. Comput.
Fuel Cell Vehicles: Fundamentals, Theory, and Design, 2nd ed. Boca Aided Control Syst. Des., Glasgow, U.K., Sep. 18–20, 2002, pp. 272–277.
Raton, FL, USA: CRC Press, 2009. [50] T. Mayer, D. Kreyenberg, J. Wind, and F. Braun, “Feasibility study of
[26] C. Mi, M. A. Masrur, and D. Gao, Hybrid Electric Vehicles—Principles 2020 target costs for PEM fuel cells and lithium-ion batteries: A two-
and Applications with Practical Perspectives. Hoboken, NJ, USA: factor experience curve approach,” Int. J. Hydrogen Energy, vol. 37,
Wiley, 2011. no. 19, pp. 14 463–14 474, Oct. 2012.
[27] A. Burke, “Batteries and ultracapacitors for electric, hybrid, and fuel cell [51] Accessed Jul. 2010. [Online]. Available: http://www.houseofbatteries.
vehicles,” Proc. IEEE, vol. 95, no. 4, pp. 806–820, Apr. 2007. com/documents/VL45E.pdf
[28] A. Burke, M. Miller, and H. Zhao, “Lithium batteries and ultracapacitors [52] Accessed Jul. 2010. [Online]. Available: http://www.maxwell.com/
alone and in combination in hybrid vehicles: Fuel economy and battery products/ultracapacitors/docs/datasheet_k2_series_1015370.pdf
stress reduction advantages,” presented at the Proc. 25th World Bat- [53] D. Rotenberg, A. Vahidi, and I. Kolmanovsky, “Ultracapacitor assisted
tery, Hybrid Fuel Cell Electric Vehicle Symp. Exhib., Shenzhen, China, powertrains: Modeling, control, sizing, and the impact on fuel econ-
Nov. 5–9, 2010, Paper EVS-25. omy,” IEEE Trans. Control Syst. Technol., vol. 19, no. 3, pp. 576–589,
[29] R. T. Doucette and M. D. McCulloch, “A comparison of high-speed May 2011.
flywheels, batteries, and ultracapacitors on the bases of cost and fuel [54] L. Gaines and R. Cuenca, “Costs of lithium-ion batteries for vehicles,”
economy as the energy storage system in a fuel cell based hybrid electric Center Transp. Res. Argonne Nat. Lab., U.S. Dept. Energy, Argonne, IL,
vehicle,” J. Power Sources, vol. 196, no. 3, pp. 1163–1170, Feb. 2011. Tech. Rep., 2000.
[30] A. Sciarretta, M. Back, and L. Guzzella, “Optimal control of parallel [55] “Technology roadmap: Electric and plug-in hybrid electric vehicles,” IEA,
hybrid electric vehicles,” IEEE Trans. Control Syst. Technol., vol. 12, Paris, France, Tech. Rep., 2011.
no. 3, pp. 352–363, May 2004. [56] A. Dreher and T. Krieger, “Diesel price convergence and mineral oil
[31] L. Guzzella and A. Sciarretta, Vehicle Propulsion Systems—Introduction taxation in Europe,” Appl. Econ., vol. 42, no. 15, pp. 1955–1961,
to Modeling and Optimization, 3rd ed. Berlin, Germany: Springer- Jun. 2010.
Verlag, 2013. [57] S. Ebbesen and L. Guzzella, “Trade-off between fuel economy and battery
[32] P. Pisu and G. Rizzoni, “A comparative study of supervisory control strate- life for hybrid electric vehicles,” in Proc. ASME Dyn. Syst. Control Conf.,
gies for hybrid electric vehicles,” IEEE Trans. Control Syst. Technol., Arlington, VA, USA, Oct. 31/Nov. 2, 2011, pp. 217–223.
vol. 15, no. 3, pp. 506–518, May 2007.
[33] L. Serrao, S. Onori, and G. Rizzoni, “A comparative analysis of energy
management strategies for hybrid electric vehicles,” Trans. ASME, J. Dyn.
Syst. Meas. Control, vol. 133, no. 3, pp. 031012-1–031012-9, Mar. 2011.
[34] P. Elbert, S. Ebbesen, and L. Guzzella, “Economic viability of battery
load-leveling in hybrid electric vehicles using supercapacitors,” in Proc. Xiaosong Hu (S’11–M’13) received the Ph.D. de-
Int. Sci. Conf. RHEVE, Rueil-Malmaison, France, Dec. 6/7, 2011, pp. 1–8. gree in automotive engineering from Beijing Institute
[35] E. Tara, S. Shahidinejad, S. Filizadeh, and E. Bibeau, “Battery storage of Technology, Beijing, China, in 2012.
sizing in a retrofitted plug-in hybrid electric vehicle,” IEEE Trans. Veh. Between 2010 and 2012, he did scientific research
Technol., vol. 59, no. 6, pp. 2786–2794, Jul. 2010. and completed the Ph.D. dissertation in the Automo-
[36] M. J. Kim and H. Peng, “Power management and design optimization tive Research Center, University of Michigan, Ann
of fuel cell/battery hybrid vehicles,” J. Power Sources, vol. 165, no. 2, Arbor, MI, USA. He is currently a Postdoctoral Re-
pp. 819–832, Mar. 2007. searcher with the Swedish Hybrid Vehicle Center and
[37] S. Ebbesen, C. Dönitz, and L. Guzzella, “Particle swarm optimization the Department of Signals and Systems, Chalmers
for hybrid electric drive-train sizing,” Int. J. Veh. Des., vol. 58, no. 2–4, University of Technology, Gothenburg, Sweden. His
pp. 181–199, Jun. 2012. research interests include optimal control and dimen-
[38] N. Murgovski, L. Johannesson, J. Sjöberg, and B. Egardt, “Component sioning of electrified powertrains and modeling and management of energy
sizing of a plug-in hybrid electric powertrain via convex optimization,” storage systems.
Mechatronics, vol. 22, no. 1, pp. 106–120, Feb. 2012. Dr. Hu was a recipient of the Beijing Best Ph.D. Dissertation Award in 2013.
HU et al.: THREE ELECTROCHEMICAL ENERGY BUFFERS APPLIED TO A HYBRID BUS POWERTRAIN 1205

Nikolce Murgovski received the M.Sc. degree in Bo Egardt (SM’90–F’03) received the M.Sc. degree
software engineering from University West, Troll- in electrical engineering and the Ph.D. degree in
hättan, Sweden, in 2007 and the M.Sc. degree in automatic control from Lund Institute of Technology,
applied physics and the Ph.D. degree in signals and Lund, Sweden, in 1974 and 1979, respectively.
systems from Chalmers University of Technology, During 1980, he was a Research Associate with
Gothenburg, Sweden, in 2007 and 2012, respectively. the Information Systems Laboratory, Stanford, CA,
He is currently a Postdoctoral Researcher with USA. From 1981 to 1989, he was with Asea Brown
the Department of Signals and Systems, Chalmers Boveri (ABB), where he was heavily involved in
University of Technology. His research activities the introduction of adaptive control in the process
include optimal control and sizing of automotive industry. Since 1989, he has been a Professor with
powertrains. Chalmers University of Technology, Gothenburg,
Sweden, where he is the Head of the Automatic Control Group. He is also the
Head of the Group of Systems and Control with the Swedish Hybrid Vehicle
Lars Mårdh Johannesson (M’12) received the Center, Chalmers University of Technology. His main areas of interest include
M.Sc. degree in automation and mechatronics and system identification, adaptive and optimal control, and applications of control
the Ph.D. degree in automatic control from Chalmers in the automotive area.
University of Technology, Gothenburg, Sweden, in Dr. Egardt is a member of the Royal Swedish Academy of Engineering
2004 and 2009, respectively. Sciences. He is also a member of the editorial board for the International
Since 2011, he has been with the Electromobility Journal of Adaptive Control and Signal Processing. He has been an Associate
Group, Viktoria Swedish ICT, Gothenburg, Sweden, Editor of the IEEE T RANSACTIONS ON C ONTROL S YSTEMS T ECHNOLOGY
working with research on powertrain control within and of the European Journal of Control.
the Chalmers Energy Initiative. His main research in-
terests include optimal control of hybrid and plug-in
hybrid electric vehicles, control of auxiliary systems
in trucks, active cell balancing, and system studies of hybrid vehicles.

You might also like