2023 ARTICLE, Module-Level Direct Coupling in PV-battery Power Unit Under Realistic Irradiance and Load

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Solar Energy 249 (2023) 233–241

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

Module-level direct coupling in PV-battery power unit under realistic


irradiance and load
Uchechi Chibuko a, c, Tsvetelina Merdzhanova a, Daniel Weigand a, Fabian Ezema c,
Solomon Agbo a, b, Uwe Rau a, d, Oleksandr Astakhov a, *
a
IEK5-Photovoltaik, Forschungszentrum Jülich GmbH, 52425 Jülich, Germany
b
UE, Forschungszentrum Jülich GmbH, 52425 Jülich, Germany
c
Department of Physics and Astronomy, University of Nigeria, Nsukka, 410001 Nsukka, Nigeria
d
Faculty of Electrical Engineering and Information Technology, RWTH Aachen University, Mies-van-der-Rohe-Straße 15, 52074 Aachen, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: A photovoltaic (PV) module, battery and consumer or load is usually tied together by a complex power elec­
PV-battery tronics, including maximum power point tracking (MPPT) device for power coupling to maximize output of the
Power coupling PV modules. At the same time, a typical battery itself can play the role of a power coupling element in addition to
Irradiance
its main energy storage function. In principle, a properly chosen PV-battery pair can maintain a high degree of
Load
Daily cycle
internal power coupling even under variable irradiance and load without MPPT electronics. This option is of
interest for e.g. module-level integration of PV and battery to cope with natural intermittency of a PV module
power output. In this work, we experimentally examine the function of a laboratory scale unit of a 7-cell silicon
heterojunction PV module directly connected to a lithium-ion battery and variable load. The unit is the simplest
PV-battery module representative for detailed study under a series of emulated realistic profiles of irradiance and
power consumption. The directly coupled PV-battery unit shows coupling efficiencies of above 99.8% at high
irradiance and approx. 98% on average through the daily cycle – a value that is comparable to modern MPPT
devices.

1. Introduction potentially advantageous solution (Fagiolari et al., 2022; Vega-Garita


et al., 2018a, b). PV modules with sufficient internal energy buffer
Photovoltaics (PV) has the potential to be a major energy source in will have much smoother power output with wide peaks of generation,
the future (Ram et al., 2019). According to the same report, PV can make mild power ramps and extended power output during evening and night
as much as 69 % of total energy generation. At the same time, due to its hours. Smooth, more predictable module output allows reducing peak
intermittent nature, PV requires large amount of storage for power power rating of PV system components, which potentially reduces cost
supply stabilization. Even at present, deployment of PV faces the prob­ of the system without loss of its reliability. Extended hours of energy
lem of the time mismatch between peaks of PV generation and electricity generation naturally eliminate or reduce the “duck curve” issue. Com­
consumption often referred to as “duck curve” problem (Howlader et al., bination of PV with battery is shown to be beneficial for stability and
2018; Kosowatz, 2018; Sheha et al., 2020). Application of electro­ efficiency of electrochemical solar fuel production via e.g. water split­
chemical batteries as an energy buffer storage is a straightforward so­ ting (Astakhov et al., 2021; Kin et al., 2022a). Finally, it is suggested that
lution for this problem of the natural temporal mismatch between PV battery can provide efficient power coupling in addition to the main
generation and electricity demand (Astakhov et al., 2021; Fagiolari storage functionality in direct coupling configuration (Astakhov et al.,
et al., 2022; Howlader et al., 2018; Kosowatz, 2018; Mohanty et al., 2020; Ayeng’o et al., 2019; Kakimoto and Asano, 2017), where power
2016; Rana et al., 2022; Sheha et al., 2020; Vega-Garita et al., 2018a, b). coupling refers to the matching of the working point of the PV to its MPP
Integration of both PV and battery technologies at PV system level is when connected to the power coupling element (in this case a battery).In
very common practice (Rana et al., 2022). Integration of PV and bat­ other words power coupling refers to the ability of the whole device to
teries at the PV module level is less explored, challenging, but utilize maximal available power of the PV module. This point is of

* Corresponding author.
E-mail address: o.astakhov@fz-juelich.de (O. Astakhov).

https://doi.org/10.1016/j.solener.2022.11.040
Received 16 September 2022; Received in revised form 16 November 2022; Accepted 24 November 2022
Available online 2 December 2022
0038-092X/© 2022 The Authors. Published by Elsevier Ltd on behalf of International Solar Energy Society. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
U. Chibuko et al. Solar Energy 249 (2023) 233–241

particular interest because PV-battery modules can be internally the actual module working point (WP) and maximum power point
matched without maximum power point trackers (MPPT) which is a power (PMPP) expressed as C = PPV /PMPP . The coupling factor is there­
very typical component of PV systems (El Fadil and Giri, 2011; Masoum fore describing how much of the power available at MPP is harnessed by
et al., 2002; Midya et al., 1996; Mohanty et al., 2016; Zavos). A MPPT a directly coupled unit. Since there is no power consuming electronics
device provides high degree of flexibility in PV system design and reli­ between PV, battery and load in the directly coupled module, the
able output of PV modules under intermittent irradiance. However, if coupling factor C in this case is compared to available literature data on
battery can provide comparable degree of power coupling, the use of the power efficiency of MPPTs (Abu Qahouq et al., 2014; Cao and Kim,
power electronic in the system can be reduced. In short, the concept of 2015; Cooley and Leeb, 2011; Uprety and Lee, 2014, 2017). Results of
the module-level integrated PV-battery device holds promise for the experiment show high potential of the direct coupled module level
simpler, cheaper, more reliable PV systems with highly stable power PV-battery unit, which shows persistently high performance over real­
output. istic operation cycle.
Several issues related to performance of the batteries under elevated
temperatures, life and cost of the battery technology are yet to be solved 2. Experiment
to realize reliable and affordable PV-battery modules (Vega-Garita et al.,
2019; Vega-Garita et al., 2017; Vega-Garita et al., 2018a, b). However, 2.1. Measurement circuit
battery technology shows rapid progress (Ma et al., 2021) and we
believe that PV-battery integration at module level is a matter of the The circuit diagram of a directly coupled PV-battery unit connected
nearest future. At the same time, parallel progress in crystalline silicon to variable load resistor is shown in Fig. 1(a). This is also the basic circuit
technology like silicon heterojunction solar cells (Chime et al., 2022; used for the setup of our experiment in the laboratory, as shown in Fig. 1
Duan et al., 2022; Yoshikawa et al., 2017) as well as multiple emerging (b). The load draws direct current (DC) from the PV-battery unit. An
PV technologies, like water-based dye sensitized solar cells (de Haro inhouse built connection board was used to realize stable and repro­
et al., 2021; Galliano et al., 2021) or sustainable Perovskite solar cells ducible current and voltage measurements between the system elements
(Luo et al., 2022; Schmitz et al., 2022; Wang et al., 2022) open new ways with minimal voltage drops due to measurement equipment. The board
for PV-battery integration at different levels and sizes for indoor (Kin and connection for the components are shown in supplementary Fig. S1.
et al., 2022b) and field applications (Ayeng’o et al., 2019). Note that there is no power management electronics or energy
In this context, in several of our previous works (Agbo et al., 2016a; management system in the circuit. We address here the simplest com­
Agbo et al., 2016b; Sandbaumhüter et al., 2017) we addressed feasibility bination of PV and battery where battery is expected to play both storage
of PV-battery combinations at different level, with particular focus on and power stabilization functions. For the experimental setup, 10 mΩ
the feasibility of the direct coupling between PV and battery devices, sense resistors are connected in series to the PV module and battery, as
addressed in detail in our recent theoretical study (Astakhov et al., shown in Fig. 1(b), which are used for sensing current from the PV and
2020). battery by recording the voltage across the 10 mΩ resistors. This helps to
In this work, the predictions of high efficiency and operation stability accurately measure the PV and battery current by minimizing power loss
in voltage matched PV-battery system are tested experimentally with provided from the conventional approach of connecting digital amme­
emulated realistic profiles of irradiance and power offtake. The focus of ters in series to the circuit. Keithley digital multimeters (DMM) are used
the study is on stability, internal power conversion efficiency, and losses for recording the voltage across the 10 mΩ resistors. The voltage of PV is
in PV-battery combination. In our experiment we study an elementary recorded by a Keithley SMU 2420 device, while the battery voltage is
PV-battery combination of a Silicon Heterojunction (SHJ) solar cell recorded by a Keithley SMM 3706A device. Using this approach, the
minimodule(Lee et al., 2020) directly coupled to a single cell commer­ current and voltage values are obtained to accuracies of ± 0.01 mA and
cial Li-ion battery In most cases battery cells have higher voltages than ± 0.01 V respectively, sufficient for our analysis.
solar cells. Several solar cells must be stacked as tandem or connected in Irradiance for PV module has been provided by an AM1.5G class-A
series as a module to match the voltage range of the battery. Commercial sun simulator from Wacom with PV module installed on the water-
Li-ion batteries typically operate in the voltage range of 3.7–4.3 V which cooled stage of the solar simulator and with temperature of approxi­
approximately corresponds to maximum power point voltage of 7 SHJ mately 25 ◦ C. Due to the constant connection of the PV-battery to the
cells connected in series. The cycling is realized in a lab experiment load, we use the term “PV-battery-load unit” to refer to the setup in the
using a pattern of global horizontal irradiance (Zubair et al., 2018) and manuscript.
load profile (Adeoye and Spataru, 2019) typical for Nigeria urban area.
“Module-level direct coupling” in this paper, refers to a direct connec­ 2.2. PV module, battery and load
tion of the PV module to a battery without use of power electronics. (not
a physical integration into one device as e.g. in (Agbo et al., 2016a; The laboratory-scale solar module made up of 7 silicon hetero­
Vega-Garita et al., 2018a, b)). As it is shown in the work (Leible and junction (SHJ) cells has been prepared inhouse. The SHJ cells, prepared
Bessler, 2021), a directly coupled PV-battery-load unit in real operation according to (Duan et al., 2022) were laser cut and interconnected with
can have significant losses supposedly due to insufficient internal power low temperature soldering in shingled configuration as described in (Lee
coupling. In view of importance of the power coupling issue, in this work et al., 2020). Each cell of the PV module has an area of 3.2 cm2 with the
we focus on coupling efficiency or coupling factor in the PV-battery combined active area of the 7-cell module equal to 22.4 cm2. The current
module and detailed power loss analysis. Experimental performance of voltage (IV) characteristics of the module measured under different
the studied PV-battery module is analyzed in the context of the literature irradiance from 1 sun to 0.019 sun are presented in Fig. 2. The trail of the
data on MPPT performance. MPPT units usually show excellent perfor­ MPP through the different intensities is correspondingly shown with
mance in tracking MPP with “tracking efficiency” or “MPPT efficiency” yellow filled circles. At standard test conditions (STC), the module has
up to 99.8 % (Belhachat and Larbes, 2018; Jiang et al., 2017; Uprety and an efficiency of 19.2 %, open circuit voltage (Voc) of 5.03 V, short circuit
Lee, 2014, 2017). The tracking efficiency, however, should be distin­ current (Isc) of 112 mA, voltage at maximum power point (Vmpp) of
guished from the energy efficiency, which usually has peak values of 4.22 V, current at maximum power point (Impp) of 101.75 mA and
93–96 % (Abu Qahouq et al., 2014; Cao and Kim, 2015; Cooley and power at maximum power point (Pmpp) of approx. 430 mW.
Leeb, 2011; Uprety and Lee, 2014, 2017) due to self-consumption in the The storage battery is represented by a 153040 DNK commercially
MPPT electronics. available lithium-ion cell with nominal capacity of 110 mAh and nom­
The key performance metrics of the PV-battery unit is the coupling inal voltage of 3.7 V. The battery is a lithium-polymer (LIP) rechargeable
factor C defined as the ratio between the output power of the PV (PPV) at cell which defines the battery chemistry. In order to study power

234
U. Chibuko et al. Solar Energy 249 (2023) 233–241

Fig. 1. (a) Circuit diagram of a directly coupled PV-Battery-Load unit; PV connected directly to load and battery without a maximum power point tracking device.
Red arrows indicate PV current and blue arrows indicate battery current. (b) Circuit diagram of experimental setup showing digital voltmeters across sense resistors
for measuring PV current (IPV) and battery current (IB), as well as devices for measuring voltage of the PV module and battery. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)

module IV characteristics. To achieve a minimum SOC for the battery,


the battery is discharged to its minimum rated cut of voltage of 3.0 V.
The maximum SOC is achieved by charging under 1 sun and integrating
the charging current with time (i.e. columbic counting) using the
(∫ ) (∫
expression that SOC [%] = 100⨯ Idt /110. The charged capacity
)
Idt in mAh is divided by the nominal full capacity of 110 mAh and
multiplied by 100 to get the SOC in %. The charging and discharge
voltage versus SOC plot for 1C rate is shown in supplementary Fig. S2
which shows an operating voltage window of 3.0–4.54 V for the battery.
From Fig. 2, the slope of the battery IV characteristics at the minimum
and maximum SOC is − 1.03 A/V and − 1.02 A/V, respectively. Note that
the slope of the battery IV plots remained approximately same under
these extremes of SOC. Overall, the cell used in the study like many other
modern commercial battery cells shows very stable linear IVs with
constant slope within the nominal range of application conditions (state
of charge and currents). One side effect of the direct coupling is potential
leakage of the battery charge through the forward biased PV module in
the dark. Detectable leakage current through the PV module was
recorded only at illuminations less than 0.053 sun for high battery SOC.
This recorded back flow of current at extremely low illumination and
ultimately in the dark is generally referred to as dark current (Id) for the
simplest case of the directly coupled unit. This dark current loss can be
omitted using a blocking diode.
The effect of the load was emulated with a manually operated Bourns
Fig. 2. Current-voltage (IV) characteristics of the silicon heterojunction module 3680 Precision Potentiometer/variable resistor of 10–1000 Ohms. We
measured under different irradiance as indicated for every IV characteristic emulated power consumption in the range of 0.54 mW to 14.8 mW per
(Red lines). Yellow filled circles mark the maximum power point (MPP) of the cm2 of PV module, where the upper power corresponds to the 77 % of
PV module. Battery IV characteristics taken at the battery min. charge and the PV capability under standard test conditions (1 sun, AM1.5). Com­
battery max. charge (blue lines). The intersections of the PV and the battery IVs mon battery and load IV characteristic is represented by the black line in
highlighted with blue closed circles are possible working points (WP) of the PV- Fig. 2. This common battery-load IV characteristic is taken at 75 % SOC
battery combination given by the intersection of the IV characteristics. The with the resistor set at a power consumption of approximately 14.8 mW
black line represents the common battery + load IV for the load resistance of per cm2 (77 % of PV 1 sun capability). Note that the black IV has nearly
14.8 mW per cm2 (77 % of PV 1 sun capability) at the battery SOC ≈ 75 %. (For
same slope as the battery IV. This is because the steep slope of the bat­
interpretation of the references to colour in this figure legend, the reader is
tery IV dominates the IV characteristic of PV-Battery-Load combination
referred to the web version of this article.)
for the whole range of practically relevant load powers.

coupling in PV-battery combination, it is instructive to measure also


current voltage characteristics of the battery. Open circuit voltage of any 2.3. Irradiance and power consumption profiles
secondary battery is a function of state of charge, therefore no single IV
can characterize the battery sufficiently. On the other hand, it is The relevance of the study for field PV applications is based on the
reasonable to evaluate battery IV characteristics under different states of realistic irradiance and power usage profiles. The blue curve in Fig. 3(a)
battery charge to get an indication of the region where the operating shows the global horizontal irradiance profile (Zubair et al., 2018) used
point of the PV-battery combination can be located with respect to the as a reference to the laboratory experiment. This shows the variation of
maximum power point of the PV module. This is important for power solar irradiance throughout a typical day time for a hot climate region,
coupling battery to a given PV module. If the region of the battery IV at having peaks of solar irradiance between 12 and 2 pm in the afternoon.
minimum and maximum SOC is sufficiently away from the MPP of the For a simple case of realistic profiles, we do not consider shading, clouds,
module at different irradiance levels, then a significant amount of power rain or other adverse weather effects on the solar irradiance. The
at MPP will be lost when the battery is connected to the module. The reference load consumption pattern for Nigeria urban area is also shown
battery IV characteristics taken at minimal SOC (min charge) and (Adeoye and Spataru, 2019). Both profiles have been reproduced in the
maximal SOC (max charge) are shown in Fig. 2 together with the PV laboratory experiment with stepped profiles shown in Fig. 3(b).

235
U. Chibuko et al. Solar Energy 249 (2023) 233–241

Fig. 3. (a) Reference profiles of global horizontal irradiance and load vs time from the literature (Adeoye and Spataru, 2019; Zubair et al., 2018), (b) Discrete step
profiles of the irradiance and load current reproducing the real time profiles in the laboratory experiment over a scaled down period of 6.4 h.

2.4. Characterization of the unit under variable irradiance, battery state operation.
of charge and load power. As highlighted from the components of the setup, the detectable
leakage current through the PV at extremely low irradiance and in the
In the experiment, the PV module, battery and load were connected dark, referred to as dark current, was present during the cycling
in parallel as shown in the circuit diagram from Fig. 1. experiment. Therefore, the dark current was blocked by an “ideal diode”
Before the cycling of the PV-battery-load unit under reproduced assembly in one experimental run involving two cycles with a shorter
realistic patterns presented in Fig. 3(b), the unit is characterized for time interval of 2.3 h per cycle.
isolated effects of irradiance, battery SOC and load on the power
coupling. For this purpose, the effect of irradiance at different battery 3. Results
SOC on the power coupling factor was studied. The solar simulator
irradiance was varied using neutral density filters from 0.053 sun to 1 3.1. Characterization of the unit: Effect of irradiance, battery state of
sun with working points recorded to evaluate coupling factor (C) at charge and load on coupling factor
different state of charge. In order to minimize effect of SOC drift, the
measurement at every irradiance has been reduced to approx. 12 s, Prior to exposing the unit to realistic profiles of irradiance and load
which results in negligible battery charging even at the highest irradi­ and periodic cycling over the realistic profiles, the dependence of
ance. As part of the unit characterization, the effect of the load power on coupling factor on solar irradiance, battery SOC and load power was
the coupling factor has also been studied under different irradiance at characterized separately.
constant SOC of approximately 75 %. At this intermediate SOC, power The dependence of coupling factor on the battery SOC and irradiance
consumption was varied by variation of the load resistor from 43 Ω to in PV-battery connection/subunit without load is shown in the colour
1000 Ω for each irradiance level from 0.053 sun to 1 sun. Working points map in Fig. 4(a). The coupling factor (C) was greater than 0.9 for illu­
recorded at every irradiance are combined with the previously recorded minations above 0.1 sun and for the whole range of the battery SOC. In
maximum power points to evaluate the coupling factor C. other words, the PV-battery connection maintains very high-power
coupling for the whole relevant energy generation range. C generally
drops with decrease in irradiance and at the highest battery SOC (close
2.5. Cycling procedure to 100 %).
Fig. 4(b) shows coupling in the PV-Battery subunit connected to the
Realistic operation cycle of irradiance presented with blue profile in load as a function of irradiance and load power at approximately con­
Fig. 3(a) has been reproduced with stepwise attenuation of the solar stant battery SOC (75 %). A dashed diagonal line in Fig. 4(b) represents
simulator irradiance. As it is illustrated with blue profile in Fig. 3(b) the the conditions where PV power (PPV) equals load power (PL), i.e., all PV
cycle starts from 0.1 sun at the beginning of the cycle and increased generated power is consumed in the load and neither charging nor
stepwise up to a maximum of 1 sun followed by a stepwise reduction till discharging of the battery takes place. In the region above the line, PPV
complete darkness. The time interval for each step was 16 min, which > PL and excess power charges the battery (battery charging conditions).
emulates the operating conditions for 1 h in a real-life situation from the In the region below the line, PPV < PL and the deficiency of PV power is
reference data. Simultaneously, the variable resistor was varied in the covered by the battery discharge (battery discharging conditions). In the
range of 121–545 Ω to suit the load pattern in the reference profile, as it real conditions PV-Battery-Load system navigates between the charging
is shown with black profile in Fig. 3(b). Each operating cycle was for 6.4 region in the peak PV generation hours and discharge region under low
h and was carried out for 6 consecutive cycles, corresponding to a total irradiance. The same effect can have highly demanding load operating
of 38.4 h of laboratory time, representing 6 days of real-time system

236
U. Chibuko et al. Solar Energy 249 (2023) 233–241

Fig. 5. (a) Current profile of performed realistic cycles of the PV-Battery-Load


unit against time. (b) Coupling factor against time. (c) Battery state of charge
against time.

Fig. 4. (a) Coupling factor of the directly coupled PV-Battery subunit as a


function of battery SOC and irradiance. (b) Coupling factor for the PV-Battery-
Load unit at ~ 75 % battery SOC with varying load power under different PV
power before running realistic profiles. The dashed blue diagonal line drawn for
PPV = PL. (For interpretation of the references to colour in this figure legend, the
reader is referred to the web version of this article.)

in pulsed mode. It can be clearly seen in Fig. 4(b) that PV-Battery-Load Fig. 6. (a) Battery capacity against time for cycling experiment. (b) Battery
voltage against time.
unit stays highly coupled (C greater than 0.9) over very wide range of PV
power and load power even for conditions with load demand several
times higher than the power generated by PV. This is the stabilizing, daytime and most of the generated power is stored in the battery as
coupling effect of the battery, where the battery is crucial for the evidenced by the positive battery current.
operation of the directly coupled PV-Battery-Load unit. The beginning of the dark period shows high battery discharge cur­
rent driving the load in the dark with load current having maximum
values corresponding to high demand. A slight decrease of PV current
3.2. Cycling of the unit under realistic irradiance and load profiles through the dark period is also noticed. This implies a back flow of
current from the battery to the forward biased PV module in the dark, as
The results from the cycling experiment of the realistic profiles are previously highlighted from the battery IV characteristics discussed
shown in Figs. 5 and 6. The module level unit of PV-Battery-Load from Fig. 2.
combination has been exposed to 6 cycles. The first cycle is zoomed The coupling factor C, over the time of the experiment, is shown in
into in Fig. 5 for a better visualization of the behavior of the unit. The Fig. 5(b). We observed C close to 100 % for the periods of the most
rest of the profiles are essentially the same and no significant deviations significant PV power generation. A significant drop of the coupling
in the behavior of the unit have been observed. Therefore, time scale for factor down to about 50 % is only observed at extremely low irradiance
the rest 5 cycles is compressed in Fig. 5. of 0.053 sun just before complete darkness.
Fig. 5(a) shows the current profiles of the PV-Battery-Load unit under The progression of the battery SOC is shown in Fig. 5(c). As previ­
(∫ )
periodic cycling. The convention used in the plot is that negative current ously highlighted, the SOC by columbic counting = 100⨯ Idt /110.
values imply a net-release of current, while positive current values imply The charged/discharged capacity
(∫ )
Idt in mAh is divided by the
a net-receipt of current. nominal full capacity of 110 mAh and multiplied by 100 to get the SOC
With focus on the first cycle, the PV current and load current both in %. The curve is smooth in contrast to other stepped profiles because
follow the emulated irradiance and consumption profiles. The load SOC is essentially a result of the battery current integration. The battery
profile taken from (Adeoye and Spataru, 2019) has a valley during the

237
U. Chibuko et al. Solar Energy 249 (2023) 233–241

state of charge increases in the beginning of the day cycle till the load, while at low PV power range the load power is wide. The dashed
“evening” when reduction of the PV generation and the increase in of the line is also drawn for PPV = PL, the critical power utilization conditions.
power consumption is covered by the battery energy. The battery is Above the line, the PV power is higher than the power consumption and
cycled from minimum to maximum capacity on purpose even though excess power is collected in the battery. Below the line the deficiency of
this is not recommended for real applications to help sustain battery life. the PV power is covered by the battery discharge power. As well as in the
However, a full charge and discharge of the battery is used for our time scale presentation in Fig. 5, the colour plot in Fig. 7 clearly shows
experiment to help mirror extreme cases of usage common to household that the unit maintains high power coupling for the most relevant region
applications. This provides a larger range of battery voltage for the PV- of significant PV generation, which ensures high overall efficiency even
Battery conditions for evaluation of coupling factor instead of a rela­ without maximum power point tracking.
tively lower voltage range in a case where the battery is not fully
charged or discharged. This therefore gives a good insight for real 3.3. Energy and efficiency in PV-Battery-Load unit in cycling experiment
conditions where full charge and discharge of the battery might be the
case. All cycles taken together show a tendency for SOC to drift to higher Evaluation of the performance of the unit in terms of energy effi­
values. This drift results to having the final two cycles show SOC to be a ciency requires analysis of the energy profiles for different components
bit greater than 100 % with a peak of 108.7 % corresponding to 119.6 of the unit. To do so, the data on power of the elements of the unit are

mAh reached for the final cycle. This drift can be related either to charge integrated with respect to time ( Pdt). The results are shown in Fig. 8.
accumulation or loss of charge in the battery. In order to distinguish The blue line at the apex of the plots shows the potentially obtainable
between these two cases, it is instructive to consider the profiles of final energy provided the PV module is always at MPP – ideal power
battery capacity and voltage shown in Fig. 6. Note that the battery rated coupling conditions. The black line shows the energy generated by the
nominal capacity is 110 mAh (at 0.5C rate of discharge). It is evident in PV module. From the start of the experiment (t = 0), the black line (PV
Fig. 6(a) that capacity increases with a slope of 4.1 mAh per cycle (24.86 energy) increases as the experiment goes on and intermittently flattens
mAh for 6 cycles). At the same time, the maxima of the battery voltage out during the dark periods. The difference between the PV energy and
stay very stable at 4.35 V for all cycles. This persistent voltage behavior MPP energy is the total energy loss related to coupling (C-Loss) – the
indicates that the drift of capacity is not related to charge accumulation difference between ideal coupling and coupling in the experiment. En­
and mostly related to Faradaic losses in the battery. If the whole capacity ergy profiles in Fig. 8 confirm very high coupling performance of the
drift is attributed to the charge loss and not charge accumulation, which directly connected PV-Battery-Load unit in realistic experiment. Final
would lead to a drift of the battery voltage, the Faradaic efficiency of the coupling efficiency is 98.2 %.
battery of approximately 96 % is calculated. The red line shows the integral of the power generated by PV module
An illustrative representation of the performance of the unit in a taken together with the energy dissipated by the module during the dark
cycling experiment can be obtained if coupling factor is presented as a due to the dark current in forward biased operation. The difference
function of PV and load powers PPV and PL. The data of the cycling between the final values of the energy generated by PV (black line)
experiment presented in Fig. 5 is presented as a colour plot in Fig. 7, minus the energy shown by the red line gives the total energy lost to dark
showing dependence of coupling factor on varying PV power and load current (Id-Loss). The dark current caused approximately 4.4 % loss of
power similar to the plot in Fig. 4(b). The clusters of black dots show the the total generated energy, which is equivalent to 4.2 % of the energy at
actual operating points of the PV-Battery-Load unit during the cycling MPP (EMPP).
experiment. Colour map is an interpolation of the coupling factor be­ The purple line shows the energy consumed by the load. This energy
tween the operating points. Unlike Fig. 4(b), the map in Fig. 7 visualizes
the region of operation of the unit in realistic cycle. It is characteristic for
the unit to operate at high PV power having low power demand from the

Fig. 8. Energy evolution with time - profiles of energy generation and con­
Fig. 7. Coupling factor for the realistic cycling experiment. Dashed blue line sumption in the different components of the unit. Percentage energy (E) losses
represents the PV and load power balance conditions, where PPV = PL. The with respect to MPP energy are shown in yellow cases. EMPP is energy at
cluster of black dot represents the PV working point (WP) showing the regions maximum power. EPV is energy of the photovoltaic module. EPVdark is energy
of operation of the unit during the experiment. (For interpretation of the ref­ dissipated by the PV module during the dark. ELoad is energy of the load. EBattery
erences to colour in this figure legend, the reader is referred to the web version is energy of the battery. (For interpretation of the references to colour in this
of this article.) figure legend, the reader is referred to the web version of this article.)

238
U. Chibuko et al. Solar Energy 249 (2023) 233–241

as expected increases from the beginning of the experiment till the end consumed by the load is almost entirely governed by the load resistance,
of the experiment with steeper slopes noticeable at the dark periods very similar to the stable electric grid.
matching higher load demand during these periods, as seen from the With a total efficiency of approximately 82.5 %, it is shown that in-
performed experiment. The difference between the final value of the red between the MPP energy and the load energy there are losses to
line (PV energy + dark current loss) and the purple line of the load coupling, Id (PV dark current), wiring and battery, as shown in Fig. 8.
energy gives internal energy losses in the unit related to the battery and The losses due to Id - the dark current loss, can be eliminated using
wiring/resistance – internal losses in the unit beyond PV module, as blocking diode. However, we intentionally omitted any blocking ele­
shown in Fig. 8. The wiring loss was equal to 4.2 % of energy at MPP and ments in the main experiment to observe the magnitude of this dark
was obtained by subtracting the energy lost to the battery from the total current loss. Blocking diodes typically have noticeable forward voltage
internal energy losses. drops which causes energy loss especially during high energy generation
Green line in Fig. 8 shows the battery energy. The energy of the periods. The dark loss without diode must be compared to the diode loss
battery oscillates; rising during charging and falling during discharging. to evaluate usefulness of the blocking diode in the PV-battery device.
The baseline of the oscillating battery energy profile with a slope of 38 Alternatively, an “ideal diode” circuit emulating diode function with
mWh per cycle (226 mWh for the 6 cycles) is related to the losses in the negligible forward voltage drop can be used. For this purpose, an ideal
battery as no significant charge accumulation has been observed (see diode circuit was tested successfully for one experimental run. The loss
Fig. 6 and related description). The energy loss to the battery was due to the Id was eliminated.
approximately equal to 7.2 % of the energy at MPP. This loss in the Battery and wiring losses are always inherent to the systems with or
battery has two components – Faradaic losses related to the loss of without MPP tracking devices. These losses are battery technology
charge and potential or voltage losses related to the charge–discharge specific and in the context of PV-battery modules also depend on the
overpotentials (Ernst et al., 2019; Kasnatscheew et al., 2016). The bat­ capacity per unit area of a PV device (Astakhov et al., 2020).
tery energy roundtrip efficiency with respect to energy supplied to the Overall design of the directly coupled PV-battery device has two
battery in our experiment was approximately 91 %. Taking into account primary aspects: voltage matching and mutual scaling of PV and battery
Faradaic efficiency of 96 %, the voltage efficiency of the battery is components. Both aspects must be addressed considering utilization
approximately 95 % in our experiment, assuming that roundtrip effi­ scenario and degree of autonomy of the designed system (Astakhov
ciency is a product of the voltage and Faradaic efficiencies. et al., 2020). To achieve proper voltage matching, the battery operating
The integral total energy efficiency calculated as a ratio of the load voltage range must overlap closely with the range of the PV maximum
energy to MPP energy amounts to approx. 82.5 %. power point voltages. For each utilization scenario MPP voltages for the
major energy generation conditions should be taken. Example of this
4. Discussion kind of analysis and optimization is given in the work of (Ayeng’o et al.,
2019).
Results of the experiments with isolated effects of irradiance and Scaling of the PV and battery components can be expressed as a PV-
load power show very good coupling factor C (above 0.9) for illumina­ specific battery capacity, or capacity per unit area of the PV device. This
tions above 0.1 sun and for all SOC of the battery (see Fig. 4). The capacity depends on the insolation, PV-efficiency, targeted degree of
coupling factor generally decreases at low solar irradiance and also autonomy, and PV-specific power consumption (Astakhov et al., 2020).
drops at high battery SOC (close to 100 %), showing that the best SOC Considering possible variability of all mentioned factors there is a range
for harnessing maximum power from the PV is at low/ intermediate of solutions. We expect that utilization of PV-battery devices is mean­
battery SOC in the studied device. This is because, at maximum battery ingful once the battery can store at least the energy produced during one
SOC, PV module working point drifts further away from MPP, especially full sun hour. This requirement for Li-ion cell (approx. 4 V) and PV ef­
at lower solar irradiance in our PV-battery unit. The load does not have ficiency of 20 % results in approx. 50 Ah/m2 (m2 refers to the area of PV
any significant impact on the coupling factor when battery is present, at module) under one standard sun.
least for the cases when load energy consumption is approximately Issues concerning adverse conditions on the unit such as rain, cloud,
balanced with PV energy production over the daily cycle. Therefore, varying temperature of PV module through the daily cycles and shading
with regards to varying load conditions, the battery acts as an efficient are not covered in this paper and could be the focus of further research.
coupling element. The size of the unit could also increase the energy losses from different
In the cycling experiment shown in Fig. 5, the unit demonstrates very components of the unit and could be an interesting study to see how the
high coupling in the most relevant range of power generating condi­ energy losses scales with size for further studies.
tions, which results in cumulative coupling efficiency of 98.23 % at the
end of the experiment. This is an excellent result, which can be 5. Conclusion
compared to best MPP solutions from literature with MPP power effi­
ciency of 93–96 % (Abu Qahouq et al., 2014; Cao and Kim, 2015; Cooley In our work we experimentally studied power coupling and energy
and Leeb, 2011; Uprety and Lee, 2014, 2017). efficiency in a directly coupled PV-battery module exposed to realistic
Fig. 5 and Fig. 6 show a successful unaided operation of the directly cycles of irradiance and power consumption. We experimentally verified
coupled PV-battery load unit under realistic conditions with a good theoretical predictions of directly coupled PV-battery module to main­
stability noticeable from the plot for battery SOC. The unit tends to keep tain high internal power coupling even under variable irradiance and
stable cycling behavior, which is evidenced by the time profile of the load without MPPT electronics. The experiment was carried out with a
battery voltage in Fig. 6. laboratory scale unit of a 7-cell silicon heterojunction PV module
The unaided operation is the key point of the study and implies that directly connected to a lithium-ion battery and variable load.
there is no power electronics nor energy management system involved. The directly coupled PV-battery unit demonstrates stable and effi­
The flow of energy is naturally governed by the balance between the cient operation with average coupling efficiency over 98 % and approx.
generation of the PV module on the one hand and load resistance of the 100 % at high irradiance, which is comparable or better than reported
other with storage in between. The stabilizing and power coupling effect MPPT solutions considering self-consumption in the power electronics.
of the battery is related to the operating voltage stabilization due to The high degree of coupling is related to the voltage stabilizing function
steep battery IV. Whenever voltage is increasing exceeds VOC of the of the battery, which makes it an effective coupling element in addition
battery at high PV generation, the battery starts charging, and dis­ to its storage function. High degree of coupling is maintained even in the
charging when the voltage is slightly lower than the battery VOC. For periods when load power is significantly higher than PV power – the
the load it means nearly constant operating voltage. The power case relevant for pulsed consumption. At the same time, in the case of

239
U. Chibuko et al. Solar Energy 249 (2023) 233–241

simplest direct connection some fraction of battery energy gets lost via Duan, W., Lambertz, A., Bittkau, K., Qiu, D., Qiu, K., Rau, U., Ding, K., 2022. A route
towards high-efficiency silicon heterojunction solar cells. Prog. Photovolt. 30 (4),
dark current in PV module at night. This loss of approx. 4.4 % in our case
384–392.
can be eliminated with use of blocking diode circuit. Other losses related El Fadil, H., Giri, F., 2011. Climatic sensorless maximum power point tracking in PV
to battery and wiring, are common for systems with or without MPPT generation systems. Control Eng. Pract. 19 (5), 513–521.
devices. Ernst, S., Heins, T.P., Schlüter, N., Schröder, U., 2019. Capturing the current-
overpotential nonlinearity of lithium-ion batteries by nonlinear electrochemical
We believe that, depending on the scale and architecture of the PV impedance spectroscopy (NLEIS) in charge and discharge direction. Front. Energy
system with batteries, direct coupling can be viable or even preferable Res. 7, 151.
option from the efficiency, cost, reliability or simplicity points of view. Fagiolari, L., Sampò, M., Lamberti, A., Amici, J., Francia, C., Bodoardo, S., Bella, F.,
2022. Integrated energy conversion and storage devices: Interfacing solar cells,
batteries and supercapacitors. Energy Storage Mater. 51, 400–434.
Galliano, S., Bella, F., Bonomo, M., Giordano, F., Grätzel, M., Viscardi, G., Hagfeldt, A.,
Declaration of Competing Interest Gerbaldi, C., Barolo, C., 2021. Xanthan-based hydrogel for stable and efficient quasi-
solid truly aqueous dye-sensitized solar cell with cobalt mediator. Solar Rrl 5 (7),
2000823.
The authors declare that they have no known competing financial Howlader, H.O.R., Sediqi, M.M., Ibrahimi, A.M., Senjyu, T., 2018. Optimal thermal unit
interests or personal relationships that could have appeared to influence commitment for solving duck curve problem by introducing CSP, PSH and demand
response. IEEE Access 6, 4834–4844.
the work reported in this paper.
Jiang, J.-A., Su, Y.-L., Kuo, K.-C., Wang, C.-H., Liao, M.-S., Wang, J.-C., Huang, C.-K.,
Chou, C.-Y., Lee, C.-H., Shieh, J.-C., 2017. On a hybrid MPPT control scheme to
Acknowledgment improve energy harvesting performance of traditional two-stage inverters used in
photovoltaic systems. Renew. Sustain. Energy Rev. 69, 1113–1128.
Kakimoto, N., Asano, R., 2017. Linear operation of photovoltaic array with directly
The German Federal Ministry of Education and Research (BMBF) is connected lithium-ion batteries. IEEE Trans. Sustain. Energy 8 (4), 1647–1657.
acknowledged for funding the YESPV-NIGBEN -African CLIENT II proj­ Kasnatscheew, J., Rodehorst, U., Streipert, B., Wiemers-Meyer, S., Jakelski, R.,
Wagner, R., Laskovic, I.C., Winter, M., 2016. Learning from overpotentials in lithium
ect (BMBF Förderkennzeichen 03SF0576A-B) under which this research ion batteries: a case study on the LiNi1/3Co1/3Mn1/3O2 (NCM) cathode.
was conducted. The authors would like to thank the silicon hetero­ J. Electrochem. Soc. 163 (14), A2943.
junction solar cell baseline; Alain Doumit for the wafer texture and Kin, L.-C., Astakhov, O., Lee, M., Haas, S., Ding, K., Merdzhanova, T., Rau, U., 2022a.
Batteries to Keep Solar-Driven Water Splitting Running at Night: Performance of a
cleaning; Silke Lynen, Volker Lauterbach, Andreas Mück for PECVD Directly Coupled System. Solar RRL.
deposition; Hildegard Siekmann for the ITO sputtering; and Dr. Kaining Kin, L.-C., Liu, Z., Astakhov, O., Shcherbachenko, S., Kungl, H., Kirchartz, T., Eichel, R.-
Ding, Dr. Andreas Lambertz and Dr. Weiyuan Duan for scientific sup­ A., Rau, U., Merdzhanova, T., 2022b. Efficient indoor light harvesting with
CH3NH3Pb (I0. 8Br0. 2) 3 solar modules and sodium-ion battery. Cell Reports Phys.
port. The authors would also like to acknowledge the contribution of Sci. 101123.
Christoph Zahren for technical support, software support and guidance. Kosowatz, J., 2018. Energy Storage Smooths the Duck Curve. Mech. Eng. 140 (06),
Thomas Birrenbach is also acknowledged for his contribution to the 30–35.
Lee, M., Ding, X., Banerjee, S., Krause, F., Smirnov, V., Astakhov, O., Merdzhanova, T.,
encapsulation of the 7-cell SHJ module.
Klingebiel, B., Kirchartz, T., Finger, F., Rau, U., Haas, S., 2020. Bifunctional
CoFeVOx Catalyst for Solar Water Splitting by using Multijunction and
Appendix A. Supplementary material Heterojunction Silicon Solar Cells. Adv. Mater. Technol. 5 (12), 2000592.
Leible, V., Bessler, W.G., 2021. Passive hybridization of photovoltaic cells with a lithium-
ion battery cell: An experimental proof of concept. J. Power Sources 482, 229050.
Supplementary data to this article can be found online at https://doi. Luo, H., Li, P., Ma, J., Han, L., Zhang, Y., Song, Y., 2022. Sustainable Pb Management in
org/10.1016/j.solener.2022.11.040. Perovskite Solar Cells toward Eco-Friendly Development. Adv. Energy Mater. 12
(30), 2201242.
Ma, J., Li, Y., Grundish, N.S., Goodenough, J.B., Chen, Y., Guo, L., Peng, Z., Qi, X.,
References Yang, F., Qie, L., Wang, C.-A., Huang, B., Huang, Z., Chen, L., Su, D., Wang, G.,
Peng, X., Chen, Z., Yang, J., He, S., Zhang, X., Yu, H., Fu, C., Jiang, M., Deng, W.,
Sun, C.-F., Pan, Q., Tang, Y., Li, X., Ji, X., Wan, F., Niu, Z., Lian, F., Wang, C.,
Abu Qahouq, J.A., Jiang, Y., Orabi, M., 2014. MPPT Control and Architecture for PV
Wallace, G.G., Fan, M., Meng, Q., Xin, S., Guo, Y.-G., Wan, L.-J., 2021. The 2021
Solar Panel with Sub-Module Integrated Converters. J. Power Electron. 14 (6),
battery technology roadmap. J. Phys. D Appl. Phys. 54 (18), 183001.
1281–1292.
Masoum, M.A., Dehbonei, H., Fuchs, E.F., 2002. Theoretical and experimental analyses
Adeoye, O., Spataru, C., 2019. Modelling and forecasting hourly electricity demand in
of photovoltaic systems with voltageand current-based maximum power-point
West African countries. Appl. Energy 242, 311–333.
tracking. IEEE Trans. Energy Convers. 17 (4), 514–522.
Agbo, S.N., Merdzhanova, T., Yu, S., Tempel, H., Kungl, H., Eichel, R.-A., Rau, U.,
Midya, P., Krein, P.T., Turnbull, R.J., Reppa, R., Kimball, J., 1996. Dynamic maximum
Astakhov, O., 2016a. Development towards cell-to-cell monolithic integration of a
power point tracker for photovoltaic applications, PESC Record. 27th Annual IEEE
thin-film solar cell and lithium-ion accumulator. J. Power Sources 327, 340–344.
Power Electronics Specialists Conference. IEEE, pp. 1710-1716.
Agbo, S.N., Merdzhanova, T., Yu, S., Tempel, H., Kungl, H., Eichel, R.-A., Rau, U.,
Mohanty, P., Muneer, T., Gago, E.J., Kotak, Y., 2016. Solar radiation fundamentals and
Astakhov, O., 2016b. Photoelectrochemical application of thin-film silicon triple-
PV system components, Solar Photovoltaic System Applications. Springer, pp. 7-47.
junction solar cell in batteries. Physica Status Solidi (a) 213 (7), 1926–1931.
Ram, M., Bogdanov, D., Aghahosseini, A., Gulagi, A., Oyewo, A., Child, M., Caldera, U.,
Astakhov, O., Merdzhanova, T., Kin, L.-C., Rau, U., 2020. From room to roof: How
Sadovskaia, K., Farfan, J., Barbosa, L., 2019. Global energy system based on 100%
feasible is direct coupling of solar-battery power unit under variable irradiance? Sol.
renewable energy–power, heat, transport and desalination sectors. Study by
Energy 206, 732–740.
Lappeenranta University of Technology and Energy Watch Group, Lappeenranta,
Astakhov, O., Agbo, S.N., Welter, K., Smirnov, V., Rau, U., Merdzhanova, T., 2021.
Berlin.
Storage batteries in photovoltaic–electrochemical device for solar hydrogen
Rana, M.M., Uddin, M., Sarkar, M.R., Shafiullah, G.M., Mo, H., Atef, M., 2022. A review
production. J. Power Sources 509, 230367.
on hybrid photovoltaic – Battery energy storage system: Current status, challenges,
Ayeng’o, S.P., Axelsen, H., Haberschusz, D., Sauer, D.U., 2019. A model for direct-
and future directions. J. Storage Mater. 51.
coupled PV systems with batteries depending on solar radiation, temperature and
Sandbaumhüter, F., Agbo, S.N., Tsai, C.-L., Astakhov, O., Uhlenbruck, S., Rau, U.,
number of serial connected PV cells. Sol. Energy 183, 120–131.
Merdzhanova, T., 2017. Compatibility study towards monolithic self-charging power
Belhachat, F., Larbes, C., 2018. A review of global maximum power point tracking
unit based on all-solid thin-film solar module and battery. J. Power Sources 365,
techniques of photovoltaic system under partial shading conditions. Renew. Sustain.
303–307.
Energy Rev. 92, 513–553.
Schmitz, F., Lago, N., Fagiolari, L., Burkhart, J., Cester, A., Polo, A., Prato, M.,
Cao, G., Kim, H.-J., 2015. A novel analog maximum power point tracker for low-cost and
Meneghesso, G., Gross, S., Bella, F., 2022. High open-circuit voltage Cs2AgBiBr 6
low-power distributed PV systems. IEEJ Trans. Electr. Electron. Eng. 10 (4),
carbon-based perovskite solar cells via green processing of ultrasonic spray-coated
474–478.
carbon electrodes from waste tire sources. ChemSusChem.
Chime, U., Wolf, L., Buga, V., Weigand, D., Gad, A., Köhler, J., Lambertz, A., Duan, W.,
Sheha, M., Mohammadi, K., Powell, K., 2020. Solving the duck curve in a smart grid
Ding, K., Merdzhanova, T., 2022. How Thin Practical Silicon Heterojunction Solar
environment using a non-cooperative game theory and dynamic pricing profiles.
Cells Could Be? Experimental Study under 1 Sun and under Indoor Illumination.
Energ. Conver. Manage. 220, 113102.
Solar RRL 6 (1), 2100594.
Uprety, S., Lee, H., 2014. 23.6 A 43V 400mW-to-21W global-search-based photovoltaic
Cooley, J.J., Leeb, S.B., 2011. Per panel photovoltaic energy extraction with multilevel
energy harvester with 350μs transient time, 99.9% MPPT efficiency, and 94% power
output DC-DC switched capacitor converters, 2011 Twenty-Sixth Annual IEEE
efficiency, 2014 IEEE International Solid-State Circuits Conference Digest of
Applied Power Electronics Conference and Exposition (APEC). pp. 419-428.
Technical Papers (ISSCC). IEEE, pp. 404-405.
de Haro, J.C., Tatsi, E., Fagiolari, L., Bonomo, M., Barolo, C., Turri, S., Bella, F.,
Uprety, S., Lee, H., 2017. 22.5 A 93%-power-efficiency photovoltaic energy harvester
Griffini, G., 2021. Lignin-based polymer electrolyte membranes for sustainable
with irradiance-aware auto-reconfigurable MPPT scheme achieving> 95% MPPT
aqueous dye-sensitized solar cells. ACS Sustain. Chem. Eng. 9 (25), 8550–8560.

240
U. Chibuko et al. Solar Energy 249 (2023) 233–241

efficiency across 650µW to 1W and 2.9 ms FOCV MPPT transient time, 2017 IEEE Wang, X., Dong, B., Feng, M., Xue, D.-J., Wang, S.-M., 2022. Sustainable management of
International Solid-State Circuits Conference (ISSCC). IEEE, pp. 378-379. lead in perovskite solar cells. J. Mater. Chem. A 10 (30), 15861–15864.
Vega-Garita, V., Garg, S., Narayan, N., Ramirez-Elizondo, L., Bauer, P., 2018. Testing a Yoshikawa, K., Kawasaki, H., Yoshida, W., Irie, T., Konishi, K., Nakano, K., Uto, T.,
PV-battery Integrated Module Prototype, 2018 IEEE 7th World Conference on Adachi, D., Kanematsu, M., Uzu, H., 2017. Silicon heterojunction solar cell with
Photovoltaic Energy Conversion (WCPEC)(A Joint Conference of 45th IEEE PVSC, interdigitated back contacts for a photoconversion efficiency over 26%. Nat. Energy
28th PVSEC & 34th EU PVSEC). IEEE, pp. 1244-1248. 2 (5), 1–8.
Vega-Garita, V., Ramirez-Elizondo, L., Bauer, P., 2017. Physical integration of a Zavos, I., Design and Modeling of Switching Battery Management System for Solar-
photovoltaic-battery system: A thermal analysis. Appl. Energy 208, 446–455. powered Storage Installations.
Vega-Garita, V., Ramirez-Elizondo, L., Narayan, N., Bauer, P., 2018b. Integrating a Zubair, M., Bilal Awan, A., Al-Ahmadi, A., Abo-Khalil, A.G., 2018. NPC based design
photovoltaic storage system in one device: A critical review. Prog. Photovolt. Res. optimization for a net zero office building in hot climates with PV panels as shading
Appl. 27 (4), 346–370. device. Energies 11 (6), 1391.
Vega-Garita, V., Hanif, A., Narayan, N., Ramirez-Elizondo, L., Bauer, P., 2019. Selecting a
suitable battery technology for the photovoltaic battery integrated module. J. Power
Sources 438, 227011.

241

You might also like