You are on page 1of 8

A Simplified battery charge controller for

safety and increased utilization in


standalone PV applications
S. G. Tesfahunegn *,**, O. Ulleberg**, T.M. Undeland*, P.J.S. Vie**
* Norwegian University of Science and Technology, Trondheim, (Norway)
** Institute for Energy Technology, NO-2027 Kjeller, (Norway)

Abstract--This paper presents a new solar/battery charge series/shunt dissipative element or PWM control to
controller that combines both MPPT and over-voltage prevent over-charge [1]. Dissipative controllers don’t
controls as single control function. A small-signal model of disconnect the PV array/battery due to over-voltage but
lead acid battery is derived in detail to design the employed rather limit the charging current by dissipating part of the
dual-loop control configuration. Two case studies are then power at constant voltage and therefore have better PV
conducted, in SIMULINK/SIMPOWER, first to evaluate
energy capture than on/off controllers. However, the need
the performance of the designed controller in terms of
transient response and voltage overshoot. Secondly, realistic for heat sinking to dissipate unused energy limits
irradiance data is used to evaluate the performance of the dissipative controllers to smaller power applications. The
developed charge controller in terms of parameters such as PWM controller uses a power electronic switch such as
PV energy utilization factor and over-voltage compared to MOSFET to regulate the amount of charge flowing into
the conventional hysteretic on/of controller. The designed the battery by chopping the current at varying duty ratio
controller is demonstrated to have good transient response enabling control over how the battery reaches full charge
with only small voltage overshoot. It is also found that the and producing less heat. The above controllers may or
developed charge controller fares better in terms PV energy may not have maximum power point tracker (MPPT) as
utilization and shows at least the same level of over-voltage
an integral part of the controller depending on PV/battery
control.
Index Terms—Battery, charge controller, MPPT, PV interface voltage compatibility and cost considerations.
This paper presents a simplified solar/battery charge
I. INTRODUCTION controller that combines both MPPT and over-voltage
controls in a single control function. The charge
The performance of batteries in actual PV controller is basically an over-voltage controller with
applications is often not as good as data presented by MPPT capability and avoids need for use of two separate
battery manufacturers which is typically based on tests units or any mode change. This allows simpler and
conducted at more favorable conditions [1]. Premature cheaper implementation of the controller and ensures
failure of batteries is therefore a big challenge and seamless transition between MPPT and power limit
attributes to very high share of the running cost in such control objectives avoiding any delay or transients. Other
applications. In addition to understanding the features of the developed controller are high damping of
performance characteristics and operational requirements voltage over-shoot and constant voltage charging
of the battery itself, choice of proper control algorithm enabling full utilization of battery capacity. Although the
and regulation set points for the charge controller is very latter feature is also true of dissipative controllers, the
crucial. The ultimate goal of a charge controller in stand- proposed controller is not dissipative and produces less
alone PV systems is to maintain the highest possible heat. The only weakness of the proposed controller seems
state-of-charge while preventing battery over-charge to be in applications which don’t use MPPT where it may
during high solar insolation and avoid over-discharging be cheaper to use the other controllers. A small signal
during low insolation and excessive loading. A good model of lead acid battery, not available in literature, is
charge controller does this with the least amount of PV also derived in detail to enable accurate design of the
energy being dumped. Charge controllers can be charge controller.
classified as on/of controllers and soft controllers. In
on/off (interruptive) charge control [2-4] which comprise II. MODELING OF SUBSYSTEMS
the majority of controllers, all or part of the PV array or
the battery is disconnected to limit the energy flow into A. PV Model
the battery during over-voltage. In this case a portion of Based on the well known one diode model of a PV cell
the PV energy may be unused as the battery may not be shown in Fig. 1, the expression for the cell current can be
fully charged because of premature disconnection due to written as (1) where IL is the photon current, RS is the
battery voltage exceeding the disconnect voltage series resistance, A is the thermal potential and IS is the
regulation setpoint particularly because of internal saturation current.
resistance induced over-voltages at higher charging
­ U + IRs ½
currents. Soft controllers, on the other hand, use either I = I L − I s ®exp( ) − 1¾ (1)
¯ A ¿
This work was supported by the Research Council of Norway

978-1-4244-8930-5/11/$26.00 ©2011 IEEE 137


For suitability in interfacing with power electronic
converters, modeling the PV array as controlled voltage
source rather than controlled current source is preferred
here.

Fig. 2. Thevenin equivalent of lead acid battery

III. CONTROL LAW OF CHARGE CONTROLLER


To find a control law that regulates the battery voltage
Fig. 1. One diode model of PV cell in the face of a current disturbance using classical control
tools, a small signal model governing how a small change
Taking the natural logarithm of both sides of (1) and in battery current dynamically affects the battery voltage
scaling current and voltage for a PV array having Ns should be analyzed. For this purpose a small signal model
modules in series and Np strings in parallel gives the of a lead acid battery with battery current i as control
expression in (2) which can be used to model the PV input and battery voltage Vbatt as the controlled state
array as controlled voltage source. IPV and VPV represent variable is derived. The following assumptions are made
the array current and voltage respectively. while driving the small signal model and subsequent
transfer functions:
§ I L − I s − I PV / N p · 1. Operating point is near full battery charge
V PV = A. N s ln¨¨ ¸
¸
which corresponds to a maximum allowed
I battery thresh-hold voltage VTH since the
© s ¹ over-voltage controller is supposed to work
I close to this point as will be pointed out
− N s PV Rs (2) later.
Np 2. Capacity change due to change in current
amplitude is assumed to be small.
B. Battery Model 3. Any variations in resistance and capacitance
The lead acid battery model used in this study is due to SOC near operating point can be
based on the thevenin equivalent circuit model [5-7] neglected.
shown in Fig. 2 where VNL is the no load voltage and R1 is 4. Variation in load current is an external
the internal resistance. The over-voltage resistance and disturbance.
5. Battery current has positive polarity during
capacitance R2 and C represent the first order dynamics of
charging.
the battery. The no load voltage VNL is modeled using a
modified Shepherd equation given by (3) [8] which is A. Small signal model of battery
suitable for empirical implementation where V0 is the Applying circuit laws to Fig. 2, the large signal
constant voltage [V], K is the polarization constant dynamic model of the battery can be written as
[V/Ah], C is total battery capacity [Ah], q is given by (4),
A is exponential zone amplitude [V], 1/B is exponential
(v batt − v nl − R1i )
zone time constant [Ah-1] and SOC0 is the initial state of
charge [Ah]. The Shepherd equation is modified to
i= +C
d
(vbatt − v nl − R1i )
R2 dt
include the dependence of the battery capacity C on (5)
amplitude of current using the Peukert effect. For
simplicity, the effect of state of charge (SOC) level on Assuming battery current as control input, perturbing
battery resistance and capacitance are neglected without the current as in (6) by a small amount around a steady
loss of accuracy since their values don’t change much state operating point I will result in the corresponding
within the operating range of most battery applications. perturbations in (7) and (8) where upper case variables
C represent steady-state operating points while variables
V NL = V0 − K q + A exp(− B .q ) (3) with ‘~’ are small perturbations.
C −q
~
q = (1 − SOC 0 ) × C − ³ idt (4) i=I+i (6)
vbatt = Vbatt + v~batt (7)
The effect of charge leakage has been included by v nl = Vnl + v~nl (8)
connecting a large resistance in shunt with the battery.

138
In line with assumption 2 the perturbation in no load generated by the external voltage controller. Both
voltage due to a small change in current can be neglected. controllers are implemented as PI (Proportional Integral)
With this assumption, substituting (6-8) into (5) and regulator. The current controller C1 is designed using
equating perturbed variables on the left side with those on linear control theory (frequency response) based on the
the right side gives the small signal representation (9) of open loop transfer function GOL1(s)=GI(s)C1(s)H1(s)
the battery with battery current as the control input and where H1 is the current sensor gain and GI is the small
battery voltage as the controlled state variable. signal DC/DC converter plant which is given in (12). L,
R, C, r are the converter parameters given in Table I
~§ R · d ~ d v~ where R is the resistive equivalent of the load presented
i ¨¨1 + 1 ¸¸ + R1C i = C v~batt + batt (9) to the converter by the battery at maximum power point
© R2 ¹ dt dt R2 (MPP). Using the parameters listed in Table I, the PI
compensator with Kp and KI values also given in Table I
is designed to give a stable response with sufficient phase
Battery current is the difference between the PV
margin and band width.
current iPV and load current iL. Assuming all load changes
can be rejected by the controller as external disturbance,
§ 1 ·
any disturbance in battery current can be directly linked ~ ¨s + ¸ (12)
i V pv © RC ¹
to disturbance in PV current as (10). The immediate GI ( s) = ~ =
consequence of this is the battery voltage can be L 2 § L + RL + rRC · § 1+ R ·
d s +¨ ¸s + r¨ ¸
controlled by directly controlling the PV current. © LRC ¹ © LRC ¹

~ ~ To simplify the design of the voltage controller C2,


i = i PV − i L Ÿ i = i PV (10) the closed current control loop gain GCL1 is approximated
by the first order transfer function given in (13) having a
Now, for control analysis using frequency response, dominant time constant ’ which is equal to the inverse of
the small signal expression in (9) is transformed into the the cross-over angular frequency of GOL1 and whose gain
Laplace domain. Rearrangement in Laplace domain is the same as the inverse of current sensor gain H1. The
eventually yields the first order transfer function from approximation will hold well as long as the voltage
battery voltage to current given in (11) with time constant controller is designed to be sufficiently slower than the
=R2xC. This transfer function can be used to design the current controller. Here, since the current controller was
desired control law using linear control tools such as bode designed to be almost ten times faster, the desired
plots to enable regulating the battery voltage by adjusting response was obtained as expected.
current input.
~ 1
v~ ( s ) ( R1 + R2 ) + R1 R2 Cs i H
Gb ( s ) = batt = GCL1 ( s ) = ~ = 1
(13)
~
i ( s) 1 + τs
(11) iref 1 + τ ' s

Fig. 3. Control diagram of over-voltage controller

B. Design of Controller
Following a similar procedure as previous, the open
A buck converter connecting the PV array to the loop gain (14) is used to design the voltage controller C2
battery acts as an active block which enables both MPP where H2 is the voltage transducer gain. The PI
tracking and over-voltage control. To control the battery compensator with K’P and K’I given in Table I gives
voltage by controlling the current, two control loops are sufficient phase margin of PM=91.80 and cross over
employed in a cascaded internal and external loop frequency Ȧc=82.2rad/s as indicated in the open loop
configuration shown in Fig. 3. The control input to the bode plot in Fig. 4.
battery is the controlled current output of the converter.
§ K ·
The internal current controller directly controls the G OL 2 = ¨ K p + I ¸ ( H 2 )(G CL1 )(G b ) (14)
battery current by tracking the current setpoint which is © s ¹

139
Bode Diagram thresh hold point and the controller automatically starts
Gm = Inf , Pm = 91.3 deg (at 81.9 rad/sec)
20 shifting the PV operating point to limit the PV power
produced as the voltage nears the thresh hold point (i.e.
over-voltage control). The maximum power point current
Magnitude (dB)

0 referred to the output (inductor) side of the DC/DC


converter is dynamically calculated by the MPPT
algorithm as a function of the instantaneous irradiance,
-20 temperature and battery voltage. This imposes a dynamic
upper limit on the current going into the battery and
enables a seamless change between MPPT and over-
-40 voltage control operations realized in a single block
-70
without the need for switching between different modes.
-75 Finally, it is important to note that the voltage control
loop will produce a large reference current due to the
Phase (deg)

-80
accumulation of error at normal operation under MPPT
-85 due to its integral action. To prevent output windup when
the over-voltage control action starts, an anti-wind up is
-90
implemented to reset the integral output. On this note, it
-95 is important to point out also that as the over-voltage
1 2 3 4
10 10 10 10 controller’s operating regime is only near the threshold
Frequency (rad/sec) point where the battery resistance and capacitance are not
Fig. 4. Bode plot of open loop gain GOL2 expected to change, our original assumption to neglect
their dependence on state of charge will not entail any
error on the choice of the controller parameters.
IV. THE PROPOSED CHARGE CONTROLLER
Fig. 5 shows a schematic of the proposed charge V. RESULTS AND DISCUSSION
controller where the current reference generated by the The studied system comprises a PV array composed
voltage control loop is dynamically limited to have an of 6 parallel strings each having 3 modules in series, i.e.,
upper value equal to maximum power point current. a total of 18 modules. The module used is the SCM210
(220Wp) type from REC SOLAR. The battery used is
sealed lead acid battery type. Battery data used such as
resistance, capacitance, etc. are the same as the sealed
lead acid battery cell studied in [11]. All the parameters
of the studied system including the selected controller
gains are given in Table I.
To evaluate the developed charge controller, two
case studies are conducted first to assess its performance
in terms of transient response and voltage overshoot.
Secondly, realistic irradiance data is used to evaluate the
performance of the charge controller in terms of
important parameters such as PV energy utilization factor
and over-voltage compared to the conventional hysteretic
on/of controller.
Fig. 5. Proposed simplified charge controller
A. Case study 1
The DC/DC converter can be thought of as a In this case study the charge controller response to a
controlled current source that injects a given amount of step increase in irradiance and fast load removal over 550
current i into the battery depending on the extent of seconds period is evaluated. The PV array defined earlier
deviation in the battery voltage from a set thresh-hold (6x3 modules) is employed. The battery bank used has a
value Vbatt,ref. If a voltage source type load is connected at nominal voltage of 48V and is composed of 21 cells
the output of a DC/DC converter, the output power can be connected in series in one string with a total capacity of
maximized by increasing the output current [9, 10]. In 8.0Ah. The thresh hold voltage of Vbatt,ref =50.5V is used.
this case since the load seen by the DC/DC converter is Fig. 6, 7, 8 and 9 show the responses obtained for a step
battery which is a voltage source type load, so long as increase in irradiance (from 0.2 to 1 kW/m2, at 200C
there is an error between the battery voltage and thresh ambient temperature) at 150s and a fast load
hold voltage setpoint, the reference current generated will disconnection at 225s both intended to cause over-
significantly increase as a result of the over-voltage voltages. It can be observed from Fig. 6 that initially the
control action. The output current will, however, never PV system operates at MPPT corresponding to about 15A
exceed the maximum current due to the dynamic (at 0.2kW/m2) seen from the battery side. During the
limitation. The PV array, therefore, will always work at sudden changes in irradiance and load, a temporary over-
MPP at battery voltages away from the over voltage voltage that arises prompts the controller to immediately

140
start over-voltage control action. The resulting voltage comparative study is the hysteretic series on/off controller
overshoot is quickly damped and is limited to within 2% as it is generally used in small and large systems. The
of the thresh hold value by the controller action. All hysteretic band used is ±0.4V which is recommended for
possible over-voltages, therefore, can be completely the type of battery used in this study. The disconnection
eliminated by having the thresh hold value setting 2% voltage is selected to be the same as the thresh hold
less, for example, than a presumed dangerous level. The voltage setpoint of the proposed controller and hence the
zoom in given in Fig. 9 demonstrates that the controller reconnection point is set to 49.7V. The on/off controller
can restore the voltage in less than 0.2 seconds which is is assumed to have MPPT for comparative purposes. A
in agreement with the band-width of the designed constant load of 500W having a constant peak of 1kW
controller. between 11AM and 9:00PM is used.

45 96

40
94
35

30 92
PV current (A)

SOC (%)
25
90
20

15 88

10
86
5

0 84
0 100 200 300 400 500 550 0 100 200 300 400 500 550
Time (s) Time (s)
Fig. 6. Response to irradiance increase (150s) and load removal (225s) Fig. 8. Battery state of charge

52

51.5 51

51
50.9

50.5
50.8
Vbatt (V)
Vbatt (V)

50
50.7
49.5

50.6
49

48.5 50.5

48 50.4
0 100 200 300 400 500 550 224.7 224.8 224.9 225 225.1 225.2 225.3
Time (s) Time (s)
Fig. 7. Battery voltage response Fig. 9. Zoom in of battery voltage

B. Case study 2
Fig. 10 and 11 show the available PV current on
In this study long time simulation is conducted based April 26 and August 11 if all was to be utilized at
on real irradiance data input obtained from Oslo to maximum power point under ideal conditions provided
evaluate the savings on the amount of PV dump the battery will not see over-voltage. In Fig. 12 and 13 the
compared to a conventional on/off charge controller. The current captured/utilized by the controllers is shown. Fig.
PV array used is the same as case study 1 but the battery 14 to 17 give the resulting battery voltage and state of
bank now comprises a parallel connection of 19 of the charge profiles during the two days. It is seen that the
string used in case 1 to give 152Ah capacity. on/off controller has to disconnect the battery when the
Performance is evaluated on two typical days-April 26 voltage reaches the disconnect voltage while the proposed
and August 11. The reference controller used for the

141
controller continues charging the battery at the thresh TABLE II
ENERGY UTILIZATION OF EACH CONTROLLER
hold voltage. Once the on/off controller disconnects the
April 26 August 11
battery, the battery has to remain disconnected for some Controller
time until the reconnect voltage is reached. In the mean kWh Uf (%) kWh Uf (%)
time some of the available PV energy will not be utilized On/off 13.06 83.5 13.68 78
by the on/off controller. Therefore, the proposed Proposed 13.66 87.4 13.9 79.38
controller will be able to capture more energy than the Only MPPT 15.64 100 17.51 100
on/off controller. This is reflected in the battery ending
the day with more charge for the proposed controller than
the on/off controller. Table II gives the total amount of Finally, it is important to note from Fig. 14 and 15 that
PV energy available during each day and the amount both controllers maintain the battery voltage at or below
captured by each controller. For comparison, a utilization the thresh hold setpoint. However, the on/off controller
factor Uf is defined to give a measure of the percentage of has one important limitation. Some applications may
utilized PV energy cause very frequent on/off chattering as the battery nears
full charge requiring many on/off controllers be built with
timed on/off control to avoid oscillation. The on/off
kWh( utilized ) controller will not be able to follow over-voltages
Uf = (15)
occurring more frequently than the timed period. On the
kWh( available )
other hand, the proposed controller doesn’t have this
limitation since it will be able to compensate any
From Table II it can be observed that the proposed transient very quickly guaranteeing better safety. Hence,
charge controller has better utilization factor than the the proposed charge controller should be the preferred
on/off controller as expected. The results also show that choice in such systems.
the actual energy utilized is 10-20% less than if only
MPPT were used which is assumed in most simulation 50
studies found in literature which often wrongly ignore the
realistic limitation of the energy captured due to battery
over-charge. 40

TABLE I
PARAMETERS OF STUDIED SYSTEM
PV current (A)

PV array Lead acid battery cell PI regulators 30


Parameter Value Param. Value para Value
Vmpp (V) 84.9 Q (Ah)/cell 8.0 Kp 100
Impp (A) 46.2 R1/cell () 0.0026 KI 0.0077 20
VOC (V) 110 R2/cell () 0.0005 K’p 0.92
ISC (A) 49.8 C/cell (F) 23 K’I 318.86
Pmpp (kWp) 3.96
10
DC/DC converter
Parameter Value Parameter Value
Vin,max(V) 150 DSS 0.57
Vout,nom (V) 48 ILSS (A) 82.5 0
fs (kHz) 20 VCSS (V) 48 0 4 8 12 16 20 24
L (H) 330 R (),r () 0.58, 0.05 Time (h)
C (F) 50 H1, H2 0.04,0.167 Fig. 10. Available PV current (April 26)
50
The studied system is characterised by a lower number 45
of batteries connected in series and more kWh capacity
was added by conneting more strings in parallel instead 40
of adding more batteries in series to maintain a nominal 35
48V bus. In other systems having a large number of
PV current (A)

battries connected in series, the internal resistance will 30


increase significantly while more current will flow 25
through each string. In the latter arrangement, specially
when battery bank is small compared with the size of the 20

PV array, there would be larger over-voltages occuring 15


due to more current flowing through bigger internal
resistances causing the voltage to reach the disconnect 10

setpoint more frequently. Consequently, the on/off 5


controller would be able to capture even much less
0
energy due to frequent disconnections. In this case, the 0 4 8 12 16 20 24
proposed controller will fare even much better than the Time (h)
on/off controller in terms of energy utilization. Fig.11. Available PV current (August 11)

142
On/off controller Proposed controller On/off controller Proposed controller
50 50
45

40 40

35

30 30

Current (A)
Current (A)

25

20 20

15

10 10

0 0
0 4 8 12 16 20 24 0 4 8 12 16 20 24
Time (h) Time (h)
Fig. 12. Current captured by each controller (April 26) Fig.13. Current captured by each controller (August 11)

On/off controller Proposed controller On/off controller Proposed controller


51.5 51

50.5 50

49.5 49

48.5 48
Vbatt (V)

Vbatt (V)

47.5 47

46.5 46

45.5 45

44.5 44

43.5 43

42.5 42
0 4 8 12 16 20 24 0 4 8 12 16 20 24
Time (h) Time (h)
Fig. 14. Battery voltages (April 26) Fig.15. Battery voltages (August 11)

On/off controller Proposed controller On/off controller Proposed controller


100 100

90 90

80 80
SOC (%)

SOC (%)

70 70

60 60

50 50

40 40

30 30
0 4 8 12 16 20 24 0 4 8 12 16 20 24
Time (h) Time (h)
Fig.16. Battery state of charge (April 26) Fig.17. Battery state of charge (August 11)

143
VI. CONCLUSIONS [8] C. M. Shepherd: Design of Primary and Secondary Cells -
Part 2. An equation describing battery discharge. Journal
A simplified solar/battery charge controller which of Electrochemical Society, vol. 112, 1965, pp. 657-664.
combines both MPPT and over-voltage controls as single [9] J. H. R. Enslin and D. B. Snyman: Simplified feed-forward
control function is proposed. A small signal model of control of the maximum power point in PV installations.
lead acid battery, not available in literature, is also Proc. of International Conf. on Ind. Electronics, Control,
Instrum., and Automation, 1992, pp. 548-553.
derived in detail to enable accurate design of the [10] A. S. Kislovski and R. Redl: Maximum-power-tracking
developed charge controller. Two case studies are using positive feedback. Proc. of 25th Annual IEEE Power
conducted first to evaluate the transient and voltage Electron. Specialists Conference, 1994, pp. 1065-1068.
overshoot response of the designed controller. Secondly, [11] B. S. Bhangu, P. Bentley, D. Stone, C. Bingham: Observer
a comparative study is made based on realistic irradiance techniques for estimating the State-of-Charge and State-of-
Health of VRLABs for Hybrid Electric Vehicles. Proc. of
data to evaluate the performance of the proposed charge Vehicle power and Propulsion conference, 2005, pp. 780-
controller in terms of energy utilization factor and over- 789.
voltage compared to the conventional series hysteretic
on/off controller.
The designed controller is shown to have very fast
transient response and very small transitory voltage
overshoot. It is also found that the proposed charge
controller shows better PV energy capture than the on/off
controller. While the studied system is characterized by
lower battery bank internal resistance, it has been inferred
that in systems with higher nominal voltages and thus
higher internal resistance and having installed battery
capacity which is small compared to the PV array, the
proposed controller will have significantly higher
utilization factor and hence should be the preferred
controller. In terms of over-voltage, both controllers were
able to maintain the battery voltage at or below the thresh
hold setpoint in the system considered here. In systems
with very rapid on/off operations near full charge the
on/off controller could risk oscillation or would have
difficulty following very frequent over-voltages if built
with timed on/off period. The proposed charge controller,
on the other hand, guarantees battery safety under all
conditions. The only weakness of the proposed controller
seems to be in applications which don’t use MPPT where
it may be cheaper to use the other controllers.

REFERENCES
[1] J. P. Dunlop and P.E.: Batteries and Charge Control in
Stand-Alone Photovoltaic Systems. Sandia National
Laboratories, Cocoa, FLJan. 15, 1997.
[2] T. D. Hund and B. Thompson: AmpHour Counting Charge
Control for Photovoltaic Hybrid Power Systems. Presented
at the 26th IEEE PVSC, Anaheim, CA, Sept. 29-Oct.
3,1997.
[3] R. P. L. Sanidad: Y. Baghzod and R. Boehm: Effect Of
On/Off Charge Controller On Stand-Alone Pv System
Performance. Center for Energy Research, University of
Nevada, Las Vegas2000.
[4] H. Yang,H.Wang, G. Chen, G. Wu: Influence of the charge
regulator strategy on state of charge and lifetime of VRLA
battery in household photovoltaic systems. Solar Energy,
vol. 80, 2005, pp. 281-287.
[5] H. L. Chan and D. Sutanto: A New Battery Model for use
with Battery Energy Storage Systems and Electric Vehicles
Power Systems.
[6] M. Dur, A. Cruden., S. Gair, J. Mcdonald: Dynamic model
of a lead acid battery for use in a domestic fuel cell system.
Journal of Power Sources, vol. 161, 2006, pp. 1400-1411.
[7] Z. M. Salameh, M. Casacca, W. Lynch: A Mathematical
Model for Lead-Acid Batteries. IEEE Transactions on
Energy Conversion, vol. 7, 1992, pp. 93-98.

144

You might also like