Sustainable Energy Technologies and Assessments 11 (2015) 126–133

Contents lists available at ScienceDirect

Sustainable Energy Technologies and Assessments

journal homepage: www.elsevier.com/locate/seta

Original Research Article

Lead-acid battery response to various formation levels – Part B: Internal

resistance
Justin Deveau ⇑, Chris White, Lukas G. Swan
Renewable Energy Storage Laboratory, Mechanical Engineering, Dalhousie University, PO BOX 15000, B3H 4R2 Halifax, Canada

a r t i c l e i n f o a b s t r a c t

Article history: In Part A of this study, eight lead-acid battery cells were formed to different levels to investigate their
Received 24 March 2015 performance in conventional and off-grid solar photovoltaic applications. In Part B of the study (this arti-
Revised 14 August 2015 cle) the objective is to investigate the internal resistance of the cells as a function of finished formation
Accepted 16 August 2015
level. Cells were formed to levels corresponding to 0.7, 1.4, 1.9, 2.3, 2.8, 4.2, 5.6, and 7.0 times their the-
oretical capacity, and cycled 10 times using a deep-cycle algorithm. A correlation between formation
level and internal resistance is shown. Higher formation levels resulted in lower internal resistance val-
Keywords:
ues for both discharging and charging processes, although they differed in magnitude. It is also shown
Lead-acid battery
Formation level
that internal resistance values of the under-formed cells start converging with the remaining cells after
Internal resistance 4 cycles as they complete their formation process. It is recommended that manufacturers of under-
Off-grid formed cells suggest higher voltage setpoints during the constant-voltage phase of the charge. This would
Solar storage overcome the high internal resistance limitations of these aforementioned cells, thus allowing for faster
formation completion, while increasing their energy efficiency in deep-cycling operation.
Ó 2015 Elsevier Ltd. All rights reserved.

Introduction electrolyte will inherently feature an electrical double layer (EDL)

As introduced in Part A of this study, the initiative to develop
rural electricity services using off-grid solar photovoltaic systems
provides new sale opportunities for lead-acid battery manufactur-
ers. The lead-acid battery (LAB) is presently the most widely used
energy storage medium for off-grid systems due to comparatively
low cost, its wide availability, and its maturity [1,2]. Part A
investigated the LAB characteristics of amp-hour (Ah) capacity,
voltage, and temperature as a function of formation level using a
constant-current formation algorithm [3]. In Part B of this study
(this article), we will investigate the LAB characteristics of internal
resistance (IR) as a function of formation level.
Background and motivation
The IR components of a battery can be modeled by an equiva-
lent electric circuit, as shown in Fig. 1. In the diagram, the current
conducting elements, such as the tabs, grids, active material, and
electrolyte, are modeled with the series resistor, Ro (i.e., ohmic
resistance), while the charge transfer reactions are modeled
with the parallel resistor, Rct . A charged electrode immersed in
⇑ Corresponding author. Tel.: +1 506 442 2563; fax: +1 902 423 6711.
E-mail address: deveau.jd@dal.ca (J. Deveau).
at the interface. This phenomenon is explained in detail in [4].
The electrostatic charge and discharge of the EDL plays an impor-
tant role in the charge/discharge mechanisms of the battery and
is typically modeled as the parallel capacitor, C dl . Diffusive mass
transfer effects are usually included in the model as well, and are
typically represented by the Warburg impedance, Z W , which is a
constant-phase-element [4]. While the electrical response of the
ohmic resistance is essentially instantaneous, both the EDL and
the diffusion effects will have significant time constants in their
electrical responses to stimuli. Time constants for the EDL, while
current dependent, are typically on the order of seconds, while
time constants for diffusion are typically on the order of minutes
to hours [5].
By considering the equivalent electrical circuit in Fig. 1, it can be
seen that an open-circuit battery in equilibrium will have a total
voltage V bat equal to the electromotive force V EMF , which is equal
to 2.041 V + (RT/nF) ln ða2H2 SO4 =a2H2 O Þ, where R is the ideal gas con-
stant (8.314 J K1 mol1), T is the temperature in K, n is the number
of moles of electrons exchanged in the electrochemical reaction (i.
e., 2 for a LAB), F is the is the Faraday constant (96,485 C mol1),
and where aH2 SO4 and aH2 O are the activity coefficients of sulfuric
acid and water, respectively [6]. When a current is applied to the
battery, the corresponding voltage change is a result of the elec-
tronic dynamics in the metallic components (i.e., voltage across

http://dx.doi.org/10.1016/j.seta.2015.08.005
2213-1388/Ó 2015 Elsevier Ltd. All rights reserved.
 J. Deveau et al. / Sustainable Energy Technologies and Assessments 11 (2015) 126–133
List of Nomenclature and Symbols
Acronyms
a.c. alternating current
CC constant-current
CV constant-voltage
d.c. direct current
EDL electrical double layer
Ro ) as well as the ionic dynamics at the reaction layer (i.e., voltage
across the C dl  Rct  Z W sub-circuit). The total battery voltage V bat
is therefore increased or decreased from V EMF depending on the
direction of the current flow. More complex equivalent circuit
models can also include inductive effects (which are important at
high frequencies), components for both positive and negative
plates, geometric capacitance, self-discharge resistance, and addi-
tional resistor–capacitor sub-circuits, as additional time constants
improve the accuracy of the model. Examples of more complex
models can be found in [7–11].
With respect to the aforementioned ohmic resistance of the LAB
(Ro in Fig. 1), its three major components (i.e., the electrode, the
electrolyte, and the separators) have the greatest influence. Wag-
ner [12] states that in order to achieve ideal battery performance
and to reduce the effect of inhomogeneous current distribution
across the plates, the overall resistance of the electrode must be
minimized. This parameter can be controlled by the grid structure
and material of the plates, and by the composition of the active
material [12,13]. Lead-antinomy alloys, which are used in a wide
range of lead-acid battery applications such as stationary power
storage and automobiles, are strong, creep resistant, and can easily
be cast into the desired shape [14]. While these benefits result in
easier handling of the material for the remainder of the manufac-
turing process, lead-antimony alloys are 3–10% less conductive (i.
e., higher resistance) than comparable calcium or tin alloys, thus
resulting in reduced performance [14]. Regarding the composition
of the active material, considerable work has been done regarding
the addition of carbon additives in the paste of the negative active
material to improve the electrical conductivity at partial state-of-
charge operation [15–18]. For instance, Fernandez et al. [17]
demonstrated that the addition of a carbon additive to the negative
active material of a LAB evenly distributes the generation of PbSO4
throughout the thickness of the plate during cycling, which is
otherwise deposited on the surface of the plate, inhibiting the acid
diffusion to the reaction pores, as well as creating a high ohmic
resistance barrier.
Fig. 1. Simple equivalent electric circuit for a battery.
127
HFR high-frequency resistance
IR internal resistance
LAB lead-acid battery
SG specific gravity
SoC state of charge
The resistivity of the electrolyte, influenced by its concentration
and temperature, plays a significant role regarding the IR of a LAB
and consequently on its power capability [19]. Pavlov [19] has
shown that electrolyte specific gravity (SG) values ranging
between 1.100 (fully discharged LAB) and 1.275 (fully charged
LAB) [20] have the lowest resistivity (e.g., 1.7 X cm to 1.3 X cm,
respectively, for an electrolyte temperature of 25 °C), where being
outside this range results in a substantial increase in resistivity.
Furthermore, he has shown that the resistivity of the electrolyte
has an inverse relationship with temperature, with electrolyte
temperatures higher than 40 °C offering the lowest resistivity,
and temperatures below 0 °C resulting in a reduction in available
power and energy for a LAB [19].
Finally, the resistance of the separator is a function of its poros-
ity and tortuosity factor (i.e., the ratio of the mean path length the
ion will travel, to the actual thickness of the membrane) [21]. Thus,
one is able to reduce the electrical resistance of the separator by
increasing its porosity and by decreasing the tortuosity factor.
With these characteristics in mind, measuring the IR of a LAB
can be used to obtain information about its power performance
and its relative ‘‘state-of-health”. This can be accomplished using
two different methods: alternating current (a.c.) impedance mea-
surements, and direct current (d.c.) impedance measurements. In
its most advanced form, the a.c. impedance is measured using elec-
trochemical impedance spectroscopy, which injects an a.c. excita-
tion current into the battery with frequencies ranging from a few
millihertz to several kilohertz. Alternatively, a single a.c. current
frequency can be injected with the goal of isolating the ohmic
resistance from the other impeding components [22]. In either
case, Ohm’s law is used to calculate the impedance based on the
change in a.c. voltage and, in some cases, the corresponding phase
shift. Huet et al. [22] measured the high-frequency resistance
(HFR) of a LAB during charge and discharge conditions with the
goal of evaluating the feasibility of state of charge (SoC) character-
ization. They developed an electronic device capable of processing
the voltage response of a 1200 Ah, 2 V cell to a high-frequency cur-
rent perturbation (10 A a.c. at 500 Hz) supplied by a galvanostat.
Using a discharge current of 10 A d.c., the authors showed minimal
change in the HFR of the cell for the first 75% of discharge (i.e.,
900 Ah), after which it increased rapidly from 0.2 mX to 1.7 mX
as the cell reached a fully discharged state. The HFR of the cell
relaxed towards 1.0 mX once the cell stopped discharging and
was in open-circuit. The authors attributed the sharp increase at
the end of the discharge to either a reduction in available active
material, or to an increase in resistivity of the electrolyte in the
pores of the active mass. In contrast, the relaxation in HFR after
discharging the cell could be attributed to the dissolution of the
PbSO4 crystals formed at the end of discharge, or to the increase
in concentration of electrolyte ions diffusing towards the pores of
the active mass. Using a charging current of 10 A d.c., the authors
showed a quick initial decrease in HFR during charging of the
LAB from 1.0 mX to 0.25 mX, explained by the rapid dissolution
of unstable PbSO4 crystals. The resistance then increased slightly
for the second part of charge due to the progressive formation of
128 J. Deveau et al. / Sustainable Energy Technologies and Assessments 11 (2015) 126–133

H2 bubbles which blocked the active mass, and then finally Table 1
decreased again at the end of charge as the average lifetime of bub- Cell characteristics [27].

bles over the active mass became shorter. While the authors
showed a direct correlation between the SoC and HFR of the LAB,
it was found to be a non-linear relationship. This conclusion was
later reaffirmed by Huet et al. [23], who investigated the HFR of
a 45 Ah LAB, and found that ‘‘Monitoring the SoC from the single
value of HFR was found to be impossible since, for different cycling
rates, distinct values of SoC may correspond to the same value of
HFR.”
The d.c. resistance is also measured using Ohm’s law, based on
the relationship between the change in d.c. voltage resulting from
a change in d.c. current. While SoC characterisation using d.c. resis-
tance measurements has the same limitations as a.c. impedance
measurements addressed above [24], it is an important parameter
in determining the power performance of a LAB. Applying a d.c.
current pulse to the equivalent electric circuit in Fig. 1 would result
in an immediate voltage drop across Ro , while C dl would have to
charge/discharge to a new equilibrium voltage before the final
voltage drop across Rct and Z W would be apparent. This reveals
the importance of the d.c. pulse duration when measuring d.c.
resistance. Thus, if a measurement is taken before the EDL has
reached its new equilibrium, the true d.c. resistance will not be
observed. Kirchev et al. [25,26] used d.c. pulse charging of LAB
electrodes to compare the response of the EDL over a range of pulse
frequencies. Using pulse durations ranging from 25 ms to 50 s, the
authors found that the EDL of the positive and negative electrodes
had their most significant responses towards the end of charge,
and that the EDL response of the positive electrode had a very dif-
ferent dependence on the pulse frequency than that of the negative
electrode. As expected, longer pulse durations provided the most
time for the EDL to reach equilibrium, but the data presented by
the authors indicated that pulse durations as short as 5 s may be
enough to enable EDL equilibrium.
Objective
The preceding review shows that IR research has been done
with regards to the influence of SoC on IR, and the influence of IR
on the pulse charging of a LAB. No studies were found which dis-
cussed the influence of formation levels on IR during cycling. Thus,
the objective of this article (Part B) is to complement the former
article (Part A) by presenting the characteristic of IR for the LAB
while cycling (for both charging and discharging processes), as a
function of formation level it had previously received. This project
uses an applied engineering evidence-based approach using exper-
imental data, so that recommendations can be provided to battery
manufacturers seeking to enhance the performance of their as-
delivered product.
Method
The following section provides a brief review of this study with
respect to the cells employed, and the formation and test cycling
algorithms. A complete description was provided in Part A of this
study [3].
Cell type Flooded lead-acid
Number of positive plates 7
Number of negative plates 8
Plate height 293 mm
Plate width 143 mm
Positive plate thickness 4.32 mm
Negative plate thickness 3.05 mm
Dry weight 12.3 kg
Rated capacity 244 Ah
Theoretical capacity 710 Ah
the theoretical capacity (TC) was calculated based on the active
material of the positive plates.
Algorithms
The cells were formed and cycled using an Arbin BT-2000
power-cycler, rated 0–20 V and ±100 A, with auxiliary T-type ther-
mocouples and cell voltage ports. The SGs of the cells during the
performance analysis were measured using a digital density meter
(Anton Paar DMA 35) by taking samples of electrolyte from the top
of the cell through the fill port, and were corrected to a tempera-
ture of 20 °C using a correction factor of 0.00088 g/cm3/K. One
should note that due to the size of the cells, SG values obtained
using this method are subjected to the effects of acid stratification.
This effect was not taken into account due to the configuration of
the cell, and the resulting difficulties in obtaining SG values at dif-
ferent height intervals. The laboratory temperature was main-
tained between 20 and 25 °C for the duration of this study.
Formation algorithm
Prior to the one-shot container formation, the cells were soaked
for 1 h using electrolyte with an SG of 1.200. The formation was
then completed using a constant-current algorithm of 20 A. Eight
cells were formed in series, each to a different formation level
(coulombic value, Ah) in order to capture a wide range of formation
levels while measuring cell voltage, outside cell temperature at
mid-height, and cell SG. Additionally, distilled water was added
to the cells during formation when deemed necessary to ensure
the plates remained submerged, while electrolyte with an SG of
1.265 was used to top-up the cells if required at the end of forma-
tion. The selected formation levels are shown in Table 2, along with
their respective ratios of the calculated theoretical capacity. The
cells are named based on their ratio of theoretical capacity.
Cycling algorithm
The deep discharge cycling algorithm which was used to test
the eight cells is shown in Table 3. This algorithm is comprised
of a 4 h discharge rate at a constant-current (CC) to a voltage cutoff
setpoint, and a three-step charging configuration: CC, constant-
voltage (CV), and CC. Rest periods of 30 min are provided between
Table 2
Selected formation levels.

Cell name Formation input (Ah) Ratio of theoretical capacity

Cells
0.7TC 500 0.7
1.4TC 1000 1.4
Table 1 highlights the characteristics of the cells used in this 1.9TC 1333 1.9
study. The cells were provided to the laboratory ‘‘green”, which 2.3TC 1667 2.3
is to say the pasted and cured plates were assembled into cells, 2.8TC 2000 2.8
but electrolyte was not added, nor was formation current applied. 4.2TC 3000 4.2
5.6TC 4000 5.6
Furthermore, one should also note that the rated capacity is based
7.0TC 5000 7.0
on a 4 h discharge rate with an electrolyte SG of 1.280 [27], while
 J. Deveau et al. / Sustainable Energy Technologies and Assessments 11 (2015) 126–133
Table 3
Cycling algorithm.
Step # Step name
1 Rest – begin
2 Discharge
3 Rest – transition
4 Charge phase 1
5 Charge phase 2
6 Charge phase 3
7 Rest – end
discharge/charge transitions, and additionally require that the cell
be less than 28 °C before proceeding to the next step.
Finally, while not shown in Table 3, important substeps were
embedded in Steps 2 (discharge) and 4 (charge) of the cycling algo-
rithm consisting of current pulses of 10 s to 100 A (discharge) and
+100 A (charge), respectively. These pulses, which occur at the 5 Ah
charge/discharge capacity point, and at every multiple of 50 Ah
charge/discharge capacity points thereafter, are used to determine
the d.c. IR of the cell. During these current pulses, cell voltages
were logged every second in order to provide a higher resolution
of IR measurements. Pulse magnitudes of 39 A for discharging,
and +39 A for charging were chosen as they correspond to the dif-
ference between the regular cycling current of ±60.93 A and the
±100 A current limits of the Arbin power-cycler. Furthermore as
stated in Section Background and Motivation, Kirchev et al. [25]
found that the EDL pulse response of a LAB electrode was able to
approach equilibrium with a pulse duration of 5 s when alternating
between a 10-h rate current and a resting state. Thus, a pulse dura-
tion of 10 s was chosen for this study with the intent of providing
enough time for the capacitive pulse responses to stabilize, with-
out greatly affecting the SoC of the battery during the respective
charging and discharging pulses. Since diffusive mass transfer pro-
cesses are much slower than ohmic and EDL processes, diffusion
effects were not taken into consideration. Finally, it should be
noted that current pulses only occurred during the CC steps of
the cycling algorithm, as this allowed for the base current to
remain at 60.93 A.
Results and discussion
Each cell was cycled 10 times using the cycling algorithm pre-
sented in Table 3. Fig. 2 shows the measured discharge capacities
of the cells through all 10 cycles. More details of these results,
along with an analysis of formation can be found in Part A of this
study [3]. The average coulombic efficiency of the eight cells was
calculated to be 87.4% after 10 cycles.
Fig. 2. Cell discharge capacities as a function of cycle and cell formation level.
129
Description
30 s rest
Constant current (60.93 A, or C/4 Rate) until cell reaches 1.75 V
Rest 30 min and until cell temperature is <28 °C
CC of 60.93 A (C/4 Rate) until cell reaches a voltage of 2.50 V
CV at 2.50 VPC until 7.5 A is reached (2% of C/20 Rate)
CC of 9.15 A (C/27 Rate) for 4 h
Rest 30 min and until cell temperature is <28 °C
As shown in Fig. 2, the discharge capacities of cells 0.7TC and
1.4TC increases significantly with each cycle, thus confirming that
these two cells are still forming. Similarly for the remaining cells,
with the exception of cell 7.0TC, a slight increase in respective dis-
charge capacities is also noticed, with all eight cells settling at a
final capacity of approximatively 250 Ah, 2.5% higher than the
rated capacity of the cell. While cells 4.2TC, 5.6TC, and 7.0TC offer
higher discharge capacities than cell 2.8TC, the difference is rela-
tively insignificant when compared to the additional formation
that is required. Based on these results, it had previously been
determined that the ideal formation level for this cell is between
2.3 and 2.8 times the theoretical capacity of the cell, and depends
on the future applications of the battery [3].
Ten seconds d.c. internal resistance results
Fig. 3 shows the resultant IR values for cycles 1, 3, 4, 6, and 10 of
discharging pulses using 10 s voltage intervals as a function of
cycle number and discharged Ah.
As shown in Fig. 3, it is clear that the higher the formation level,
the lower the IR when respective discharging capacities are com-
pared. It should also be noted that the cell IR values of the different
formation levels start converging after 4 cycles, which coincides
with the convergence of total capacities as shown in Fig. 2. In other
words, as the cells continue their formation process due to their
respective deep-cycling operation, increasing amounts of PbO
(1012 X cm [28]) and PbSO4 (0.3  1010 X cm [29]) are formed
into PbO2 (105 to 1.2  106 X cm [28]) for the positive plates,
and Pb (20.6  106 X cm [30]) for the negative plates, which
both have ohmic resistances lower than PbO and PbSO4, thus
reducing the overall IR of the respective cells. Furthermore, it is
important to note that the positive active material of a LAB is com-
prised of a-PbO2 and b-PbO2 crystal zones, and hydrated gel zones
(i.e., Pb(OH)2) [31]. The a-PbO2, which is created during the first
stage of formation and forms the active mass skeleton, has a speci-
fic resistance of 8.4  104 X cm, while b-PbO2, which is created
130 J. Deveau et al. / Sustainable Energy Technologies and Assessments 11 (2015) 126–133
Fig. 3. Discharge IR with 10 s voltage intervals (cycles 1, 2, 4, 6, and 10).
during the second stage of formation and is electrochemically
active, has a specific resistance of 1.4  104 X cm [31,32]. Thus,
in fully formed batteries, the b-PbO2 crystal structure is simply
consumed/rebuilt over the already existing skeleton of the positive
plates during discharging and charging operation, respectively.
Although changes do occur to the primary structure of the positive
active material (i.e., a-PbO2) throughout the cycle life of the bat-
tery, these changes are much slower [33]. On the other hand, the
gel zones, which are electron and proton conductors, and which
facilitate the discharge reaction resulting in PbSO4 and H2O, have
relatively high resistance values when compared to the crystal
zones [34,35].
The shape of the IR curves shown in Fig. 3 can be explained as
follows. The first two values of IR, 5 and 50 Ah, are higher than
the IR values measured at a discharge capacity of 100 Ah, with
the exceptions of the cells which have not yet finished forming.
This is due to the oxygen layer which forms on the positive plates
of the cell during the overcharge of the charging phase. As stated
by Shukla et al. [36], this oxygen layer acts as a barrier during
the discharge of the cell, which limits its performance. This effect
would be less dominant if the cells were rested for an extended
period of time before discharging, instead of the prescribed
30 min rest period. The cells which are achieving rated capacity
will at this point experience their lowest measured IR values at a
discharge capacity of 100 Ah. The IR for these cells then increases
significantly until the end of discharge due to a reduction in avail-
able active material due to the conversion of Pb and PbO2 to PbSO4,
and a reduction in electrolyte SG (see Fig. 5 for an example).
Finally, it is important to note that IR during discharge is a
direct representation of the available power in the cell. In other
Fig. 4. Charge IR with 10 s voltage intervals (cycles 1, 2, 4, 6, and 10).
words, the higher the IR, the lower the available power in the cells,
which in turns adversely affects its energy efficiency.
Fig. 4 shows the results of the calculated IR values for cycles 1,
3, 4, 6, and 10 of charging pulses using 10 s voltage intervals. It
should be noted that as the charging current pulses only occurred
during the first phase of the charging process (CC phase only, see
Table 3 – Step 4), IR values obtained at a charging capacity of
200 Ah correspond to approximatively 73% of the total Ah input
received by these cells during their respective charging processes.
Furthermore, it is important to note that these charging IR values
are dependent on the status of the previous discharge. Specifically,
as each cell had previously attained a different discharge capacity
value, the corresponding Ah depletion values for the Charge (Ah)
values listed in Fig. 4 are different for each cell. Nonetheless, one
should note that from the perspective of a fully discharged state,
the cells are all on the same basis.
As shown in Fig. 4, the IR charging trend is relatively similar to
the discharging trend discussed above albeit with higher magni-
tudes of IR values (e.g., 0.94 mO discharging vs. 1.3 mO charging,
equivalent to a discharging/charging ratio of 72%, for cycle 1 of cell
2.8TC at the 100 Ah value). As was observed during discharge, the
higher formation levels correspond to lower values of IR when
respective charging capacities are compared. Furthermore, cell IR
values once again start converging during charge after 4 cycles. It
is important to note that the higher the IR during the CC bulk
charge (see Table 3 – Step 4), the higher the cell voltage. This
results in a premature shift to the absorption phase of the charging
algorithm (see Table 3 – Step 5), which can thus increase total
charging time. As such, for cells that are under-formed, the voltage
setpoint indicating the end of the bulk charge should be increased,
 J. Deveau et al. / Sustainable Energy Technologies and Assessments 11 (2015) 126–133
Fig. 5. SG, IR, and temperature as a function of Ah depletion where D = discharge and C = charge (cell 2.8TC, cycle 10).
as this would overcome said IR limitations and allow for faster for-
mation completion of these cells.
The first IR value during charging (i.e., 5 Ah) is shown to be
higher than the next two IR values for all eight cells, which is
due to the presence of PbSO4 on the positive and negative plates.
The measured IR values reach their lowest point during charging
at a capacity of 50 Ah due to the dissolution of the unstable PbSO4
crystal structure. IR values are then increasing until charge com-
pletion due to the increasingly limited reaction rate as the bulk
of the electrode materials become fully converted to their charged
state. This corresponds to a significance increase in the charge
transfer resistance (see Rct in Fig. 1), which will neglect the
decrease in ohmic resistance (see Ro in Fig. 1) associated with the
conversion of PbO and PbSO4, to PbO2 and Pb. The progressive
appearance of H2 bubbles which block the active material, as pre-
viously stated by Huet et al. [22], will also contribute to an increase
in IR towards the end of charge.
An example of the influence of electrolyte SG and cell tempera-
ture on IR as a function of Ah depletion is shown in Fig. 5 for cycle
10 of cell 2.8TC. It should be noted that the cell temperature has
been divided by 10 for presentation purposes.
As shown in Fig. 5, the SG of the cell decreases linearly during
discharge, yet does not start increasing significantly during charge
until an Ah depletion value of 50 Ah, corresponding to the CV
phase of the charging process (see Table 3 – Step 5). However, it
is believed that the SG values did increase at the start of the charg-
ing process, yet as the electrolyte samples were limited to being
taken from the top of the cell, the increase in electrolyte SG was
not noticeable until the gassing process became more pronounced
(i.e., CV phase of the charging process). Thus, it difficult to deter-
mine the influence of electrolyte SG on the rise in IR which occurs
between amp-hour depletion values of 100 Ah and 50 Ah. Addi-
tionally, one should note that the cell temperature increases by
about 4 °C during discharge, and by a further 4 °C during charge.
Effect of pulse duration on d.c. internal resistance results
In order to verify that the selected pulse duration of 10 s was
appropriate in determining the d.c. IR, a follow-up test was per-
formed on an additional LAB of the same type as all other cells
tested in this study. This cell was discharged and charged using
the same algorithm outlined in Section Cycling Algorithm, with
the exception of smaller pulse intervals (i.e., every 25 Ah rather
131
than every 50 Ah), a longer pulse duration (i.e., 20 s rather than
10 s), and a faster data logging rate during the pulse (i.e., 50 Hz
rather than 1 Hz). The voltage responses during the discharging
and charging pulses are shown in Fig. 6. During a given current
pulse, the initial ohmic voltage response occurs instantaneously,
after which the EDL is charged or discharged until a new equilib-
rium is achieved, or until the pulse terminates. Ideally, the EDL
would reach its new equilibrium before a d.c IR measurement is
made, as explained in Section Background and Motivation. Since
the EDL can be modeled as a capacitor, the charging rates of the
EDL during a pulse can be approximated by taking the time deriva-
tive of the voltage response. Actual EDL charging rates can be cal-
culated if the value of the EDL capacitance is known, as shown by
Kirchev et al. [25]. However, this requires more complex methods
involving electrochemical impedance spectroscopy and was not
addressed in this study.
A 50 Hz logging rate of the voltage values during the pulses
allowed for moving averages to be applied to the time derivative
of the voltage response, as shown in Fig. 6. Since charging pulses
were only applied during the CC portion of the charge, the highest
SoC shown in Fig. 6 corresponds to the end of the bulk charge, or
approximatively 80% SoC. For comparison purposes, the differen-
tial voltages for discharging and charging pulses are plotted on
the same scale, revealing that the overall capacitive effects are
much less pronounced during discharge than on charge, and most
pronounced at the end of the bulk charge. These results are in
agreement with Kirchev et al. [25].
Thus, the results in Fig. 6 show that the capacitive effects during
discharging pulses are dissipated within the first 2–3 s for all SoCs.
In contrast, the capacitive effects during the charging pulses are
shown to dissipate within the first 4–5 s for lower SoCs (i.e.,
<50% SoC), but require more time to dissipate at higher SoCs (i.e.,
>70% SoC). In fact, the capacitive effects during the pulse at the
highest SoC (i.e., 225 Ah charging pulse) required the majority of
the 20 s pulse duration to fully dissipate. This means that any IR
measurements taken near the end of the bulk charge may have
been slightly underestimated by using the prescribed 10 s pulse
duration. However, since the IR trends have already been shown
to increase rapidly at the end of the bulk charge, correcting for a
slight underestimation would not provide any additional informa-
tion regarding the trends of IR during deep-cycling operation.
To further explore the impact of the EDL on the observed IR
trends during cycling, IR values taken after 1 s of pulsing were
132 J. Deveau et al. / Sustainable Energy Technologies and Assessments 11 (2015) 126–133

Fig. 6. voltage and differential voltage values during discharging (left) and charging (right) pulses.

Fig. 7. IR trends calculated after 1 s and 10 s of pulsing (discharge & charge, cycle 10).
 J. Deveau et al. / Sustainable Energy Technologies and Assessments 11 (2015) 126–133
compared to those taken after 10 s of pulsing. An example of these
results is shown in Fig. 7, which gives a comparison of the mea-
sured IR values for cycle 10 using the two time intervals for the dis-
charging and charging processes, respectively. It is shown that the
overall magnitudes of IR values for both discharging and charging
processes are smaller when taken after 1 s of pulsing when com-
pared to 10 s of pulsing. This is to be expected after previously
establishing that the EDL requires more than 1 s to reach an equi-
librium state regardless of SoC and charging direction (Fig. 6).
However, Fig. 7 does indicate that despite the clear dependence
of IR magnitude on the pulse duration, the overall IR trend is rec-
ognizable after both 1 s and 10 s of pulsing, results which were
consistent for all 10 deep-discharge cycles.
Conclusions and recommendations
This study has shown a correlation between formation level and
IR, with higher formation levels resulting in lower IR values for
both discharging and charging processes. Furthermore, IR values
of the different formation levels start converging after 4 cycles,
coinciding with the convergence of total capacities as the under-
formed cells (i.e., cells 0.7TC, 1.4TC, and 1.9TC) finish their forma-
tion process. These results were shown to be relatively indepen-
dent of electrolyte SG, cell temperature, and IR measurement
intervals. This study could be further developed by measuring
the IR of the cells during the second and third phases of charging,
which would provide more details on the effect of the electrolyte
on the cells.
Based on these results, it is recommended for cells which have
been under-formed, that the voltage setpoint during the CV phase
of the charge be increased. This would overcome the high IR limi-
tations and allow for faster formation completion of these afore-
mentioned cells. Consequently, these cells would be left with
lower IR values throughout future use, increasing their energy effi-
ciency and allowing them to reach their rated capacities faster than
would be enabled by a normal charging algorithm.
Acknowledgements
The authors acknowledge Surrette Battery Company, and Pascal
Ferron, Plant Manager at Surrette Battery Company, who provided
us with the un-formed green cells and support to carry out this
research project. Additionally, the authors acknowledge the Natu-
ral Sciences and Engineering Research Council of Canada who also
provided funding for this research through a collaborative research
and development grant.
References
[1] Jossen A, Garche J, Sauer DU. Operation conditions of batteries in PV
applications. Sol Energy 2004;76:759–69.
[2] Pavlov D. Chapter 3 – formation of lead-acid batteries and structure of positive
and negative active masses. In: Parker DAJRGTMD, editor. Valve-regulated
lead-acid batteries. Amsterdam: Elsevier; 2004. p. 37–108.
[3] Deveau J, White C, Swan LG. Lead-acid battery response to various formation
levels – Part A: recommended formation levels for off-grid solar and
conventional applications. Sustain Energy Technol Assess 2015;11:1–10.
[4] Kurzweil P. CAPACITORS | electrochemical double-layer capacitors. In: Garche
J, editor. Encyclopedia of electrochemical power sources. Amsterdam: Elsevier;
2009. p. 607–33.
[5] Abu-Sharkh S, Doerffel D. Rapid test and non-linear model characterisation of
solid-state lithium-ion batteries. J Power Sources 2004;130:266–74.
[6] Pavlov D. Fundamentals of lead-acid batteries. Lead-Acid Batteries: Sci Technol
2011:29–114.
[7] Min C, Rincon-Mora GA. Accurate electrical battery model capable of
predicting runtime and I–V performance. IEEE Trans Energy Convers
2006;21:504–11.
133
[8] Kirchev A, Delaille A, Perrin M, Lemaire E, Mattera F. Studies of the pulse
charge of lead-acid batteries for PV applications: Part II. Impedance of the
positive plate revisited. J Power Sources 2007;170:495–512.
[9] Buller S, Thele M, Karden E, De Doncker RW. Impedance-based non-linear
dynamic battery modeling for automotive applications. J Power Sources
2003;113:422–30.
[10] Rodrigues S, Munichandraiah N, Shukla AK. A review of state-of-charge
indication of batteries by means of a.c. impedance measurements. J Power
Sources 2000;87:12–20.
[11] Karden E, Buller S, De Doncker RW. A frequency-domain approach to
dynamical modeling of electrochemical power sources. Electrochim Acta
2002;47:2347–56.
[12] Wagner R. SECONDARY BATTERIES – LEAD-ACID SYSTEMS | Electrode Design.
In: Garche J, editor. Encyclopedia of electrochemical power
sources. Amsterdam: Elsevier; 2009. p. 599–609.
[13] Pavlov D. Chapter 7 – additives to the pastes for positive and negative battery
plates. In: Pavlov D, editor. Lead-acid batteries: science and
technology. Amsterdam: Elsevier; 2011. p. 311–61.
[14] Prengaman RD. SECONDARY BATTERIES – LEAD-ACID SYSTEMS | Lead Alloys.
In: Garche J, editor. Encyclopedia of electrochemical power
sources. Amsterdam: Elsevier; 2009. p. 648–54.
[15] Křivík P, Micka K, Bača P, Tonar K, Tošer P. Effect of additives on the
performance of negative lead-acid battery electrodes during formation and
partial state of charge operation. J Power Sources 2012;209:15–9.
[16] Boden DP, Loosemore DV, Spence MA, Wojcinski TD. Optimization studies of
carbon additives to negative active material for the purpose of extending the
life of VRLA batteries in high-rate partial-state-of-charge operation. J Power
Sources 2010;195:4470–93.
[17] Fernández M, Valenciano J, Trinidad F, Muñoz N. The use of activated carbon
and graphite for the development of lead-acid batteries for hybrid vehicle
applications. J Power Sources 2010;195:4458–69.
[18] Pavlov D, Nikolov P, Rogachev T. Influence of carbons on the structure of the
negative active material of lead-acid batteries and on battery performance. J
Power Sources 2011;196:5155–67.
[19] Pavlov D. SECONDARY BATTERIES – LEAD-ACID SYSTEMS | Electrolyte. In:
Garche J, editor. Encyclopedia of electrochemical power
sources. Amsterdam: Elsevier; 2009. p. 610–9.
[20] Pavlov D. H2SO4 electrolyte – an active material in the lead-acid. Cell
2011:117–48.
[21] Boehnstedt W, Whear JK. SECONDARY BATTERIES – LEAD-ACID SYSTEMS |
Separators. In: Garche J, editor. Encyclopedia of electrochemical power
sources. Amsterdam: Elsevier; 2009. p. 620–31.
[22] Huet F, Nogueira RP, Torcheux L, Lailler P. Simultaneous real-time
measurements of potential and high-frequency resistance of a lab cell. J
Power Sources 2003;113:414–21.
[23] Huet F, Nogueira RP, Lailler P, Torcheux L. Investigation of the high-frequency
resistance of a lead-acid battery. J Power Sources 2006;158:1012–8.
[24] Waag W, Sauer DU. SECONDARY BATTERIES – LEAD-ACID SYSTEMS | State-of-
Charge/Health. In: Garche J, editor. Encyclopedia of electrochemical power
sources. Amsterdam: Elsevier; 2009. p. 793–804.
[25] Kirchev A, Perrin M, Lemaire E, Karoui F, Mattera F. Studies of the pulse charge
of lead-acid batteries for PV applications: Part I. Factors influencing the
mechanism of the pulse charge of the positive plate. J Power Sources
2008;177:217–25.
[26] Kirchev A, Mattera F, Lemaire E, Dong K. Studies of the pulse charge of lead-
acid batteries for photovoltaic applications: Part IV. Pulse charge of the
negative plate. J Power Sources 2009;191:82–90.
[27] Surrette Battery Company, S-480 Spec Sheet, in, <http://rollsbattery.com/
public/specsheets/S-480.pdf/> (Accessed on 26.11.14).
[28] Pavlov D. Lead alloys and grids. Grid Des Princ 2011:149–221.
[29] Barak M. Electrochemical Power Sources: primary and secondary batteries,
Institution of Engineering and Technology, 1980.
[30] American Elements, Lead Sponge, in, www.americanelements.com/lead-
sponge.html/ (Accessed on 03.06.15).
[31] Pavlov D. A theory of the grid/positive active-mass (PAM) interface and
possible methods to improve PAM utilization and cycle life of lead/acid
batteries. J Power Sources 1995;53:9–21.
[32] Fan N, Sun C, Kong D, Qian Y. Chemical synthesis of PbO2 particles with
multiple morphologies and phases and their electrochemical performance as
the positive active material. J Power Sources 2014;254:323–8.
[33] Pavlov D. Chapter 12 – technology of formation. In: Pavlov D, editor. Lead-acid
batteries: science and technology. Amsterdam: Elsevier; 2011. p. 501–31.
[34] Dimitrov M, Pavlov D. Influence of grid alloy and fast charge on battery cycle
life and structure of the positive active mass of lead acid batteries. J Power
Sources 2001;93:234–57.
[35] Pavlov D. Formation of Positive Lead-Acid Battery Plates. Lead-Acid Batteries:
Sci Technol 2011:443–79.
[36] Shukla AK, Ganesh Kumar V, Munichandraiah N, Srinath TS. A method to
monitor valve-regulated lead acid cells. J Power Sources 1998;74:234–9.