IECON2015-Yokohama

November 9-12, 2015

SOC and SOH Characterisation of Lead Acid

Batteries
Jacques Marchildon, Mamadou Lamine Doumbia, Kodjo Agbossou
Université du Québec à Trois-Rivières
Institut de recherche sur l’hydrogène
Département de génie électrique et génie informatique
[email protected]; [email protected]; [email protected]

Abstract – Lead acid batteries are an important part of most I. INTRODUCTION

energy storage applications. Managing these applications is
complicated by the inherent difficulty of following the evolution of Lead acid type batteries are a very common energy storage
important battery parameters; more specifically, the State of device and are an important part of the self production
Charge (SOC), State of Health (SOH) and Ampere-hour capacity applications. Althought they have low energy densities, they
(AHC). This article shows that the Two-Pulse Load Test is a are low in cost, reliable and readily available. In addition, they
simple and reliable method to determine SOC and SOH of a Lead
acid storage battery. are easily recycled. According to an Environment Canada
The Two-Pulse Load Test is based upon the hypotheses that report dating 2009 [1], 99.2% of all Lead acid batteries are
certain voltage readings are linear in regard to the SOC and SOH. recycled each year.
The results presented in this article validate these hypotheses for a
180 ampere hour lead acid battery. Storage batteries are found in most alternative energy
Although the battery characterisation requires some time systems and on telecommunications sites. Alternative energy
consuming preliminary testing, SOC and SOH can thereafter be sources may consist of solar panels, wind turbines, and internal
determined in less than 5 minutes. combustion generators. The particularity of these alternate
Index energy systems is that the supplied energy, as well as the load,
varies constantly and unpredictally. In contrast,
AHCNom : Nominal ampere hour capacity telecommunications applications have very predictable load
CR : Discharge capacity rating
characteristics.
EMFMin : Electromotive force at full discharge
VEMF : Voltage of the Electromotive The central component of all of these systems is the storage
VMax : Open Circuit voltage prior to the first impulse battery. They store the instantaneous surplus energy, and
ΔV1 : Voltage difference following the first impulse render it available as an energy source when the main sources
ΔV2 : Voltage difference following the second impulse are unable to meet the demand.
RΩ: Ohmic resistance
Because of the importance of their role in the continuous
Rct: Load transfer resistance
operation of the systems they support, the storage battery
Cct: Load transfer capacitance
becomes an important part in the management of such systems.
α et β : Empirical parameters Characterising a battery is considered complete when we are
δ et γ : Empirical parameters able to define their three main parameters;
SOC : State of Charge • AHC- Amp Hour Capacity
SOH : State of Health
• SOC- State of Charge and
• SOH- State of Health
Definitions In this paper two-pulse load method is investigated for 180
amperes hour lead acid battery to determine its SOC and SOH.
AHC : Quantity of electricity that a battery can debit under
Once the battery is characterized, its SOC and SOH can be
certain conditions of temperature, discharge current
found in 5 minutes.
and final voltage.
CR: The current load required to totally discharge a
battery during R hours. II. STATE OF THE ART OF BATTERIES MODELS
SOC : Available battery capacity, usually expressed as a Different techniques have been proposed in order to
percentage of nominal capacity. characterize storage batteries. Each obtains their roots through
an appropriate modelling of the storage battery.
SOH : Ratio between the actual AHC and the AHCNom.
A. CHEMICAL MODEL

-1-
978-1-4799-1762-4/15/$31.00 ©2015 IEEE 001442
 The chemical model is based upon the chemical reaction manufacturer's specifications. Its popularity is that it is
equation of the Lead-acid battery. essentially the quality assurance test that is performed on every
⎯Déch
⎯⎯ arg e
⎯→ new battery in order to determine its inherent AHC. In a
PbO2 + Pb + 2 H 2 SO4 2 PbSO4 + 2 H 2O (1)
nutshell, the battery is discharged using its CR current rating
←⎯ ⎯⎯
ch arg e
and the battery voltage is measured during the n hours required
to perform the test. If the battery voltage, at the end of the test,
The chemical analysis of the storage battery is performed
is higher than the specified minimal battery voltage, the battery
using equation 1 as well as a detailed knowledge of the
is then recognised as having the CR rating.
phenomena is in place. Kuhn, Forgez and Friedrich [2]
Manufacturers include this graph in their specifications.
proposed a model for Ni-mH storage batteries in a hybrid
More elaborated specifications will include several graphs
electrical vehicle. They considered several phenomena such as:
according to different CR ratings. For example, C2, C4, C8, C10
- The chemical reaction,
etc.
- Chemical and electrochemicial potentials,
SOC estimations are based upon the measured battery
- Equivalent resistance of the electrolyte and the
voltage as compared to the supplied graph. Although this
connectivity,
appears simple and straightforward, there are three main
- Charge transfer, and
inconsistencies in using this method for SOC estimation. The
- Diffusion.
first is that the provided graphs show battery voltage under a
By measuring all of the above phenomena, it is able to
specific load. Extrapolations are necessary if the actual load is
determine the exact state of the reaction. The measurements
different from the provided data. The second is that the graphs
could then be translated into the instantaneous values of AHC,
do not give any insight on battery voltage in a no load
SOC and SOH. The chemical model ends up being rather
situation. For example, under a C20 load, the battery voltage
complex in its application to research since a very large series
will be much lower than the battery voltage under no load, and
of instantaneous values are required in order to establish the
for the same SOC.
model. In addition, specialised instruments are required to
Finally, this model does not consider the batteries' SOH. The
perform the required measurements.
AHC of a newly manufactured battery will increase over a
B. ELECTRICAL MODEL certain number of cycles to attain its' maximum level. After
which, the AHC will decrease. Two discharge graphs are
Each storage battery can be considered as a single electrical
therefore necessary to ensure proper analysis, one for the
circuit. There are several useful electrical models that are
empirical data at the time of manufacturing and one for the
adopted according to the required detail of analysis. For
maximum AHC expectation according to the manufacturers'
example, a simple Thevenin equivalent circuit may be useful in
specifications. Taking into account these three inconsistencies
certain studies whereas a more elaborated model may be
adds to the imprecision associated with model.
required in other studies.
Although the electrical parameters of the battery can be
modelled by different methods and according to the proposed
III. PROPOSED MODEL– TWO-PULSE LOAD TEST
theory of operation, they are all based upon equivalent
electrical circuits. However, most electrical models do not M. Coleman, W.G. Hurley and C.K. Lee [4] have opted for a
include parameters capable of modelling the SOH of the rather simple electrical model for the storage battery. This
battery. model is also used by K.S. Ng, C.S. Moo, Y.P. Chen et Y.C.
RC Kroeze and PT Krein [3] have proposed a rather Hsich [5] and is presented in Figure 1.
complex model for Dynamic Electric Vehicle Simulations.
This model has the advantage of integrating the SOC as a
parallel capacitor on the internal resistances of the battery’s
anode and cathode.
The various electrical models share an important practical
characteristic in that there are a multitude of simulators that
can perform detailed analysis on electrical circuits. This leads
to a large advantage over other models.
However, the variety of different electrical models gives
hold to the argument that each model is specific to the related Figure 1 Electrical model of Lead Acid battery
application and that a comprehensive or all including electrical
model may be elusive. In their article, K.S. Ng, C.S. Moo, Y.P. Chen et Y.C. Hsich
show that there is a linear relationship between the dynamic
C. DISCHARGE MODEL
open circuit voltage of a storage battery and its' SOC.
The discharge model is arguably the most common model In the case of the Two-Pulse Load Test the authors have
used for storage batteries as it is the basis for the based their model upon previous theory and experimentation

-2-
001443
 [5] and [6] and the hypothesis that the first pulse (see figure 2) 144AH and 119AH respectively.
serves as a sort of "historical eraser" where the batteries'
Each battery was then subjected to a series of three pulses,
history, preceding usage, charges and discharges, is eliminated
the sequence of which are presented in Figure 3.
as a factor in the SOC and SOH characterisation.
The second pulse, when applied to the storage battery,
renders the required data to plot the graphs of VMAX versus V1MAX V2MAX
...
SOC and CR vs ΔV2.

Current (A)
Voltage (V)
ٛ V2
VMin
...
V1MAX V2MAX
... ... I Discharge

Current (A)
Voltage (V)

Pulses #1 #2 #3
...
ٛ V2 t1 t2 t3 t4 t5 t6 Time (s)
VMin
Figure 3 Sequence of applied pulses
The first two pulses were those required for the Two-Pulse
I Discharge
Load Test. The third pulse was applied in order to discharge
Pulses #1 #2
the battery by a known percentage. The sequence was repeated
...
t1 t2 t3 t4 Time (s)
until the battery voltage was below the allowable level
t5
recommended by the manufacturer i.e. 6.9V.
It is important to note that the total discharge time, at 50A,
was an average of 85 minutes. This results in an AHC of only
Figure 2 Voltage and current values
70Ah. This is far below the rated capacity of 183AH.
The values of VMAX and ΔV2, are represented in Figure 2,
As discussed previously, the AHC depends upon the SOH of
Different values of VMAX and ΔV2 are obtained for each series
the battery. Being brand new, we could expect an AHC of
of two pulses. The battery is of course discharged by a known
close to 70% of AHCNOM. In our case, the AHCNOM should be
amount between each series of two pulses in order to determine
between 119Ah and 144Ah. At 70%, this should be within
the associated SOC.
Considering the electrical model of Figure 1, we may 83Ah and 100Ah.
represent the linear relations as: We must therefore consider that these batteries, although
with identical AHCNOM were not in optimal condition.
V MAX + β − EMFMIN (2)
SOC =
α
C R = δ ( ΔV2 ) + γ (3) IV. EXPERIMENTAL SETUP
In order to reduce, and hopefully eliminate, any
Subsequent analysis is performed with the following manipulations errors, an automatic test bed was setup to apply
definitions for AHC and SOH the pulse sequences and measure the battery voltage. As shown
1 (4) in Figure 4, the storage batteries were connected to a Dynaload
AHC =
CR programmable load.
AHC ACTUEL (5)
SOH =
AHC NOM

A. NOMINAL AHC
In order to determine the SOC of a battery, we must first
establish the nominal AHC of the battery. Since the AHC
varies with battery SOH, we have taken three identical and
brand new batteries and performed a discharge test using a 50A
load. We use US Batteries, model 8VGCHC XC. The
published C20 rating [7] for this battery is 183Ah. This battery
should therefore be capable of discharging 6.1A during 20 hrs.
The data sheet also gives us tow more capacity ratings, 5.76
hours at 25A and 1.59 hours at 75A for a total capacity of Figure 4 Experimental setup

-3-
001444
 Labview virtual instruments were then used to control the The values of α, β, et EMFMIN can then be determined for each
charging and discharging of the batteries. As shown in Figure battery using a linear regression algorithm. The average values
3, the pulse sequence applied two 50A pulses for 10s with a of the resulting linear regressions are:
10s delay. A 30s delay after the second pulse was introduced in
α = 0.01 V
order to enable the battery to stabilise. The third pulse, of 60s
duration enabled a faster discharge of the battery. The battery β = 0.25 V
was then allowed to stabilise for 60s prior to applying the next
EMFMIN = 11.18 V
cycle of impulses.
It is very important that the timing of the cycles be as And inserted into Equation 2 to render :
accurate as possible as the voltage values vary in time. Should
the time between pulses or the duration of the pulses not be the VMAX + 0,25 − 11,18 (6)
SOC =
same, the linearity of the resulting data would be 0,01
compromised.
Equation 3 is the relation of CR versus ΔV2. In order to
establish this relation we must chose a specific value of R. For
V. EXPERIMENTAL RESULTS example, we have chosen R to be 20 and CR represents C20. Its
Figures 5, 6 and 7 represent the graphs of V1MAX versus numerical value is therefore 183AH/20hr = 9.15A. The actual
SOC for each battery. values of CR needed to determine Equation 3 are the relative
values of the discharge current as compared to C20 or 9,15A.
Table 1 shows the relation between the discharge currents
and the CR value of the following graphs.
Table 1 Determination of CR values
Discharge Current CR value
10A 1.093
25A 2.732
35A 3.825
50A 5.464

Figure 5 V1MAX vs SOC Battery #1 65A 7.104

80A 8.743

Figures 8, 9 and 10 represent the graphs of CR versus ΔV2

for each battery.

Figure 6 V1MAX vs SOC Battery #2

Figure 8 CR vs ΔV2 Battery #1

Figure 9 CR vs ΔV2 Battery #2

Figure 7 V1MAX vs SOC Battery #3

-4-
001445
 C.K. Lee [4] by applying the Two-Pulse Load Test to a series
of identical batteries. This was done in order to determine its'
appropriateness in characterising a single battery type and
subsequent use as a quick and precise SOC and SOH
determination method for electrical vehicles.
Three identical and brand new US 8VGCHC XC lead acid
storage batteries were subjected to the same series of discharge
pulses. The results rendered quasi linear relations for VMAX vs
SOC and CR vs ΔV2. In addition, these linear relations are
quasi identical from one battery to another.
Figure 10 CR vs ΔV2 Battery #3 The experimentation on three identical batteries has also
provided insight into the possible real time applications of the
The values of δ et γ, can then be determined for each battery Two-Pulse Load Test in real life applications. Once the
using a linear regression algorithm. The average values of the batteries are characterised with their empirical values of α, β, δ
resulting linear regressions are: and γ, we are able to determine SOC and SOH in less than 5
minutes.
δ = 1.06Ah/V
Continuing experimentation will be required in order to
γ = - 18Ah further prove the "historical eraser" effect of the first pulse by
repeating the discharge cycles under various rates and
And inserted into Equation 3 to render :
sequences.
C R = 1,06ΔV2 − 0,18 (7) ACKNOWLEDGMENTS
At the present time, the validity of Equation 5 is not able be My acknowledgement to the Institut de Recherche sur
validated. However, we are able to confirm that the AHCNOM l’Hydrogène, UQTR for providing technical and experimental
of the tested batteries, in their present condition, is 70Ah at support. Thank you to the graduate student Nilson Henao for
50A continuous discharge rate. helping programming the Matlab virtual instrument that was
AHC ACTUEL essential for this study.
SOH =
70 (8)
An important aspect of the Two-Pulse Load Test is that it REFERENCES
can be performed on a battery in less than 5 minutes. We must [1] Le Recyclage des piles au Canada -mise à jour de 2009, Environnement
consider of course that the battery has been subjected to the Canada, préparé par Kelleher Environmental.
[2] E. Kuhn, C. Forgez, G. Friedrich, « Modèle pour accumulateurs Ni-mH
preliminary characterising testing and that we have reliable en vue d’une application pour véhicule hybride électrique », Génie
values to include into equations. Électrique-Vie et Qualité, Marseille, février 2002.
In order to obtain the preliminary characterising results, [3] R.C. Kroeze and P.T. Krein, « Electrical Battery Model for Use in
there are many hours of testing to be performed. First of all, we Dynamic Electric Vehicle Simulations » University of Illinois at
Urbana-Champaign Department of Electrical and Computer
need to perform a total discharge of the battery. According to Engineering, 2008.
the projected application, this could be between 5 and 20 [4] M. Coleman, W.G. Hurley and C.K.Lee, “An Improved Battery
hours. Thereafter, we must apply a sequence of load pulses that Characterization Method Using a Two-Pulse Load Test”, IEEE
will enable us to determine the empirical coefficients of α, β, δ Transactions on Energy Conversion, Vol. 23, No. 2, juin, 2008, pages
708-713.
and γ. [5] K.S. Ng, C.S. Moo, Y.P. Chen et Y.C. Hsich, “State-of-Charge
Once the empirical parameters have been determined, same Estimation for Lead-Acid Batteries Based in Dynamic Open-Circuit
type batteries can be tested for SOC and SOH within 5 Voltage s”, 2nd Annual IEEE Power and Energy Conference, PECon, 1-
3 décembre 2008, pages 972-976, Johor Baru, Malaisie.
minutes. The time required to apply the Two-Pulse Load Test [6] C.B. Zhu, M. Coleman et W.G. Hurley, “State of Charge Determination
is less than 3 minutes and calculating SOC and SOH from the in a Lean-Acid Battery : Combined EMF Estimation and Ah-balance
equations 6 and 8 should not be longer than 2 minutes. Approach”, 34th Annual IEEE Power Electronics Specialists
This leads us to believe that this method could be utilised in Conference, PESC'03, 20-25 juin 2004 Page(s): 1908 - 1914 Vol.3,
Acapulco, Mexico.
determining SOC and SOH in a real time application such as [5] K.S. Ng, C.S. Moo, Y.P. Chen et Y.C. Hsich, “State-of-Charge
electrical vehicles and alternate energy systems. Estimation for Lead-Acid Batteries Based in Dynamic Open-Circuit
Voltage s”, 2nd Annual IEEE Power and Energy Conference, PECon, 1-
3 décembre 2008, pages 972-976, Johor Baru, Malaisie.
VI. CONCLUSION [2] US Battery, 2012. "Battery specifications",
http://www.usbattery.com/usb_us8vgchcxc.html, Consulted, 4/3/2013.
In this paper we have continued the previous
experimentation undertaken by M. Coleman, W.G. Hurley and

-5-
001446