Available online at www.sciencedirect.

com

ScienceDirect

Modeling of rechargeable batteries

Resmi Suresh, Hemanth Kumar Tanneru and
Raghunathan Rengaswamy

Development of accurate, real time models for rechargeable improved efficiency. Figure 1 describes what an optimal
batteries is essential for efficient battery management and their battery operation implies, factors that need to be consid-
safe operation. This paper presents a short review on the ered and tools needed to achieve optimal battery opera-
various existing rechargeable battery models. We propose a tion. Optimal operation of battery results in prolonged
new classification of the models as being first principles based, battery life with maximum capacity by reducing capacity
surrogate first principles based, data-based and/or a hybrid of losses while ensuring safety. This optimal operation is
first principles and data based approaches. While we describe achieved by operating a battery with optimal charge
the state of the art in modeling, avenues for future research discharge cycles (provided by an optimizer), through a
work in this area are also identified. We also evaluate these controller. A model (representative of a battery) is in-
models from the perspective of their use in various operational volved as an integral component in the optimizer, con-
tasks and their application potential. troller, as well as state of charge (SOC) estimator.
Estimation of SOC is usually performed through an
Address
estimator (Kalman filter) for linear models and UKF or
Department of Chemical Engineering, IIT Madras, Chennai 600036, India
EKF for nonlinear models. Online data collection is
Corresponding author: Rengaswamy, Raghunathan ([email protected]) needed for the process engineering tasks that maintain
the battery operation to be optimal.
Current Opinion in Chemical Engineering 2016, 13:63–74
There are different states associated with batteries such
This review comes from a themed issue on Energy and as, state of charge (SOC), voltage, current, capacity etc.
environmental engineering
These states have to be monitored and maintained at
Edited by Vasilios I Manousiouthakis and Vilas Pol their desired levels for optimal battery operation, and real
For a complete overview see the Issue and the Editorial time battery models can help realize such desirable
Available online 20th August 2016 performance. For a model to be useful, it should include
http://dx.doi.org/10.1016/j.coche.2016.08.005
all the significant phenomena that occur inside a re-
chargeable battery. A reasonably comprehensive list of
2211-3398/# 2016 Published by Elsevier Ltd.
such phenomena would include: dynamic V–I character-
istics, nonlinear capacity, SOC, aging effect, recovery
effect (present in Lead acid batteries), hysteresis effect
(present in Ni–Cd batteries), memory effect (present in
Ni–Cd batteries, recently found to exist for Li ion bat-
teries as well [1]), thermal effects, cycling effects, and
Introduction
capacity fade mechanisms (passive film formation, active
Rechargeable batteries or secondary cells are those bat-
material dissolution and so on).
teries which can be charged and discharged multiple
times. They have gained importance because of their
Dynamic behavior in a battery can be instantaneous, fast
significant application potential, acceptable recharge cost,
or slow depending on the physical phenomena that is at
and reduced environmental impact when compared to
the root of the behavior [2]. Instantaneous effects due to
non-rechargeable batteries. Based on the chemistry and
ohmic resistance are likely to happen in a time-scale of
materials used, there are different kinds of batteries
microseconds. Charge transfer and double layer formation
available in the market. Lead acid, nickel cadmium
are examples of fast transient processes, whose time-scale
(NiCd), nickel metal hydride (NiMH) and lithium ion
can be in the order of few milliseconds up to several
are some of the well-known rechargeable batteries. The
minutes. Slow transient processes influence the battery
performance characteristics of each of these batteries are,
characteristics in several hours, days, or even years. For
naturally, different and depend on their respective chem-
example, mass transfer across the electrolyte and aging
istries and operational strategies. Li-ion batteries, which
are slow transient effects.
have high energy density and low self-discharge, are
currently quite popular and have been considered for a
Figures 1 and 2 explain the perspective of this review
wide range of applications.
article, which covers the following aspects
Models for rechargeable batteries can be leveraged in (a) Classify the various models that are available in the
developing strategies for optimal battery operation with literature (Model philosophy branch).

www.sciencedirect.com Current Opinion in Chemical Engineering 2016, 13:63–74

64 Energy and environmental engineering

Figure 1

Optimizer
Optimized
Optimum Battery charging and
Operation Set Point
discharging cycles
• Prolonged battery life
• Reduced charging time Controller
• Larger capacity Safety control:
• Reduced capacity losses Ensure operation SOC Model
• Reduced heating ... in safe zone

SOC estimator

Data

Diagnostic tool
Current Opinion in Chemical Engineering

Framework for optimal battery operations.

Figure 2

Battery
models

Model Utility Application

Philosophy View View

Battery
SOC Multi-cell
Diagnosis Control behavioral
estimation design
studies

First Surrogate
first Data based Hybrid
principles
principles models models
models models Automotive Space Portable
Storage
applications applications devices
Equivalent Lumped
circuit characteriz-
models ation
Empirical Stochastic
models models
Numerical Analytical

Current Opinion in Chemical Engineering

Overall perspective of the present review.

Current Opinion in Chemical Engineering 2016, 13:63–74 www.sciencedirect.com

 Modeling of rechargeable batteries Suresh, Kumar Tanneru and Rengaswamy 65

(b) Discuss pros and cons of these models when applied et al. [6]. This is a pseudo two-dimensional (P2D) model
for specific utilities such as control and diagnosis developed based on porous electrode and concentration
(utility view branch). electrode theories. The name pseudo 2D refers to the
(c) Present details on why and which type of models best pseudo dimension ‘r’ resulting from the assumption that
fits particular application such as portable devices and each electrode is composed of spherical particles. The
storage etc. (application view branch). improvements to the model presented in [6] such as
extending it to non-isothermal case by including energy
In this paper, we review the various modeling strategies balance, incorporation of mechanisms to describe relaxa-
and existing models for rechargeable batteries, focusing tion effects, and discussion of the limitations of the original
more on the recent developments. We categorize the model for higher discharge rates and possible mitigation
modeling strategy/models based on three broad perspec- strategies thereof are described in [7]. These early models
tives, viz., modeling philosophy, utility, and applications. did not consider many of the significant characteristics like
The details of this typology are described in Figure 2. cycling effects and capacity fade mechanisms. Another
disadvantage of the earlier models is the computational
Modeling philosophy based classification effort for solving the governing set of partial differential
Based on the modeling philosophy, several classification equations. Subsequently, the research focus has shifted to
schemes have been proposed in the past. A general address these issues. First principle models with hysteresis
approach is to classify models as being electrochemical, effects are not included in this review as the focus is mainly
equivalent circuit and hybrid [3]. There are other classi- on Li-ion batteries where hysteresis is negligible.
fications also. For example, Rao et al. [4] classified battery
models as physical, empirical, abstract, and mixed mod- Cycling performance
els, while Ramadesigan et al. [5] classified models into With cycling, the capacity of a battery degrades due to
four classes: empirical, electrochemical engineering, mul- various reasons. To accurately predict the lifetime of
tiphysics and molecular models. In this paper, we take a batteries, it is necessary to include cycling effects in
slightly different perspective and have classified various the model. Ning et al. [8] developed a generalized ap-
models based on modeling philosophy into four types as proach to include cycling effects into the P2D model by
follows: first principles models, surrogate first principles increasing the film resistance of the battery and by
models, data based models, and hybrid models. In first developing a model to account for the loss of active
principles modeling, mass and energy balance equations lithium ions with cycling. As this model is computation-
are used to derive a model for the battery type of interest. ally time consuming, a simplified P2D model with poly-
In a surrogate first principles modeling, a surrogate physi- nomial approximation (PP approach) for concentration
cal system is conceptualized for the battery and mass and inside each spherical particle is considered in [9]. This
energy balances are written for the surrogate system. model could simulate battery behaviors with cycling in
Though the parameters are estimated from data in these much shorter time compared to P2D model with no loss of
two model types, the modeling philosophy is first prin- accuracy. Though computational time is reduced consid-
ciples driven. In contrast, there are a different set of erably, it is still not possible to use this model for real time
battery models that are completely data-based. Hybrid simulations.
models, as the name suggests, are models derived from a
combination of these two (first principles and data based) Capacity fade mechanisms
modeling strategies. Capacity fade mechanisms are phenomena that occur
inside a battery under various operating conditions result-
First principles models ing in the degradation of the battery capacity. Arora
First principles models for batteries are derived using et al. [10] presented a detailed review of various capacity
material and energy balances taking into account the fade mechanisms, the corresponding side reactions, their
electrochemical reactions that are at the heart of the causes and effects along with an approach to incorporate
battery chemistry. In this section, we present a brief review these side reactions into the battery model. The kinetics
of the electrochemical modeling of rechargeable batteries, of side reactions are used to determine the current density
followed by recent advances in the development of first lost due to the capacity fade and the effective current
principle models for rechargeable batteries. The focus is density is calculated by using linear superposition princi-
on first principles models developed for Li-ion batteries ple. Solid electrolyte interface (SEI) layer formation is
due to the recent increased research interest in Li-ion one of the most significant side reactions to be modeled
battery systems. We conclude this section with the merits by the battery community. A model to incorporate the
and demerits of the first principle models and future combined effect of SEI formation and gas evolution is
directions. Early work on modeling lithium ion systems presented in [11]. The model developed can quantify
was the development of a mathematical model to study the the loss of electrolyte from SEI formation and gas evo-
galvanostatic charge discharge of Li | separator | Li-ion lution independently along with the cycling effects. Re-
insertion electrode system presented in Ramadesigan cently, Kang et al. [12] and Liu and Zhu [13] made

www.sciencedirect.com Current Opinion in Chemical Engineering 2016, 13:63–74

66 Energy and environmental engineering

attempts to include SEI layer formation in a first princi- developed a combined comprehensive model which con-
ples model. Kang et al. [12] developed a new model to siders electrical-thermal and mechanical interactions si-
capture electrolyte decomposition and SEI layer forma- multaneously. This model was able to distinguish the
tion, through the introduction of a new variable, volume advantages and disadvantages of tests used to detect short
averaged capacity loss (Qloss), and the governing equations circuits in batteries and the separator failure short circuits.
for the former in conjunction with the model is presented The model presented in [6] assumes that the solid phase
in [6]. The parameters whose values change during of the electrode composite is made up of spherical par-
cycling are handled directly by updating them as ticles. An extension of the model is provided by Lueth
a function of Qloss. The model developed in [13] is a et al. [20] considering the agglomerates of the active
non-isothermal electrochemical model which is extended material. In this approach, the electrolyte is assumed to be
with the governing equations for SEI layer. The model inside the agglomerate and the transport in the agglom-
equations were solved based on COMSOL Multiphysics erate is accounted for through additional governing equa-
battery module. tions. It is assumed that the kinetic reactions occur both
inside and at the outer surface of the agglomerate. Ag-
A model for the formation and dissolution (FD) of the SEI glomerate based approach predicts higher discharge ca-
layer can be found in [14]. This model is different from pacity in comparison to the spherical particle based model
the others in that it also considers dissolution of the SEI due to reduction in transport losses and increase in active
layer while most of the models focus on SEI layer forma- surface area. But this higher performance of battery as
tion alone. The capacity fade is directly correlated to the predicted by these types of models still needs to be
FD rate constants and the model predicts the experimen- validated using experiments. The inclusion of governing
tal data quite well. Ramadesigan et al. [15] presented a equations for modeling the cycling behavior and capacity
different approach to include various capacity fade mech- fade mechanisms for the agglomerate models is one
anisms to the P2D model. Instead of developing a sepa- possible avenue for future work. In large scale battery
rate mathematical model for each and every side reaction, stacks which are necessary for storage, electric vehicles
they propose that the capacity fade be captured through (EV) and space applications, the batteries interior to the
changes in the kinetic and transport parameters. stack will experience different operating conditions as
opposed to the ones in the 1D model (P2D). Such
Model reformulations situations change the battery working and will affect its
Model reformulations are mathematical approaches that performance. A non-isothermal, pseudo 3D model [21]
are used to solve the governing equations with ease and for lithium ion battery has been developed recently to
less computational intensity. Subramanian et al. [16] address issues of this nature.
attempted various model reformulation techniques for
the P2D model, like finite difference method, finite Surrogate first principles models
element method, collocation method, etc. and concluded Surrogate first principles models represent the battery
that finite difference and polynomial approximation tech- using a surrogate, like an equivalent electrical circuit, and
niques are the most effective techniques. A review of first principles model equations are derived for the surro-
various reformulation techniques can be found in [17]. gate. Surrogate models can again be classified based on
Though considerable reduction in computational time is the surrogate used: equivalent circuit and lumped
achieved using these reformulations, these require con- conceptualization.
siderable mathematical effort and thus will become non-
standard when additional effects and capacity fade mech- While we gain on computational time using surrogate
anisms have to be included to the P2D model. Also, these models compared to first principles model, there is defi-
reformulations cannot be applied for the whole domain of nitely a compromise on accuracy for both state estimation
operation. For example, polynomial approximation is and prediction. Surrogate models, like single particle
effective only for low discharge rate applications [17] model (SPM), are valid only for low discharge rates, while
and Galerkin collocation method works well only for the accuracy and applicability of the equivalent circuit
linear diffusivities [17]. There are other reformulations models depends on the training data set used while
like orthogonal collocation method, which can handle estimating the parameters of the model. If the training
nonlinear diffusivities and can be applied for higher data set covers a wide range of temperature and discharge
discharge rates, but are restricted by computational com- rates, then the accuracy of the model will be reasonable in
plexity [18]. these ranges. However, if the model is trained in a limited
range, accuracy while predicting outside these ranges will
Recent advances in the first principles modeling be lesser. Similarly, whether a surrogate model predicts
Effects due to mechanical stress and volume changes internal variables or not depends on the chosen model.
during intercalation and de-intercalation of Li ions Surrogate models like SPM, where the surrogate is a
are considered to be negligible in all models based on physical characterization, internal variables like pore wall
P2D model. Zhang et al. [19] in their recent work flux can be predicted. However, if the surrogate is an

Current Opinion in Chemical Engineering 2016, 13:63–74 www.sciencedirect.com

 Modeling of rechargeable batteries Suresh, Kumar Tanneru and Rengaswamy 67

electrical circuit, like equivalent circuit models, internal parameter estimation by comparing impedance data of
variables that can be predicted will be related to the the circuit with that of electrochemical impedance spec-
electrical circuit rather than the actual battery. Hence, a troscopy (EIS) of battery.
model has to be chosen based on the requirements and
applications. There is currently no model available which Recently, Jung and Kang [33] developed a multi-dimen-
can be applied for all the utilities with the best possible sional model to consider the effect of temperature, state
accuracy and with smallest computational time. of health (SOH), SOC, battery shape, etc. on voltage and
current characteristics of a battery. Equivalent circuit
Equivalent circuit models model parameters are expressed as functions of these
Equivalent circuit models assume that battery character- states using bilinear interpolation. Multi-dimensional
istics can be described using an electrical circuit with effects and local degradations are captured in this model
components ranging from simple resistors and capacitors by using localized equivalent circuit models. Though this
to more complex Warburg [22] and Zarc elements [23]. An model is able to capture the effects of various parameters
ideal voltage source is used to represent the open circuit on battery performance, the model is still complex, thus
voltage (OCV) of the battery. This OCV is not a constant, imposing barriers for online simulation. Wang et al. [34]
but varies depending on SOC. A resistor is used to proposed a discharge capacity estimation technique that
represent internal resistance of the system and one or can estimate discharge capacity under various operating
more parallel resistor–capacitor (RC) subcircuits are conditions (temperature, cycles, C-rate) using particle
employed to model slow diffusion processes occurring filters. However, this technique is found to have around
inside a battery [24]. Many equivalent circuit models, like 10–30% error in estimation, which is not desirable. A few
Randles circuit [22], use a nonlinear Warburg element to non-integer order equivalent circuit models are available
model diffusion. But this increases the computational in the literature [23,35,36] and are found to have good
complexity of the equivalent circuit model as it cannot accuracy with lesser number of parameters. Fractional
be represented by a simple differential equation and is order circuits have higher potential for capturing memory
difficult to simulate. Thus multiple parallel RC networks effect of the battery [37].
are used to replace this Warburg element. However, to
exactly capture the effect of Warburg element, infinite A slightly different modeling approach was proposed
RC subcircuits are necessary. However, a small number of recently by Balasingam et al. [25] in which four differ-
RC pairs can often characterize the diffusion effects for a ent modes of battery operation are identified and
small frequency range [22,24]. slightly different dynamic ECMs are identified for each
operating mode. This model does not require any
Simplified equivalent circuits are computationally fast initialization and also includes hysteresis effect (mod-
and are easy to embed with other electrical circuits and eled as error). This approach can be used for seamless
simulation tools. Hence, they are a viable alternative for SOC tracking. However, several modifications are still
battery design and for simulation along with other elec- necessary for this model to be able to include other
trical systems. However, these equivalent circuit models battery characteristics. In summary, while there has
consider only linear capacity effects and thus are limited been significant effort expended in equivalent circuit
in characterizing nonlinear capacity effects and operation- modeling, further work is needed to develop accurate
al aspects. The equivalent circuit model structure is model concepts, which include all the important char-
decided based on the dynamic effects to be studied. acteristics without significant compromise on computa-
Computational time will be more for a more complex tional speed.
model with a large number of elements. Hysteresis effect
is modeled only in a few existing equivalent circuit Lumped characterization — numerical
models. All other models either ignore the hysteresis Surrogate lumped characterization refers to those battery
effects or model it as an error from actual prediction models which are conceptualized based on a simplified
[25]. Hu et al. [26] provide a detailed comparison of surrogate to the actual battery. A typical example is the
twelve different equivalent circuit models based on ac- SPM [9] that considers the electrode as a single spherical
curacy, robustness and complexity. They show that a first particle, having a surface area equal to that of actual
order RC model is best for LiNMC and first order RC electrode surface area. The material and energy balances
model with hysteresis is best for LiFePO4. Equivalent are then written for two spherical particles (surrogates of
circuit models are not only characterized by the circuit anode and cathode). An extension of the SPM for study-
structure, but also by the parameter values and the ing the thermal behavior of batteries [38] has been
relationship between these parameters, voltage and developed by incorporating the energy balance to the
SOC. Various techniques like forgetting factor least model proposed in [9].
squares minimization [27], Kalman filter [28–30], Particle
Swarm–Nelder–Mead optimization [31] and experimen- Recently, Coman et al. [39] studied thermal runaway of
tally determined look up tables [32] can be used for lithium ion battery using a lumped parameter model.

www.sciencedirect.com Current Opinion in Chemical Engineering 2016, 13:63–74

68 Energy and environmental engineering

The model developed calculates the heat generated Data based models
inside the battery from the heat generated from the Data based models available for rechargeable batteries
dissolution of the SEI layer, the active materials of can be classified as stochastic and empirical models.
anode, cathode and electrolyte by ordinary differential Stochastic models use various stochastic approaches, like
equations. In this work, a review of earlier efforts on discrete time transient Markov process or Monte Carlo
modeling thermal runaway and their inability in ade- technique, to describe battery behaviors. Probabilities are
quately predicting the exact thermal runaway behavior expressed in terms of parameters or states related to
are presented. It is argued that the prediction of the physical characteristics of a battery. Stochastic models
onset of thermal runaway is much more accurate are used to model specific effects like discharging effects
through models that consider venting compared to or recovery effects. For example, the stochastic model by
those that do not. The reason for this is attributed to Chiasserini and Rao [43] uses discrete time transient
the inability to include endothermic effects in models Markov process for modeling, focusing only on the recov-
that do not consider venting. Capacity fade mechanisms ery effect. Further, the model can handle only pulsed
have also been incorporated in lumped models with the discharge of battery. Kinetic Monte Carlo method [5] is
assumption that there are no spatial concentration var- used to study the effect of electrode structure on lithium
iations in the electrodes. A new approach to incorporate mobility in Li-ion batteries by modeling discharge be-
capacity fade mechanisms from a formal reaction engi- havior. As stochastic models focus on a very specific
neering framework is presented in [40]. battery characteristic and do not provide overall charac-
teristics of battery, their applications are limited.

Lumped characterization — analytical Empirical models characterize a battery using equations

There are some analytical models in the literature, where- developed intuitively or based on data, whose parameters
in the surrogate conceptualization is such that the resul- are fitted to match experimental data. For example,
tant first principles model equations for the surrogate are Peukert’s law [4,44] represents an empirical relation
analytically solvable. A brief review of the analytical between capacity, current and lifetime of a battery.
models that were developed, and their inability to predict Weibull fit model [4] describes voltage as a function of
the battery performance in case of variable load profiles is capacity using Weibull distribution. Although all these
provided in [41]. Rakhmatov and Vrudhula [41] devel- empirical models are much easier to configure and are
oped a model which represents real battery as a simple computationally faster compared to any other model,
equivalent battery with diffusion of ions being considered their applications are very limited as their parameters
as the only prominent phenomena that describes battery do not have any physical significance and thus they do not
performance. A diffusion model which uses one-dimen- provide significant insight into the battery behavior.
sional diffusion of species to predict battery behavior is These parameters will have to be estimated repeatedly
also developed in [38]. The proposed model has the for different conditions and configurations of battery,
ability to characterize the recovery effect and the battery which restrict their application.
life-time. The governing equations can be solved analyt-
ically and the expressions which relate battery capacity, Hybrid models
parameters of the model and time to failure can be solved Hybrid models are used to integrate the features of any of
at a very low computational cost as opposed to the first the above models and to derive a model of higher accu-
principles models. racy and lower complexity. Most common hybrid models
[3,45–47] are combinations of equivalent circuit models
Another analytical model based on the kinetic processes, and analytical models. This hybridization helps in com-
kinetic battery model (KiBaM) is presented in [42]. The bining the dynamic V–I characteristics derived using
basic idea used in developing this model is the splitting of equivalent circuit model with the nonlinear effects cap-
available battery capacity across two wells; first, available tured using analytical models. But these models do not
charge well which can supply charge to load applied and include other features like aging, memory effect or re-
second, bound charge well which can only transfer the covery effect. Zhang et al. [45] made an attempt to include
charge to the available charge well. The dynamics of recovery effect using a hybrid model which is a combina-
charge transfer between these wells are formulated in tion of equivalent circuit model and Rakhmatov’s diffu-
such a way that rate capacity and recovery effects are sion model [41]. But this model is quite complex and thus
modeled. The death of battery occurs when the charge in is not suitable for real time applications. However, this
the available charge well is exhausted. The Ki-BaM model can be used for large scale multi-cell battery
model was developed for lead acid batteries which have modeling as shown in [48].
flat discharge profiles and are not applicable as such to Li-
ion batteries, which have steep discharge profiles. Devel- Wijewardana et al. [46] proposed a hybrid model combining
oping the Ki-BaM type models for Li-ion batteries will be second order equivalent circuit model and thermal model
an interesting work to pursue in future. (energy balance) for understanding battery dynamics.

Current Opinion in Chemical Engineering 2016, 13:63–74 www.sciencedirect.com

 Modeling of rechargeable batteries Suresh, Kumar Tanneru and Rengaswamy 69

Table 1

An overview of different phenomena that occur in rechargeable batteries along with model types that attempt to model these phenomena.
A ‘tick’ represents that considerable work has been done in literature for that specific phenomena using that model type. A ‘cross’
represents that very few models are available for the same and thus there are possibilities for further research

model type Phenomena

Dynamic Non-linear SOC Aging Recovery Hysteresis Memory Thermal Capacity Cycling
V–I capacity effect effect effect effect fade effects
characteristics mechanisms
First principles model U U U U U a b U U U
Surrogate first U U U U U U U U U U
principles model
Data based model U U U U c   U  U
Hybrid model U U U  d   U  
a
Hysteresis effect is not considered as the focus is on Li-ion batteries.
b
Memory effect is recently found to exist in Li-ion batteries: but there are no first principles models focusing on this phenomenon [1].
c
Extended Peukart’s law captures the recovery effect.
d
[47] captures recovery effect.

Table 2

Comparison of computational times for various first principles models and surrogate first principles models

Model Computational time (s) Comments

First principle P2D + SEI layer formation [9] 1.12e5 for 800 cycles Information on platform used, CPU and
RAM is not provided.
First principles P2D with reformulation [9] 9600 for 800 cycles –
Surrogate first principles (SPM) [9] 27 for 800 cycles –
First principle P2D + Solvent reduction [8] 5.4e4 for 2000 cycles Compaq Fortran environment. PC with
Pentium1 2.0 GHz CPU and 512MB RAM
First principles P2D with reformulations [17] 4.5 s for discharge simulation for Information on platform used, CPU and
galerkin based reformulation RAM is not provided.
at 5C discharge rate. Not modeled capacity fade and cycling
1.38 s for mixed FD based
reformulation at 5C discharge rate
First principle P2D 3.6e5 for 1000 cycles COMSOL Multiphysics
Combined effect of SEI layer Single core i-7 processer, 8 GB RAM
formation and gas evolution. [11]
First principles 20.38 for a discharge curve at 1C MAPLE using IDA solver
2D model
(P3D) [21]

Though this model could predict the battery V–I profile improvements to the current state of battery modeling,
and its temperature dependencies, all other effects are parameterization of these models are given in supple-
ignored in this model. Though hybrid models are a com- mentary material. Additional information on various criti-
promise between model accuracy and computational com- cal issues present while developing an accurate and
plexity, there are very few models that exist in the computationally efficient model along with strategies
literature which can capture most of the important features for addressing these can also be found in [5].
of battery in real time; this is an interesting area for future
research.
Utility view
A summary of the models available in the literature can Rechargeable batteries have a wide variety of applications
be found in Table 1. A comparison of computational and based on the application, the design specifications
times for representative models is provided in and operational constraints will vary. To achieve opti-
Table 2. Surrogate models are much simpler than the mum battery operation and optimum battery designs,
first principles models and thus can be solved much faster models should be useful in the following tasks: SOC
than the first principles models. It can be seen that SPM is estimation, control, diagnosis, battery behavior analysis,
around 4000 times faster than P2D model for simulating and multi-cell design. Each of the above tasks requires a
800 cycles in FORTRAN [9]. An evolution of the models battery model which meets certain requirements. For
and our view of future developments can be found in example, diagnosis of a battery requires a model which
Figure 3. Information on merits, demerits, possible captures most of the prominent side reactions inside a

www.sciencedirect.com Current Opinion in Chemical Engineering 2016, 13:63–74

70 Energy and environmental engineering

Figure 3

Thermal runaway in lumped models

Equivalent circuits based
on mode of operation
Model reformulations
P3D model
Stochastic models
Multidimensional equivalent circuits
Energy balance (1st principles)
Cycling effects in 1st principles
Diffusion model
Capacity fade in 1st principles
Single particle model
Capacity fade and hysterisis in
KiBaM equivalent circuit models
st
First model for Li-ion using 1 principles
Initial equivalent circuit models Non integer order
equivalent circuits
Energy balance in 1st principles Mechanical stress modeling

Agglomerate model

1990 2000 2010 2020

Current Opinion in Chemical Engineering

Timeline of various modeling strategies for lithium ion batteries. Solid line represents research during the time period shown while dashed line
represents possible future research directions.

battery. If the interest is in online diagnosis, then the computationally fast. Equivalent circuit models provide
model should be simple and computationally fast as well. the dynamic V–I characteristics, but there is no direct
There are various first principles models [12,13,14] information about SOC. A polynomial relation is usually
which include different capacity fade mechanisms. How- assumed between SOC and OCV [27,52,53], and the
ever, diagnosis using these models will be time consum- parameters are estimated. Various SOC estimation algo-
ing and are thus not suitable for online diagnosis. Though rithms have been used in the literature such as extended
several attempts have been made to improve diagnosis Kalman filter [46], adaptive sliding mode observer [36]
techniques and capture various side reactions using and adaptive Cubature Kalman filter [27]. A detailed
equivalent circuit models [49,50], all these techniques comparison of these different types of SOC estimation
are still far from being used as online diagnosis tool. techniques can be found in [51]. Each method has its own
advantages and disadvantages. There is no single method
A battery has to be maintained within a safe zone of which meets all the requirements. Few attempts have
voltage, current and temperature to ensure safe opera- been made in literature to develop optimal control strat-
tions. Moreover, to improve durability of battery and egies like charging time minimization [54] and maximiz-
reduce the capacity fade mechanisms in the battery, ing stored charge with reduced capacity fade mechanisms
charging and discharging cycles of battery should be [55]. However, considerable work remains to be done in
controlled. For achieving this, an efficient and fast con- the development of control schemes that can optimally
troller and a SOC estimator are necessary. SOC can be control charging and discharging cycles and at the same
estimated by many techniques [51] that do not use a time ensure that the battery operates in a safe zone. An
model, like the discharge test method, ampere-hour important thing to note here is whenever we discuss
integral method and OCV method. But none of these optimization, there are two approaches: sequential and
techniques are precise and/or fast enough to be used for simultaneous. When modeling is viewed separately from
online SOC estimation [51]. Thus SOC estimation tech- optimization, an easy framework to understand optimi-
niques based on battery models that can simulate their zation is the so called sequential framework (model is
characteristics in real time are commonly used. Either repeatedly solved within an optimizer). Of course, simul-
equivalent circuit models or analytical models are pre- taneous optimization is also possible with benefits with
ferred for online control and state estimation as they are respect to computational performance. However, setting

Current Opinion in Chemical Engineering 2016, 13:63–74 www.sciencedirect.com

 Modeling of rechargeable batteries Suresh, Kumar Tanneru and Rengaswamy 71

up of a sequential approach is much easier when detailed is presented in [59]. Markov chain based Monte Carlo
models are involved. More details can be found in [56]. simulation approaches are used for parameter estimation.
A study to verify robustness of different SOC estimation
First principles models are the best suited for battery algorithms for the modeling errors and uncertainties is
behavioral studies as they are the most accurate models. presented in [60] using an equivalent circuit model for
A well-trained electrochemical model can be used to lithium ion battery for electric vehicle application. The
understand how a battery might respond to various con- ability of a reformulated physics based electrochemical
ditions. This study is required to gain knowledge about its model to predict the battery behavior across varied dis-
safe zone of operation, dependency on various parameters, charge profiles of EVs is discussed in [61]. A review on the
etc. For most of the battery applications, power from a use of lithium ion batteries for EVs in cold temperature
single cell is not enough; thus, multi-cell batteries are countries and the need for development of efficient
generally used. For designing these, an accurate model is thermal management strategies by incorporating the de-
necessary. However, single cell models cannot be directly tailed mechanisms of capacity fade that are predominant
used for multi-cell batteries as they do not consider cell in cold temperatures are discussed in [62].
connections and cell status variations. Parallel cell connec-
tion is more challenging as current should be unevenly Space applications
distributed among cells so as to maintain equal voltage in Specific requirements of batteries that can be used in
each branch. A general iterative procedure for parallel space applications are presented in [63]. An analysis on
connection as explained in [57] is computationally com- predicting the Li-ion battery performance in the orbital
plex and time consuming, thus restricting its application to by coupling existing thermo-electrochemical models
small scale multi-cell batteries. Zhang et al. [48] proposed a along with the thermal desktop (TD), a software to
new iterative procedure for which computational com- calculate the environment conditions based on vehicles
plexity increases linearly with number of cells and hence position, is presented in [64]. As an extension to this
might be applicable for large scale multi-cell batteries. analysis, TD based thermo-electrochemical model of Li-
ion batteries in conditions present in orbital environments
Application view for space applications can be found in [65].
The application of rechargeable batteries can be classified
into four major areas: First, automotive; second, space; Portable applications
third, portable devices and fourth, storage. Details of each For portable applications, as the major interest is in
application and the type of model used in their battery accurately predicting the available runtime, online
management systems (BMS) are presented next. SOC estimation is a major concern. Equivalent circuit
models will suffice for this purpose according to [66]. Ki-
Automotive applications BaM and diffusion models were tested to predict battery
Recent mathematical studies in analyzing the use of Li- life-time for portable wireless communication devices
ion batteries for EV and plug-in hybrid electric vehicles [67] and it was concluded that any of these two models
(PHEV) are presented in [58]. Usage patterns of EVs and can be used for life-time predictions. The inclusion of
PHEVs were modeled based on reformulated P2D model stochastic nature for the workload models along with Ki-
which takes SEI layer formation into account. The effects BaM is shown to work well.
of different charging rates and driving profiles (discharge
profiles) on the performance and guidelines to minimize Storage
capacity fade based on charge discharge cycles are dis- Use of equivalent circuit models for modeling the storage
cussed. A new metadata model, which takes the uncer- system for stationary application is presented in [68]. The
tainties in the measurements to obtain battery parameters challenges involved in modeling a micro-grid storage

Table 3

Overview of various models used for different applications and utilities

Utility Application
Automotive Space Portable devices Storage

SOC estimation and life time predictions [23,27,36 ,46,52,58,60] – [41,42] [72,73]
Diagnosis and capacity fade modeling [30,58,59] – –a [72]
Control [28,29,32,61] – –b [69,70,72]
Battery behaviors [58,61] [64,65] [68]
Multi-cell design [28,29] – – [72]
a
Though specific models for diagnosis and capacity fade modeling are not available in literature, any equivalent circuit model considering capacity
fade mechanisms could be used for diagnosis.
b
Specific applications for portable devices are not available in literature. But, data based models can be generally used.

www.sciencedirect.com Current Opinion in Chemical Engineering 2016, 13:63–74

72 Energy and environmental engineering

system is presented in [69]. A comprehensive review on 8. Ning G, White RE, Popov BN: A generalized cycle life model of
rechargeable Li-ion batteries. Electrochim Acta 2006, 51:2012-
battery energy storage systems and the BMS for storage 2022 http://dx.doi.org/10.1016/j.electacta.2005.06.033.
applications is provided in [70]. The difficulties involved 9. Santhanagopalan S, Guo Q, Ramadass P, White RE: Review of
in development of battery models to use in BMS of hybrid models for predicting the cycling performance of lithium ion
renewable energy systems is discussed in [71]. An online batteries. J Power Sources 2006, 156:620-628 http://dx.doi.org/
10.1016/j.jpowsour.2005.05.070.
management for a pack of lithium ion batteries, based on
10. Arora P, White RE, Doyle M: Capacity fade mechanisms and
equivalent circuit models incorporating capacity fade side reactions in lithium-ion batteries. J Electrochem Soc 1998,
mechanisms is developed in [72]. A summary of the 145:3647-3667 http://dx.doi.org/10.1149/1.1838857.
usefulness of models for various tasks in the four identi- 11. Rashid M, Gupta A: Mathematical model for combined effect of
fied application areas can be found in Table 3.  SEI formation and gas evolution in Li-ion batteries. ECS
Electrochem Lett 2014, 3:A95-A98 http://dx.doi.org/10.1149/
2.0041410eel.
The authors developed a mathematical model for Li-ion cell including
Conclusion effects of SEI layer formation and gas evolution. Authors concluded that
Modeling concepts for rechargeable batteries are due to accumulation of lithium on negative electrode, there will be more
capacity reduction at higher cycling rates.
reviewed in this paper. Different models available in
the literature are analyzed based on the modeling philos- 12. Kang J, Conlisk AT, Rizzoni G: Integration of capacity fading in
 an electrochemical model of Li-ion batteries. J Solid State
ophy, utility, and application. This paper focuses more on Electrochem 2014, 18:2425-2434 http://dx.doi.org/10.1007/
the recent developments in this field. Although consid- s10008-014-2479-y.
Authors presented a new method of including side reactions in the battery
erable modeling attempts exist in the literature, signifi- model. Equilibrium potentials of parasitic reactions are identified as the
cant work is still required in the development of most influencing parameters in modeling. Reducing overcharging is
identified as a protective way to prevent battery aging.
comprehensive, yet computationally tractable models
that can be used in a wide variety of tasks and applica- 13. Liu L, Zhu M: Modeling of SEI layer growth and electrochemical
impedance spectroscopy response using a thermal-
tions. electrochemical model of Li-ion batteries. ECS Trans 2014,
61:43-61 http://dx.doi.org/10.1149/06127.0043ecst.
14. Ramesh S, Krishnamurthy B: A mathematical model to study
Appendix A. Supplementary data  capacity fading in lithium ion batteries: formation and
Supplementary data associated with this article can be dissolution reactions. J Electrochem Soc 2015, 162:A545-A552
http://dx.doi.org/10.1149/2.0221504jes.
found, in the online version, at http://dx.doi.org/10.1016/j. Authors have considered both FD reactions of SEI layer at the two
coche.2016.08.005. electrodes to model capacity fading, which is an interesting viewpoint.
Authors proved with the simulations that SEI layer dissolution is a critical
phenomenon that contributes more to the capacity fading at higher
temperatures.
References and recommended reading
Papers of particular interest, published within the period of review, 15. Ramadesigan V, Chen K, Burns Na, Boovaragavan V, Braatz RD,
have been highlighted as: Subramanian VR: Parameter estimation and capacity fade
analysis of lithium-ion batteries using reformulated models.
 of special interest J Electrochem Soc 2011, 158:A1048 http://dx.doi.org/10.1149/
 of outstanding interest 1.3609926.
16. Subramanian VR, Boovaragavan V, Ramadesigan V, Arabandi M:
1. Sasaki T, Ukyo Y, Novák P: Memory effect in a lithium-ion Mathematical model reformulation for lithium-ion battery
battery. Nat Mater 2013, 12:569-575 http://dx.doi.org/10.1038/ simulations: galvanostatic boundary conditions. J Electrochem
nmat3623. Soc 2009, 156:A260 http://dx.doi.org/10.1149/1.3065083.
2. Garcı́a-Plaza M, Serrano-Jiménez D, Eloy-Garcı́a Carrasco J, 17. Ramadesigan V, Boovaragavan V, Pirkle JC, Subramanian VR:
Alonso-Martı́nez J: A Ni–Cd battery model considering state of Efficient reformulation of solid-phase diffusion in physics-
charge and hysteresis effects. J Power Sources 2015, 275:595- based lithium-ion battery models. J Electrochem Soc 2010,
604 http://dx.doi.org/10.1016/j.jpowsour.2014.11.031. 157:A854 http://dx.doi.org/10.1149/1.3425622.
3. Dung L-R, Yuan H-F, Yen J-H, She C-H, Lee M-H: A lithium-ion 18. Northrop PWC, Ramadesigan V, De S, Subramanian VR:
battery simulator based on a diffusion and switching Coordinate transformation, orthogonal collocation, model
overpotential hybrid model for dynamic discharging behavior reformulation and simulation of electrochemical-thermal
and runtime predictions. Energies 2016, 9:51 http://dx.doi.org/ behavior of lithium-ion battery stacks. J Electrochem Soc
10.3390/en9010051. 2011:158 http://dx.doi.org/10.1149/2.058112jes.
4. Rao R, Vrudhula S, Rakhmatov DN: Battery modeling for energy 19. Zhang C, Santhanagopalan S, Sprague MA, Pesaran AA: A
aware system design. IEEE Comput Soc 2003, 36:77-87.  representative-sandwich model for simultaneously coupled
5. Ramadesigan V, Northrop PWC, De S, Santhanagopalan S, mechanical-electrical-thermal simulation of a lithium-ion cell
Braatz RD, Subramanian VR: Modeling and simulation of under quasi-static indentation tests. J Power Sources 2015,
lithium-ion batteries from a systems engineering perspective. 298:309-321 http://dx.doi.org/10.1016/j.jpowsour.2015.08.049.
J Electrochem Soc 2012:159 http://dx.doi.org/10.1149/ This article presents a coupled electrical–mechanical–thermal model
2.018203jes. capable of predicting the effects of mechanical stress on batteries like
fracture of cell structure. Safety of lithium ion batteries during various
6. Doyle M, Fuller TF, Newman J: Modeling of galvanostatic charge mechanical abuse conditions were studied using the model.
and discharge of the lithium/polymer/insertion cell.
J Electrochem Soc 1993, 140:1526 http://dx.doi.org/10.1149/ 20. Lueth S, Sauter US, Bessler WG: An agglomerate model of
1.2221597.  lithium-ion battery cathodes. J Electrochem Soc 2015,
163:A210-A222 http://dx.doi.org/10.1149/2.0291602jes.
7. Gomadam PM, Weidner JW, Dougal R, White aRE: Mathematical Authors introduced a more realistic approach to model electrodes of Li-
modeling of lithium-ion and nickel battery systems. J Power ion batteries, by considering agglomerate structure inside active material
Sources 2002, 110:267-284 http://dx.doi.org/10.1016/S0378- of electrodes, which is generally assumed as a solid spherical particle in
7753(02)00190-8. most of the models reported earlier.

Current Opinion in Chemical Engineering 2016, 13:63–74 www.sciencedirect.com

 Modeling of rechargeable batteries Suresh, Kumar Tanneru and Rengaswamy 73

21. Northrop PWC, Pathak M, Rife D, De S, Santhanagopalan S, batteries. Commun Nonlinear Sci Numer Simul 2015, 24:127-144
 Subramanian VR: Efficient simulation and model reformulation http://dx.doi.org/10.1016/j.cnsns.2014.12.015.
of two-dimensional electrochemical thermal behavior of A state estimation algorithm for a fractional order equivalent circuit model
lithium-ion batteries. J Electrochem Soc 2015, 162:A940-A951 is developed using adaptive sliding mode observer. The proposed method
http://dx.doi.org/10.1149/2.0341506jes. was shown to be accurate and effective using various experimental
Psuedo 3D model for a lithium ion battery is developed. POD is developed studies.
as a viable model reformulation technique to minimize computational
efforts. 37. Igor P: Fractional Differential Equations. Academic Press; 1999.

22. Kularatna N: Dynamics and modeling of rechargeable 38. Guo M, Sikha G, White RE: Single-particle model for a lithium-
batteries: what electrochemists? Work tells the electronic ion cell: thermal behavior. J Electrochem Soc 2011, 158:A122
engineers. IEEE Power Electron Mag 2014:23-33. http://dx.doi.org/10.1149/1.3521314.

23. Xu J, Mi CC, Cao B, Cao J: A new method to estimate the state 39. Coman PT, Rayman S, White RE: A lumped model of venting
of charge of lithium-ion batteries based on the battery during thermal runaway in a cylindrical lithium cobalt oxide
impedance model. J Power Sources 2013, 233:277-284 http:// lithium-ion cell. J Power Sources 2016, 307:56-62 http://
dx.doi.org/10.1016/j.jpowsour.2013.01.094. dx.doi.org/10.1016/j.jpowsour.2015.12.088.

24. Plett GL: Equivalent-Circuit Cell Models. 2014. 40. Suresh RMP, Rengaswamy R: Modeling failure modes in Li-ion
battery. 2015 AIChE Annu. Meet.. 2015.
25. Balasingam B, Avvari GV, Pattipati B, Pattipati KR, Bar-Shalom Y:
 A robust approach to battery fuel gauging, part I: Real time 41. Rakhmatov DN, Vrudhula SBK: An analytical high-level battery
model identification. J Power Sources 2014, 272:1142-1153 model for use in energy management of portable electronic
http://dx.doi.org/10.1016/j.jpowsour.2014.07.034. systems. 2001 IEEE/ACM Int Conf Comput Des. 2001:488-493
A real time approach for parameter estimation is developed which can be http://dx.doi.org/10.1109/ICCAD.2001.968687.
used for obtaining better SOC estimate. Four different battery models are
used to represent four different operational models of battery unlike the 42. Manwell JF, McGowan JG: Lead acid battery storage model for
conventional method of using a single model for all battery operations. hybrid energy systems. Sol Energy 1993, 50:399-405 http://
dx.doi.org/10.1016/0038-092X(93)90060-2.
26. Hu X, Li S, Peng H: A comparative study of equivalent circuit
models for Li-ion batteries. J Power Sources 2012, 198:359-367 43. Chiasserini C-F, Rao RR: Energy efficient battery management.
http://dx.doi.org/10.1016/j.jpowsour.2011.10.013. IEEE J Sel Areas Commun 2001, 19:1235-1245 http://dx.doi.org/
10.1109/49.932692.
27. Xia B, Wang H, Tian Y, Wang M, Sun W, Xu Z: State of charge
estimation of lithium-ion batteries using an adaptive cubature 44. Doerffel D, Sharkh SA: A critical review of using the Peukert
Kalman filter. Energies 2015:5916-5936 http://dx.doi.org/ equation for determining the remaining capacity of lead-acid
10.3390/en8065916. and lithium-ion batteries. J Power Sources 2006, 155:395-400
http://dx.doi.org/10.1016/j.jpowsour.2005.04.030.
28. Plett GL: Extended Kalman filtering for battery management
systems of LiPB-based HEV battery packs. J Power Sources 45. Zhang J, Ci S, Sharif H, Alahmad M: An enhanced circuit-based
2004, 134:252-261 http://dx.doi.org/10.1016/ model for single-cell battery. Conf Proc – IEEE Appl Power
j.jpowsour.2004.02.031. Electron Conf Expo – APEC. 2010:672-675 http://dx.doi.org/
10.1109/APEC.2010.5433597.
29. Plett GL: Sigma-point Kalman filtering for battery management
systems of LiPB-based HEV battery packs. J Power Sources 46. Wijewardana S, Vepa R, Shaheed MH: Dynamic battery cell
2006, 161:1369-1384 http://dx.doi.org/10.1016/ model and state of charge estimation. J Power Sources 2016,
j.jpowsour.2006.06.004. 308:109-120 http://dx.doi.org/10.1016/j.jpowsour.2016.01.072.

30. Andre D, Nuhic A, Soczka-Guth T, Sauer DU: Comparative study 47. Kim T, Member S, Qiao W: A hybrid battery model capable of
of a structured neural network and an extended Kalman filter capturing dynamic circuit characteristics and nonlinear
for state of health determination of lithium-ion batteries in capacity effects. IEEE Trans Energy Conver 2011, 26:1172-1180
hybrid electric vehicles. Eng Appl Artif Intell 2013, 26:951-961 http://dx.doi.org/10.1109/TEC.2011.2167014.
http://dx.doi.org/10.1016/j.engappai.2012.09.013.
48. Zhang J, Ci S, Sharif H, Alahmad M: Modeling discharge
31. Mesbahi T, Khenfri F, Rizoug N, Chaaban K, Bartholomeüs P, Le P: behavior of multicell battery. IEEE Trans Energy Convers 2010,
Dynamical modeling of Li-ion batteries for electric vehicle 25:1133-1141.
applications based on hybrid Particle Swarm–Nelder–Mead
(PSO – NM) optimization algorithm. Electr Power Syst Res 2016, 49. Erdinc O, Vural B, Uzunoglu M: A dynamic lithium-ion battery
131:195-204 http://dx.doi.org/10.1016/j.epsr.2015.10.018. model considering the effects of temperature and capacity
fading. 2009 Int Conf Clean Electr Power. 2009:383-386 http://
32. Li X, Xu L, Hua J, Lin X, Li J, Ouyang M: Power management dx.doi.org/10.1109/ICCEP.2009.5212025.
strategy for vehicular-applied hybrid fuel cell/battery power
system. J Power Sources 2009, 191:542-549 http://dx.doi.org/ 50. Liaw BY, Jungst RG, Nagasubramanian G, Case HL, Doughty DH:
10.1016/j.jpowsour.2009.01.092. Modeling capacity fade in Lithium-ion cells. J Power Sources
2005, 140:157-161 http://dx.doi.org/10.1016/
33. Jung S, Kang D: Multi-dimensional modeling of large-scale j.jpowsour.2004.08.017.
lithium-ion batteries. J Power Sources 2014, 248:498-509 http://
dx.doi.org/10.1016/j.jpowsour.2013.09.103. 51. Lu L, Han X, Li J, Hua J, Ouyang M: A review on the key issues for
lithium-ion battery management in electric vehicles. J Power
34. Li J, Wang L, Lyu C, Zhang L, Wang H: Discharge capacity Sources 2013, 226:272-288 http://dx.doi.org/10.1016/
estimation for Li-ion batteries based on particle filter under j.jpowsour.2012.10.060.
multi-operating conditions. Energy 2015, 86:638-648 http://
dx.doi.org/10.1016/j.energy.2015.04.021. 52. Xia B, Chen C, Tian Y, Sun W, Xu Z, Zheng W: A novel method for
state of charge estimation of lithium-ion batteries using a
35. Riu D, Montaru M, Bultel Y: Time domain simulation of Li-ion nonlinear observer. J Power Sources 2014, 270:359-366 http://
 batteries using non-integer order equivalent electrical circuit. dx.doi.org/10.1016/j.jpowsour.2014.07.103.
Commun Nonlinear Sci Numer Simul 2013, 18:1454-1462 http://
dx.doi.org/10.1016/j.cnsns.2012.06.028. 53. Lee S, Kim J, Lee J, Cho BH: State-of-charge and capacity
Authors developed and validated a new algorithm to simulate more estimation of lithium-ion battery using a new open-circuit
accurate non-integer order battery models with no physical approxima- voltage versus state-of-charge. J Power Sources 2008,
tion. 185:1367-1373 http://dx.doi.org/10.1016/
j.jpowsour.2008.08.103.
36. Zhong F, Li H, Zhong S, Zhong Q, Yin C: An SOC estimation
 approach based on adaptive sliding mode observer and 54. Klein R, Chaturvedi NA, Christensen J: Optimal charging
fractional order equivalent circuit model for lithium-ion strategies in lithium-ion battery. Am Control Conf 2011:382-387.

www.sciencedirect.com Current Opinion in Chemical Engineering 2016, 13:63–74

74 Energy and environmental engineering

55. Suthar B, Northrop PWC, Braatz RD, Subramanian VR: Optimal desktop. J Power Sources 2014, 269:486-497 http://dx.doi.org/
charging profiles with minimal intercalation-induced stresses 10.1016/j.jpowsour.2014.07.020.
for lithium-ion batteries using reformulated pseudo 2-
dimensional models. J Electrochem Soc 2014, 161:F3144-F3155 65. Walker W, Yayathi S, Shaw J, Ardebili H: Thermo-
http://dx.doi.org/10.1149/2.0211411jes. electrochemical evaluation of lithium-ion batteries for space
applications. J Power Sources 2015, 298:217-227 http://
56. Methekar RN, Ramadesigan V, Braatz RD, Subramanian VR: dx.doi.org/10.1016/j.jpowsour.2015.08.054.
Optimum charging profile for lithium-ion batteries to
maximize energy storage and utilization. ECS Trans 2010, 66. Spotnitz: Battery modeling. Electrochem Soc Inerface 2005:39.
25:139-146.
67. Jongerden MR, Haverkort BR: Which battery model to use? IET
57. Wu MS, Lin CY, Wang YY, Wan CC, Yang CR: Numerical Softw 2009, 3:445 http://dx.doi.org/10.1049/iet-sen.2009.0001.
simulation for the discharge behaviors of batteries in series
and/or parallel-connected battery pack. Electrochim Acta 2006, 68. Chen SX, Tseng KJ, Choi SS: Modeling of lithium-ion battery for
52:1349-1357 http://dx.doi.org/10.1016/j.electacta.2006.07.036. energy storage system simulation. 2009 Asia-Pacific Power
Energy Eng. Conf., IEEE. 2009:1-4 http://dx.doi.org/10.1109/
58. Lawder MT, Northrop PWC, Subramanian VR, Model-Based SEI: APPEEC.2009.4918501.
Layer growth and capacity fade analysis for EV and PHEV
batteries and drive cycles. J Electrochem Soc 2014, 161:A2099- 69. Gu W, Wu Z, Bo R, Liu W, Zhou G, Chen W et al.: Modeling,
A2108 http://dx.doi.org/10.1149/2.1161412jes. planning and optimal energy management of combined
cooling, heating and power microgrid: A review. Int J Electr
59. Lee J, Sung W, Choi J: Metamodel for efficient estimation of Power Energy Syst 2014, 54:26-37 http://dx.doi.org/10.1016/
capacity-fade uncertainty in Li-ion batteries for electric j.ijepes.2013.06.028.
vehicles. Energies 2015.
70. Lawder BMT, Suthar B, Northrop PWC, De S, Hoff CM,
60. Li X, Jiang J, Zhang C, Wang LY, Zheng L: Robustness of SOC
Leitermann O et al.: System (BESS) and battery management
estimation algorithms for EV lithium-ion batteries against
system (BMS) for grid-scale applications. Proc IEEE 2014:102.
modeling errors and measurement noise 2015. Math Problems
Eng 2015. 71. Mohammed YS, Mustafa MW, Bashir N: Hybrid renewable
61. Sung W, Shin C: Electrochemical model of a lithium-ion battery energy systems for off-grid electric power: review of
implemented into an automotive battery management system. substantial issues. Renew Sustain Energy Rev 2014, 35:527-539
Comput Chem Eng 2015. http://dx.doi.org/10.1016/j.rser.2014.04.022.

62. Jaguemont J, Boulon L, Dubé Y: A comprehensive review of 72. Feng X, Gooi H, Chen S: Capacity fade-based energy
lithium-ion batteries used in hybrid and electric vehicles at management for lithium-ion batteries used in PV systems.
cold temperatures. Appl Energy 2016. Electr Power Syst Res 2015, 129:150-159 http://dx.doi.org/
10.1016/j.epsr.2015.08.011.
63. Bugga R, Smart M, Whitacre J, West W: Lithium ion batteries for
space applications. 2007 IEEE Aerosp Conf. 2007:1-7 http:// 73. Riffonneau Y, Bacha S, Barruel F, Ploix S: Optimal power flow
dx.doi.org/10.1109/AERO.2007.352728. management for grid connected PV systems with batteries.
IEEE Trans Sustain Energy 2011, 2:309-320 http://dx.doi.org/
64. Walker W, Ardebili H: Thermo-electrochemical analysis of 10.1109/TSTE.2011.2114901.
lithium ion batteries for space applications using thermal

Current Opinion in Chemical Engineering 2016, 13:63–74 www.sciencedirect.com