logoGlobal Energy Interconnection




      Global Energy Interconnection

      Volume 4, Issue 2, Apr 2021, Pages 145-157

      State of charge and health estimation of batteries for electric vehicles applications: key issues and challenges

      Samarendra Pratap Singh1 ,Praveen Prakash Singh2 ,Sri Niwas Singh3 ,Prabhakar Tiwari4
      ( 1.Department of Electrical Engineering, IET, Dr Rammannohar Lohia Avadh University, Ayodhya, India , 2.Department of Electrical Power Engineering and Mechatronics Tallinn University of Technology, Estonia , 3.Department of Electrical Engineering, Indian Institute of Technology Kanpur, India , 4.Department of Electrical Engineering, Madan Mohan Malaviya University of Technology Gorakhpur, India )


      Using electric vehicles (EVs) for transportation is considered as a necessary component for managing sustainable development and environmental issues.The present concerns regarding the environment, such as rapid fossil fuel depletion, increases in air pollution, accelerating energy demands, global warming, and climate change, have paved the way for the electrification of the transport sector.EVs can address all of the aforementioned issues.Portable power supplies have become the lifeline of the EV world, especially lithium-ion (Li-ion) batteries.Li-ion batteries have attracted considerable attention in the EV industry, owing to their high energy density, power density, lifespan, nominal voltage, and cost.One major issue with such batteries concerns providing a quick and accurate estimation of a battery’s state and health; therefore, accurate determinations of the battery’s performance and health, as well as an accurate prediction of its life, are necessary to ensure reliability and efficiency.This study conducts a review of the technological briefs of EVs and their types, as well as the corresponding battery characteristics.Various aspects of recent research and developments in Liion battery prognostics and health monitoring are summarized, along with the techniques, algorithms, and models used for current/voltage estimations, state-of-charge (SoC) estimations, capacity estimations, and remaining-useful-life predictions.

      0 Introduction

      One major focus of today’s world concerns decarbonizing the transport sector.Decarbonizing transport is proving to be one of the largest R&D projects in the early 21st century.There are approximately one billion automobiles in use worldwide, satisfying many needs for mobility in daily life.In 2003-2004, the road transport sector contributed to 94.5% of the total CO2 emissions in India [1].In view of the above, India has acceded to the Paris Agreement and promoted electric vehicles (EVs), based on the framing scheme “Faster Adoption and Manufacturing EVs” and the “National Mission on Electric Mobility [2].” EV markets are growing for new car buyers, as more people are now interested in switching from petrol/diesel vehicles to EVs.

      Electric car charging stations have been growing exponentially, based on the assumption/knowledge that environmentally friendly cars will soon become a necessity.The ecosystem of EVs depends on the market demand (high capital costs and consumer perceptions), government policies (taxation on vehicles and components, subsidies), available infrastructure (charging and battery swapping infrastructures), and technical solutions (e.g., those concerning range, charging time, and safety).The rapid and widespread adoption of EVs on a large scale will be made possible by the availability of sustainable and easily accessible rechargeable energy storage batteries and improvements in regards to the driving range, reliability, safety, and power management systems.

      Providing an appropriate state-of-charge (SoC) estimation is one of the most important issues in battery applications.A precise SoC estimation of the battery power can avoid unpredicted system interruptions and prevent the batteries from becoming over-charged or over-discharged, which may cause permanent damage to the internal structures of such batteries [3].Some researchers have attempted to develop mathematical models for capturing the features of the battery chemical reactions at the cell level, aiming to explain, e.g., battery side reactions, chemical degradation, losses of active materials, thermal dynamics during resting, and increases in internal resistance.

      In general, current battery prognostic technologies are mainly based on advanced battery performance modeling methods and supplementary data-driven tools.This study critically reviews the battery prognostic technologies and elaborates on the major unmet needs in battery health monitoring for battery manufacturers, car designers, and EV drivers.Several approaches have been developed for monitoring battery health status and performance; in addition, further evolution of prognostics modeling methods has been reported.The primary goal of this review of these methodologies is to identify feasible and cost-effective solutions for managing battery life issues under dynamic operational conditions.

      This paper is divided into seven sections.Section 2 discusses EVs and their charging.Section 3 presents the electrochemical cell, along with its rating terminologies.The working and associated terms of the lithium-ion (Li-ion battery) are explained in Section 4.Section 5 describes the different modeling approaches for battery predictive technology, whereas Section 6 discusses the health monitoring of batteries.Section 7 examines different safety battery tests.

      1 Electric vehicle (EV)

      EVs are vehicles that use one or more traction motors for propulsion.An EV may be powered through a selfcontained system for converting fuel to electricity, with, e.g., a battery, solar panels, fuel cells, or an electric generator [4].Fig.1 shows a schematic of a battery-operated EV.

      Fig.1 Battery electric vehicle (BEV)

      There are three main types of EVs, classified based on the degree to which electricity is used as the energy source: battery EVs (BEVs), plug-in hybrid EVs (PHEVs), and hybrid EVs (HEVs).Owing to their large battery capacities, BEVs are capable of charging at “level 3,” i.e., a DC fast charge, as described below.The different types of EVs are shown in Fig.2 and Table 1, along with their energy sources and propulsion systems [5].HEVs and PHEVs combine an internal combustion engine with the electric drive system.

      Fig.2 Types of electric vehicles (EVs)- hybrid EV (HEV), plug-in HEV (PHEV), BEV

      EV charging requires plugging into a charger at a charging station connected to an electric grid.This charger is also denoted as the “EV supply equipment.” Based on the maximum amount of power that the charger provides power to the battery from the grid, the chargers can be categorized as follows.

      · Level 1: This level of charger is used for home charging.It provides charging through a 120-V AC wall socket plug, and does not require the installation of additional charging equipment.It can deliver 2-5 miles of range per hour of charging.It is most often used in homes and workplaces.

      Table 1 Types of electric vehicles (EVs) and propulsion systems

      S.No Types of EV Energy Source Propulsion 1.Hybrid EV (HEV) Gasoline/ Diesel Electric Drive+ internal combustion engine (ICE)2.Plug-in-HEV (PHEV)Gasoline/ Diesel+ on board charger for battery Electric Drive+ ICE 3.Battery EV (BEV/EV)Electrical Charging Pure Electric Drive

      · Level 2: This level of charger provides a platform for home and public charging through a 240-V plug and, requires the installation of additional charging equipment.It can deliver 10-20 miles of range per hour of charging.It is used in homes, workplaces, and public charging stations.

      · Level 3: This level of charger concerns a DC fast charger that provides charging through a 480-V AC input, and requires highly specialized, high-powered equipment, as well as special equipment in the vehicle itself.These systems typically provide up to an 80% charge in just 20-30 min [6].

      2 Electrochemical cell

      Electrochemical energy storage (via a battery storage system) offers the greatest flexibility in regard to energy and power capacity (MW and MW/h ratings) across the vast majority of application markets.The continuously decreasing cost of Li-ion batteries (LIBs) further reduces the installation cost.LIBs provide a quick response, enabling fast energy management.Advancements in power electronics and power conversion systems have resulted in the better utilization of battery-based storage systems.Electrochemical cells can generally be classified as shown in Fig.3.An equivalent circuit model of a corresponding battery and the V-I characteristic of the cell [8] are shown in Fig.4.

      2.1 Cell voltage

      The cell voltage is the potential difference between the two electrodes (anode and cathode), and is measured in volts (V).Under a no-load condition, the cell has an open-circuit voltage (OCV), which is usually close to the theoretical voltage.The IR drop in the cell voltage is owing to the current flowing across the internal resistance of the battery.

      · Activation polarization - This term refers to the various retarding factors inherent to the kinetics of an electrochemical reaction, such as the work function that the ions must overcome at the junction between the electrodes and electrolyte.

      · Concentration polarization - This factor considers the resistance faced by the mass transfer (e.g., diffusion) process by which ions are transported across the electrolyte from one electrode to another [8].

      Fig.3 Classification of electrochemical cells

      Fig.4 Equivalent Circuit Model of Batteries and V-I Characteristic of Cells [8]

      2.2 Capacity

      The total charge (Q) transported by the current during a specific time period indicates the capacity of a cell.The theoretical capacity is defined as the current density passing through the cell until the cut-off voltage is reached.The charge-delivering capability of an electrochemical cell depends on several factors, such as the cell chemistry, amount of active electrode material, discharge rate, voltage level, and temperature.The capacity variation depends on the discharge rate of the cell.

      2.3 Charge and discharge rates

      The rate at which charging, and discharging occurs is defined as the charge and discharge rate, or C-rate.The C-rate expresses the current capabilities of an electrochemical cell.The effects of different discharge rates on the voltage profile are shown in Fig.5.

      Fig.5 (a) Relationship between discharge rate and capacity (b) Capacity variation with discharge time [8]

      2.4 Energy and power (Cell)

      The energy of a cell refers to the amount of useful work that the cell can perform until the cut-off voltage is reached.The energy stored in a cell depends on the voltage and charge stored.The power of the cell refers to the rate at which this work can be performed.This can also be understood as follows:

      · the energy determines how far the vehicle can go, i.e., the distance;

      · the power determines how fast the vehicle can go, i.e., its speed.

      2.5 Specific energy and power

      The specific energy is the amount of electrical energy stored for every kilogram of battery mass (Wh/kg).This is also known as the gravimetric energy density.The power of the battery is the amount of electrical power obtained from every kilogram of battery mass.It is a highly variable quantity, as the power output of the battery depends on the load connected to it, rather than on the battery itself.

      3 Battery: Li-Ion

      The battery is the heart of an EV.EVs can be used as portable energy storage devices.Rechargeable battery systems are the most important energy storage systems, and provide readily convertible chemical energy.Over the past decade, many new technologies have improved the battery capacity and power density.However, further developments are needed to improve the storage capacity, and a significant balancing of systems is required to meet functional requirements and to reduce catastrophic failures [7].The different layers of a Li-ion cell are shown in Fig.6.The working principle is described below.

      A Li-ion cell comprises a positive electrode (anode), negative electrode (cathode), separator, and two current collectors.Li+ is transferred from the anode to the cathode through an electrolytic separator to complete the discharging cycle.The negative electrode is generally formulated from graphite, and the anode generally contains one of the following materials: Li-ion manganese oxide (LMO), Li-iron-phosphate, or Li-nickel-manganese-cobalt-oxide.Diethyl carbonate or ethylene carbonate is used as the electrolyte.Aluminium and copper are used as the positive and negative current collectors, respectively.The chemical reactions of LMO/graphite have been reported as examples of both charging and discharging [11].

      Fig.6 Different layers of Tesla cell (Li-ion) [9]

      In comparison to other cell technologies used for various applications, such as nickel-metal hybrids, nickel-cadmium, lead-acid, or super capacitors, the major advantage of Liion technology is its high energy density (approximately 180 Wh/kg) and reasonable power density (approximately 9 kW/kg).Only supercapacitors can deliver higher power densities (30 kW/kg); nevertheless, they have limited feasibility, owing to their limited energy density (3-5 Wh/kg) [11].

      Fig.7 Simplified working diagram of a Li-ion battery [6]

      4 Battery modelling techniques

      Some existing battery models are discussed in this section.Aging, fading, and other issues are also discussed in [64].

      4.1 Physical modelling approach

      A simple internal resistance battery model (Rint model) was tested in the National Renewable Energy Laboratory and implemented on an advanced vehicle simulator (ADVISOR) [30].It comprised an ideal battery with an OCV and constant internal resistance (Rint), as shown in Fig.8.

      Fig.8 Rint model [31]

      The SoC for the Rint model was estimated by performing ampere counting, including considering the Coulombic efficiency losses during charging, as shown in Eq.(1) as follows:

      For A >0, Discharge and for A <0, Charge [31].

      The measurable terminal voltage (Vo) can be acquired from the open-circuit voltage, and (R) can be obtained from an open-circuit measurement and one extra measurement with the load associated at the terminal when the battery is fully charged [32, 33].The Rint model’s voltage predictions are accurate to within 3%-12%.Although this model has been extensively used owing to its simplicity, it does not consider the varying characteristics of the internal impedance of the battery with varying SoCs, electrolyte concentrations, and sulfate formations.Such a model is only valid under steady-state load conditions, as its voltage response to load changes is too fast.The observed limitations of the Rint model are that the model’s voltage response to load changes is excessive, and that the internal resistance does not change as a function of the current magnitude [31].

      4.2 Resistance-capacitance model

      The resistance-capacitance (RC) battery model was implemented in ADVISOR in 2001 [31].A schematic of the electrical model is presented in Fig.9.The electrical model consists of two capacitors (Cb and Cc) and three resistors (Re, Rc, and Rt).Capacitor Cb is very large, and represents the ample capability of the battery to chemically store charge.Capacitor Cc is small, and mostly represents the surface effects of a cell.The parameters vary with the SoC and temperature (T).

      Fig.9 Resistance-capacitance (RC) model [31]

      The SoC for the RC model was estimated using the voltages of the two capacitors, as given in Eq.(3).The estimator weighed VCb heavily, as it represented the bulk energy in the battery.The RC SoC estimator [32-33] was constructed as follows:

      4.3 Thevenin model

      As batteries are being applied to various and more dynamic electrical loads, the Rint model is not sufficient for accurately forecasting the behaviors of batteries.To simulate the dynamic behaviors of batteries, the Thevenin model was developed based on the Thevenin theory, as shown in.Fig.10.It includes resistors and an RC network in series, and is used to predict the battery response at a particular SoC, assuming that the open-circuit voltage is constant [10].The accuracy of the Thevenin model for predicting dynamic behaviors can be improved by adding additional RC networks.A Thevenin model with three pairs of RC circuits can simulate a short-term transient response (in seconds), mediate-term transient response (in minutes), and long-term transient response (in hours) [35,36].

      Fig.10 Thevenin’s model [34]

      In this study, after a deep analysis of the batteries, it was found that the basic Thevenin model was not capable of explaining all of the batteries’ behaviors.For instance, the resistors in the basic Thevenin model are assumed to be constant for both the charge and discharge processes, which may not be correct in the real world.In 1992, Salameh improved the basic Thevenin model by adding ideal diodes in series with the resistors, aiming to differentiate their values in the charge and discharge processes [34].The Zener diode model was also developed based on the Thevenin model, by adding a Zener diode in parallel with the long-term response RC network.

      Abu-Sharkh and Doerffel [37] proposed a rapid test method for measuring the open-circuit voltage and internal resistance of a battery.During their experiments, they observed a constant voltage change during a long-term transient response between 10% and 90% SoC levels in both the charging and discharging processes.This phenomenon was very similar to that of a Zener diode, and the constant voltage drop could be modelled as the Zener knee voltage (Fig.11) [37].

      Fig.11 Improved Thevenin model [37]

      4.4 Runtime-based electrical model

      As batteries are being applied to more energy-intensive systems such as EVs, information regarding batteries’ capabilities and duration has become very important.Having exact information regarding such capabilities and duration has become crucial for users.In 1993, Hageman proposed a runtime-based model as shown in Fig.12, and simulated the tools using “PSPICE” [38].

      The runtime base model used a complicated circuit network to create the battery runtime and DC voltage response for a constant discharge current.An advantage of this model is its ability to simulate capacity weakening owing to factors such as thermal effects and aging [38].However, this model cannot accurately predict runtimes nor voltage responses under dynamic load conditions [22].

      Fig.12 Runtime model [10]

      4.5 Combined electric model

      To simulate the voltage response and battery runtime while considering the impacts of battery degradation and thermal effects, Chen et al.proposed a combined electrical model in 2006.As shown in Fig.13, this model is a combination of a runtime model (left part) and RC networks similar to those in Thevenin-based models (right part).

      Fig.13 Combined electrical model [10]

      In the combined electrical model, the capacitor and current-controlled current source are used to model the capacity, SoC, and runtime of the battery.The RC networks are capable of simulating the transient response of the battery’s terminal voltage under dynamic load conditions.In this model, the battery’s usable capacity is no longer considered as a constant, but rather as a function of the cycle numbers, battery temperature, and storage time (selfdischarge) [39,40].As the OCV directly depends on the SoC, the voltage-controlled voltage source is used to modify its value according to varying SoC values.This model considered two RC networks to represent the transient voltage responses of two different time constants, so as to balance the complexity and accuracy.

      Despite its ability to predict the runtime and dynamic behavior of a battery with relatively high accuracy, this model is still unable to predict a battery’s of health (SoH) or to self-update its parameters.Hence, the prediction results are less accurate as the battery degrades over time.While some information regarding a battery can be obtained from direct measurements, hidden information such as the SoC and SoH are buried in a massive number of signals, and these signals need to be decoded in terms of the parameters.Before intelligent algorithms can be applied to estimate this invisible information, a good understanding of the battery behavior should be established; this is the main purpose of this study.The process of developing an equivalent circuit model is one of the paths for understanding battery behaviors.The different battery models described here have built up the foundation for developing linear space equations for battery prognostics.To make decisions on which model to use based on practical analysis, one should consider for which specific battery types a model can be applied, and the desired balance between accuracy and simplicity.

      4.6 Data-driven approach

      In data-driven techniques, a wise decision is generally suggested based on results learned from the historical data of the system.In this method, an important assumption is that the data conditions and regime remain constant until a system (e.g., a battery) fails [41].Pecht and Jaai classified these techniques into three categories, based on the type of labelled data available: supervised learning when both healthy and faulty data are accessible; semi-supervised learning, when only one of the classes is known; and unsupervised learning, when no labelled data are available [42].One of the benefits of data-driven approaches for EV batteries is that they are capable of learning the behavior of the battery based on monitored vehicle data; thus, they can be applied as black-box models, and do not demand battery chemical modelling and knowledge.The SoC and SoH can be predicted using a variety of data-driven techniques.Several practical techniques are reviewed in detail below.

      (a) Neural network: The main advantage of neural network (NN) methods is that they are established automatically by training, without the need for the detection of model parameters and coefficients.A feedforward network has been categorized as a type of nonlinear autoregressive (AR) model, and a recurrent network as a non-linear AR moving average (ARMA) model.This comparison suggests that recurrence models have more advantages over feedforward models, just as ARMA models have advantages over AR models [43].Researchers have utilized this method for both SoC estimations and available capacity predictions.Yamazaki used four neurons as inputs, indicating the terminal voltage, battery temperature, internal resistance, and discharge current, respectively.[44].Shen used seven neurons as an input layer; the first five represented the discharge capacity for five current ranges, respectively, the sixth neuron represented the regenerative capacity for a regenerative current, and the seventh neuron represented the temperature[45].

      As battery performance degrades over time, there is a relationship between the existing SoCt and previous SoCt-1.Therefore, the previous SoCt-1 can be interpolated as an input variable [46].Linda et al.[47] followed the same logic using voltage and current.Charkhgard et al.[48] introduced a radial basis function for NNs.The same logic was used to estimate the remaining useful life of a battery by inputting the capacity of a previous cycle to predict the capacity for the next cycle, according to a laboratory experimental data set.Capizzi et al.[49] used an RNN to predict the SoC and battery terminal voltage, and then to estimate mathematical battery model parameters.Monfared et al.[50] implemented an RNN to estimate the parameters for an equivalent battery model for a lead acid battery.

      (b) Support vector machine: According to [51], prediction and analysis based on regression is more complicated than classification using a support vector machine (SVM).The basic idea of using an SVM to determine a regression model for state prediction is to map the data in the input space, and then to transform them into a higher-dimensional feature space using a non-linear transfer function.The sample data in the feature space using a linear function [52, 53].In the case of batteries, it is very important to formulate these terms based on battery variables and outputs [54].The principle of using SVM for SoC prediction is to model the nonlinear dynamics of the battery for both the charge and discharge processes.An SVM developed to use the current, temperature, and SoC as inputs to predict the output of the load voltage yielded a maximum relative error of 3.61%.

      (c) Fuzzy Logic Approach: A simple fuzzy logic learning system has been developed and implemented for SoC estimations for rechargeable batteries [24, 55-57].For example, if a battery has a 30% SoC, then there may be two corresponding concepts for customers, e.g., full or empty.Various methods have been employed to provide appropriate data for fuzzy logic models.Singh et al.[58] acquired electrochemical impedance spectroscopy data for NieMH batteries over 28 cycles, comprising real and imaginary values.The adaptive neuro-fuzzy inference system (ANFIS) is a method that can be applied to any type of battery under different operating conditions, such as under constant discharging and partial discharging, if trained before use [59].The main advantage of the ANFIS method is that it can use the exact solution of a NN, as well as the heuristic knowledge of fuzzy logic; however, each of these methods cannot be applied individually [60].

      (d) Fusion Approach: A critical issue with data-driven techniques is that when the data availability is not satisfied or the data are biased, the results can be imprecise, or even entirely incorrect.Physical model methods, in contrast, are not as flexible as data-driven methods, as they rely on the establishment of a physical model, which usually requires a large amount of expert knowledge and system testing.Data-driven and physical-model methods are potentially complementary to each other; therefore, it is desirable to develop a fusion model based on combining the two approaches to achieve an optimal battery prognostics and health management solution [41].Several stochastic filtering techniques, which can be classified as fusion approaches, have been applied to battery SoC and SoH estimations.Table 2 shows a comparison of the various modelling approaches.

      5 Key issues and challenges of electric vehicles

      5.1 Uncertainty of battery behaviour and internal characteristics

      The internal electrochemical process of a battery is nearly impossible to observe, owing to its nonlinear and time-variable systems.Most battery state parameters, which control the battery performance during design, manufacturing, and usage, are usually subject to interference from a dynamic ambient environment.The basic charge and discharge processes are affected by environmental conditions in terms of their phase change reactions and chemical materials.A misunderstanding of the battery characteristics and/or performance state can cause substantial issues for battery management, and impacts the mobility of EVs or HEVs.The uncertainty in the internal characteristics of batteries owing to complex operating environments consequently makes the performance observed at the cell level hardly translatable to that observed at the system level.Although the models discussed in Section 5 attempt to build a good fundamental understanding of the electrochemical process, they are still far from being applied in practice [10].

      Table 2 Advantages and disadvantages of different approaches of state-of-charge (SoC) estimation [10, 62]

      Technique Advantage Drawbacks SOC Estimation Error Thevenin model Simple and easy to implement Resistance and capacitor assumed to be constant Max 1.2%Runtime-based electrical model Simulate capacity fading owing to factors Cannot predict capacity fading, not good in dynamic load conditions Max 4.32%Combined electrical model Considers thermal effects and aging, accurate in dynamic load conditions Weak in self-updating model parameters Mean 1.43%Neural Networks Matches with other techniques Needs large amount of training data, depends on historic data set Max < 4%Support vector machine (SVM) Suitable for different battery applications Needs large amount of training data Max < 3.61%Fuzzy logic Formulated in human thinking way, easily combined with neural network Not sufficiently accurate Max < 10%Kalman filter Accurate estimation of SoC, no initial SoC needed, can easily filter noise on data Large amount of calculations needed, complicated 0.7%Sliding mode observer Simple control structure and robust tracking performance under uncertain environments, fast SoC estimation, high accuracy Slow time-varying observer for state of health (SoH) Max < 3%

      5.2 Safety issues

      New LIBs have a higher energy density; therefore, they are more prone to safety issues if any faults occur.When a high amount of electricity is flowing and the battery temperature increases to several hundred degrees within seconds, the increase in temperature will spread to neighbouring cells, resulting in the battery catching on fire or exploding [11].One of the most critical steps for developing battery prognostics solutions is to establish a battery model for simulating battery behaviors and interpreting battery issues in a form that can be understood by users and designers.In the last several decades, many different models have been developed, with limited accuracy [12-13].Other researchers have taken advantage of theoretical explanations of battery behaviors from electrochemical models and integrated them with battery system issues in SoC and SoH modelling; this approach has been claimed to provide more accurate results than simple mathematical theories [16-18].

      Models are generally established based on batterydependent variables, such as the concentration, potential, reaction rate, and current density.Under a concentrated solution theory, John Newman used Stefane-Maxwell equations to explain the mass and energy transport of each species for each phase and component of the battery cell [14].This model could explain the side effects and heat released during relaxation.It also helped to understand the shape of the discharge curve.Regarding macroscopic information, other well-known electrochemical models (such as Peukert’s Law and Shepherd’s model) describe battery electrochemical behaviors in terms of voltage and current changes [15].The development of electrochemical models for battery systems has been making great progress based on simulating system responses under nominal operating conditions.The models support investigations of battery failure diagnoses and thermal effects, specifically for cell assemblies and battery pack systems.

      5.3 Modelling issues

      In contrast to the goal of achieving a profound understanding through the modelling of complicated electrochemical processes at the cell level, researchers are also trying to simply model battery performance using electrical circuits [19].Over the last 20 years, researchers have created many equivalent circuit models for batteries.All of these models use an arrangement of voltage sources, resistors, and capacitors to simulate battery performance [20-21].Other models employ a capacitor in parallel with a resistor (to show internal resistance) to model the transient response of the battery voltage, which a steady-state model would not predict [22].However, when encountering the issues of capacity fade, thermal influence, and energy density changes, most often these models neglect the impact of degradation, and are not able to aid in understanding the interactions between components.In view of the above, most of these models do not consider the reliability of the electrochemical devices [13].Reference [63] provides a review of modelling issues.

      As one example, the influence of uncertainty in the load profile during dynamic battery operation is not considered in typical electrical circuit models.This limits their contributions to real EV applications, although they are widely employed in system design.The consideration of the dynamic operations and the complexities of circuit models are critical for EV applications.Another approach concentrates on taking full use of the battery data during operation to estimate the battery energy storage level based on a simplified electrical circuit model.Many researchers have started to focus on combining circuit models with advanced data-driven prognostic techniques [23].

      5.4 Issue and challenges in state of health (SoH)

      Issues regarding battery aging processes occur owing to irreversible changes in the characteristics of the electrolyte, anode, and cathode, as well as in the structure of the components used in the battery.Battery aging processes can be classified into two categories: aging processes that involve gradual degradation over time (which can be monitored), and those that do not have any specific mode or observable sign until a major problem or rapid changes in battery performance occur.An example of this is dendrite formation in lithium-ion batteries [24].When a battery ages, tiny particles of lithium create a fiber structure known as a dendrite that can cause a battery fire.Once this happens, short circuits can occur, causing a sudden rise in temperature and catastrophic failure of the battery.Such sudden changes can be considered an important safety issue in batteries.In contrast, [25,26] reviewed the gradual performance losses in batteries.Several studies have reported on battery SoH.

      Various techniques using artificial NNs, expert systems, and fuzzy logic [55]-based approaches have been utilized to obtain the SoH of batteries [3, 34, 36].There is still a need for the development of an effective and accurate health monitoring system for batteries.This will also help in regard to the safe operation of the battery system.

      5.5 Estimation of state of charge (SOC)

      The battery SoC estimates the amount of energy remaining in a cell relative to the energy it had when it was fully charged, and gives the user an indication of how long a battery will last before it needs recharging.Accurate SoC estimations have always been a critical and important concern in the design of battery management systems in EVs.In EV applications, the SoC works like a fuel gauge in a car.However, SoC estimation is not an easy task, and depends on the battery’s chemistry and condition.Typically, SoC estimations are classified into two main categories, i.e., direct and indirect estimations [27].

      Providing an accurate SoC estimation of a Li-ion battery is a difficult task, because it cannot be directly assessed using any physical sensor [28].Currently, Li-ion battery SoC estimation is a popular topic for researchers.Several SoC estimation techniques have been reported in the last decade.A summary of classes of SoC measurement techniques with examples is shown in Fig.14.Reference [64] provides a good review of SoC estimations.

      Fig.14 Classification of state-of-charge (SoC) measurements [29, 30]

      5.6 Prospects of future trends and research

      Researchers and manufacturers are working very hard to increase the gravimetric and volumetric energy densities of batteries.Fast charging and battery swapping mechanics are also being explored in various countries.Smart converters and their control for charging and discharging are a matter for further improvement.Moreover, researchers are considering the roles of EVs in providing ancillary services for ensuring secure and reliable power grids.It is also possible to use suitable market mechanisms to make money based on the idle/parking conditions of EVs.

      6 Different kinds of tests for batteries

      In an EV, it should be ensured that no harm can be done to any person through high currents and voltages.Therefore, when choosing a battery technology and notably, the special chemistry of the cells, very close attention must be paid to the suitability of the technology, so as to provide an intrinsically safe battery system.

      (a) Safety test: The main transportation and safety tests are listed as follows.

      - Forced discharge: A fully charged cell is discharged to 0 V with 1/3 C.

      - Overcharge: A cell is charged with 1 C to a cell voltage of 5 V or for 1.5 h, and it is ensured that the cells do not show an explosion, fire, or leakage.

      - External short circuit: An external short circuit with R≤0.005 Ω is applied to a fully charged cell for 600 s.Additionally, if the cell module is exposed to an external short circuit at a temperature of 55± 2 °C on the outer housing with R≤0.1 Ω, until the cell module reaches a temperature of 55±2 °C on the outer housing again or after 1 h, it must be ensured that no leakage, no opening, no disassembly, no crack in the cell module, no higher temperature than 170 °C on the outer housing of the cell module, and no fire (thermal event) occur within the test duration, and for an additional 6 h [61].

      (b) Thermal test: The battery is stored at a temperature of 80 ± 2 °C for a maximum of 1 h.During this time, the maximum hazard level cannot exceed Hazard Level 2 (i.e., no leakage; no venting, fire, or flame; no rupture; no explosion; no exothermic reaction or thermal runaway; no irreversible damage to the cell).As requested, with the test article fully charged, the temperature is increased in increments of 5 °C (hold time at each temperature step: 30 min) until a thermal effect is detected twice, until the temperature reaches 200 °C above the operating temperature of the cell, or until a catastrophic event (e.g., venting or major damage to the device) occurs.A test is conducted at 150 °C for 30 min, and the 100% SoC must pass Hazard Level 4 (i.e., no fire or flame; no rupture; no explosion; weight loss ≥ 50% of electrolyte weight).

      If the cell module is exposed to an ambient pressure of 11.6×103 Pa for at least 6 h at a temperature of 20 ± 5 °C, no leakage, no opening, no disassembly, no crack in the cell module, no open-circuit voltage of less than 90% of its voltage immediately prior to this procedure (if the cell module is not fully discharged), and no fire (thermal event) in the cell module may occur.

      (c) Vibration test: For all cells or cell modules with a gross mass of less than or equal to 12 kg, the following observations must be valid:

      · The vibration is a sinusoidal waveform with a logarithmic sweep between 7 and 200 Hz and back to 7 Hz, as traversed in 900 s.

      · This cycle is repeated 12 times for a total of 3 h for each direction of the vehicle coordinate system.

      · The logarithmic frequency sweep must meet from 7 Hz a peak.

      (d) Crush test: The tested device should be crushed between a resistance and crush plate with a force of at least 100 kN (but not exceeding 105 kN) with an onset time of less than 3 min and hold time of at least 100 ms (but not exceeding 10 s).The test should end with an observation period of 1 h at ambient temperature conditions in the test environment.

      (e) Nail penetration test: The cell is penetrated with a steel (conductive) pointed rod at a rate of 8 cm/s or less.The diameter of the rod is 3 mm.The penetration orientation is perpendicular to the electrode plates.

      In addition to the aforementioned tests for safety, environmental tests are also required, for example, those regarding chemical resistance, noxious gases, and salt spray tests.The test standards and protocols published/used by national and international standardization bodies such as the International Energy Commission, Institute of Electrical and Electronics Engineers, and Bureau of Indian Standards are generally considered for performance analyses of manufactured LIBs and for batch acceptance before being supplied to customers.Different international, national, and regional standards for testing Li-ion batteries can be found in [65].

      7 Conclusion

      EVs will comprise the future modes of transport systems.There are several concerns being addressed by policymakers, researchers, and manufacturers.Battery systems are very important components of EVs.In general, three main approaches, i.e., physical-model approaches, data-driven approaches, and fusion model approaches are employed for battery modelling and prognostics.This paper provides a comprehensive review of the types of EV and Li-ion battery systems, along with their modelling.The key issues and challenges in EVs are presented and critically discussed, providing very useful information to researchers in this area.Various safety tests are also discussed.


      This work was financially supported by Department of Science and Technology, New Delhi (Indo-Norway consortium) project entitled “Integrated Renewable Resources and Storage Operation and Management” program.

      Declaration of Competing Interest

      We have no conflict of interest to declare.


      1. [1]

        Pollet, B G, et al (2014) Fuel-cell (hydrogen) electric hybrid vehicles.In: Richard Folkson (ed) Alternative Fuels and Advanced Vehicle Technologies for Improved Environmental Performance, Wood head, Cambridge, p 685-735 [百度学术]

      2. [2]

        Alok Kumar (2020) The Gazette of India, Department of Heavy Industry.https://fame2.heavyindustry.gov.in/.Accessed 15 Sept 2020 [百度学术]

      3. [3]

        Ng K S, Moo C S, Chen Y P et al (2009) Enhanced coulomb counting method for estimating state-of-charge and state-of-health of lithium-ion batteries.Applied Energy, 86(9): 1506-1511 [百度学术]

      4. [4]

        Electric Vehicle (2020) Wikipedia https://en.wikipedia.org/wiki/Electric_vehicle.Accessed 19 Sept 2020 [百度学术]

      5. [5]

        EDI (2020) Engineered Design Insider.https://www.ediweekly.com/about-edi-weekly-engineered-design-insider/.Accessed 10 Oct 2020 [百度学术]

      6. [6]

        EVgo (2020) Types of Electric Vehicles.https://www.evgo.com/why-evs/types-of-electric-vehicles/.Accessed 10 Oct 2020 [百度学术]

      7. [7]

        Eletimesworld (2020) ELECtric Vehicles: Technology Brief.https://www.eletimes.com/electric-vehicles-technology-brief.Accessed 9 Oct 2020 [百度学术]

      8. [8]

        Saha, Bhaskar, Kai Goebel (2009) Modelling Li-ion battery capacity depletion in a particle filtering framework.Paper presented at annual conference of the prognostics and health management society, San Diego, 27 September-1 October 2009 [百度学术]

      9. [9]

        Lesics(2020).Lithium-Ion Battery, How Does It Work.https://youtu.be/VxMM4g2Sk8U.Accessed 10 Sept 2020 [百度学术]

      10. [10]

        Rezvanizaniani S M, Liu Z C, Chen Y et al (2014) Review and recent advances in battery health monitoring and prognostics technologies for electric vehicle (EV) safety and mobility.Journal of Power Sources, 256: 110-124 [百度学术]

      11. [11]

        Jacoby M (2007) Burning batteries.Chemical & Engineering News Archive, 85(51): 26-28 [百度学术]

      12. [12]

        Santhanagopalan S, Guo Q Z, Ramadass P et al (2006) Review of models for predicting the cycling performance of lithium ion batteries.Journal of Power Sources, 156(2): 620-628 [百度学术]

      13. [13]

        Chandrasekaran R, Bi W, Fuller T F (2008) Robust design of battery/fuel cell hybrid systems—Methodology for surrogate models of Pt stability and mitigation through system controls.Journal of Power Sources, 182(2): 546-557 [百度学术]

      14. [14]

        Newman J, Thomas K E, Hafezi H et al (2003) Modelling of lithiumion batteries.Journal of Power Sources, 119-121: 838-843 [百度学术]

      15. [15]

        Doerffel D, Sharkh S A (2006) A critical review of using the Peukert equation for determining the remaining capacity of leadacid and lithium-ion batteries.Journal of Power Sources, 155(2): 395-400 [百度学术]

      16. [16]

        Blood P J, Sotiropoulos S (2002) An electrochemical technique for state of charge (SOC) probing of positive lead-acid battery plates.Journal of Power Sources, 110(1): 96-106 [百度学术]

      17. [17]

        Dees D W, Battaglia V S, Bélanger A (2002) Electrochemical modeling of lithium polymer batteries.Journal of Power Sources, 110(2): 310-320 [百度学术]

      18. [18]

        Schmidt A P, Bitzer M, Imre Á W et al (2010) Experimentdriven electrochemical modeling and systematic parameterization for a lithium-ion battery cell.Journal of Power Sources, 195(15): 5071-5080 [百度学术]

      19. [19]

        González-Longatt, F.M.(2006).Circuit based battery models: A review.Congreso Iberoamericano de estudiantes De Ingenieria Electrica.Puerto La Cruz.3-7 Abril 2006 [百度学术]

      20. [20]

        Roscher M A, Sauer D U (2011) Dynamic electric behavior and open-circuit-voltage modeling of LiFePO4-based lithium ion secondary batteries.Journal of Power Sources, 196(1): 331-336 [百度学术]

      21. [21]

        Chen M, Rincon-Mora G A (2006) Accurate electrical battery model capable of predicting runtime and I-V performance.IEEE Transactions on Energy Conversion, 21(2): 504-511 [百度学术]

      22. [22]

        Lee J, Ni J, Djurdjanovic D et al (2006) Intelligent prognostics tools and e-maintenance.Computers in Industry, 57(6): 476-489 [百度学术]

      23. [23]

        Wenzl H, Baring-Gould I, Kaiser R et al (2005) Life prediction of batteries for selecting the technically most suitable and cost effective battery.Journal of Power Sources, 144(2): 373-384 [百度学术]

      24. [24]

        Dubarry M, Svoboda V, Hwu R et al (2007) Capacity and power fading mechanism identification from a commercial cell evaluation.Journal of Power Sources, 165(2): 566-572 [百度学术]

      25. [25]

        Vetter J, Novák P, Wagner M R et al (2005) Ageing mechanisms in lithium-ion batteries.Journal of Power Sources, 147(1-2): 269-281 [百度学术]

      26. [26]

        Ulrich L (2012) State of charge.IEEE Spectrum, 49(1): 56-59 [百度学术]

      27. [27]

        Rakhmatov D (2009) Battery voltage modeling for portable systems.ACM Transactions on Design Automation of Electronic Systems, 14(2): 1-36 [百度学术]

      28. [28]

        Piller S, Perrin M, Jossen A (2001) Methods for state-of-charge determination and their applications.Journal of Power Sources, 96(1): 113-120 [百度学术]

      29. [29]

        Singh P, Vinjamuri R, Wang X Q et al (2006) Design and implementation of a fuzzy logic-based state-of-charge meter for Li-ion batteries used in portable defibrillators.Journal of Power Sources, 162(2): 829-836 [百度学术]

      30. [30]

        Johnson V H (2002) Battery performance models in ADVISOR.Journal of Power Sources, 110(2): 321-329 [百度学术]

      31. [31]

        Xu L, Wang J P, Chen Q S (2012) Kalman filtering state of charge estimation for battery management system based on a stochastic fuzzy neural network battery model.Energy Conversion and Management, 53(1): 33-39 [百度学术]

      32. [32]

        Domenico D, Stefanopoulou A, Fiengo G (2010) Lithium-ion battery state of charge and critical surface charge estimation using an electrochemical model-based extended Kalman filter.Journal of Dynamic Systems, Measurement, and Control, 132(6): 061302 [百度学术]

      33. [33]

        Salameh Z M, Casacca M A, Lynch W A (1992) A mathematical model for lead-acid batteries.IEEE Transactions on Energy Conversion, 7(1): 93-98 [百度学术]

      34. [34]

        Zhang J L, Lee J (2011) A review on prognostics and health monitoring of Li-ion battery.Journal of Power Sources, 196(15): 6007-6014 [百度学术]

      35. [35]

        Salameh Z M, Casacca M A, Lynch W A (1992) A mathematical model for lead-acid batteries.IEEE Transactions on Energy Conversion, 7(1): 93-98 [百度学术]

      36. [36]

        Xing Y J, Miao Q, Tsui K L et al (2011) Prognostics and health monitoring for lithium-ion battery.Proceedings of 2011 IEEE International Conference on Intelligence and Security Informatics.Beijing, China.IEEE, 242-247 [百度学术]

      37. [37]

        Abu-Sharkh S, Doerffel D (2004) Rapid test and non-linear model characterisation of solid-state lithium-ion batteries.Journal of Power Sources, 130(1-2): 266-274 [百度学术]

      38. [38]

        Hageman, Steve C.(1993) Using PSpice to simulate the discharge behavior of common batteries.EDN Mag, 260-286 [百度学术]

      39. [39]

        Kroeze R C, Krein P T (2008) Electrical battery model for use in dynamic electric vehicle simulations.2008 IEEE Power Electronics Specialists Conference.Rhodes, Greece.IEEE, 1336-1342 [百度学术]

      40. [40]

        Kim I S (2008) Nonlinear state of charge estimator for hybrid electric vehicle battery.IEEE Transactions on Power Electronics, 23(4): 2027-2034 [百度学术]

      41. [41]

        Pecht, Michael G (2010) A prognostics and health management roadmap for information and electronics-rich systems." IEICE ESS Fundamentals Review 3(4): 4_25-4_32 [百度学术]

      42. [42]

        Wang S Q, Verbrugge M, Wang J S et al (2011) Multi-parameter battery state estimator based on the adaptive and direct solution of the governing differential equations.Journal of Power Sources, 196(20): 8735-8741 [百度学术]

      43. [43]

        Wang W Q, Golnaraghi M F, Ismail F (2004) Prognosis of machine health condition using neuro-fuzzy systems.Mechanical Systems and Signal Processing, 18(4): 813-831 [百度学术]

      44. [44]

        Yamazaki T, Sakurai K, Muramoto K (1998) Estimation of the residual capacity of sealed lead-acid batteries by neural network.INTELEC - Twentieth International Telecommunications Energy Conference (Cat.No.98CH36263).San Francisco, CA, USA.IEEE, 210-214 [百度学术]

      45. [45]

        Shen W X (2007) State of available capacity estimation for leadacid batteries in electric vehicles using neural network.Energy Conversion and Management, 48(2): 433-442 [百度学术]

      46. [46]

        Piao C H, Fu W L, Jin-Wang et al (2009) Estimation of the state of charge of Ni-MH battery pack based on artificial neural network.INTELEC 2009-31st International Telecommunications Energy Conference.Incheon, Korea (South).IEEE, 1-4 [百度学术]

      47. [47]

        Linda O, William E J, Huff M et al (2009) Intelligent neural network implementation for SOCI development of Li/CFx batteries.2009 2nd International Symposium on Resilient Control Systems.Idaho Falls, ID, USA.IEEE, 57-62 [百度学术]

      48. [48]

        Charkhgard M, Farrokhi M (2010) State-of-charge estimation for lithium-ion batteries using neural networks and EKF.IEEE Transactions on Industrial Electronics, 57(12): 4178-4187 [百度学术]

      49. [49]

        Capizzi G, Bonanno F, Tina G M (2011) Recurrent neural network-based modeling and simulation of lead-acid batteries charge-discharge.IEEE Transactions on Energy Conversion, 26(2): 435-443 [百度学术]

      50. [50]

        Abolhassani Monfared N, Gharib N, Moqtaderi H et al (2006) Prediction of state-of-charge effects on lead-acid battery characteristics using neural network parameter modifier.Journal of Power Sources, 158(2): 932-935 [百度学术]

      51. [51]

        Hansen T, Wang C J (2005) Support vector based battery state of charge estimator.Journal of Power Sources, 141(2): 351-358 [百度学术]

      52. [52]

        Álvarez Antón J C, García Nieto P J, de Cos Juez F J et al (2013) Battery state-of-charge estimator using the SVM technique.Applied Mathematical Modelling, 37(9): 6244-6253 [百度学术]

      53. [53]

        Wu X, Mi L, Tan W et al (2011) State of charge (SOC) estimation of Ni-MH battery based on least square support vector machines.Advanced Materials Research, 211-212: 1204-1209 [百度学术]

      54. [54]

        Chen Y Q, Long B, Lei X (2011) The battery state of charge estimation based weighted least squares support vector machine.2011 Asia-Pacific Power and Energy Engineering Conference.Wuhan, China.IEEE, 1-4 [百度学术]

      55. [55]

        Tinnemeyer, Jörn.(2006) Fuzzy logic method and apparatus for battery state of health determination.U.S.Patent No.7,072,871.4 July 2006 [百度学术]

      56. [56]

        Malkhandi S (2006) Fuzzy logic-based learning system and estimation of state-of-charge of lead-acid battery.Engineering Applications of Artificial Intelligence, 19(5): 479-485 [百度学术]

      57. [57]

        Li I H, Wang W Y, Su S F et al (2007) A merged fuzzy neural network and its applications in battery state-of-charge estimation.IEEE Transactions on Energy Conversion, 22(3): 697-708 [百度学术]

      58. [58]

        Singh P, Fennie C Jr, Reisner D Jr (2004) Fuzzy logic modelling of state-of-charge and available capacity of nickel/metal hydride batteries.Journal of Power Sources, 136(2): 322-333 [百度学术]

      59. [59]

        Cai C H, Du D, Liu Z Y (2003) Battery state-of-charge (SOC) estimation using adaptive neuro-fuzzy inference system (ANFIS).The 12th IEEE International Conference on Fuzzy Systems, 2003.FUZZ '03.St.Louis, MO, USA.IEEE, 1068-1073 [百度学术]

      60. [60]

        Shen W X, Chan C C, Lo E W C et al (2002) Adaptive neurofuzzy modeling of battery residual capacity for electric vehicles.IEEE Transactions on Industrial Electronics, 49(3): 677-684 [百度学术]

      61. [61]

        Plett G L (2004) Extended Kalman filtering for battery management systems of LiPB-based HEV battery packs: Part 3.State and parameter estimation.Journal of Power Sources, 134(2): 277-292 [百度学术]

      62. [62]

        Lu L G, Han X B, Li J Q et al (2013) A review on the key issues for lithium-ion battery management in electric vehicles.Journal of Power Sources, 226: 272-288 [百度学术]

      63. [63]

        Hussein, Ala Al-Haj, and Issa Batarseh (2011) An overview of generic battery models.IEEE Power and Energy Society General Meeting.IEEE.Detroit, MI, USA, 24-28 July 2011 [百度学术]

      64. [64]

        Ali M U, Zafar A, Nengroo S H et al (2019) Towards a smarter battery management system for electric vehicle applications: A critical review of lithium-ion battery state of charge estimation.Energies, 12(3): 446 [百度学术]

      65. [65]

        Shwetha Katari J, R Sneha, Vinayaka K U, Kuldeep Rana, (2020) A Concise review of different standards for performance testing of Lithium-ion Batteries for Electric Vehicle applications, IEEE International Conference on Power Systems Technology (POWERCON), Bangalore, 14-16 Sept, 2020 [百度学术]

      Fund Information

      financially supported by Department of Science and Technology, New Delhi (Indo-Norway consortium) project entitled “Integrated Renewable Resources and Storage Operation and Management” program;


      • Samarendra Pratap Singh

        Samarendra Pratap Singh is Assistant Professor in the Department of Electrical Engineering at Dr Rammnohar Lohiya Avadh University, Ayodhya, India.His research interest includes power system, smart grid, electric vehicles and renewable energy sources.

      • Praveen Prakash Singh

        Praveen Prakash Singh obtained his BTech in electrical degree from GLA University Mathura (India) and MTech degrees from Aalborg University, Denmark.He was a Research Associate in the department of Electrical Engineering, IIT Kanpur, India.Presently he is working for PhD degree in energy systems, Tallinn University of Technology (TalTech), Estonia.His research interest includes power systems, smart grid, electric vehicles, electricity market and renewable energy sources.

      • Sri Niwas Singh

        Sri Niwas Singh received M.Tech and Ph.D.from Indian Institute of Technology Kanpur, India in 1989 and 1995 respectively.Presently, he is Professor in the Department of Electrical Engineering, Indian Institute of Technology Kanpur.His research interests include power system restructuring, FACTS, power system optimization & control, security analysis, power system planning etc.He is a fellow of IEEE, IET, INAE, IE(I) and IETE (India).

      • Prabhakar Tiwari

        Prabhakar Tiwari is an Associate Professor in the Department of Electrical Engineering at Madan Mohan Malaviya University of Technology, Gorakhpur, India.His research interest includes power systems, smart grid, electric vehicles and renewable energy sources.

      Publish Info




      Reference: Samarendra Pratap Singh,Praveen Prakash Singh,Sri Niwas Singh,et al.(2021) State of charge and health estimation of batteries for electric vehicles applications: key issues and challenges.Global Energy Interconnection,4(2):145-157.

      (Editor Dawei Wang)
      Share to WeChat friends or circle of friends

      Use the WeChat “Scan” function to share this article with
      your WeChat friends or circle of friends