Abstract

Battery technology has been a hot spot for many researchers lately. Electrochemical researchers have been focusing on the synthesis and design of battery materials; researchers in the field of electronics have been studying the simulation and design of battery management system (BMS), whereas mechanical engineers have been dealing with structural safety and thermal management strategies for batteries. However, overcoming battery limitation in only one or two domains will not design an efficient battery pack as it requires an integrated framework. So far, there are few research studies that circumscribed all the multidisciplinary aspects (cell material selection, cell-electrode design, cell clustering, state of health (SOH) estimation, thermal management, cell monitoring, and recycling) simultaneously for battery packs in electric vehicles (EVs). This article presents a holistic engineering design and simulation strategy for a future advanced battery pack and its parts by assimilating paradigmatic solutions for cell material selection, component design, cell clustering, thermal management, battery monitoring, and recycling aspects of the battery and its components. The developed framework has been proposed based on density functional theory (DFT)-based cell material selection, topology design-based cell-electrode design, machine learning (ML)-based SOH estimation along with multidisciplinary design optimization-based liquid cooling system. The proposed framework also highlights the optimal configuration of cells using ML algorithms and multi-objective optimization of cell-assembly parameters. The role of digital twins for real-time and faster acquisition of data has been highlighted for the advanced and futuristic battery pack designs. Furthermore, a preliminary investigation of robot-assisted disassembly and recycling of battery packs has been summarized. Each proposed methodology has been discussed in detail along with advantages and limitations. Critical research orientations are also discussed in the end.

1 Introduction

In recent years, electric vehicles (EVs) and the associated battery technology have gained much prominence. This advancement in the field of the automotive industry is to satisfy the stringent emission standards and to improve vehicle efficiency. With the advent of higher energy density batteries, EVs are undoubtedly the possible successor of the conventional internal combustion engine-operated vehicles. However, one of the most crucial aspects of EV design is the on-board energy storage system. Depending on the degree of hybridization of the vehicle, the energy can be stored in batteries, fuel cells, ultra-capacitors, or ultra-flywheels. Among the various available energy storage systems, the battery is the most popular choice due to its ease of application and availability, compact and robust construction, safety considerations, and ability to power a diverse range of applications [1]. However, the EV battery undergoes degradation with time and hence is one of the major setbacks in the full-scale commercialization of EVs. The performance maintenance of the battery and its dependability for a specified period without failure is the current research focus in battery advancement technologies.

The life and performance of a battery depend on several internal and external factors. The choice of material design at the cell level, the cell-electrode design of the module, cell-electrode assembly, and optimization of its associated parameters may improve its performance and life substantially [2].

Lithium-ion batteries (LIBs) are the most commonly used batteries owing to their lighter weight, excellent electrochemical properties, and high energy density. The major limitation in using lithium-ion batteries lies in the fact that they always need to be operated within the safe operating area (SOA) as they are highly sensitive toward under charge and over charge [3]. However, the research work has been initiated toward sodium ion, zinc–air, or graphene-based modification of the battery. Not only the cell-electrode material but also the electrode design has an impact on the life and performance of the cell. It has been observed that the cell design affects fast charging and its capability [4]. It affects the speed and the quantity of lithium ion accumulated, which directly affects the energy density, power density, and capacity of the cell.

The state of health (SOH) of a battery is the measure to store and deliver electrical energy with respect to its ideal conditions, i.e., the time at which the battery was newly manufactured. The research on SOH estimation focused on model-based methods, direct or experimental methods, adaptive filter approaches, and data-driven techniques [5]. The data-model hybrid approach is also a hot orientation of SOH estimation.

Clustering of cells in a battery pack is also an important aspect in battery designing. The cells used in series and parallel configuration in a module show variation in performance owing to some manufacturing defects. This variation in cell characteristics leads to incomplete charging–discharging profile and nonuniform distribution of temperature, which directly affects the battery capacity and its life [6]. To solve this problem, the cells with similar performance should be clustered together to develop a module having superior electrochemical performance.

While taking care of the nonuniform distribution of heat, the thermal management of the lithium-ion battery uses air-cooled, fluid-cooled, or phase-changing material-based cooling system [7]. The air-cooled method is most widely used because of its simplicity but it is not very efficient. The fluid-cooled method is more efficient and faster. However, there is a requirement of auxiliary components like cooling fans and fluid/liquid cooling needs a pump [8]. Phase-changing material has superior heat dissipation, but its inherent thermal conductivity is low.

Not only the cell-designing aspect but also operation monitoring of the battery pack is important for good performance and long life. This is accomplished by the battery management system (BMS), which is an electronic controller that manages a rechargeable battery. It measures voltage, current, and temperature in real time and necessary cell protection by estimating state of charge (SOC) and SOH. However, the contributions of external parameters such as vibrations, temperature, and unanticipated usage profile are not taken into account, which give rise to uncertainties in the real-time prediction of parameters [9]. These problems are not addressed by the existing BMS.

With the increasing demand for EVs, battery manufacturing has shown tremendous growth over the recent years. When the capacity of the battery degrades to 80%, it is rendered unfit to use in EVs. It is expected that nearly 750,000 tons of batteries need to be disposed of by the year 2025 [10]. Therefore, the effective and safe recycling of discarded batteries will be a mammoth challenge in the coming years.

So far, the aforementioned studies have been carried out individually on a single outlook. However, few researches have been carried out on the multidisciplinary aspects of the battery like cell material selection, thermal design, SOH estimation, and so on. This article aims at proposing a comprehensive framework for a next-generation advanced battery module design software that may combine interfaces of different software architecture and design an entire battery module starting from the selection of battery electrode module, electrode design, SOH estimation, cell clustering, thermal design, BMS design, and recycling aspects of the battery.

The remainder of the article is as follows: Section 2 presents the different research aspects in designing an efficient and sustainable battery pack. Eight research problems have been undertaken. Section 3 highlights the problem statement and the innovation contribution of the article. The proposed methodology and results have been discussed in Sec. 4. Section 5 presents the implications and future research directions. Section 6 presents the conclusion.

2 Areas of Research

The research areas have been highlighted for an efficient design of battery pack by integrating the quintessential choice of material at cell level, the electrode design, SOH estimation, cell-assembly, thermal design, battery monitoring, and battery recycling. The proposed methodology covers the battery production, battery assembly, battery operation, and battery recycling phase. The choice of material for cell electrode and the design means is important for battery capacity. Also, it is important to consider the cell consistencies during the battery assembly. While in operation, the SOH of the battery needs to be estimated to prevent its thermal runaway and improve battery life. In addition, it is essential to carefully dismantle the battery pack, estimate remaining the life, and reuse the cell by using advanced recycling technologies. The research problem is divided into eight parts as follows.

2.1 Role of Cell Material Selection for Battery.

The cell material selection for the battery electrode is relevant for improving its life, performance, and stability. The optimal material selection may even slow down the process of aging. The most commonly used materials for making cathodes are lithium-cobalt oxide, lithium-ion phosphate, V2O5, Fe S2, and other electronic polymers [11]. The most typically used materials for anodes are tin-based alloys, lithium-titanate, graphite, hard carbon, and silicon-based materials. The materials like LiAsF6, LiPF6, and so on are used as electrolytes. There are materials apart from these components that are used as flame retardant, gel, binder, solvent, and precursor [12]. Despite having an appreciable energy density of LIBs, the cell material selection has to be considered for improving cycle life and power density. Penick et al. reviewed the traditional methods of selecting the cellular materials for designing and divided them into three design levels of tessellation, type of element, and connectivity [13]. The biomemetic approach was used for decision-making at all the three levels. However, the work was confined to the analysis of the structural domain, but the proposed methodology was not applied for thermal and optical analyses. Bhate provided the framework about cell selection, spatial variation of the cell size, selection of optimal parameters, and the best possible way of integrating cellular material into large structures [14]. Different state-of-art techniques have been discussed for the same. However, the failure of the material due to extreme loading conditions and in the multi-physics environment was not taken into account. Hence, it is inferred from the aforementioned studies that the development of newer electrode material and advancement in electrode fabrication technology may have the potential to have more promising LIBs for EV applications. The characteristics of some of the commercial battery electrodes are presented in Table 1.

Table 1

Commercially used battery electrodes with advantages and disadvantages

ElectrodeApplicationSpecific capacity (mAh/g)AdvantagesDisadvantages
LiCoO2Anode140
  • Superior performance

  • Costly

  • Resource limitation of Co

  • Low capacity

LiNi0.8Co0.15Al0.05O2Anode180–200
  • High capacity

  • High voltage

  • Superior performance

  • Less reliable

  • Costly

  • Resource limitation of Co and Ni

Li Ni1/3Co1/3O2Anode160–170
  • High voltage rating

  • Resource abundance of Mn

  • Economical

  • Superior performance

  • Low capacity

  • Limited life cycle

LiFePo4Anode170
  • Reliable

  • Good cycle life

  • Low cost

  • Abundance of Fe

  • Nontoxic

  • Low energy density

  • Low capacity

GraphiteCathode372
  • Good cycle life

  • Abundance in nature

  • Low energy density

  • Inefficiency in solid electrolyte formation

Li4Ti5O12Cathode175
  • Material with zero strain

  • Good cycle life

  • High efficiency

  • Low capacity

  • Low energy density

  • Low voltage

ElectrodeApplicationSpecific capacity (mAh/g)AdvantagesDisadvantages
LiCoO2Anode140
  • Superior performance

  • Costly

  • Resource limitation of Co

  • Low capacity

LiNi0.8Co0.15Al0.05O2Anode180–200
  • High capacity

  • High voltage

  • Superior performance

  • Less reliable

  • Costly

  • Resource limitation of Co and Ni

Li Ni1/3Co1/3O2Anode160–170
  • High voltage rating

  • Resource abundance of Mn

  • Economical

  • Superior performance

  • Low capacity

  • Limited life cycle

LiFePo4Anode170
  • Reliable

  • Good cycle life

  • Low cost

  • Abundance of Fe

  • Nontoxic

  • Low energy density

  • Low capacity

GraphiteCathode372
  • Good cycle life

  • Abundance in nature

  • Low energy density

  • Inefficiency in solid electrolyte formation

Li4Ti5O12Cathode175
  • Material with zero strain

  • Good cycle life

  • High efficiency

  • Low capacity

  • Low energy density

  • Low voltage

2.2 Role of Cell-Electrode Design.

Electrode design in lithium-ion battery is a very important parameter to be considered as it determines the amount of lithium to be stored and also its speed of accumulation. The cell-electrode design and its fabrication technique have significant impact on the cell electrochemical and safety performance [15,16]. Mei et al. experimentally verified that with the increase in thickness of the electrode material and the associated the active material volume, there was a consequent increase in energy density, heat generation, and polarization of the cell [17]. The degradation in power density was improved by performing the multi-objective optimization of energy density and power density [17]. However, the effect of external parameters like temperature was not considered in the estimation. Ye et al. performed the structural optimization of three-dimensional porous electrodes for high-performance Li-ion batteries [18]. By optimizing the pore size and the thickness of active material, it has been observed that there is a significant improvement in the power performance of the electrode. In addition, the voltage drop across the electrode–electrolyte interface has been minimized [18]. All the experiments were performed at ambient temperature, and the effect of temperature rise was not considered on the energy density.

The electrode design parameters such as thickness, volume of active material, and particle size influence the energy density, power density, and thermal characteristics. Generally, it is seen that the utilization of novel materials having high specific capacity, and thicker electrode design could enhance the high energy density [19]. However, there is a tradeoff in cell design between power and energy requirements. Therefore, multi-objective and multiparameter optimization should be performed for better thermal and electrochemical performance.

2.3 Role of Optimal Cell-Electrode Assembly and Its Parameters.

Not just the design, the cell assembly also affects the performance and life of the cell. Reutar et al. analyzed the balanced cell having amorphous silicon thin film anode and NCM811 cathode by varying the area capacity ratio of cathode to anode [20]. It has been shown experimentally that the N/P ratio affected the electrochemical performance of the cell [20]. However, this result was estimated for 50 cycles and may change for the higher number of galvanostatic cycles that were not discussed in the article. Lopez-Chavéz and Cuentas-Gallegos demonstrated the effect of binder in electrode material for improvement in capacitance. The technique of cyclic voltammetry was used to investigate the two-electrode cell assembly [21]. The electrochemical properties of the cell showed stable behavior up to 7000 charging–discharging cycles. However, due to the slurry nature of the binder material, slight degradation in the cell performance was observed.

2.4 Role of State of Health Estimation.

The accurate estimation of SOH of a battery is essential for the effective health management of a battery. It is characterized by the remaining available capacity to the time it was new and is estimated by the relative change in the cell internal resistance. The SOH of a battery can be estimated by direct methods (coulomb counting, open circuit voltage (OCV), and electrochemical impedance spectroscopy), model-based techniques (electrochemical method and equivalent circuit model), adaptive filter approach (Kalman filter, particle filter, unscented Kalman filter, and extended Kalman filter), and data-driven methods (Fuzzy logic, state vector machine algorithm, and artificial neural network) as presented in Table 2.

Table 2

Effective methods of SOH estimation with advantages and disadvantages

MethodAdvantagesDisadvantages
Equivalent circuit models [22]
  • Simple

  • Good dynamic response

  • Easy to implement on real-time systems

  • Error in parameter identification increases continuously

  • Prediction accuracy is not very high

Electrochemical model [23]
  • High precision

  • Internal parameters such as aging is taken into consideration

  • Parameter identification is difficult

  • No control on internal decay mechanism

Mathematical model [24]
  • Simple modeling

  • Wide range of application

  • Sensitive to external factors like temperature, pressure, and so on

Data-driven models [25]
  • Simple

  • High accuracy

  • No requirement of ageing analysis

  • Large computational data set is required

  • Update efficiency is low

  • Requires hardware and storage technology

Hybrid method [26]
  • Good prediction accuracy

  • Complex computation

  • Depends on experimental data

Incremental current analysis method [27]
  • Reaction mechanism of battery can be analyzed

  • Need to combine with other methods for good accuracy

MethodAdvantagesDisadvantages
Equivalent circuit models [22]
  • Simple

  • Good dynamic response

  • Easy to implement on real-time systems

  • Error in parameter identification increases continuously

  • Prediction accuracy is not very high

Electrochemical model [23]
  • High precision

  • Internal parameters such as aging is taken into consideration

  • Parameter identification is difficult

  • No control on internal decay mechanism

Mathematical model [24]
  • Simple modeling

  • Wide range of application

  • Sensitive to external factors like temperature, pressure, and so on

Data-driven models [25]
  • Simple

  • High accuracy

  • No requirement of ageing analysis

  • Large computational data set is required

  • Update efficiency is low

  • Requires hardware and storage technology

Hybrid method [26]
  • Good prediction accuracy

  • Complex computation

  • Depends on experimental data

Incremental current analysis method [27]
  • Reaction mechanism of battery can be analyzed

  • Need to combine with other methods for good accuracy

Harting et al. performed the SOH identification of LIB based on the nonlinear frequency response analysis [28]. A support vector regression-based degradation model was derived from nonlinear frequency response data set. The testing of the cell was done at 25 °C, and a correlation measure was used for frequency identification [28]. Zhou et al. performed the SOH prognostics for LIBs by Gaussian regression with neural network [29]. It has been experimentally verified that the prediction accuracy was improved, and uncertainty was lowered. However, the results were uncertain for higher driving cycles. Tessier et al. performed the on-board SOH estimation by measuring the internal resistance of the battery cells [30]. The equivalent model has been developed considering the degradation influencing parameters. However, the effect of temperature is not compensated, while estimating the degradation of the cell. Saha et al. proposed an incremental voltage difference-based technique for the online state of health estimation of LIBs [31]. The proposed methodology uses partial charging data to prepare a training data set by extrapolation of charging voltage readings. The results obtained are fairly accurate but are tested for a lesser number of driving cycles.

All the earlier discussed methods use voltage, current, and temperature for the SOH estimation. It is clearly evident that the model-based techniques are accurate but are not easily applicable to batteries. Although statistical methods are not very accurate, they can be easily applied to batteries. So the real-time estimation of SOH in a LIB is a challenging task due to the complex battery chemistry. For real-time monitoring of the LIBs, a new and advanced methodology needs to be devised keeping in view the accurate prognostics, suitability in real-life situations, and its cost-effective implementation in BMS.

2.5 Role of Optimal Configuration of Cells for Battery Module.

An enormous amount of power is required to operate an EV. So a large number of cells need to be clustered to form a module, and the cluster of modules form a pack. The LIBs are made up of series parallel connected cell structure. The parallel structure increases the battery current, whereas the series configuration increases the overall voltage of the battery pack [32]. However, the difference in the dynamic properties of the cells creates problems in battery usage cycles. As a result, when the weakest cell is exhausted, the battery pack would stop discharging and the other cells remain underutilized. Similar condition happens when the weakest cell is not fully charged [32]. In the worst scenario, the overcharging and undercharging of the cell may cause the thermal runaway, resulting in the degraded usage cycle and life of the battery. To avoid such a situation, the homogeneous battery cells can be sorted out to make a uniform battery pack.

Clustering of homogeneous cells brings the high cell uniformity and extended life span of the battery. Traditionally, the charging and discharging characteristics of the cells are considered for battery clustering [33]. Sun et al. performed the model-based dynamic multipower calculation to estimate the available current and terminal voltage of the cell accurately [34]. Raspa et al. considered total capacity, open-circuit voltage, and equivalent circuit model parameters to estimate the vector distance [35]. They obtained the clustering results by using self-organizing neural network maps. Li et al. considered internal resistance, capacity, terminal voltage, thickness, and discharging time of cells for clustering [36]. Mauger et al. proposed that the dynamic characteristics of the cell such as charging–discharging profiles were the better ways to cluster cells [37]. Wang et al. proposed a fuzzy logic-based cell classification system based on automatic characteristic recognition [38]. Wang et al. considered correlation coefficient and Euclidean distance methodology to differentiate between cell characteristics [39]. However, the aforementioned research methodologies are based on cell charge/discharge characteristic, which is a time-consuming process. However, the accuracy of the earlier discussed methodologies is not reliable. Some efficient, faster, and reliable cell clustering mechanism needs to be developed, which would circumvent any cell discrepancy to produce a more uniform battery pack.

2.6 Role of Thermal Management of a Lithium-Ion Battery.

In the design of LIBs, the thermal management system is an integral part because it ensures that the battery always operates within its SOA. An optimally designed thermal management system ensures prolonged life, improved performance, and higher battery capacity [40]. Poor thermal management not only deteriorates the cycle life of the battery but also influences charging–discharging rates. The condition of thermal runaway may exist in severe conditions, rendering battery permanently damaged. Prominent battery thermal management system uses air-cooled battery thermal management system (BTMS), fluid-cooled BTMS, and phase-change material-based BTMS [41]. Hybrid systems can also be generated by integrating any of the earlier discussed methods.

Chen and coworkers experimentally analyzed the thermal behavior of LIBs using phase change material (PCM). Twelve LiFePO4 batteries were subjected to the temperature of 54 °C and 12 °C by charging–discharging cycle at 2C and by natural convection [42]. Cooling with heat dissipation fins has not showed very promising results. The cooling was appreciable in the case of PCM as the temperature decreased about 2 °C. However, it has been inferred from the study that PCMs are having lesser latent heat and relatively lower heat conduction performance [42]. Bhattacharjee et al. designed an immersion-based liquid cooling system under different discharging conditions to ensure maximum heat dissipation [43]. It has been observed experimentally that the maximum temperature has been reduced from 49.76 °C to 28 °C at 2C discharging rate after using the proposed design. The experimentation has been carried out at an ambient temperature at the outlet. The initial conditions were assumed at room temperature, which is practically not the case of LIBs. Wang et al. performed the thermal investigation of LIBs with different PCM schemes [44]. They have verified experimentally that the composite materials provide a promising solution for temperature control and maintain the uniform temperature throughout the module. Kiani et al. presented an integrated numerical experimental approach for hybrid BTMS using PCM, nanofluid, and metal foam [45]. Han et al. compared the liquid versus air cooling mechanisms in LIB thermal management in terms of heat transfer parameters and basic heat flow [46]. Yi et al. performed a multidisciplinary optimization on the air cooling BTMS incorporating the parameter uncertainties [47]. All the earlier discussed methods have their own pros and cons and so an integration of two or more thermal management techniques. A multidisciplinary design optimization approach could be implemented for efficient and cost-effective cooling of LIBs. Moreover, the earlier discussed researches mainly focused on one aspect or discipline of the BTMS, but in practice, the thermal behavior of LIBs depends on different disciplines or components involved. Therefore, a multidisciplinary design optimization framework needs to be developed that significantly reduces the battery temperature and the material volume.

2.7 Role of Battery Management System/Battery Controller Design.

In LIBs, there are no precise design requirements that are universally accepted by all stakeholders, including manufacturers, developers, users, and integrators. The BMS designs are generic, overly simplistic, or overly complex. The inconsistent BMS design may often produce missing or conflicting battery requirements and systems [48]. The increased cost is also a matter of concern for efficient design of BMS as designing, validation, and testing are required for both performance and safety at cell level and module level.

Gao et al. proposed a novel design and implemented it on a LIB system with real-time fault diagnosis [49]. The SOC of a battery is defined as the percentage of releasable capacity to the rated capacity and is given by Eq. (1).
SOC=CreleasableCrated×100%
(1)
The maximum capacity, which differs from the releasable capacity, was used in SOH estimation. Hence, SOH of the battery is given by Eq. (2).
SOH=CmaxCrated×100%
(2)
At the time of discharging, the depth of discharging of a battery is defined as the percentage of discharge capacity, relative to rated capacity as shown in Eq. (3).
DOD=CratedCreleasableCrated×100%
(3)
The difference in the DOD by time period T with the measured current I is given by Eq. (4).
ΔDOD=tt+TIdtCrated×100%
(4)
SOC in terms of SOH is given in Eq. (4)
DOD(t)=1SOC(t)
(5)
where Creleasable is the releasable capacity, Crated is the rated capacity, Cmax is the maximum capacity, Crated is the rated capacity, t is the time at which the charging starts, T is the time period of the charging cycle, SOH is the state of health, and DOD is the depth of discharge of the battery.

For improved convergence performance of SOC, the authors implemented a self-initialization method to provide prior configuration for over-voltage, over-temperature, and open-wire fault diagnosis [49]. However, the authors did not consider the transient conditions occurring due to regenerative braking and fast charging on the BMS.

Nizam et al. developed a BMS using Arduino uno microcontroller. The BMS was able to monitor the voltage with an accuracy of 99% and precision of 99.7% [50]. The protection feature includes the over-voltage and excessive temperature control; however, the BMS was unable to predict the SOH of the battery in the proposed design. Cabrera et al. proposed a design of a reconfigurable BMS [51]. The proposed design consisted of three printed circuit boards (PCBs): the BMS, control area network (CAN) bus, and the burnt resistors. The research work focused on the thermal management, nullifying the capacitive and inductive effects, and preventing electromagnetic disturbances while designing the BMS [51]. However, the computational capacity of BMS, real-time estimation of SOH and SOC, and fault prediction other than over-temperature were not taken into account.

All the aforementioned BMS design studies focused mainly on one or two aspects of BMS like over-current protection, thermal management, and so on, but most of the research work fails in real-time estimation of SOC and SOH. In addition, the computational data required for efficient BMS become cumbersome to process and may result in unrealistic cost. So an optimal approach toward BMS design needs to be implemented, covering the overall domain of cell balancing, over-voltage and over-current protection, thermal management, and accurate SOC and SOH estimation.

2.8 Role of Battery Recycling.

The market for EVs has been growing rapidly to meet the global targets of reducing greenhouse gas emission. However, rapidly increasing EVs and hence their batteries pose a serious waste management challenge for the recyclers toward the end of life of batteries. It has been predicted that by the year 2030, the used LIBs will hit 2 million per year [52]. The battery pack contains lithium, nickel, and cobalt, which can be processed, recovered, and reused, but only 2–3% of LIBs are recycled at present [53]. However, the life cycle analysis of batteries, technical constraints, logistic issues, and economic gaps in battery recycling need to be focused.

Gains focused on material separation technologies in battery recycling for her research work [54]. The study mainly focused on the extraction of lithium, nickel, and cobalt by pyro-metallurgy or smelting, hydrometallurgy or leaching, and direct recycling by physical processes [54]. The various operating conditions and the involved chemical reactions were discussed in detail. However, the process design and estimation of reminiscent energy in battery cells have not been focused. Mossali et al. conducted a review of opportunities and issues of recycling treatments for LIBs. The research focused on the ongoing technological solutions and existing industrial processes [55]. A typical LIB recycling process is shown in Fig. 1.

Fig. 1
A typical LIBs recycling process after battery post-use
Fig. 1
A typical LIBs recycling process after battery post-use
Close modal

The battery recycling is a complex and significantly challenging by the manual process. The earlier discussed researches failed to provide an intelligent holistic framework for battery recycling process, ensuring safe, and efficient battery recycling system. Therefore, an advanced recycling framework requires to be proposed over traditional methods in a cost-effective manner so as to limit the battery waste.

3 Problem Statement and Innovation Contribution

The problem statement considered in this study is the design of the holistic and efficient battery pack, which considers the optimal scenario, right from the cell material selection to the last stage of battery recycling. All the research aspects, as stated in section 2, are closely related to one another, as an improvement in any one or two of them will not result in an efficient battery pack. In futuristic design methodology, the cell material needs to be selected at the beginning of the production stage to obtain high capacity. Along with the cell material selection, the design of the electrodes and their assembly based on binder thickness and N/P ratio need to be considered. This is specifically important to reduce any inconsistency in performance due to the difference in cell capacity.

Once the cell is successfully manufactured, in-operando estimation of SOH becomes very necessary. The real-time estimation of SOH is important to predict the battery degradation and remaining useful life by analyzing the cell performance, which is also affected by its optimal configuration in series and parallel to form a battery module. After the cell module assembly, its thermal management becomes the matter of concern as the series parallel configured cells are of inherently different capacity, resulting in inconsistent performance and temperature difference. So an efficient cooling mechanism, which is easier to implement and has a better cooling effect, needs to be adopted.

The BMS monitors the state of function and the temperature in real time, so faster acquisition of data is required in real-time operations. After the battery is rendered unfit for use, it is important to recycle the battery economically, ensuring safety to operating personnels and causing less environmental damage. Considering all the aforementioned aspects, a comprehensive framework of the advanced battery pack has been proposed, covering the holistic manufacturing, operating, and recycling environment. Also, the future critical research directions are discussed toward the end of the literature. An optimal engineering design for a battery module in EVs based on density functional theory (DFT), topology optimization, ML, multi-objective optimization, and digital twins has been proposed in detail in section 4 based on the aforementioned research aspects.

4 Proposed Framework

4.1 Material Design at Cell Level Using Density Functional Theory Method.

With the development of modern materials, there is a growing need to understand the physical properties and processes taking place at atomic level during cell material selection. The laws of quantum mechanics govern the interactions between the electrons and atoms, hence efficient and accurate methodologies for solving the basic quantum mechanics equations for a complex large electrons/large atom are required to be developed. The DFT is a function of three spatial coordinates and is an efficient computational theory based on electron density [56]. The application of DFT at cell material selection level involves three steps: (1) adaption of the engineering problem to a computative atomistic model, (2) computation or study of physical and chemical properties, and (3) affirmation of the computational or simulation results by comparing them with laboratory experiments/testing data [57]. The many-body Schrödinger equation is used to determine physical and chemical properties of a system given by Eq. (6)
HΨi(r,R)=EΨi(r,R)
(6)
where Ψi is the wavefunction of the system, Ei is the Eigen value or allowed energy states, and H is the Hamiltonian operator that forms the pre-basis of DFT. The energy of the system in terms of electron density is given in Eq. (7).
ETF[ρ(r)]=310(3Π2)2/3ρ5/3(r)drZρ(r)rdr+12ρ(r1)ρ(r2)r12dr1dr2
(7)
where ρ is the electron density, the first term represents the electron kinetic energy, the second term represents attractive nuclear force between nuclei and electrons, and the third term represents coulomb repulsive forces between electrons [58]. It has been observed that DFT presents an efficient and systematic approach toward cell-electrode material design. The wavelength calculation over energy density makes it more preferable and economical over classical design methods. The ab initio methods like local density approximate (LDA) and general gradient approximate (GGA) functions do not consider the excitation energy. Hybrid functionals like time dependent density functional theory (TD-DFT) gives fairly accurate results for small electrode design but are not suitable for larger electrodes. DFT is a promising approach for material screening and also for efficient mixed metal identification. The hybrid methods/functionals also provide accurate results and reduce the discrepancy between experimental and theorical calculations. The exchange correlation energy of functional and the electron correlation of the system under study are major parameters to be considered for the cell material design.

Nashed et al. presented the use of DFT to design solar energy-based renewable systems. The authors designed sensitizers for dye-sensitized solar cells using DFT and compared the simulation results with LDA and GGA results [59]. The recent advancements in battery technology require efficient battery storage elements. DFT can be an efficient tool to understand the electrochemical working mechanisms of the cell material before its industrial applications. DFT can be used in the cell material selection for the following estimations [60]:

  • Estimation of structural stability: The structural stability of a cell material is a key decisive factor for its cyclic performance. The stability of the material can be determined by the calculation of Gibb’s free energy, formation energy, cohesive energy, and dispersion spectrum of phonon.

  • Theoretical capacity prediction and estimation of reaction voltage: The power density of the battery is considered as one of the most important communication index and is proportional to the equilibrium voltage. DFT methodology can be used to study the electrochemical reaction occurring during the cycling process of a battery and also to predict the voltage of the new battery.

  • Electronic structure of an electrode/electrolyte: The electronic structure of an electrode or electrolyte plays an important role in battery performance. The DFT calculations are based on electron density rather than the wavelength, so the electronic structure of the cell material can be analyzed using DFT calculations. Atomic properties like band structures, state density, molecular orbitals, and distribution of charge can be used to analyze the electronic structure of a cell electrode or electrolyte.

  • Transport kinetics of ions/molecules: The ionic diffusion plays a very important role in the performance of electrode materials that affect the stability, cyclic performance, capability, and other electrochemical properties of the battery. By estimating the activation energy along the ion diffusion paths, the transport mechanism ions can be simulated by employing DFT in batteries.

  • Adsorption kinetics of electrode: The cyclic performance and battery capacity are greatly influenced by the reactivity of electrodes, which is dependent on the adsorption of ions. The storage of Li ions is partially contributed by the adsorption of Li ions of the electrode surface; hence, adsorption kinetics must be taken into account when considering a new electrode material for the battery.

Although DFT is a powerful computational tool for cell material selection, there are certain inconsistencies associated with the accuracy of employed density functions and should not be employed without experimental validation. Density functions like GGA, LDA, GGA + U, and hybrid functions are suitable for different applications [61]. LDA and GGA functions, although effective, may produce nonnegligible errors during electron transfer and may underestimate the lithium transfer voltage by the fraction of cells. The semi-local DFTs may show some deficiencies while dealing with the correlation effects of electrons. The earlier discussed cell material selection criteria based on DFT are summarized in Fig. 2.

Fig. 2
Parameters for material design at cell level using DFT
Fig. 2
Parameters for material design at cell level using DFT
Close modal

4.2 Topology Design Method for Cell-Electrode Design.

The intrinsic structure of an electrode material is also one of the major factors that affect the battery chemistry and its performance, along with the choice of the cell material. The property and structural optimization of electrode materials for the battery is required for a highly efficient storage system. The structure of the components such as separators and electrodes are specified in terms of their tortuosity and porosity. The ionic resistivity, associated voltage losses, and the discharge capacity are determined by the ratio of porosity and tortuosity, which yields the transport coefficient of ions in the pore space filled by electrolyte [62].

Topology optimization is an efficient mathematical approach that considers certain sets of loading and boundary conditions based on the finite element method. The approach of battery design based on topology optimization of steady-state models consists of the following steps. Step 1 involves optimization of electrode shape for better conduction current in steady-state conditions. Step 2 involves time-dependent simulation analysis of the optimized electrode shape [63]. Topology optimization uses the mathematical model that focuses on the relation between terminal potential, current density, and material resistance. Table 3 presents the parameters for the mathematical model for topology optimization.

Table 3

Parameters for mathematical modeling for topology optimization

Modeling parameterUnit
Electrolyte potentialV
Electrode potentialV
Electronic conductivityS/m
Ionic conductivity of electrolyteS/m
Times
Number of transported positive ions
System temperatureK
Radius of spherical particlesμm
Gas constantMol K
Salt activity in electrolyteMol/m3
Current density of electrodeA/m3
Current density of electrolyteA/m3
Diffusion coefficient of electrodem2/s
Concentration of lithium ion in electrodeMol/m3
Concentration of lithium ion in electrolyteMol/m3
Modeling parameterUnit
Electrolyte potentialV
Electrode potentialV
Electronic conductivityS/m
Ionic conductivity of electrolyteS/m
Times
Number of transported positive ions
System temperatureK
Radius of spherical particlesμm
Gas constantMol K
Salt activity in electrolyteMol/m3
Current density of electrodeA/m3
Current density of electrolyteA/m3
Diffusion coefficient of electrodem2/s
Concentration of lithium ion in electrodeMol/m3
Concentration of lithium ion in electrolyteMol/m3

The topology optimization method of cell-electrode design consists of the following steps.

4.2.1 Mathematical Modeling.

The current density in the electrode and electrolyte phase is given in Eqs. (8) and (9).
Is=σΦs
(8)
Il=kΦl
(9)
The charging collector current that needs to be optimized is given as in Eq. (10)
I=B0BnσΦs×dl
(10)
where σ is the electronic conductivity, Φs is the electronic potential, Φl is the electrolytic potential, Is is the electrode current density, Il is the electrolytic current density, k is the electrolytic constant, Bo is the initial flux density, and Bn is the final flux density [64].

4.2.2 Formulating the Objective Function.

The main goal of formulating the objective function is to optimize the shape of the electrode so as to minimize the internal resistance of the battery or to maximize the conductivity of the electrode [64].

4.2.3 Constraint Formulation.

The conductivity of an electrode depends on the nature of the material used. The total amount of the electrode material area consisting of the porous material in electrolyte is given in Eq. (11) as follows:
W=ΣρA
(11)
where ρ is the electronic conductivity and A is an area of the electrode.
For co-existence of both electrode and electrolyte, there need to be an upper bound in the total amount of electrode material used and is given by Eq. (12) [65]:
W=ΣρAWU
(12)

The steps involved in the topology optimization method for cell-electrode design are shown in Fig. 3.

Fig. 3
Flowchart for the topology optimization of cell-electrode design
Fig. 3
Flowchart for the topology optimization of cell-electrode design
Close modal

4.3 Multi-Objective Optimization of Cell-Electrode Assembly and Its Parameters (Binder Thickness; N/P Ratio).

After the optimal designing of cell electrode, the effective cell-electrode assembly plays an important role in cell manufacturing as the latter ensures high volumetric energy density and economical manufacturing costs. Bigger electrode dimensions not only have benefits like higher capacity but also produce increased heat generation due to large diameter. So proper optimization of cell-electrode assembly is quite challenging for battery design and performance. The objectives of the multidisciplinary design optimization are to maximize specific energy and discharge specific power with the minimized capacity loss [66]. The design variables like binder thickness, N/P ratio, thickness of electrodes, particle size, and porosity can be optimized that significantly affects the battery performance [67]. The flowchart depicting the nondominated sorting genetic algorithm (NSGA-II) is shown in Fig. 4.

Fig. 4
Flowchart of NSGA-II optimization
Fig. 4
Flowchart of NSGA-II optimization
Close modal
The main purpose of the optimization is to minimize the capacity loss over the cycle and to improve the travel range per charge. The objective function for maximizing specific discharge energy, specific discharge power, and capacity are shown in Eqs. (13)(15), respectively:
F(x)=max{Ecell=1Mcell0dischargetimeVcell(t).Idt}
(13)
=max{Pcell=1t0.Mcell0t0Vcell(t)dt}
(14)
=max{Qcell=1McellIdt}
(15)
where Mcell is the cell mass, Vcell is the cell potential, I is the cell current, Ecell is the specific discharge energy, Pcell is the specific discharge power, t0 is the initial time, and Qcell is the capacity at the cell end.

4.4 Cell-Level State of Health Prediction by Machine Learning Algorithms.

The SOH of the battery quantifies the aging level in terms of power fade or capacity fade and is an important parameter to be estimated, while the cell is in operation in real-time conditions. When the battery capacity drops to 80% of its initial rated value, the battery is rendered unfit for use [68,69]. Similarly, when the internal resistance of the battery increases to the significantly high value resulting in power loss, the battery should be considered for replacement.

The conventional method for SOH estimation includes open-circuit voltage method, coulomb-counting method, ampere hour counting method, impedance-based methods, fuzzy-based methods, and model-based estimation methods. Among all these methods, model-based techniques are extensively used for their ability to be used for online applications and low computational demand. However, the accuracy of model-based methods is limited by the extent to which the model has been parameterized. The physics-based model, which is the advanced model-based estimation technique, provides insights to the internal dynamics at the cell level. However, these models are not very useful for online applications owing to complicated governing equations and high computational cost [70]. Moreover, the aforementioned methods do not take into account the material characteristics, which greatly influence the degradation behavior related to SOH and remaining useful life.

Machine learning (ML) algorithms for cell-level SOH estimation accurately model over length and time, allowing the real-time SOH estimation. Material parameters and domain knowledge can be incorporated to an ML model whose fidelity largely depends on the quality and size of the dataset. Extensive experimentation and computation in controlled conditions can produce a high volume of precise data. DFT-based multiscale modeling can be used to understand the molecular dynamics and the associated degradation mechanisms taking place in the battery [71].

To design, understand, and predict the complete battery behavior, the variables that fully capture the battery behavior must be incorporated in the battery modeling. To simplify the model behavior, some parameters are either assumed constant or ignored. The possible input variables for an ML-based model at cell level-based SOH estimation can be classified as follows:

  • Continuous variable: Electrode geometry, battery internal structure, cell temperature, and current flow

  • Integer variables: Number of charge/discharge cycles

  • Categorial parameters: Type of battery such as lead–acid, Li ion or Nickel–cadmium

The ML-based model approach is used for predicting the change in battery characteristics over charge/discharge cycles. The current, SOC ion concentration, and electrode defects are tracked, which address the remaining useful life of the battery [72]. Figure 5 depicts the typical ML approach for predicting SOH and remaining useful life (RUL) of the LiBs. Table 4 presents the summary of recent work on ML algorithms for SOH estimation in LIBs.

Fig. 5
A typical ML approach for SOH estimation of LIBs in EVs
Fig. 5
A typical ML approach for SOH estimation of LIBs in EVs
Close modal
Table 4

Summary of recent work on ML algorithms for SOH estimation in Li-ion batteries

MethodFeature set inputsProsCons
Neural network [73]Voltage, current, cycle number, capacity
  • Provides fairly accurate results for nonlinear battery behavior

  • Increased complexity with more number of battery parameters while estimating SOH

Support vector machine [74]Voltage, temperature, cycle number, capacity
  • Effective when data is scare

  • Considerable error in SOH estimation while considering environmental and loading conditions

Gaussian/Bayesian [75]Current, electrode geometry
  • Fairly accurate

  • High computational cost

Regression [76]Current, voltage, cycle number, geometry
  • Simple

  • Robust fit

  • Faster computational time

  • Since the battery characteristic is nonlinear, quadratic and higher order terms need to be included in the fit by Taylor series analogy

Random forest/tree [77]Current, voltage, temperature, power
  • Accurate and easy to train dataset

  • Robust against outliners

  • Improve the quality of the fit to the large extent, even if the dataset is scarce

  • Function obtained is discrete rather than smooth

  • Costly computational demands

Kalman filter [78]Current, voltage, temperature
  • Good estimation accuracy

  • High convergence rate

  • More computational time is required

  • Does not take into account of the battery nonlinearity

MethodFeature set inputsProsCons
Neural network [73]Voltage, current, cycle number, capacity
  • Provides fairly accurate results for nonlinear battery behavior

  • Increased complexity with more number of battery parameters while estimating SOH

Support vector machine [74]Voltage, temperature, cycle number, capacity
  • Effective when data is scare

  • Considerable error in SOH estimation while considering environmental and loading conditions

Gaussian/Bayesian [75]Current, electrode geometry
  • Fairly accurate

  • High computational cost

Regression [76]Current, voltage, cycle number, geometry
  • Simple

  • Robust fit

  • Faster computational time

  • Since the battery characteristic is nonlinear, quadratic and higher order terms need to be included in the fit by Taylor series analogy

Random forest/tree [77]Current, voltage, temperature, power
  • Accurate and easy to train dataset

  • Robust against outliners

  • Improve the quality of the fit to the large extent, even if the dataset is scarce

  • Function obtained is discrete rather than smooth

  • Costly computational demands

Kalman filter [78]Current, voltage, temperature
  • Good estimation accuracy

  • High convergence rate

  • More computational time is required

  • Does not take into account of the battery nonlinearity

The ML-based models need to be validated once the data have been fitted. ML models are susceptible to overfitting like other fitting optimizable functions. With the advancement of mathematical algorithms and computational techniques, data-driven ML techniques are quite promising in the field of battery technology for SOC/SOH estimation.

4.5 Optimal Configuration of Cells in Series and Parallel for Forming Battery Module Using Machine Learning Approaches.

A module represents cell package assembly embedded with the safety features like voltage, current, and temperature monitoring, recorded by the battery management system. Based on the voltage and capacity needs, the cells are configured in series and parallel. The battery configuration is a very important aspect for studying its behavior during deep discharges and abnormal operating conditions. The performance and longevity of the pack also depend on the optimal configuration of the cells in series and parallel. Battery packs under different configurations having variables of series and parallel connections have different capacities [79]. Battery manufacturing defects result in variation in cell performance used in series and parallel. Incomplete charging and discharging of battery and nonuniform distribution of temperature results in reduced battery capacity and cycle life [79].

ML-based approaches can be utilized to observe the reason for the difference in capacities observed from a theoretical calculation from the practical observations, so as to construct an optimal pack configuration for achieving maximum capacity. The ML algorithms can be classified into two categories: supervised learning and unsupervised learning. Classification of cells can be carried out by supervised learning, and clustering can be performed by unsupervised learning [80]. The clustering algorithms group data according to the different sets of rules and divide the data into various classes. The data that belong to the same class resemble each other, whereas the data that belong to other classes differ by a set of rules [80]. The charging–discharging tests can be performed on cell to obtain discharge voltage, discharge capacity, discharge temperature, charge voltage, charge capacity, and charge temperature. The main steps involved in k-means clustering algorithm are shown in Fig. 6.

Fig. 6
Steps involved in k-means clustering algorithm used for series parallel configuration of cells
Fig. 6
Steps involved in k-means clustering algorithm used for series parallel configuration of cells
Close modal

The state vector algorithm (SVM) is an unsupervised learning clustering algorithm that maps data space to high-dimension space feature by using the Gaussian function. It does not require any rule-set or training-set [81]. The data obtained from charge–discharge test can be used as inputs to perform the clustering analysis. The output obtained is the clustering result, which can be further verified by performing experimental validation. However, these ML algorithms are not only simple and easier to implement but also have certain limitations. The reasonable value for “k” number of cluster is difficult to choose. Furthermore, the randomness while selecting the initial cluster center can generate an error in the clustering results. These algorithms are prone to sensitivity and noise [82]. However, with appropriate training, the ML clustering algorithms can be an efficient way for optimal clustering of cells in series or parallel for maximum battery capacity.

4.6 Liquid Cooling Systems in Battery Module by Multidisciplinary Design Optimization.

The thermal management techniques used in LIBs involve the complex process that uses optimal temperature control strategies and heat dissipation methods to maintain safe temperature conditions, thus maintaining the safety and lifetime of the battery.

The efficiency of liquid cooling-based BTMS depends on the effects layout, thickness, spacing, and diameter of the cooling plate/tube [83]. The associated heat dissipation by computational fluid dynamics can be an effective method for minimizing the mass and the volume of the cooling plate, followed by the multi-objective optimization that allows the optimum temperature with the minimum mass of the material. Figure 7 shows the effects of heat dissipation factors on the thermal performance of the cooling plate.

Fig. 7
Effect of heat dissipation factors on thermal performance of cooling plate
Fig. 7
Effect of heat dissipation factors on thermal performance of cooling plate
Close modal

Since the battery thermal performance depends on various components and parameters, multidisciplinary design optimization in liquid cooling systems can be an efficient way of optimizing the BTMS. The multidisciplinary optimization of the liquid cooling system basically involves two steps: (1) to comprehend the effect of coupling between different components and subsystems and (2) to develop the system objective function and its constraints for the thermal management system and its subsystems. Figure 8 shows the hierarchical decomposition of the thermal management system used in the battery.

Fig. 8
The hierarchical decomposition of BTMS using liquid cooling-based thermal management
Fig. 8
The hierarchical decomposition of BTMS using liquid cooling-based thermal management
Close modal
The multi-objective design optimization on the cooling structure for the liquid cooling system reduces the weight and volume of cooling fins. The heat transfer design becomes the optimization problem with multipurpose features. First, to implement the optimization with discrete variables, the range of variables needs to be defined. Numerical simulations to analyze the performance of different combinations of discrete variables on heat transfer are to be performed. After result validation, the optimization-based design scheme is developed. Due to the discrete nature of the design variables used in multi-objective optimization of cooling plates, they have uncertain discrete values. The mathematical expressions for obtaining the maximum or minimum value of the objective function are expressed in Eq. (16) [84]
{F(x)=[f1(x),f2(x)fN(x)]i=1,2.Ns.t.gj(x)0,j=1,2MhK(x)=0,K=1,2Qx=[x1,x2..xD]TxLxxU
(16)
where F(x) is the multi-objective function vector, hK(x) and gj(x) are equality and inequality constraint functions, xU and xL are upper and lower bounds, respectively. In the optimization problem, the maximum temperature needs to be minimized and also the weight of the cooling plates needs to be reduced [85,86]. So the objective function becomes as shown in Eq. (17):
{F(x,y,z,ξ)=[Tmax(x,y,z,ξ),weight(x,y,z,ξ)]x1xx2y1yy2z1zz2ξ(n1,n2,n3,n4,n5,n6)
(17)
where x, y, z, and ξ are wall thickness of the tube, tube spacing, tube diameter, and tube layout of the cooling plate, respectively. nij is the jth discrete value of the ith design variable. The design parameters are series of discrete variables in the designing structure of the cooling tube, so the design variable in discrete variable has to be selected.
The constraint values can be added to the objective values by the penalty to obtain the corrected response. The characteristic function gets modified as shown in Eq. (18):
mF=w1Tmax0Tmax+w2weight0weight
(18)
where Tmax0 is highest temperature and weight0 is the maximum weight of the optimized cooling plate. w1 and w2 are the weights of the objective function. The characteristic function can be estimated by Eq. (18) for each model.

Multidisciplinary design optimization for the liquid cooling system is an efficient way of BTMS, but more optimization parameters make the problem more complex, with increased computational time and costs.

4.7 Battery Management System/Battery Controller Design: Concept of Digital Twin for Faster Acquisition of Data.

The Li-ion cell configuration requires an electronic acquisition board controller (BMS) that contains safety circuitry to prevent cell damage. It is the weakest cell in the configuration that establishes the overall capacity and voltage of the module, so it becomes important to monitor real-time SOC and SOH at the cell level. Due to the complex and nonlinear behavior of LIB, its real-world control becomes quite challenging. Recent advancements toward battery aging, degradation, modeling, and diagnostics, the digital twin can be a smart solution for real-time monitoring, fault diagnosis, energy management, data collection, and its processing. This cyber–physical system is an intelligent interaction of the physical and digital embodiment of the battery that circumscribes the use of sensors, Wi-Fi, 5G, data-driven frameworks, AI-based methods, and the essential hardware [87]. Figure 9 shows the cyber–physical elements of the battery digital twin.

Fig. 9
Cyber–physical system-based digital twin of a LIB battery
Fig. 9
Cyber–physical system-based digital twin of a LIB battery
Close modal

With the increased categorization of IoT devices and their low-cost sensing, it is possible to have a remote sensing of a LIB, combined with the cloud-based models, which monitors, senses, and creates the virtual representation of a physical battery system [88]. Individually, a battery or its digital equivalent is not able to provide statistically significant data-driven SOC or SOH estimation. However, when integrated together along with the fusion of an ML-based model, the real-time updation with the closed-loop optimizer can be performed. In short, the digital twin in battery systems provides benefits in four potential aspects, resulting in increased system performance and reliability, as shown in Fig. 10.

Fig. 10
Potential aspects in digital twin of battery management systems
Fig. 10
Potential aspects in digital twin of battery management systems
Close modal

Cyber–physical hierarchical and interactive network-based framework consists of the following interdependent hierarchical layer, which includes the following functions [88]:

  • Integration of digital models: Seamless multiscale mapping of digital models from time-consuming experimentation to achieve efficient and flexible data.

  • Linking cyber-physical systems: The important physical and electrochemical cell parameters can be uploaded on a cloud-based server during manufacturing to perform closed-loop optimization. The models can be easily assessed and quickly trained, thus providing the real-time upgradation of existing assets.

  • Battery health management and prognostics: Health prognostics can be realized by capturing states from battery modules, second life applications. and automotive use from cloud-based servers, where data are shared and uploaded for analysis

  • Multilayer collaboration or “end-edge cloud”: Data processing can occur either on-board or on cloud-based servers. Optimization of data in this network can not only provide faster data processing but also ensure the necessary data securities.

4.8 Battery Recycling.

The traditional battery recycling technologies rely mainly on manual processes, which involve both mechanical and chemical operations. The first step involves taking out the battery pack out of the case, separating into particles and then recycling through chemical processes. The mechanical process involves disassembly (battery to module/ module to battery), crushing, and material sorting. The next phase involves extracting materials like cobalt, nickel, and lithium from used batteries and uses them for further applications [89].

The existing battery recycling frameworks, which involve mechanical dismantling and chemical recovery, are mainly conducted on the manual basis. The process is not only time consuming and costly but also suffers from potential safety hazards due to the inflammable nature of the battery [55,90,91]. Therefore, it becomes necessary to establish an intelligent solution (automatic or semi-automatic) for the safe disassembly and efficient recovery of retired cells from the discarded batteries. Figure 11 highlights the major challenges in the battery recycling process.

Fig. 11
Major challenges in battery recycling process
Fig. 11
Major challenges in battery recycling process
Close modal

The proposed battery recycling framework can be classified into following processes:

  • Digital detection: Digital detection consists of robot-based monitoring that performs relevant operations through hand–eye system. Coordinate transformation from camera local to robot local, the specific local of Li-ion battery can be estimated and robotic arms are enabled by position coordinates accordingly. Figure 12 shows advanced automatic removal tools used after digital detection.

  • Robotic operation: After automated digital detection, identification and detection of the targeted object through positioning and image acquisition takes place through robot assembly. It will reduce the operating time and safety risks.

  • Robotic disassembly: The mechanical disassembly process involves cell sorting and material recovery. This can be accomplished by intelligent recognition based on vision sensors. Figure 13 shows some of the key research objectives in the battery recycling framework.

Fig. 12
Advanced automatic removal tools used during battery recycling
Fig. 12
Advanced automatic removal tools used during battery recycling
Close modal
Fig. 13
Summary of key research objectives in the battery recycling process
Fig. 13
Summary of key research objectives in the battery recycling process
Close modal

5 Implications and Future Critical Research Directions

5.1 Implications.

Previously existing research works on Li-ion batteries mainly focused on battery materials, estimation of SOC/SOH, and thermal management individually. These studies have been applied on the cell level or pack level. This research article circumscribes a comprehensive battery pack and its multidisciplinary component design that includes cell material design, cell-electrode design, optimization of the electrode assembly, SOH prediction, optimal configuration of cells, thermal design, battery controller design, and battery recycling. Figure 14 shows the advanced battery pack design based on the proposed methodology

Fig. 14
Battery pack design based on proposed methodology
Fig. 14
Battery pack design based on proposed methodology
Close modal

This article highlights four main aspects of battery design: production, assembly, operation, and its recycling to form a holistic and overall design methodology as illustrated in Fig. 15. In the cell production aspect, the DFT method has been proposed for material design at the cell level. The topology design method is considered for cell-electrode design. Multi-objective optimization of cell-electrode assembly and its parameters can result in efficient performance of the pack. ML-based algorithms (k-means clustering and SVM) can be used for the optimal configuration of cells in series and parallel for forming an effective battery module. The operation phase highlights the use of ML for SOH prediction at the cell level. The thermal design involves multidisciplinary design optimization for liquid cooling systems in a battery module. The advanced BMS controller involves the use of digital twin for real-time and faster acquisition of data. In battery recycling, the secondary use of battery involving mechanical process and recycling of battery material involving chemical processes involves robotic and artificial intelligence (AI)-based operation for safer and intelligent recovery of battery cells.

Fig. 15
Production, assembly, operation, and recycling aspects of a lithium-ion battery
Fig. 15
Production, assembly, operation, and recycling aspects of a lithium-ion battery
Close modal

5.2 Critical Research Directions.

Some of the future critical research directions are summarized as follows.

5.2.1 Role of Digital Twins.

The digital twins-based battery controller is not only used for remote monitoring and health prognostics but can also be used to reflect design change, customer feedback, charging type, type of operation, system pre-warning, energy management, smart charging, state of X (SOX) estimation, remote battery diagnostics, cell balancing, algorithm application, data collection, and processing. Hence, the cyber-physical-based battery digital twins can be employed as a promising approach toward battery health characterization and monitoring.

5.2.2 Standardization of Battery Packs.

The type of batteries used in EV has different shapes and sizes. Some batteries are manufactured by the battery suppliers, and some are produced by the car manufacturers themselves. Since the battery production differs, there are still no standardized battery packs. The countries have their own standardized battery packs, in terms of shape, size, and dimensions. The standardization of battery packs will not only provide a cost-effective solution for the battery manufacturers but also to the EV consumers. Standardization of battery packs in terms of weight, size, and dimensions will also help in simplifying the complex battery recycling process where the discarded batteries are sorted as per their size and shape. This will make the process less time consuming and less costlier. The upcoming researchers in this field may work on AI-based techniques to select the optimal battery dimensions for the standardization of battery packs in the future.

5.2.3 Role of Topology Design.

Structural optimization is very important in vehicle designing as lighter structures have a favorable influence on fuel economy, speed, and power requirements. Conversely, stable and strong structures are required for vehicle safety conditions and also for the vehicle static and dynamic behavior. Because of the contradicting requirements, optimization is indispensable. The topology optimization can be a robust design technique for chassis and casing structure optimization. This design is extremely helpful that addresses a task such as reducing casing dimension, yet maintaining the performance targets such as safety, vehicle handling, and so forth. Topology optimization aims at selecting the most suitable distribution of material and addresses problems solved by the finite element analysis approach. By topology optimization, optimal structural topology and the most suitable dimension of the truss components in chassis and casing layout can be determined.

5.2.4 Use of Nondestructive Testing Approaches for Real-Time Condition Monitoring.

Lithium-ion batteries are prone to failure at each stage of their life due to abuse conditions, aging, or manufacturing defects. These failures may be (a) mechanical, leading to internal short circuit due to dendrites; (b) electrical, resulting in over-voltage, causing temperature rise, and gassing; and (c) thermal, causing flame exposure. The nondestructive testing (NDT) methods can be applied either at the manufacturing stage or even at the product stage to predict the degradation or failures, if any. The stress and frequency sensors may result in inaccurate measurements. Hence, NDT methods using X-rays, optic sensors, near infrared (NIR) spectroscopy, infrared thermoscopy, EchM spectroscopy, and cycle test for real-time monitoring of battery failure and SOC/SOH estimation. Table 5 presents the description of NDT testing methods. Therefore, advanced research is required to detect the failures in advance, while in operation.

Table 5

Description of NDT methods showing principle of measurement and causes of failure

MethodPrinciple of measurementCause of failure
Electrochemical impedance spectroscopyAC currentFracture in lithium plating and welded parts
Neutron radiographyNeutronsGassing
Cycle testDC currentEstimation of SOC/SOH
X-ray tomographyElectromagnetic (EM) wavesInternal short circuit
Light optical systemEM wavesFracture in separator
NIR spectroscopyEM wavesPresence of moisture
High-precision weighingWeightPresence of leakage
Infrared thermoscopyInfrared raysDetection of foreign particles
MethodPrinciple of measurementCause of failure
Electrochemical impedance spectroscopyAC currentFracture in lithium plating and welded parts
Neutron radiographyNeutronsGassing
Cycle testDC currentEstimation of SOC/SOH
X-ray tomographyElectromagnetic (EM) wavesInternal short circuit
Light optical systemEM wavesFracture in separator
NIR spectroscopyEM wavesPresence of moisture
High-precision weighingWeightPresence of leakage
Infrared thermoscopyInfrared raysDetection of foreign particles

6 Conclusions

This article proposes a comprehensive and holistic perspective for an advanced battery pack design and its components. The proposed methodology integrated the optimal scenario for cell-electrode material selection, electrode design, battery assembly configuration, liquid cooling thermal design optimization, AI-based SOH prediction, digital twin-based BMS for faster data acquisition, and battery recycling aspects. The illustrated battery design framework has been provided based on the extensive literature review for the last 10 years. The proposed methodology is a general framework, applicable for all battery designs. However, there are certain limitations associated with the proposed framework, which have been discussed in the earlier sections. Some critical research gaps from the extensive literature review have been identified, and based on that, some future research directions are discussed: (a) role of digital twin, (b) standardization of battery packs, (c) role of topology design, and (d) use of nondestructive testing approaches for real-time condition monitoring.

Conflict of Interest

There are no conflicts of interest.

References

1.
Bijoy
,
S.
,
Kalita
,
P.
,
Garg
,
A.
,
Niu
,
X.
,
Zhang
,
X.
,
Peng
,
X.
, and
Bhattacharjee
,
D.
,
2019
, “
A Review of State of Health Estimation of Energy Storage Systems: Challenges and Possible Solutions for Futuristic Applications of li-Ion Battery Packs in EV
,”
ASME J. Electrochem. Energy Convers. Storage
,
16
(
4
), p.
040801
.
2.
Shimamura
,
O.
,
Watanabe
,
K.
,
Ohsawa
,
Y.
, and
Horie
,
H.
,
2007
, “
Research and Development Work on Lithium-Ion Batteries for Environmental Vehicles
,”
World Electr. Vehicle Association J.
,
1
(
1
), pp.
251
257
.
3.
Melhem
,
F. Y.
,
Grundur
,
O.
,
Hammoudan
,
Z.
, and
Moubayed
,
N.
,
2017
, “
Optimization & Energy Management in Smart Home Considering Photo-Voltaic, Wind and Battery Storage System With Integration of Electric Vehicles
,”
Canad. J. Electr. Comput. Eng.
,
40
(
2
), pp.
128
138
.
4.
Uddin
,
K.
,
Moore
,
A. D.
,
Barai
,
A.
, and
Marco
,
J.
,
2016
, “
Effects of High Frequency Current Ripple on Electric Vehicle Battery Performance
,”
Appl. Energy
,
178
, pp.
142
154
.
5.
Tang
,
X.
,
Xiaofang
,
M.
,
Lin
,
J.
, and
Koch
,
B.
,
2011
, “
Capacity Estimation for Li-Ion Batteries
,”
American Control Conference
,
San Francisco, CA
,
June 29–July 1
, pp.
947
952
.
6.
Wang
,
Q.
,
Cheng
,
X. J.
, and
Wang
,
J.
,
2017
, “
A New Algorithm for a Fast Testing and Sorting System Applied to Battery Clustering
,”
Renewable Energy Resources Impact: 2017 6th International Conference on Clean Electrical Power (ICCEP)
,
Santa Margherita Ligure
,
June 27–29
, pp.
397
402
.
7.
Peng
,
X.
,
Garg
,
A.
,
Zhang
,
J.
, and
Shui
,
L.
,
2017
, “
Thermal Management System Design for Battery Packs of Electric Vehicles: A Survey
,”
2017 Asian Conference on Energy, Power and Transportation (ACEPT)
,
Singapore
,
Oct. 24–26
, pp.
1
5
.
8.
Reitler
,
C.
,
Wasssiliadis
,
N.
, and
Lien Kamp
,
M.
,
2019
, “
Design of Thermal Management Systems for Battery Electric Vehicles
,”
2019 Fourteenth International Conference on Ecological Vehicles and Renewable Energies (EVER)
,
Monte-Carlo, Monaco
,
May 8–10
, pp.
1
10
.
9.
He
,
H.
,
Zhang
,
X.
,
Xiong
,
R.
,
Xu
,
Y.
, and
Guo
,
H.
,
2012
, “
Online Model Based Estimation of State of Charge and Open-Circuit Voltage of Li-Ion Batteries in Electric Vehicles
,”
Energy
,
30
(
1
), pp.
320
318
.
10.
Zhu
,
F.
,
Liu
,
G.
,
Tao
,
C.
,
Wang
,
K.
, and
Jiang
,
K.
,
2017
, “
Battery Management System for li-Ion Battery
,”
J. Eng.
,
2017
(
13
), pp.
1437
1440
.
11.
Li
,
W.
,
Garg
,
A.
,
Xiao
,
M.
,
Peng
,
X.
,
Le Phung
,
M. L.
,
Tran
,
V. M.
, and
Gao
,
L.
,
2020
, “
Intelligent Optimization Methodology of Battery Pack for Electric Vehicles: A Multidisciplinary Perspective
,”
Int. J. Energy Res.
,
44
(
12
), pp.
9686
9706
.
12.
Mishra
,
A.
,
Mehta
,
A.
,
Basu
,
S.
,
Maloda
,
S. J.
,
Shetti
,
N. P.
,
Shukla
,
S. S.
,
Nadagouda
,
M. N.
, and
Aminabhavi
,
T. M.
,
2018
, “
Electrode Materials for Lithium-Ion Batteries
,”
Mater. Sci. Energy Technol.
,
1
(
2
), pp.
182
187
.
13.
Penick
,
C.
,
Bhate
,
D.
,
Ferry
,
L.
, and
Lee
,
C.
,
2019
, “
Classification and Selection of Cellular Materials in Mechanical Design: Engineering and Biometric Approaches
,”
Designs
,
3
(
1
), p.
19
.
14.
Bhate
,
D.
,
2019
, “
Four Questions in Cellular Material Design
,”
Materials
,
12
(
7
), p.
1060
.
15.
Heard
,
D.
, and
Lennox
,
A.
,
2020
, “
Electrode Materials in Modern Organic Electrochemistry
,”
Angew. Chem. Int. Ed.
,
59
(
43
), pp.
18866
18884
.
16.
Xu
,
Z.
,
Yang
,
J.
,
Zhang
,
T.
,
Nuli
,
Y.
,
Wang
,
J.
, and
Hirano
,
S.
,
2018
, “
Silicon Microparticle Anodes With Self-Healing Multiple Network Binder
,”
Joule
,
2
(
5
), pp.
950
961
.
17.
Mei
,
W.
,
Chen
,
H.
,
Sun
,
J.
, and
Wang
,
Q.
,
2019
, “
The Effect of Electrode Design Parameters on Battery Performance and Optimization of Electrode Thickness Based on Electrochemical-Thermal Coupling Model
,”
Sust. Energy Fuels
,
2019
(
3
), pp.
148
165
.
18.
Ye
,
J.
,
Baumgaertal
,
A. C.
,
Wang
,
Y. M.
,
Biener
,
J.
, and
Biener
,
M. M.
,
2014
, “
Structural Optimization of 3D Porous Electrodes for High-Rate Performance in Li-Ion Batterise
,”
CAS Nano
,
9
(
2
), pp.
2194
2202
.
19.
Mei
,
W.
,
Chen
,
H.
,
Sun
,
J.
, and
Wang
,
Q.
,
2018
, “
Numerical Study on Tab Dimension Optimization of Lithium-Ion Battery From the Thermal Safety Perspective
,”
Appl. Therm. Eng.
,
142
, pp.
148
165
.
20.
Reutar
,
F.
,
Baasner
,
A.
,
Pampel
,
J.
,
Piwko
,
M.
,
Dorfler
,
S.
,
Althues
,
H.
, and
Kaskel
,
S.
,
2019
, “
Importance of Capacity Balancing on the Electrochemical Performance of NCM811 Silicon Fuel Cells
,”
J. Electrochem. Soc.
,
166
(
14
), p.
A3265
.
21.
Lopez-Chavéz
,
R.
, and
Cuentas-Gallegos
,
A. K.
,
2013
, “
The Effect of Binder in Electrode Materials for Capacitance Improvement and ELDC Binder Free Cell Design
,”
J. New Mater. Electrochem. Syst.
,
16
(
3
), pp.
197
202
.
22.
Noura
,
N.
,
Boulon
,
L.
, and
Emei
,
S.
,
2020
, “
A Review of Battery State of Health Estimation Methods: Hybrid Electric Vehicle Challenges
,”
World Electr. Vehicle J.
,
11
(
4
), p.
wevy 11040066
.
23.
Bartlett
,
A.
,
Marciki
,
J.
,
Onori
,
S.
,
Rizzoni
,
G.
, and
Guang
,
X.
,
2016
, “
Electrochemical Model-Based State of Charge and Capacity Estimation for a Composite Electrode Li-Ion Battery
,”
IEEE Trans. Control Syst. Technol.
,
24
(
2
), p.
1
.
24.
Andre
,
D.
,
Appel
,
C.
,
Soczka-Guth
,
T.
, and
Sauer
,
D. U.
,
2013
, “
Advanced Mathematical Methods of SOC & SOH Estimation for Li-Ion Batteries
,”
J. Power Sources
,
224
, pp.
20
27
.
25.
Li
,
Y.
,
Liu
,
K.
,
Foley
,
A. M.
,
Zulke
,
A.
,
Berecibar
,
M.
,
Maury
,
E. N.
,
Van Mierlo
,
J.
, and
Hoster
,
H. E.
,
2019
, “
Data-Driven Health Estimation & Lifetime Prediction of Lithium-Ion Batteries: A Review
,”
Renew. Sust. Energy Rev.
,
113
, p.
109254
.
26.
Zhao
,
L.
,
Wang
,
Y.
, and
Cheng
,
J.
,
2019
, “
A Hybrid Method for Remaining Useful Life Estimation of Lithium-Ion Battery With Regeneration Phenomena
,”
Appl. Sci.
,
9
(
9
), p.
1890
.
27.
Riviere
,
E.
,
Sari
,
A.
,
Venet
,
P.
,
Meniere
,
F.
, and
Bultei
,
Y.
,
2019
, “
Innovative Incremental Capacity Analysis Implementation for LiPeO4 Cell State of Health Estimation in Electric Vehicles
,”
Batteries
,
5
(
37
), pp.
1
13
.
28.
Harting
,
N.
,
Wollf
,
N.
,
Roder
,
F.
, and
Krewar
,
U.
,
2019
, “
State of Health Diagnostic of Li-Ion Batteries Using Non-Linear Frequency Response Analysis
,”
J. Electrochem. Soc.
,
166
, p.
A277
.
29.
Zhou
,
D.
,
Yin
,
H.
,
Fu
,
P.
, and
Song
,
X.
,
2018
, “
Prognostics for SOH of Lithium-Ion Batteries Based on Gaussian Process Regression
,”
Math. Prob. Eng.
,
2018
, pp.
1
11
.
30.
Tessier
,
A. O.
,
Dubois
,
M. R.
, and
Trovao
,
J. P.
,
2016
, “
Real-Time Estimator Li-Ion Cells Internal Resistance for EV Applications
,”
World Electr. Vehicle J.
,
8
(
2
), pp.
410
421
.
31.
Saha
,
A.
,
Han
,
S.
,
Agarwal
,
S.
, and
Guha
,
A.
,
2020
, “
An Incremental Voltage Difference Based Technique for Online State of Health Estimation of Li-Ion Batteries
,”
Sci. Rep.
,
10
(
1
), p.
9526
.
32.
Yun
,
L.
,
Shui
,
L.
,
Gao
,
L.
,
Fun
,
Z.
,
Ruhatiya
,
C.
,
Wang
,
C. T.
, and
Garg
,
A.
,
2019
, “
Experimental Combined Grouping Analysis Approach for Robust Battery Pack Design for EV With Higher Performance
,”
IOP Conference Series: Earth and Environmental Science
,
1
, pp.
1
10
.
33.
Piao
,
C.
,
Wang
,
Z.
,
Cao
,
J.
,
Zhang
,
W.
, and
Lu
,
S.
,
2015
, “
Lithium-Ion Battery Cell-Balancing Algorithm for Battery Management System Based on Real-Time Outlier Detection
,”
Math. Prob. Eng.
,
2015
, pp.
1
12
.
34.
Sun
,
F.
,
Xiong
,
R.
,
He
,
H.
,
Li
,
W.
, and
Aussems
,
J. E. E.
,
2012
, “
Model-Based Dynamic Multi-Parameter Method for Peak Power Estimation of Lithium-Ion Batteries
,”
Applied Energy
,
96
, pp.
77
86
.
35.
Raspa
,
P.
,
Frinconi
,
L.
,
Mancini
,
A.
, and
Cavalletti
,
M.
2011, “
Selection of Lithium-Cells for EV Battery Pack Using Self-Organisation Maps
,”
Batteries
.
36.
Li
,
R.
,
Zhang
,
H.
,
Li
,
W.
,
Zhao
,
X.
, and
Zhou
,
Y.
,
2020
, “
Towards Group Applications: A Critical Review of the Classification Strategies of li-ion Batteries
,”
World Electr. Vehicle J.
,
11
(
3
), p.
58
.
37.
Mauger
,
A.
,
Julien
,
C.
,
Paolella
,
A.
, and
Armand
,
M.
,
2019
, “
Building Better Batteries in the Solid State: A Review
,”
Materials
,
12
(
23
), p.
3892
.
38.
Wang
,
D.
,
Yang
,
F.
,
Gan
,
L.
, and
Li
,
Y.
,
2019
, “
Fuzzy Prediction of Power Lithium-Ion Battery State of Function Based on Fuzzy C-Means Clustering Algorithm
,”
World Electr. Vehicle J.
,
10
(
1
), p.
1
.
39.
Wang
,
H.
,
Ullah
,
M. M.
,
Klaser
,
A.
,
Laptev
,
I.
, and
Schmid
,
C.
,
2009
, “
Evaluation of Local Spatio-Temporal Features of Action Recognition
,”
British Machine Vision Conference
,
London
,
Sept. 8–10
, pp.
1
11
.
40.
Bernard
,
A. A. O. T.
, and
Zhang
,
W.
,
2015
, “
Thermal Management of Lithium-Ion Battery Pack With Liquid Cooling
,”
2015 31st Thermal Measurement, Modeling & Management Symposium (SEMI-THERM)
,
San Jose, CA
,
Mar. 15–19
, pp.
298
302
.
41.
Wang
,
Q.
,
Jiang
,
B.
,
Li
,
B.
, and
Yan
,
Y.
,
2016
, “
A Critical Review of Thermal Management Models and Solutions of Li-Ion Batteries for the Development of Pure Electric Vehicles
,”
Renew. Sust. Energy Rev.
,
64
, pp.
106
128
.
42.
Ouyang
,
D.
,
Chen
,
M.
,
Huang
,
Q.
,
Weng
,
J.
,
Wang
,
Z.
, and
Wang
,
J.
,
2019
, “
A Review on the Thermal Hazards of the Li-Ion Battery and the Corresponding Counter Measures
,”
Appl. Sci.
,
1
(
1
), pp.
1
10
.
43.
Bhattacharjee
,
A.
,
Mohanty
,
R.
, and
Ghosh
,
A.
,
2020
, “
Design of an Optimized Thermal Management System for Li-Ion Batteries Under Different Discharging Conditions
,”
Energies
,
13
(
21
), pp.
1
21
.
44.
Wang
,
Z.
,
Li
,
X.
,
Zhang
,
G.
,
Youfu
,
L. V.
,
Wang
,
C.
,
He
,
F.
,
Yang
,
C.
, and
Yang
,
C.
,
2017
, “
Thermal Investigation for Lithium-Ion Battery Module With Different Phase-Changing Materials
,”
RSC Adv.
,
7
(
68
), pp.
42909
42918
.
45.
Kiani
,
M.
,
Ansari
,
A.
, and
Houshfar
,
E.
,
2020
, “
Hybrid Thermal Management of Li-Ion Batteries Using Nao-Fluid, Metal-Foam & PCM: An Integrated Numerical Experimental Approach
,”
J. Therm. Anal. Calorim.
,
141
(
39
), pp.
1703
1715
.
46.
Han
,
T.
,
Khaligi
,
B.
,
Yen
,
C.
, and
Kaushik
,
S.
,
2018
, “
Lithium-Ion Pack Thermal Management-Liquid vs Air-Cooling
,”
ASME J. Therm. Sci. Eng. Appl.
,
11
(
2
), p.
021009
.
47.
Yi
,
Y.
,
Li
,
W.
,
Xiao
,
M.
, and
Gao
,
L.
,
2019
, “
A set Strategy-Based Approach for Multidisciplinary Robust Design Optimization Under Interval Uncertainty
,”
Adv. Mech. Eng.
,
11
(
1
), p.
1687814018820383
.
48.
Brandl
,
M.
,
Gall
,
H.
, and
Wenger
,
M.
,
2012
, “
Batteries and Battery Management System for Electric Vehicles
,”
Design, Automation and Test in Europe Conference & Exhibition Germany.
49.
Gao
,
Z.
,
Chin
,
C.
,
Chiew
,
J.
,
Jia
,
J.
, and
Zhang
,
C.
,
2017
, “
Design and Implementation of a Smart Lithium-Ion Battery System With Real-Tie Fault Diagnosis Capability for EVs
,”
Energies
,
10
(
10
), pp.
1
15
.
50.
Nizam
,
M.
,
Maghfiroh
,
H.
,
Rosadi
,
R. A.
, and
Kusumputri
,
K. D. U.
,
2020
, “
Battery Management System Design for Lithium-Ion Batteries
,”
AIP Conf. Proc.
,
2217
, p.
030157
.
51.
Cabrera
,
J.
,
Vega
,
A.
,
Tobojas
,
F.
, and
Deniz
,
V.
,
2014
, “
Design of a Reconfigurable Li-ion Battery Management System
,”
Technologies Applied to Electronics Teaching (TAEE)
,
Bilbao, Spain
,
June 11–13
, pp.
1
6
.
52.
Harper
,
G.
,
Sommerville
,
R.
,
Kendrik
,
E.
, and
Driscoll
,
L.
,
2019
, “
Recycling Lithium-Ion Batteries From Electric Vehicles
,”
Nature
,
575
(
7781
), pp.
75
86
.
53.
Pan
,
L.
,
Zhan
,
R.
, and
Oldenburg
,
Z.
,
2018
, “
Recovery of Active Cathode Materials From Lithium-Ion Batteries Using Froth-Floatation
,”
Sust. Mater. Technol.
,
17
, p.
e00062
.
54.
Gains
,
L.
,
2019
, “
The Future of Automotive Lithium-Ion Battery Recycling: Charting a Sustainable Course
,”
Sust. Mater. Technol.
,
1
(
2
), pp.
2
7
.
55.
Mossali
,
E.
,
Picone
,
N.
,
Gentilini
,
L.
,
Rodriguez
,
O.
,
Perez
,
J. M.
, and
Colledani
,
M.
,
2020
, “
Lithium-Ion Batteries Towards Circular Economy: A Literature Review of Opportunities and Issues of Recycling Treatments
,”
J. Environ. Manage.
,
264
, p.
110500
.
56.
Hafner
,
J.
,
Wolvuton
,
C.
, and
Ceder
,
G.
,
2006
, “
Towards Computational Material Design: The Impact of Density Functional Theory on Material Research
,”
MRS Bull.
,
31
(
9
), pp.
659
668
.
57.
Butler
,
K. T.
,
Gautam
,
G. S.
, and
Canepa
,
P.
,
2019
, “
Designing Materials in Energy Materials Applications With First Principle Calculations
,”
NPJ Comput. Mater.
,
5
(
19
), pp.
1
12
.
58.
Nashed
,
R.
,
Ismail
,
Y.
, and
Allam
,
N. K.
,
2103
, “
Recent Advances in the Use of Density Functional Theory to Design Efficient Solar Energy Based Renewable Systems
,”
J. Renew. Sust. Energy
,
5
(
2
), p.
022701
.
59.
Neugebauer
,
J.
, and
Hickel
,
T.
,
2013
, “
Density Functional Theory in Material Science
,”
WIREs Comput. Mol. Sci.
,
3
(
5
), pp.
438
448
.
60.
Schleder
,
G. R.
,
Padhila
,
A. C. M.
,
Acosta
,
C. M.
, and
Fazzio
,
A.
,
2019
, “
From DFT to Machine Learning: Recent Approaches to Material Science—A Review
,”
J. Phys. Mater.
,
2
(
3
), p.
032001
.
61.
Aliano
,
A.
, and
Cicero
,
G.
,
2012
, “Ab-Initio DFT Simulations of Nanostructures,”
Encyclopaedia of Nanotechnology
,
Springer
,
Dordrecht
.
62.
Gupta
,
D. K.
,
Lanelaar
,
M.
,
Barink
,
M.
, and
Van Keulen
,
F.
,
2014
, “
Topology Optimization : An Effective Method for Designing Front Metallization Patterns of Solar Cells
,”
Photovoltaic Specialist Conference (PVSC)
,
Denver, CO
,
June 8–13
, pp.
2471
2475
.
63.
Mei
,
W.
,
Chen
,
H.
,
Sun
,
J.
, and
Wang
,
Q.
,
2019
, “
The Effect of Electrode Design Parameters on Battery Performance and Optimization of Electrode Thickness Based on Electrochemical-Thermal Coupling Model
,”
Sust. Energy Fuels
,
3
(
1
), pp.
148
165
.
64.
Kim
,
J.-S.
,
Lee
,
D.-C.
,
Lee
,
J.-J.
, and
Kim
,
C.-W.
,
2020
, “
Optimization for Maximum Specific Energy Density of a Lithium-Ion Battery Using Progressive Quadratic Response Surface Method and Design of Experiments
,”
Sci. Rep.
,
10
(
1
), p.
15586
.
65.
Yoon
,
G. H.
, and
Park
,
J.
,
2010
, “
Topological Design of Electrode Shapes for Dielectrophoresis Based Devices
,”
J. Electrostat.
,
68
(
6
), pp.
475
486
.
66.
Kim
,
C.
,
Jeong
,
K. M.
,
Kim
,
K.
, and
Yi
,
C. W.
,
2015
, “
Effects of Capacity Ratios Between Anode and Cathode on Electrochemical Properties of Lithium Polymer Batteries
,”
Electrochem. Acta
,
155
, pp.
431
436
.
67.
Liu
,
C.
, and
Liu
,
L.
,
2017
, “
Optimal Design of Li-Ion Batteries Through Multi-Physics Modelling and Multi-Objective Optimization
,”
J. Electrochem. Soc.
,
164
(
11
), pp.
E3254
E3264
.
68.
Schmitt
,
J.
,
Raatz
,
A.
,
Dietrich
,
F.
,
Droder
,
K.
, and
Hesselbach
,
J.
,
2014
, “
Process and Performance Optimization by Selective Assembly of Battery Electrodes
,”
CIRP Annals
,
63
(
1
), pp.
9
12
.
69.
Ng
,
M. F.
,
Zhao
,
J.
,
Yan
,
Q.
,
Conduit
,
G.
, and
Sen
,
Z.
,
2020
, “
Predicting the State of Charge and Health of Batteries Using Data-Driven Machine Learning
,”
Nat. Machine Intelligence
,
2
, pp.
161
170
.
70.
Severson
,
K. A.
,
Attia
,
P. M.
,
Jin
,
N.
,
Perkins
,
N.
,
Jiang
,
B.
,
Yang
,
Z.
,
Chen
,
M. H.
,
Aykol
,
M.
,
Herring
,
P.
,
Harris
,
S.
,
Bazant
,
M.
,
Chueh
,
W. C.
, and
Braatz
,
R. D.
,
2019
, “
Data Driven Prediction of Battery Cycle Life Before Capacity Degradation
,”
Nat. Energy
,
4
(
5
), pp.
383
391
.
71.
Lu
,
L.
,
Han
,
X.
,
Li
,
J.
,
Hua
,
J.
, and
Ouyang
,
M.
,
2013
, “
A Review on the Key Issues for Lithium-Ion Battery Management in Electric Vehicles
,”
J. Power Sources
,
226
, pp.
272
288
.
72.
Johnson
,
V. H.
,
2002
, “
Battery Performance Models in ADVISOR
,”
J. Power Sources
,
110
, pp.
321
329
.
73.
Yang
,
D.
,
Wang
,
Y.
,
Pan
,
R.
,
Chen
,
R.
, and
Chen
,
Z.
,
2017
, “
A Neural Network Based SOH Estimation of Lithium-Ion Battery in Electric Vehicles
,”
Energy Proc.
,
105
, pp.
2059
2064
.
74.
Klass
,
V.
,
Behm
,
M.
, and
Lindbergh
,
G.
,
2014
, “
A Support Vector Machine Based State of Health Estimation Method for Lithium-Ion Batteries Under EV Operation
,”
J. Power Sources
,
270
, pp.
262
272
.
75.
Jia
,
J.
,
Liang
,
J.
,
Shi
,
Y.
, and
Wen
,
J.
,
2020
, “
SOH & RUL Prediction of Lithium-Ion atteries Based on Gaussian Process Regression With Indirect Health Indicators
,”
Energies
,
13
(
2
), pp.
1
20
.
76.
Lee
,
H.
,
Park
,
J.
, and
Kim
,
J.
,
2019
, “
Incremental Capacity Curve Peak Points Based Regression Analysis for the SOH Prediction of a Retired LiNiCoAlO2 Series/Parallel Configured Battery Pack
,”
Electronics
,
8
(
10
), p.
1118
.
77.
Li
,
Y.
,
Zou
,
C.
,
Berecibar
,
M.
,
Nanini-Maury
,
E.
,
Chan
,
J. C. W.
,
van den Bossche
,
P.
,
Van Mierlo
,
J.
, and
Omar
,
N.
,
2018
, “
Random Forest Regression for Online Capacity Estimation of Lithium-Ion Batteries
,”
Appl. Energy
,
232
, pp.
197
210
.
78.
Pop
,
V.
,
Bergveld
,
H. J.
,
Notten
,
P. H. L.
, and
Regtein
,
P. P. L.
,
2005
, “
State of the Art of Battery State of Charge Determination
,”
Meas. Sci. Technol.
,
16
(
12
), pp.
93
110
.
79.
Li
,
W.
,
Chen
,
S.
,
Peng
,
X.
,
Xiao
,
M.
,
Gao
,
L.
,
Garg
,
A.
, and
Bao
,
N.
,
2019
, “
A Comprehensive Approach for the Clustering of Similar Performance Cells for the Design of a Li-Ion Battery Module for Electric Vehicles
,”
Engineering
,
5
(
4
), pp.
795
802
.
80.
Ye
,
M.
,
Song
,
X.
,
Xiong
,
R.
, and
Sun
,
F.
,
2019
, “
A Novel Dynamic Performance Analysis and Evaluation Model of Series-Parallel Connected Battery Pack for Electric Vehicles
,”
IEEE Access
,
99
, pp.
1
1
.
81.
Bruen
,
T.
, and
Marco
,
J.
,
2016
, “
Modelling and Experimental Evaluation of Parallel Connected Li-Ion Cells for an Electric Vehicle Battery System
,”
J. Power Sources
,
310
, pp.
91
101
.
82.
Wang
,
X.
,
Li
,
M.
,
Liu
,
Y.
,
Sun
,
W.
,
Song
,
X.
, and
Zhang
,
J.
,
2017
, “
Surrogate Based Multi-Disciplinary Design Optimization of Li-Ion Battery Thermal Management System in Electric Vehicles
,”
Struct. Multi-Discip. Optimization
,
56
(
6
), pp.
1555
1570
.
83.
Xie
,
J.
,
Zang
,
M.
,
Wang
,
S.
, and
Ge
,
Z.
,
2017
, “
Optimization Investigation on the Liquid Cooling Heat Dissipation Structure for the Lithium-Ion Battery Package in Electric Vehicles
,”
Proc. Inst. Mech. Eng. Part D: J. Automobile Eng.
,
231
(
13
), pp.
1735
1750
.
84.
Wang
,
N.
,
Li
,
C.
,
Li
,
W.
,
Huang
,
M.
, and
Qi
,
D.
,
2021
, “
Effect Analysis on Performance Enhancement of a Novel Air Cooling Battery Thermal Management System With Spoilers
,”
Appl. Therm. Eng.
,
192
, p.
116932
.
85.
Li
,
W.
,
Peng
,
X.
,
Xiao
,
M.
,
Garg
,
A.
, and
Gao
,
L.
, “
Multi-Objective Design Optimization for Mini-Channel Cooling Battery Thermal Management System in Electric Vehicle
,”
Int. J. Energy Res.
,
43
(
8
), pp.
3668
3680
.
86.
Li
,
W.
,
Gao
,
L.
,
Garg
,
A.
, and
Xiao
,
M.
,
2020
, “
Multidisciplinary Robust Design Optimization Under Parameter and Metamodeling Uncertainties
,”
Eng. Comput.
, pp.
1
18
.
87.
Wu
,
B.
,
Widanage
,
W. D.
,
Yang
,
S.
, and
Liu
,
X.
,
2020
, “
Battery Digital Twin: Perspective on the Fusion of Models, Data and Artificial Intelligence for Smart Battery Management Systems
,”
Energy AI
,
1
, p.
100016
.
88.
Li
,
W.
,
Kwiecien
,
M.
,
Badeda
,
J.
,
Jöst
,
D.
,
Schulte
,
D.
, and
Sauer
,
D. U.
,
2020
, “
Digital Twin for Battery Systems: Cloud Battery Management System With Online State of Charge and State of Health Estimation
,”
J. Energy Storage
,
30
, p.
101557
.
89.
Gaines
,
L.
,
2018
, “
Lithium-Ion Battery Recycling Process Research Towards Sustainable Course
,”
Sust. Mater. Technol.
,
17
, p.
e00068
.
90.
Zhou
,
L.
,
Garg
,
A.
,
Zheng
,
J.
,
Gao
,
L.
, and
Ch
,
K. Y.
,
2020
, “
Battery Pack Recycling Challenges for the Year 2030: Recommended Solutions Based on Intelligent Robotics for Safe and Efficient Battery Disassembly, Residual Energy Detection and Secondary Utilization
,”
Energy Storage
,
3
(
3
), pp.
1
12
.
91.
Shui
,
L.
,
Chen
,
F.
,
Garg
,
A.
,
Peng
,
X.
,
Bao
,
N.
, and
Zhang
,
J.
, “
Design Optimization of Battery Pack Enclosure for Electric Vehicles
,”
Struct. Multi-Discip. Optimization
,
58
(
1
), pp.
337
347
.