A Review of Lithium-Ion Battery Thermal Management Based on Liquid Cooling and Its Evaluation Method

Abstract

Electric vehicles (EVs) provide a feasible solution for the electrification of the transportation sector. However, the large-scale deployment of EVs over wide working conditions is limited by the temperature sensitivity of lithium-ion batteries (LIBs). Therefore, an efficient and reliable battery thermal management system (BTMS) becomes essential to achieve precise temperature control of batteries and prevent potential thermal runaway. Owing to their high heat-transfer efficiency and controllability, liquid-based cooling technologies have become a key research focus in the field of BTMS. In both design and operation, BTMSs are required to comprehensively consider thermal characteristics, energy consumption, economics, and environmental impact, which demands more scientific and rational evaluation criteria. This paper reviews the latest research progress on liquid-based cooling technologies, with a focus on indirect-contact and direct-contact cooling. In addition, existing evaluation methods are summarized. This work proposes insights for future research on liquid-cooled BTMS development in EVs.

1. Introduction

With the intensifying global warming effects caused by greenhouse gas emissions and the rising frequency of extreme weather events, the urgency of reducing such emissions has become increasingly prominent [1,2,3]. To address this global challenge, carbon neutrality targets for 2050 and 2060 have been established by the European Union and China, respectively [4,5]. In 2022, the transportation sector contributed approximately 14% of global greenhouse gas emissions [6]. Against this background, the electrification of road transport as a replacement for internal combustion engines is regarded as a popular and feasible approach to reducing greenhouse gas emissions [7,8,9]. The International Energy Agency projects that the total fleet of EVs will reach approximately 240 million by 2030 [10], indicating their significant potential in mitigating environmental issues. Battery packs are the core component of EVs [11,12]. Owing to their long lifespan, high energy density, and low self-discharge rate, LIBs are widely adopted as the main power source in EVs [8,13,14].

In recent years, the demand for longer driving ranges and shorter charging times in the EV market has continued to grow worldwide. To meet these requirements, the energy density of LIBs has increased from 80 Wh/kg in 1991 to 400 Wh/kg in 2020 [15]. Meanwhile, LIB technology has achieved extreme fast charging (XFC), enabling the state of charge (SOC) to reach 80% within 15 min, while capacity loss remains below 20% after 500 XFC cycles [16]. However, the significant increase in energy density and operating current intensifies the internal chemical reactions, resulting in considerable heat generation and accelerated temperature rise within the battery [17,18]. LIBs are highly sensitive to the operating temperature. When the temperature exceeds the optimal working range, high temperature can impair some active materials, which is one of the main causes of capacity degradation [19,20,21]. When the temperature exceeds 60 °C, the solid-electrolyte interphase (SEI) on the electrode may decompose and produce considerable heat, while continuous reformation of SEI leads to a vicious cycle [22]. If the battery temperature exceeds 80 °C, thermal runaway could be triggered, leading to fire or explosion [23,24]. Moreover, temperature non-uniformity among individual batteries or within battery packs can impact the overall performance [17,25]. For instance, a temperature difference of 5 °C may result in 1.5–2% capacity loss in the battery pack [26]. Therefore, LIBs should be maintained within a suitable temperature range. Researchers recommend an optimal operating temperature for LIBs between 20 °C and 45 °C [27,28], while the temperature difference among batteries in a pack should be less than 5 °C [29]. Strict temperature requirements present significant challenges for the widespread adoption of high-energy-density and XFC technologies. To address the high-temperature sensitivity of LIBs, the BTMS serves as a critical component in EVs to achieve precise temperature control and temperature uniformity, which play a crucial role in prolonging battery-pack lifespan and reducing the risk of thermal runaway [30,31].

In this work, the current progress on BTMS with a focus on liquid cooling is comprehensively evaluated. Perspectives on the future design and operation of liquid-cooled BTMS are discussed. The structure of this paper is presented as follows: Section 2 classifies and reviews BTMS based on the type of cooling medium. Section 3 further categorizes liquid-based cooling systems into indirect and direct-contact cooling, and summarizes the latest research progress in this field. Section 4 reviews current evaluation criteria for BTMS design and operation. Section 5 discusses the future development trends of liquid-based cooling BTMS. Finally, Section 6 presents the main conclusions.

2. BTMS

The most critical function of a BTMS is to maintain the battery-pack temperature within the optimal range and to minimize the temperature differences among individual cells. On this basis, a BTMS should also possess features such as energy efficiency, compact structure, lightweight, reasonable cost, and environmental friendliness. According to the type of cooling medium, BTMS can be classified as air cooling, liquid-based cooling, phase-change material (PCM) cooling, and heat pipe cooling [32,33]. The various forms of BTMS are depicted in Figure 1.

Air cooling achieves heat dissipation by means of convective heat transfer between the battery casing and the surrounding air. This approach offers advantages such as simple configuration, low cost, low energy consumption, and lightweight. As a result, air cooling was widely applied in the early development of EVs [43,44]. However, due to the limitations of the thermophysical properties of air, the heat-transfer capacity is limited, which makes it difficult for air cooling to meet the requirements for heat dissipation and temperature uniformity in extremely high-temperature environments and under high-power conditions [35,45,46]. PCM is a thermal management component that can absorb a large amount of latent heat during the solid-liquid phase transition. PCM cooling achieves heat transfer through direct contact between the PCM and the battery. Since the temperature remains nearly constant during the phase transition, this technique is particularly effective in enhancing temperature uniformity [47,48]. However, the thermal storage capacity of PCM depends on the amount used. Under high-power loads of the battery pack, achieving sufficient heat capacity may require a large quantity of PCM, which can significantly increase the system weight and volume [20,49,50]. In addition, PCMs typically cost around USD 1–8 per kilogram [20], which further raises the overall system cost. The heat pipe consists of three main components: shell, wick, and working fluid. The cooling mechanism of the heat pipe involves the working fluid absorbing heat and evaporating into vapor at the evaporation section. Driven by the slight pressure difference between the evaporation and condensation sections, the vapor flows to the condensation zone, where it returns to the liquid phase. The liquid then returns to the evaporation section by capillary action, thus achieving efficient heat transfer [51,52]. Owing to advantages such as long service life, flexible geometry, and high compactness, the heat pipe is highly favored in BTMSs [52,53]. However, heat pipes are expensive. Cost and weight are the main challenges hindering the application of heat pipes in large-scale battery packs [20]. Liquid-based cooling achieves heat exchange through direct or indirect contact between the coolant and the battery, offering higher heat-transfer efficiency compared to air cooling [54,55,56]. Under the same energy usage, the maximum temperature of batteries using liquid-based cooling can be approximately 3 °C lower than that under air cooling [35]. It is worth noting that coolant costs vary greatly between different cooling approaches, which can significantly affect the economics of liquid-based cooling systems. Wang et al. [20] provide reference prices for commonly used coolants: for indirect-contact cooling systems, fluids such as water cost only about USD 0.1–0.2 per liter, while the refrigerant R1234yf costs around USD 80 per liter; for direct-contact cooling systems, mineral oil is about USD 1.5–3 per liter, whereas NovecTM 7000 is as high as USD 239.2 per liter. Despite the disadvantages of structural complexity and increased weight resulting from additional components, the liquid-based cooling system offers precise temperature control and strong heat-transfer capability, thereby demonstrating the best overall performance among various BTMSs [57,58,59].

Table 1. Comparison of various BTMS technologies.

Type of BTMS	Advantage	Disadvantage
Air-cooling system	Simple structure Low cost Low energy consumption Lightweight	Poor temperature uniformity Not suitable for high-power batteries
Liquid-based cooling system	High-temperature control accuracy High-heat-transfer efficiency	Complex structure High weight
PCM cooling system	Temperature uniformity High heat storage capacity	Heavy Large volume High cost
Heat pipe cooling system	Long lifespan High compactness Flexible in various shapes High-heat-transfer efficiency	High cost Heavy

lists the advantages and disadvantages of various BTMS technologies.

3. Method of Liquid-Based BTMS

According to whether the liquid has direct contact with the battery, liquid-based cooling can be divided into two categories: indirect-contact cooling and direct-contact cooling. In direct-contact cooling systems, the battery is completely immersed in a non-conductive dielectric coolant, and heat is removed from the battery through fluid flow or flow boiling. In indirect-contact cooling systems, the coolant flows through intermediate heat-conducting components to transfer heat from the battery to the coolant, thereby achieving effective heat removal. Both direct and indirect-contact cooling systems can be further classified into single-phase liquid and two-phase liquid systems, depending on whether the cooling medium undergoes a phase transition.

3.1. Indirect-Contact Liquid Cooling

3.1.1. Single-Phase Liquid Indirect Cooling

The performance of the cooling fluid directly impacts the heat-transfer capacity and system energy consumption. Common single-phase coolants used in indirect-contact cooling systems include water [60], ethylene glycol [61], and glycol aqueous solutions [62], as listed in

Table 2. Performance parameters of single-phase liquid coolants in the indirect-contact cooling system.

Coolant	Density, kg/m3	Specific Heat Capacity, J/(kg·K)	Thermal Conductivity, W/(m·K)	Dynamic Viscosity, kg/(m·s)
Water [60]	998.2	4182	0.6	0.0010003
Ethylene glycol [61]	1111.4	2415	0.252	0.0157
50% glycol aqueous solution [62]	1069	3319	0.373	0.002940

. Due to excellent thermal conductivity, high specific heat capacity, and favorable fluidity, these coolants are widely used in BTMSs. Water is widely used as a coolant due to its high thermal conductivity and low dynamic viscosity. However, its relatively high freezing point leads to its solidification in cold climates where ambient temperatures drop below 0 °C, thereby impairing the normal operation of cooling systems [50,63]. In contrast, ethylene glycol exhibits a much lower freezing point. Moreover, although it has a higher dynamic viscosity than water, it can achieve superior cooling performance at the same Reynolds number [61]. By mixing ethylene glycol with water, the freezing point of the coolant can be effectively reduced. For example, a 50% glycol aqueous solution has a freezing point as low as −36.8 °C [64]. Specifically, glycol aqueous solutions provide better temperature uniformity compared to water [62,65]. Zhang et al. [66] investigated the effects of glycol solution concentration (ranging from 0% to 80%) on the maximum battery temperature and the pressure drop across cold plate channels. Their results indicated that increasing the ethylene glycol concentration leads to increases in both the maximum battery temperature and channel pressure drop. Therefore, the selection of glycol solution concentration should comprehensively consider antifreeze performance, heat-transfer efficiency, and energy consumption. Coolants containing nanoparticles can further reduce the maximum temperature difference and the peak temperature, but will also increase the system pressure drop [67,68,69]. However, these coolants are not electrically insulating, and any direct contact with the battery may cause short circuits. Therefore, indirect-contact cooling systems typically use cold plates or cooling tubes as intermediate heat-conducting components to enable indirect heat exchange between the coolant and the battery.

The cold plate denotes a flat surface equipped with internal cooling channels, suitable for prismatic and pouch batteries. The cold plate is typically made of aluminum [70], and its spatial arrangement and internal channel configuration are the main factors influencing cooling performance. As shown in Figure 2, the spatial arrangement of cold plates can be classified into four types: on the bottom of the battery pack [38], between adjacent batteries [71], on the side of the battery pack [60], and on multiple surfaces [72]. Increasing the contact area can improve the cooling effect [72], but it may also lead to an increase in the overall weight of the BTMS.

The mini-channel cold plate is a common channel configuration. Monika et al. [73] investigated the effect of channel width and the number of channels in straight mini-channel cold plates. The study indicates the presence of a design pinch point for channel width. Once this value is exceeded, further increasing the width has barely any effect on the control of temperature rise in the battery pack. The number of channels requires a balance between heat-transfer performance and pressure drop, as blindly increasing the channel number results in higher manufacturing costs. Chen et al. [74] employed multi-objective optimization techniques, taking the channel dimensions of mini-channel cold plates as design parameters, and the study found that the maximum temperature, temperature standard deviation, and energy cost were reduced by 2.1%, 23.7%, and 26.9%, respectively. Bionic concepts have been widely applied in the optimization of cold plate design, such as serpentine channels [75], honeycomb-like flow channels [76], human blood vessel structures [77], leaf vein branch channels [78], spider web flow channels [79], and fishbone channels [80], as shown in Figure 3.These structures can significantly enhance the heat-transfer area and temperature uniformity of cold plates. Monika et al. [81] studied six channel configurations with constant channel volume, finding that serpentine and hexagonal channels provided better temperature uniformity during 3C discharge, while pumpkin-shaped flow channels maintained lower pressure drop and pump power consumption. Ding et al. [82] explored the effects of cross-sectional geometry of channels and revealed that square channels were superior to circular channels in reducing the maximum battery temperature. Adding V-shaped ribs with square, triangular, semicircular, or trapezoidal cross-sections inside the flow channels can destroy the boundary layer and promote secondary flow, thereby enhancing the heat transfer between the cold plate and the coolant [83,84]. In addition to the structural optimization, key factors influencing cooling performance include the coolant flow direction, coolant temperature, and inlet coolant velocity [85,86,87]. When the coolant flows in opposite directions within two adjacent cold plates, the overall temperature uniformity can be further improved [87]. Reducing the inlet temperature and increasing the inlet velocity generally help reduce the maximum battery temperature [85,86]. However, an excessively low coolant inlet temperature may actually increase the maximum temperature difference on the battery surface [88]. In cold plate design optimization, research not only focuses on improving cooling performance but also emphasizes system compactness and lightweight [89]. For example, a simplified cold plate structure consisting of only two cold plates and one lightweight aluminum plate can reduce the weight of cooling components to 16.4 wt% [70]. Arranging a liquid-cooled plate on the short side surfaces of prismatic cells, as opposed to the traditional long side cooling configuration, can reduce the cold plate weight by 57% while maintaining similar cooling performance [65].

3.1.2. Two-Phase Liquid Indirect Cooling

To address the significant heat generation resulting from high-energy-density batteries and XFC, BTMSs must enhance their cooling capabilities. However, single-phase liquid cooling is subject to strict constraints on system lightweight and compactness, so it is not practical to simply enlarge the size and capacity of the cooling system to improve cooling performance. Compared with single-phase liquids, two-phase refrigerants exhibit nearly isothermal phase-change characteristics, which provide much higher heat-transfer efficiency and better temperature uniformity [90,91], demonstrating great potential in BTMSs [92,93]. Traditional single-phase cooling systems typically require a secondary loop to facilitate heat exchange between the coolant and the refrigerant. In contrast, a two-phase system uses refrigerant directly as the cooling medium, with the cold plate serving as an evaporator integrated into the vapor compression refrigeration cycle [94]. This method significantly reduces system volume and weight, thereby facilitating high spatial utilization and lightweight design in BTMS [92,93,94]. Common two-phase refrigerants include R134a [95,96], R1233zd [97], R1234yf [98], R141b [99], R717 [100], and R290 [101]. Wang et al. [102] used R134a as the refrigerant and found that under 3C discharge conditions, both the maximum battery temperature and heat-dissipation efficiency were significantly improved compared with single-phase liquid cooling. Hong et al. [103] found that two-phase refrigerant cooling provides considerably higher convective heat transfer than single-phase liquid cooling. Under an ambient temperature of 35 °C and a 2C charging rate, this system effectively limits the maximum temperature of LIBs and keeps the temperature under 45 °C.

The structural layout of the cold plate, the flow-path configuration, and the physical properties of the refrigerant together determine the cooling efficiency [92]. Tang et al. [94] found that when the cooling plate is installed at the pack bottom, a marked vertical temperature gradient was observed. As shown in Figure 4, the optimized configuration that combines main and side wall cooling effectively decreases temperature non-uniformity within the pack and limits the maximum temperature difference to within 5 °C. Refrigerants exhibit strong heat-transfer capabilities in the two-phase region [104]. However, severe heat-transfer deterioration is observed in the superheated vapor regions within the cold plate [90]. The formation of local vapor zones and non-uniform refrigerant distribution are the root causes of local overheating [105]. Therefore, reasonable distribution of the refrigerant and optimization of the flow channels are crucial. Lian et al. [105] employed refrigerant direct cooling plates with multi-splitting-merging channels. By adjusting the cross-sectional area of microchannels in different regions, they balanced the flow distribution and increased the heat absorption capacity in locally overheated areas. Lu et al. [39] developed a parameter model for a refrigerant-cooled multi-channel evaporator (MCE) and validated its effectiveness through experiments (Figure 5a). This model provides accurate predictions of the thermal non-uniformity of the battery under MCE cooling conditions, providing a reliable simulation tool for this promising cooling solution. Considering the volumetric expansion characteristic of two-phase fluids, Gao et al. [93] proposed a variable-area flow channel cold plate structure based on refrigerant cooling, which ensures thermal control consistency under discharge rates from 1C to 3C (Figure 5b). Tang et al. [106] effectively improved the thermal performance of the cold plate and the temperature uniformity of the battery by increasing the number of fins. The operational parameters of the refrigerant must be comprehensively optimized according to overall thermal management requirements. Although increasing the refrigerant flow rate can further lower battery temperature, it also causes a rise in system pressure drop and may slightly increase the temperature difference [102]. Conversely, appropriately increasing the inlet temperature of the refrigerant with decreased flow rate can improve the temperature uniformity of the battery [106]. In addition, as the mass flow rate of the coolant increases, the proportion of latent heat of evaporation may decrease, resulting in a reduction in the contribution of phase-change heat transfer to battery thermal management [107]. To address this issue, Ren et al. [107] proposed a variable flow rate cooling strategy. Compared to the constant refrigerant flow rate strategy, the proposed variable flow strategy better matches the heat generation pattern of the battery and enhances the contribution of latent heat-transfer on battery cooling, achieving an average latent heat-transfer proportion of 0.38. This approach effectively reduces both coolant and pump power consumption in the BTMS while providing the same or even better cooling performance. Jin et al. [108] proposed a new type of BTMS called the dual flow medium system, in which refrigerant and coolant are used in parallel to cool the battery simultaneously (Figure 5c). Compared to using refrigerant alone, the optimal coolant flow control method can reduce the maximum temperature difference between batteries by 31.94%.

3.2. Direct-Contact Liquid Cooling

3.2.1. Single-Phase Liquid Direct Cooling

Due to the electrical conductivity of liquids such as water, ethylene glycol-water solutions, and refrigerants, these liquids can potentially serve as pathways for electronic transmission [109,110]. To prevent short circuits within battery packs, the heat exchange between the coolant and the battery requires an intermediate thermal conductor. This introduces additional thermal resistance in the heat exchange path, thereby reducing the heat-transfer efficiency of the BTMS [30,111]. In direct-contact cooling systems, the batteries are completely immersed in an insulating dielectric fluid, enabling direct heat exchange between the heat source and the cooling medium. This unique cooling approach maximizes the surface area for heat transfer and provides a uniform path for heat transfer, significantly enhancing the temperature regulation capabilities and temperature homogeneity within the battery pack [112,113,114], which is especially suitable for high-rate charge and discharge situations [55,115,116].

The dielectric fluid employed in direct-contact cooling can be divided into single-phase and two-phase coolants depending on whether phase change occurs. For a single-phase system, a pump circulates the dielectric fluid between the battery and the external cooling unit, facilitating convective cooling at the battery surface. The absorbed heat is then transferred to an external heat exchanger for cooling [116]. A secondary loop is employed for the heat exchanger cooling fluid. The selection of coolant is critical since it directly influences the cooling system’s stability and safety. Single-phase coolants can be categorized into several types, including fluorinated fluids, hydrocarbons, esters, silicone oils, and water-based fluids [30,109,110]. These fluids possess a comprehensive set of advantages in the following key areas:

(1)

Safety characteristics: Non-flammable, low pour point, and high flash point.

(2)

Thermal performance: High thermal conductivity, high specific heat capacity, and good thermal stability.

(3)

Flow properties: Low density and low kinematic viscosity.

(4)

Stability: Good material compatibility and long service life.

(5)

Environmental factors: Low global warming potential (GWP), good degradability, and recyclability.

(6)

Availability: Capable of large-scale production.

It should be noted that excessive dynamic viscosity of the fluid will result in more significant pressure drops and higher system power consumption [115]. Tripathi et al. [117] conducted a sensitivity analysis and found that the thermal conductivity and specific heat capacity of the coolant primarily determine the effectiveness of immersion cooling, while the influences of fluid density and viscosity are limited. Xin et al. [109] performed a systematic evaluation of multiple typical dielectric fluids based on seven aspects: thermal conductivity, specific heat capacity, density, dynamic viscosity, pour point, flash point, and GWP. The results show that water-based dielectric fluids stand out in overall performance owing to their excellent comprehensive performance and cost-effectiveness. Zhou et al. [118] used deionized water as the direct-contact coolant and applied polyurethane coating as an electrical insulation layer. This approach significantly reduced the cooling costs of the system. However, it should be noted that the insulating layer adds thermal resistance, which may affect the cooling efficiency.

The effectiveness of single-phase cooling depends on the heat-transfer coefficient at the battery surface. Therefore, proper design of the flow rate is critical to improve the effectiveness of BTMS [55]. Increasing flow rate can notably improve heat-dissipation performance, thereby reducing the battery surface temperature rise [119]. Luo et al. [55] investigated the optimal volumetric flow rates for immersion BTMS using synthetic ester fluid under 2C and 3C conditions. The optimal volumetric flow rates were identified as 20 mL/min and 70 mL/min, resulting in maximum battery temperatures of 38.64 °C and 39.72 °C, respectively. In addition, the results indicated that heat transfer is influenced by both forced and natural convection. With increasing discharge rate, the contribution of natural convection grows, while forced convection dominates at high flow rates. Tang et al. [116] used flow guide plates in a parallel-flow immersion-cooling BTMS to enhance heat transfer. Wahab et al. [120] found that forced immersion cooling with optimized structure and flow field can significantly reduce maximum temperatures and enhance temperature uniformity, effectively alleviating local hotspots and ensuring battery safety. It is noteworthy that the heat generation from the battery tabs significantly affects the thermal distribution of the battery and must be considered in practical engineering designs. Patil et al. [113] applied forced air convection for auxiliary cooling at the tab based on immersion cooling (Figure 6a). At 5C discharge, the maximum temperature could be maintained below 40 °C. Ye et al. [121] proposed a novel cylindrical battery immersion-cooling system utilizing the Tesla valve principle (Figure 6b). By dividing the coolant flow into curved branches surrounding the battery and straight branches, a rational distribution of the coolant is achieved, significantly improving heat-transfer uniformity and cooling efficiency. Under 6C discharge conditions, the Tesla valve channel increases the convective coefficient to 689 W/(m²·K) and decreases the pressure drop to 86 Pa, which is far superior to conventional channels. Huang et al. [122] introduced a novel battery-pack cooling approach that integrates finned heat pipes with single-phase static immersion liquid, thereby enhancing cooling efficiency and temperature uniformity (Figure 6c). Bao et al. [114] proposed a static composite immersion system for batteries utilizing micro-heat pipe arrays (MHPAs), where the battery is immersed in static deionized water to use natural convection for enhanced thermal balance (Figure 6d). Simultaneously, the deionized water contacts the MHPAs on both sidewalls, using forced convection in dynamic regions to improve cooling efficiency. Tousi et al. [123] introduced indirect/direct coolant cooling strategies which achieved lower maximum temperatures and better temperature uniformity while utilizing less expensive dielectric fluid (Figure 6e). At an ambient temperature of 25 °C, during 5C and 7C discharges, the maximum temperatures recorded were 31.77 °C and 33.17 °C, respectively, with a temperature difference of below 2 °C for both discharge rates.

Table 3. Thermal performance of single-phase liquid direct cooling systems under high-rate charge and discharge.

Reference	Number of Batteries	Dielectric Fluid	Charge/Discharge C-Rate	Initial Temperature	Max Temperature	Max Temperature Difference
Donmez et al. [126]	16S1P	Engineered fluid	4C discharge	27 °C	307.28 K	<6 K
Williams et al. [127]	1	Novec 7000 (3M, Maplewood, MN, USA)	4C charge 4C discharge	\	294.6 K 300 K	2.5 K 6.8 K
Zhong et al. [128]	4S1P	EBC160 (The Karamay Petrochemical Company of China National Petroleum Corporation, Karamay, China)	4C discharge 5C discharge	25 °C	40.682 °C 43.476 °C	1.87 °C 2.25 °C
Patil et al. [113]	14S1P	mineral oil	5C discharge	25 °C	<40°C	\
Tripathi et al. [117]	1	deionized water mineral oil	5C discharge	25 °C	32 °C 39 °C	\ \
Wahab et al. [120]	16S1P	MIVOLT DF7 (MIDEL & MIVOLT Fluids Ltd, Manchester, UK)	8C discharge	20 °C	306.41 K	4.13 K

summarizes recent research on single-phase direct-contact cooling with charge or discharge rates of 4C or higher. Failure analysis of immersion-cooling systems helps improve overall system reliability. Borujerd et al. [124] applied Failure Modes and Effects Analysis to an immersed battery cooling system and found that poor sealing methods and inadequate battery temperature control are among the failure sources. Daniels et al. [125] deployed 135 temperature sensors within the fluid domain for fault detection of a 4S4P immersion-cooled battery module. They used Pearson correlation coefficients to select sensors and determine an optimal layout, thereby constructing a training dataset for machine learning. Four machine-learning algorithms were used to predict fault locations, and the Long Short-Term Memory (LSTM) network achieved 98.18% prediction accuracy on the specific external test cases. The study also compared the optimal sensor counts and placements under mineral-oil cooling versus air cooling, attributing the differences mainly to the distinct thermal properties of the two media, which lead to different flow patterns and heat diffusion behaviors.

3.2.2. Two-Phase Liquid Direct Cooling

In two-phase cooling systems, a dielectric fluid with an appropriate boiling point within the operating temperature range of the battery is selected. During the phase-change process, this fluid can absorb a large amount of vaporization latent heat. Compared with single-phase cooling systems, this design demonstrates superior temperature regulation capability when the battery operates at high power [109]. The cooling mechanism of two-phase systems can be divided into evaporation and boiling stages. In the evaporation stage, the dielectric fluid absorbs heat through natural convection, with its temperature gradually increasing to the boiling point [129]. During the boiling stage, the dielectric fluid undergoes a phase change at the battery surface, and bubbles form at the vapor-liquid interface. These bubbles disrupt the local temperature boundary layer, significantly enhancing heat transfer [130,131]. As the bubbles continue to grow and escape from the liquid phase, the produced vapor is transported to the condenser, where it is cooled and returns to the liquid phase, realizing a continuous vapor–liquid cycle [132]. In addition, immersion cooling can enable LIBs to maintain narrow temperature fluctuations during dynamic charging, reducing the complexity of cooling control [131].

To ensure that the boiling point of the fluid remains within the operational limits of the battery, choosing an appropriate two-phase coolant is crucial. Dielectric fluids suitable for two-phase systems include HFE7100 and SF33 [41,130,131]. In two-phase immersion-cooling systems, the fluid must be heated to its boiling point so that it can absorb the latent heat associated with phase change. Therefore, maintaining high heat flux within the battery is critical to achieving this process [109]. This makes two-phase immersion cooling especially suitable for rapid charge and discharge applications [130]. In terms of simulation and modeling, Jiang et al. [133] addressed the high computational cost of two-phase immersion-cooling simulations at the large battery-pack level by proposing a model that simplifies two-phase convective heat transfer into single-phase convective heat transfer. By correcting the control equation source terms and employing piecewise functions for variable properties during phase change, the model achieves 98% computational accuracy while reducing computation time by 84%. Several studies have compared two-phase liquid direct cooling with other cooling systems. Liu et al. [134] used HFE-7000 as the two-phase coolant and compared a microchannel liquid-cooling plate with a two-phase immersion-cooling system for the thermal management of a single 40 Ah battery. They found that the microchannel cold plate is limited under high loads, whereas the two-phase immersion system offers superior performance in lowering peak temperature and maintaining temperature uniformity. Lin et al. [135] evaluated air cooling, single-phase liquid indirect cooling, single-phase immersion cooling, and two-phase immersion cooling under high-rate discharge and demonstrated that two-phase immersion performs best in suppressing temperature rise and improving temperature uniformity. Goodarzi et al. [136] found that two-phase direct liquid cooling effectively removes heat from the battery pack, with complete immersion in the coolant yielding the most effective cooling. Conversely, when batteries are only partially immersed, the temperature difference within individual batteries becomes relatively large. Wang et al. [132] used SF33 as the two-phase immersion fluid and found that it could effectively keep the battery temperature below 34 °C, even at a 10C discharge rate. To further improve the cooling performance, Jung et al. [137] proposed the laser-induced graphene coating to enhance the immersion-cooling technology for batteries. Under an ambient temperature of 33.5 °C and a 5C discharge rate, the battery temperature remained far below the safety threshold, with uniform temperature distribution. Wu et al. [138] used NOVEC 7000 as the coolant in a two-phase boiling system and investigated three operating modes—static mode, continuous flow mode, and intermittent flow mode—for the thermal management of a 20 Ah pouch lithium battery. The results showed that the static mode can significantly suppress temperature rise and improve temperature uniformity during a 4C discharge, but it suffers from coolant loss during continuous cycling. The continuous flow mode can continuously supply the coolant, but it produces significant local temperature differences at high flow rates. To balance coolant supply, temperature control, and low pump power, the authors proposed the intermittent flow mode. Under 2C charge–discharge cycles, this mode can keep the maximum temperature below 36 °C and the maximum temperature difference within 2 °C.

4. Evaluation Indicator

The design and operational performance of BTMS are typically assessed by multiple thermal performance indicators, which can generally be divided into two types: temperature indicators and indicators of heat-transfer rate. Temperature indicators emphasize the extreme temperature levels of the battery pack, as well as the uniformity of the temperature distribution among individual batteries. The main indicators include the average temperature $T_{ave}$, maximum temperature $T_{\max }$, maximum temperature difference $\Delta T_{\max }$, and temperature standard deviation $T_{std}$ of the battery pack [139], as shown in Equations (1)–(4). Among these, the maximum temperature directly indicates the cooling effectiveness of the BTMS and is critical for ensuring the safe operation of the battery pack. The average temperature serves as a crucial indicator for assessing overall thermal management performance. The maximum temperature difference and temperature standard deviation are primarily employed to assess temperature uniformity among individual batteries. Reducing the temperature difference helps reduce performance variations between batteries during long-term operation, which improves the overall lifespan and reliability of the system.

$$T_{ave}={\displaystyle \frac{{\displaystyle \sum _{i=1}^n{T}_i}}{n}}$$

(1)

$$T_{\max }=\max \left(T_i\right)$$

(2)

$$\Delta T_{\max }=\max \left(T_i\right)-\min \left(T_i\right)$$

(3)

$$T_{std}=\sqrt{{\displaystyle \frac{{\displaystyle \sum _{i=1}^n{\left(T_i-T_{ave}\right)}^2}}{n}}}$$

(4)

In experiments, the temperature of a battery can be measured by placing thermocouples on the surface or in the interior. The arrangement of internal thermocouples is utilized to monitor the significant differences between core temperature and surface temperature. Fan et al. [140] proposed an embedded sensing system that can be mounted on the jelly rolls with a non-intrusive design to monitor the battery’s internal temperature in real time and transmit the data via wireless signals. This testing method offers the advantages of accurate sensing, wireless transmission, and miniaturized size. For large-format batteries, multiple thermocouples need to be positioned at various measurement points to calculate an average temperature. Due to the high cost of temperature sensors, adding multiple measurement points significantly increases expenses. Therefore, Deng et al. [141] proposed a physics-dominated neural network that utilizes the electric-thermal coupling equations of the battery as prior knowledge, allowing for the prediction of the temperature distribution across the entire space of the battery with only one temperature sensor. Meanwhile, Broatch et al. [142] reduced the number of temperature sensors deployed at the battery-pack level. Their study employed a combination of single value decomposition, thermodynamic models, and Kalman filters to optimize sensor placement and reconstruct the temperature field of the battery pack. The research demonstrated that three sensors are sufficient to effectively predict the temperature distribution of 20 batteries. In simulation analysis, the use of volumetric average temperature can serve as an effective representation of the temperature of an individual battery. In addition to the basic temperature indicators, the statistical Z-score method is also used for evaluation [143], as shown in Equation (5):

$$ZY={\displaystyle \frac{(X-\overline{X})}{S}}$$

(5)

where $ZY$ is the Z-score of the factor, $X$ is the index value of the factor, $\overline{X}$ is the average value of the factor, and $S$ is the standard deviation of the factor. However, the aforementioned temperature indicators primarily reflect static conditions, which makes these indicators insufficient for capturing the time-varying characteristics of temperature during the battery charging and discharging processes. To address this limitation, Zeng et al. [144] proposed using the standard deviation of the time percentage of the hottest battery position to evaluate the actual regulation effect of the BTMS throughout the entire operating cycle. The calculation formulas are shown in (6) and (7):

$$x_i={\displaystyle \frac{t_{hottest,i}}{t_{discharge}}}$$

(6)

$$\gamma =\sqrt{{\displaystyle \frac{{\displaystyle \sum _{i=1}^n{\left(x_i-\overline{x}\right)}^2}}{n-1}}}$$

(7)

where $\gamma$ is the temperature distribution coefficient, and $x_i$ is the time percentage of the hottest cell in the position i. The n denotes the total number of cell positions.

Regarding heat-transfer rate indicators, the cell cooling coefficient (CCC) is used to directly reflect the heat-transfer capability of the BTMS. The CCC is defined as the ratio of the heat-dissipation rate to the temperature difference along the battery heat-transfer path, as illustrated in Equation (8). A higher CCC value means that a smaller temperature difference is required for heat transfer [145]. Furthermore, there have been many extensions based on CCC, including considerations of cell capacity, geometry normalizations to aspect ratios, and thermal normalization factors [146].

$$CCC={\displaystyle \frac{Q}{\Delta T_{cell}}}$$

(8)

Several studies based on the second law of thermodynamics analyze the irreversibility in the heat-transfer process through entropy analysis. Anqi [147] systematically examined the impact of fluid flow rate and nanofluid volume fraction on thermal entropy generation and viscous entropy generation in serpentine microchannels from the perspective of entropy generation, as shown in Equations (9) and (10):

$$S_{th}={\displaystyle \frac{k}{T^2}}\left[{\left({\displaystyle \frac{\partial T}{\partial x}}\right)}^2+{\left({\displaystyle \frac{\partial T}{\partial y}}\right)}^2+{\left({\displaystyle \frac{\partial T}{\partial z}}\right)}^2\right]$$

(9)

$$S_f={\displaystyle \frac{\mu }{T}}\left[2\left[{\left({\displaystyle \frac{\partial u}{\partial y}}\right)}^2+{\left({\displaystyle \frac{\partial v}{\partial x}}\right)}^2+{\left({\displaystyle \frac{\partial w}{\partial z}}\right)}^2\right]+{\left({\displaystyle \frac{\partial u}{\partial y}}+{\displaystyle \frac{\partial v}{\partial x}}\right)}^2+{\left({\displaystyle \frac{\partial u}{\partial z}}+{\displaystyle \frac{\partial w}{\partial x}}\right)}^2+{\left({\displaystyle \frac{\partial w}{\partial y}}+{\displaystyle \frac{\partial v}{\partial z}}\right)}^2\right]$$

(10)

where $S_{th}$ is the thermal entropy generation, and $S_f$ is the viscous entropy generation.

The pressure drop caused by fluid flow is an important physical quantity for assessing the mechanical energy of the fluid pumping work. Some researchers have proposed using the j/f factor to comprehensively evaluate the heat-transfer performance and energy consumption of BTMS [148], as shown in Equations (11)–(13):

$$J={\displaystyle \frac{Nu}{\mathrm{Re}\cdot {\mathrm{Pr}}^{1/3}}}$$

(11)

$$F={\displaystyle \frac{\Delta P}{\frac{1}{2}{\rho }_l{u_{inlet}}^2}}\cdot {\displaystyle \frac{D_c}{4L}}$$

(12)

$$j/f={\displaystyle \frac{J}{F}}\cdot {10}^6$$

(13)

where $Nu$, $\mathrm{Re}$, and $\mathrm{Pr}$ represent the Nusselt number, Reynolds number, and Prandtl number, respectively. ${\rho }_l$ is the density (kg/m³), $D_c$ is the hydraulic diameter (mm), $\Delta P$ is the pressure drop (Pa), $L$ is the total length of the channel (m), and $u_{inlet}$ is the inlet velocity of the coolant (m/s).

The energy consumption of the BTMS is closely correlated with the driving range of EVs. Some researchers have proposed the cooling efficiency factor [149,150], as illustrated in Equation (14):

$$\beta ={\displaystyle \frac{Q_{diss}}{P_{BTMS}}}$$

(14)

where $\beta$ is the cooling efficiency factor, $Q_{diss}$ is the rate of heat taken (W), and $P_{BTMS}$ is the power consumption of the BTMS (W).

To achieve lightweight battery systems, energy density is utilized as an evaluation indicator, as demonstrated in Equation (15) [151]:

$${\rho }_{NRG}={\displaystyle \frac{E_{cells}}{m_{cells}+m_{BTMS}}}$$

(15)

where ${\rho }_{NRG}$ is the energy density (W/kg), $E_{cells}$ is the sum of all battery energies (W), $m_{cells}$ is the sum of the mass of the cells (kg), and $m_{BTMS}$ is the sum of the mass of the BTMS (kg). In addition, there is the packing efficiency rate based on volumetric energy density [152], as illustrated in Equation (16):

$$\mathrm{effciency}\mathrm{rate}={\displaystyle \frac{\mathrm{Volumetric}\mathrm{energy}\mathrm{density}\mathrm{of}\mathrm{the}\mathrm{moudule}}{\mathrm{Volumetric}\mathrm{energy}\mathrm{density}\mathrm{of}\mathrm{the}\mathrm{cell}}}\times 100$$

(16)

The economic aspect is a major concern for EV manufacturers. Rawat et al. [153] provide the calculation formula for the total average system cost of the BTMS, which comprehensively considers investment, operation, and maintenance, as well as energy consumption, as shown in Equation (17):

$$T_c=T_{battery}+T_{in}+T_{cs}+T_{loss}$$

(17)

where $T_c$ is the total average system cost, $T_{battery}$ is annual battery-pack depreciation cost, $T_{in}$ is installation cost, $T_{cs}$ is cooling system cost, and $T_{loss}$ is miscellaneous parasitic losses.

With increasing environmental protection requirements, the role of BTMS in controlling carbon emissions has garnered significant attention. Extreme temperatures or temperature non-uniformity can shorten battery-pack lifespan, resulting in more frequent replacements [154]. Lander et al. [154] tracked the carbon footprint of EVs from production to operation, including the carbon footprint of battery production, the carbon footprint of vehicle production, the carbon footprint of a kWh of electricity, and the carbon footprint of maintenance. Furthermore, based on the carbon footprint analysis of EVs, the study evaluated the impact of the BTMS. It was demonstrated that by optimizing the air-cooling system, lifecycle carbon emissions could be reduced by as much as 25%.

The purpose of maintaining battery temperature and temperature uniformity within the optimal operating range is to prevent battery performance degradation. From this perspective, the state of health (SOH) of the battery can be directly used as an evaluation indicator for the design and operation of BTMS. The definition of SOH is given by Equation (18) [155]:

$$SOH={\displaystyle \frac{Q_{now}}{Q_0}}\times 100\%$$

(18)

where $Q_{now}$ represents current battery capacity (Ah), and $Q_0$ is initial capacity (Ah). In addition, battery aging can also cause changes in thermal performance [155]. The internal resistance of the battery increases with aging, resulting in a significant rise in the average temperature and temperature difference of the battery pack [156]. Yang et al. [156] developed an SOH-based airflow control strategy. When the SOH is above 87%, the cooling airflow is increased according to the cooling demand. Conversely, when the SOH is below 87%, constant cooling airflow is maintained to balance temperature and energy consumption. Furthermore, production tolerances and long-term operation can also cause inconsistencies in the SOC during charging and discharging, which can be determined as shown in Equation (19) [157]. Chen et al. [157] proposed a staged reciprocating airflow cooling strategy for battery modules considering inconsistencies in SOC and SOH. The heat generation rate is calculated based on the SOC and SOH of the battery, which are estimated by a dual extended Kalman filter using recursive least squares. Compared with unidirectional flow, the maximum temperature and temperature difference of the battery module with SOH inconsistency are decreased by 4% and 32%, respectively. Similarly, for the battery module showing SOC inconsistency, the maximum temperature and temperature difference are decreased by 11% and 42%, respectively.

$$SOC=SO{C}_0+{\displaystyle \frac{1}{C_N}}{\displaystyle \underset{0}{\overset{t}{\int }}\eta Id\tau }$$

(19)

where $SO{C}_0$ is the initial state of the SOC, $C_N$ is the actual capacity of the battery (Ah), and $\eta$ is the charging or discharging efficiency of the battery. $I$ is the current (A).

Table 4. Summary of evaluation indicators.

Evaluation Indicator	Suitable System Type	Function	Limitation
Maximum temperature	Indirect-contact cooling/Direct-contact cooling	Reflects the maximum temperature; Directly related to thermal safety	Cannot reflect the temporal and spatial characteristics of local hotspots
Average temperature	Indirect-contact cooling/Direct-contact cooling	Reflects the overall heat-transfer performance of the system
Maximum temperature difference	Indirect-contact cooling/Direct-contact cooling	Evaluates the impact of thermal management on consistency of battery performance
Temperature standard deviation	Indirect-contact cooling/Direct-contact cooling	Evaluates the impact of thermal management on consistency of battery performance
Z-score of the factor [143]	Indirect-contact cooling/Direct-contact cooling	Retains the original distribution shape of battery temperatures; suitable for detecting abnormal temperature values	\
The temperature distribution coefficient [144]	Indirect-contact cooling/Direct-contact cooling	Evaluates the temporal and spatial distribution of the hottest spot	Not suitable as an objective function for optimization design
Cell cooling coefficient [145]	Indirect-contact cooling/Direct-contact cooling	Evaluates the heat-transfer capability along the heat-transfer path	Heat transfer between adjacent batteries can affect the assessment results
Thermal entropy and viscous entropy [147]	Indirect-contact cooling	Evaluates the entropy generation of the fluid flowing through the cold plate	\
j/f factor [148]	Indirect-contact cooling	Comprehensively evaluates heat-transfer performance and energy consumption	\
Cooling efficiency factor [149,150]	Indirect-contact cooling/Direct-contact cooling	Evaluates BTMS cooling efficiency	\
Energy density [151]	Indirect-contact cooling/Direct-contact cooling	Evaluates lightweighting of the BTMS	\
Total average system cost [153]	Indirect-contact cooling/Direct-contact cooling	Economic evaluation based on investment, operation, and maintenance	\
SOH [155]	Indirect-contact cooling/Direct-contact cooling	Evaluates battery SOH and performance consistency	\
SOC [157]	Indirect-contact cooling/Direct-contact cooling	Evaluates battery SOC and charging/discharging consistency	\

summarizes the suitable system types, functions, and limitations for the evaluation indicators.

5. Challenges and Outlooks

Liquid-based cooling technologies have demonstrated outstanding overall performance in various BTMSs and are regarded as highly promising cooling solutions for EVs. However, liquid-based cooling systems still have several potential technical challenges in the future that require further improvement. The main aspects are as follows:

5.1. Coolant Selection

The choice of coolants is crucial to the performance of liquid-based cooling systems. In indirect-contact cooling systems, water and its mixtures are limited by their inherent thermal properties, making it difficult to further improve heat-transfer capacity. Adding nanoparticles to the coolant to enhance heat transfer seems to be a promising solution. However, nanofluids generally exhibit relatively high dynamic viscosity, which leads to increased energy consumption. This phenomenon arises because dynamic viscosity is closely related to internal fluid friction, and an increase in viscosity intensifies flow resistance, resulting in a larger pressure drop. The long-term stability of nanofluids is also elusive, which impedes their applications. Future studies should investigate the types and concentrations of nanoparticles, aiming to improve cooling performance and its stability while balancing energy consumption and economic feasibility. In direct-contact cooling systems, dielectric fluids are expensive and typically cost more than water and its mixtures. This has become a major limiting factor in the development and application of direct-contact cooling systems, potentially offsetting the cost advantages gained from simplifying system structures. Therefore, it is necessary to develop new dielectric fluids with low costs and favorable thermal properties. The development of novel dielectric fluids requires multidisciplinary knowledge, involving materials chemistry, electrical engineering, and thermal sciences. The design process should consider safety characteristics, thermal performance, flow properties, stability, and environmental friendliness. Compared with the complex, costly development of new dielectric fluids, using deionized water as the working fluid in direct-contact cooling systems is a viable alternative due to its low cost and low viscosity. However, to ensure electrical safety, the battery surfaces must be insulated well. The insulating coating needs to have excellent dielectric strength and waterproofing properties. Its thickness is one of the key factors determining breakdown voltage, but its increase also raises thermal resistance and reduces cooling performance. Therefore, the long-term effectiveness of coating materials should be continuously monitored and evaluated under immersion, and an optimal coating thickness that balances electrical safety and thermal management should be determined over the product lifecycle.

5.2. Cold Plate Design and Optimization

In indirect-contact cooling systems, cold plates commonly face three primary challenges: non-uniform flow distribution that degrades the temperature-field uniformity; reductions in peak temperature and improvements in temperature uniformity often incur increased pressure drop with higher pump power; and the lack of available trade-off criteria for multi-objective optimization. Bionic channel designs based on the “optimal solutions” of long-term biological evolution have been shown to enhance the overall heat transfer. Future research should systematically reveal the mechanisms by which bionic heat-dissipation structures enhance heat transfer and use sensitivity analysis to identify the geometric parameters that have a significant impact on thermal performance. On this basis, a multi-objective optimization framework can be established. The strict constraints on maximum temperature and maximum temperature difference can be imposed. The objective weights according to the application scenario can be allocated. For example, prioritize the temperature indicator for fast-charging scenarios and emphasize the energy consumption indicator for long-range applications. Indicators such as cost, mass, and carbon emissions are handled similarly. In a two-phase refrigerant cooling system, achieving uniform fluid distribution inside the cold plate is key to ensuring temperature uniformity. However, current understanding of gas-liquid distribution mechanisms during two-phase boiling is insufficient, and issues such as bubble blockage remain unresolved. Therefore, microscale experimental techniques such as high-speed imaging and microscale sensors can be applied in future work. Characterization methods like infrared thermography and scanning electron microscopy can be used to reveal the multiscale mechanisms of two-phase boiling. High spatiotemporal-resolution data can help refine the mechanistic understanding of two-phase boiling and support the development of physical models as well as the application of data-driven models, thereby guiding channel geometry optimization. The multi-objective optimization approach for cold plates in two-phase systems is similar to that for single-phase systems. How to further improve the refrigerant flow distribution and prevent the formation of overheated regions is the main focus of current designs.

5.3. Coolant Leakage and Long-Term Reliability

Over long-term operation, heat pump systems in electric vehicles face a risk of refrigerant leakage, which leads to performance degradation, exacerbates greenhouse effects, and introduces safety hazards (e.g., the flammability and explosiveness of R290 and the toxicity of R717) [158]. Quantitative studies have shown that, without a high-pressure receiver and a low-pressure gas–liquid separator, a 20% reduction in refrigerant charge leads to about a 20% decline in both system efficiency and cooling capacity [159]. However, the leakage impacts specific to two-phase refrigerant BTMSs remain insufficiently studied. In particular, the mechanisms governing refrigerant distribution within cold plates, the formation and evolution of two-phase and superheated zones under leakage, and their effects on heat transfer are not yet well understood. Meanwhile, research on immersion cooling is still in its early stages. Existing studies emphasize cooling performance, with limited investigation into the long-term chemical compatibility of immersion coolants with battery components and sealing materials, as well as the temporal stability of coolant properties. In addition, there is a lack of system-level analyses of leakage risks in immersion systems. In light of these gaps, future work should conduct continuous monitoring and modeling of coolant property evolution over the service life, perform systematic compatibility and reliability assessments of sealing materials in contact with coolants, and clarify the potential interfacial failure mechanisms. Furthermore, integrating machine learning with multi-point sensing allows researchers to perform parameter sensitivity analysis to optimize sensor placement, enabling intelligent leak detection from abnormal temperature and pressure data.

5.4. Intelligent Control Strategy

In order to achieve precise temperature control and low-energy operation of BTMSs, the cooling capacity should be dynamically matched to the real-time heat generation characteristics of the battery. By monitoring the real-time temperature state of the battery pack and utilizing machine-learning algorithms to adjust system operating parameters, heat-transfer efficiency can be further improved. Furthermore, vehicle-to-cloud technology can be introduced to predict the vehicle speed and energy consumption. By combining the advantages of artificial intelligence in data-driven approaches and real-time interaction, it can achieve in-depth integration with the BTMS for intelligent control. In addition, indirect cooling of two-phase refrigerants coupled with air conditioning systems for integrated vehicle thermal management currently faces challenges in achieving optimal real-time control to satisfy both battery cooling and cabin thermal comfort simultaneously. Future studies may explore the time threshold for human thermal comfort delay and combine it with vehicle speed prediction to allocate cooling capacity for battery pre-cooling in advance, thereby meeting the cooling demands of both the cabin and the battery at the same time.

5.5. Localized Hot Spots Reduced by PCM Combined Cooling

In two-phase refrigerant cooling systems, heat transfer occurs between the cold plate and the battery through thermal exchange. However, there are currently localized vapor regions inside the cold plate that lead to localized hot spots. High temperatures at localized hot spots create large temperature differences between that area and other regions, and persistent high temperatures can cause performance differences between batteries. To address this challenge, the introduction of PCM into the system can be considered to assist in cooling by absorbing heat during the peak power operation of the battery. Currently, there are still relatively few papers on cooling with PCM combined with two-phase refrigerants. Future research directions can be pursued from two main aspects. More in-depth exploration of the causes and characteristics of the overheating regions is needed to facilitate the integration of PCM within the cold plate for enhanced design. Using the temperature distribution coefficient as an evaluation indicator for the hottest battery cell, identify the hotspot location and the duration of high temperature. Based on this information, determine the area where PCM is needed to mitigate local overheating. Additionally, investigating the external configuration of PCMs in the battery module space and implementing rational spatial arrangement designs is important. By exploring these two directions, new effective solutions for BTMS can be provided.

6. Conclusions

This work systematically discusses the progress of the latest research on current BTMS with a focus on liquid-cooling techniques. Compared to conventional air cooling, PCM cooling, and heat pipe cooling strategies, liquid-based cooling demonstrates the best overall performance in terms of heat-transfer efficiency and precise temperature regulation. The latest research progress in both indirect-contact and direct-contact liquid-cooling technologies is comprehensively illustrated. Comprehensive summaries of current evaluation indicators for the design and operation of BTMSs are also summarized to support the future development of BTMS.

For indirect-contact cooling, the improvement of thermal performance through cold plate structural optimization has reached a research bottleneck. Future research should pay more attention to the integration of technologies such as PCM and heat pipes, and the development of hybrid cooling methods that combine the advantages of various technologies, especially for high-power applications under XFC conditions. In addition, efficient and intelligent control of hybrid cooling systems will be a key challenge for future studies. In terms of direct-contact cooling technologies, the development of coolants and the enhancement of heat transfer remain critical issues. It is necessary to further improve the thermal properties of coolants while reducing their cost. Two-phase boiling techniques are becoming attractive for the future two-phase cooling design of BTMS to reduce the temperature difference among the battery packs.

With the sharply increased demand for higher power density and charge/discharge rates of EVs, BTMS systems become increasingly complex. Future optimization research should adopt a multidimensional evaluation approach, particularly by incorporating lightweight design and economic factors as important considerations.