Experimental Study of Thermal Runaway Process of 50 Ah Prismatic Nickel-Rich Battery

Abstract

Hazardous combustion and explosions during thermal runaway (TR) processes in lithium-ion batteries (LIBs) present a significant limitation to their widespread adoption. The objective of this study was to quantitatively reveal the eruption characteristics of LIBs. A commercially available prismatic cell with a capacity of 50 Ah was employed, featuring Li(Ni_0.6Co_0.2Mn_0.2)O₂ as the cathode material and graphite as the anode material. The investigation focused on the thermal runaway behavior at 100% state of charge (SOC). Three replicates of thermal runaway experiments were conducted within a 1000 L lithium battery adiabatic experimental chamber (AEC) under a nitrogen atmosphere, and the thermal runaway moments were captured using a high-speed camera. The ejection velocity of smoke during the opening of the safety valve was approximately 40 m/s; within an extremely short time frame following the opening of the safety valve, the jet stream temperature reached a peak value of 340.6 ± 42.0 °C; the duration of the ultra-high-speed jet was 12.0 ± 1.0 s, the high-speed jet lasted 9.9 ± 2.2 s, and the slow-speed jet persisted for 32.1 ± 3.0 s, resulting in an overall ejection duration of 53.9 ± 6.0 s.

1. Introduction

In recent years, there has been an exponential increase in the utilization of lithium-ion batteries (LIBs). Due to their multifunctional design, LIBs have extensive applications across various domains. Typically favored for their high energy and power density, lightweight nature, and prolonged lifespan, LIBs are becoming increasingly prevalent in portable consumer electronics, electric vehicles, and grid storage systems, resulting in a surge in scientific research pertaining to lithium-ion batteries [1].

Battery thermal runaway, one of the primary energy failure modes, can occur in batteries with various chemical characteristics [2,3,4,5]. Theses thermal runaway events can have severe consequences, including fires and explosions. In order to address this issue, Battery Thermal Management Systems (BTMS) have emerged as effective solutions for monitoring and controlling battery temperatures [2,3]. By continuously monitoring the temperature of batteries in real time and promptly implementing measures to mitigate overheating, BTMS ensure that battery temperatures remain within a suitable range, effectively reducing the risk of thermal runaway [4,5]. BTMS can dynamically adjust charging and discharging strategies, such as charging current and voltage, based on the battery’s energy density and temperature, maintaining safe operating conditions within the battery [6,7].

The consequences of thermal runaway in lithium-ion batteries are significant. During this process, an increasing amount of gas is generated within the battery. Once the internal pressure of the battery reaches a certain threshold, the safety valve ruptures, and the battery’s contents are ejected [8]. The ejection of lithium-ion batteries entails the release of thermal runaway byproducts, including the expulsion of both liquid and solid materials in addition to the gas jet emission. The ejected gas serves as a primary combustible component for fire formation, while the ejection of high-temperature particles may act as potential ignition sources for fires [9,10,11].

To better understand the behavior of lithium-ion batteries during thermal runaway, several studies have been conducted. Ping et al. investigated the flame temperature of lithium-ion batteries in an open-air environment. Their study revealed multiple temperature peaks in the fire zone, with the highest temperatures recorded reaching 1500 °C, 1020 °C, and 1091 °C at different states of charge [12,13]. Another investigation by Somandepalli et al. explored the quantity and composition of gases released from lithium-ion batteries in a controlled environment. Their experiments demonstrated that the highest temperature of gases inside a closed container was approximately 150 °C, significantly lower than the peak temperature on the battery surface [14]. Similar experiments reported in the literature studied the ejection process of thermal runaway byproducts and measured the temperature in the jetting zone. These studies provided insights into the behavior of lithium-ion batteries during thermal runaway events, with peak jet temperatures reaching around 400 °C [15,16].

Based on comprehensive analysis, the following conclusions can be drawn: In an open-air environment, the combustion phenomenon of battery jet flames results in relatively high temperatures (exceeding 1000 °C) for the ejected materials. Therefore, the primary objective of employing this method is to investigate the fire characteristics of batteries rather than studying the pre-reaction temperatures between the jetted materials and the surrounding air. In other words, this approach cannot directly provide the original temperature of the jetted materials. To obtain the original temperature of battery jet flames, a confined space with an inert atmosphere is required [14,17]. Few studies have conducted in-depth comprehensive investigations into the characteristics of the initial eruption of thermal runaway, including the ejection velocity of smoke, peak temperature of the jet, and duration of different velocity stages during the ejection process.

The primary focus of this study is the ejection characteristics of high-capacity nickel-rich batteries during thermal runaway (TR). A commercially available prismatic battery with a capacity of 50 Ah and a Li(Ni_0.6Mn_0.2Co_0.2)O₂ cathode (the experimental battery being a pristine, uncycled cell) was employed. The experiments were conducted within a 1000 L adiabatic chamber under a nitrogen atmosphere. Thermal runaway was induced by lateral heating, and the moment of thermal runaway was captured using a high-speed camera. Quantitative results were obtained for the ejection characteristics, including the ejection velocity of smoke, peak temperature of the jet, and duration of different velocity stages during the ejection process. These results provide more precise and detailed information about the behavior of LIBs during thermal runaway, contributing to a better understanding of the hazards associated with the TR process.

2. Materials and Methods

2.1. Battery Sample

A commercially available battery was employed in this study, featuring a cathode material of Li(Ni_0.6Co_0.2Mn_0.2)O₂. According to the manufacturer’s specifications, the battery has a rated capacity of 50 Ah and operates at a voltage of 3.65 V. For more detailed information, please refer to

Table 1. Detailed technical specifications of the test cell.

Parameters	Specifications
Cell mass (g)	900
Width (mm) × Thickness (mm) × Height (mm)	148 × 26.7 × 98
Nominal capacity (Ah)	50
Nominal voltage (V)	3.65
Minimum voltage (V)	2.75
Maximum voltage (V)	4.25
Main components of electrolyte	DMC, EMC
Cathode active material	Li(Ni0.6Co0.2Mn0.2)O2
Cathode coating thickness (µm)	61
Anode active material	Graphite
Anode coating thickness (µm)	73
Cathode current collector	Aluminum foil
Cathode current collector thickness (µm)	16
Anode current collector	Copper foil
Anode current collector thickness (µm)	10
Shell material	Aluminum alloy

and Appendix A.

2.2. Experimental Equipment

The experimental setup employed in this study consists of four main components: the experimental chamber, the heating system, the signal acquisition system, and the inert gas displacement system [18,19].

The experimental chamber is an adiabatic chamber with a volume of 1000 L, capable of withstanding a maximum pressure of 2 MPa. The maximum axial dimension of the chamber is 1000 mm, maintaining a length-to-diameter ratio of 1:1. The chamber’s door is hydraulically driven, which enables complete internal sealing.

The heating system comprises a constant power heating plate with a power output of 550W, which is of the same size as the battery, to induce lateral heating to trigger battery thermal runaway. A quartz plate is employed as a thermal insulation pad. Additionally, a battery fixture is utilized to provide preloading force and support.

The signal acquisition system consists of several components. Within the experimental chamber, a pressure sensor (HM90) with ±0.25% FS accuracy continuously monitors the pressure (P). A voltage sensor is employed to detect the battery voltage in real time. For temperature monitoring, eight K-type thermocouples (WRNK191) with accuracy of ±2.5 °C or ±0.75% tabs are strategically placed to measure the temperature at various points within the battery and the ambient environment. Specifically, one thermocouple is positioned at the center of the battery’s large surface to measure the surface temperature (T_S), while another thermocouple measures the positive electrode column temperature (T_P) and a separate one measures the negative electrode column temperature (T_N). A thermocouple is positioned 4 cm directly above the battery safety valve to measure the temperature of the jetting region (T_E). Three thermocouples are placed at a distance of 40 cm from the battery, at three different locations within the experimental chamber, to measure the ambient temperature (T_A1, T_A2, and T_A3). The average value of the data collected from these three thermocouples provides the average ambient temperature (T_A). The sensors operate at a sampling frequency of 10 Hz. Furthermore, the lithium-ion battery’s thermal runaway and jetting process were recorded using a high-speed camera (model: ACS-3) from Nac, Japan, at a frame rate of 5000 frames per second [15].

The inert gas displacement system operates as follows: Nitrogen gas (N2) is introduced into the experimental chamber through an intake pipeline, while the chamber’s gas is expelled through an exhaust pipeline. A vacuum pump is employed to facilitate the replacement of gases within the chamber. For a comprehensive depiction of the experimental setup, including the detailed configuration (excluding the positions of TN and TP), please refer to Figure 1 and the corresponding references [16,17].

2.3. Experimental Equipment

Prior to the commencement of the experiment, the battery was charged using a battery testing system (NEWARE CT-4008, 5V/6A) under a constant current–constant voltage (CC-CV) mode at 1/3 C until the charge reached 4.25 V. Subsequently, a one-hour rest period was observed, followed by discharge under a constant current (CC) mode at 1/3 C until the charge reached 2.75 V, followed by another one-hour rest period. The discharge and charge testing process was repeated three times. The battery was subsequently charged to 100% state of charge (SOC) using the CC-CV mode. Afterward, the battery was left undisturbed for 24 h and weighed to mitigate the influence of internal heating between the charging and discharging processes. The front and back walls of the battery casing were mechanically restrained using mica fixtures.

Experimental Setup: The battery was positioned within the AEC center, and thermocouples were strategically placed and inserted into the heating plates. Simultaneously, the fixture’s preloading force was adjusted accordingly. The experimental fixture structure followed a sequence of metal fixture → thermal insulation pad → heating plate → battery cell → thermal insulation pad → metal fixture, with a thermal insulation pad placed at the bottom of the battery, as illustrated in Figure 1.

Instrumentation Inspection: The heating plate circuitry was examined to ensure proper functionality. The temperature and pressure acquisition systems were also inspected for accurate operation. The AEC chamber door was securely closed, and the vacuuming process was repeated three times, achieving a pressure of 15 kPa. This was followed by nitrogen injection to reach a residual pressure of 106 kPa, ensuring an oxygen content below 1% within the test environment (confirmed through sampling tests). A 5 min stabilization period was observed after each vacuuming and nitrogen injection cycle to allow the temperature and pressure inside the AEC chamber to reach a stable state.

Thermal Runaway Triggering: The heating device was activated, maintaining a heating rate of 2 °C/min for the heating plate. The occurrence of thermal runaway was determined based on the point at which the battery voltage dropped to zero (this was considered the onset of thermal runaway in this study). Once thermal runaway was initiated, the heating process was halted, and the moment of thermal runaway was captured using a high-speed camera.

After the completion of the experiment, the remaining portion of the battery (solely the core) was photographed and weighed. Subsequently, it was securely sealed in a sample bag for preservation. Additionally, the ejected materials (including powders, particles, aluminum foils, electrode sheets, etc.) dispersed from the battery were collected and weighed.

3. Results

3.1. Thermal Runaway Eruption Phenomenon

During the thermal runaway (TR) process, significant amounts of combustible materials and gases are generated within the battery. The internal chemical reactions of the battery lead to a gradual increase in temperature and pressure. As excessive internal pressure accumulates beyond the structural stress and external pressure limits of the safety valve, the valve ruptures, and a substantial quantity of gases and combustible materials is ejected from the battery, resulting in fires and explosions [20,21].

Taking the first experiment as an example, the experimental results are presented. This study showcases two distinct phenomena of battery ejection, as illustrated in Figure 2: (a) smoke jetting and (b) spark jetting. The process of lithium-ion battery thermal runaway ejection was recorded using a high-speed camera. The total recording time for each experiment was 2.745 s, resulting in a total of 27,449 captured frames.

Due to the vertical jetting nature of the spray stream propelled upwards from the safety release valve, the velocity in the jetting zone above the battery’s safety release valve can to some extent characterize the speed of the spray stream. The photos were analyzed, and two images taken at different time intervals were selected. By calculating the distance traveled by the smoke within a certain time range, the ejection velocity of the smoke or spark can be determined. In this regard, the velocity of the smoke eruption at the moment the safety valve opened (averaged within a range of 0.15 m directly above the safety valve) in Figure 2a was calculated to be approximately 40 m/s, offering valuable insights for fire hazard prediction.

3.2. Jetting Temperature and Battery Surface Temperature

The temperature characteristics of the battery are primarily determined by the combined effects of the heat generation rate from the internal physicochemical reactions of the battery and the rate at which the heat dissipates to its surroundings [1]. As the jet stream is vertically ejected upward from the safety release valve, the temperature in the jetting zone above the battery’s safety release valve can to some extent reflect the temperature of the jet stream. Figure 3 displays the temporal variations of T_E and T_S in the three conducted experiments, with the reference point (t = 0) aligned with the moment the voltage dropped to zero as the indicator of thermal runaway onset.

Taking the first experiment as an example, the temperature trend above the battery’s safety release valve (T_E) can be observed from the curves in Figure 3. Over time, T_E gradually increases under the influence of the heating element until it reaches the first ejection temperature of 80.1 °C at −127.9 s. The maximum value of T_E is reached at 13.1 s, measuring 306.78 °C. From Figure 3, T_E,max is determined to be 340.6 ± 42.0 °C. The battery surface temperature (T_S) reaches its peak value of 639.1 °C at 65.5 s. According to Figure 3, T_S,max is calculated to be 593.4 ± 45.6 °C. The slight decrease in the battery surface temperature after the first ejection is attributed to the reduction in the vapor pressure of the internal electrolyte below the saturation vapor pressure upon the opening of the safety release valve, leading to significant evaporative cooling effects [16]. It should be noted that the differences in these characteristic temperatures, as mentioned in references [21,22], might be influenced by factors such as environmental pressure, ambient temperature, heating methods, battery quantity, and battery type.

Figure 4 presents the photographs taken before and after the thermal runaway of the battery. From the images, it is evident that the aluminum on the battery surface has melted. The experimentally measured value of T_S,max exceeds T_E,max by 150 °C. Therefore, when designing Battery Thermal Management Systems (BTMS), it is crucial to consider both the surface temperature of the battery and the thermal load imposed by the battery jetting temperature on the battery pack.

4. Discussion

4.1. Pressure Change Rate

Reference [23] and Figure 5 illustrate the general trend of pressure increase within the experimental chamber over time. Initially, there is a slow and steady rise in chamber pressure due to the temperature increase. This is followed by a rapid pressure surge, which then decreases quickly and gradually reaches a nearly constant value. The rate of pressure increase resulting from battery ejection is closely related to the volume of the experimental chamber. Therefore, it is necessary to normalize the pressure increase rate, and this requirement can be fulfilled using the explosion index (K_g), as indicated in Equation (1) [15]:

K_g = (dP/dt)_maxV^1/3

(1)

(dP/dt)_max represents the maximum rate of change of pressure with respect to time during the explosion process within the volume V.

The equation is derived based on an idealized treatment of gas explosions, assuming that K_g is independent of the volume of the chamber. This theoretical framework is widely applied to measure the severity of gas or dust explosions and has been adopted in standards by the International Organization for Standardization (ISO) and the British Standards Institution (BSI) [20].

In order to obtain a standardized pressure rise rate curve for LIB ejections, this study introduces the concept of LIB Ejection Index (K_LIB) based on findings from the literature [16]. Formula (2) is employed to calculate K_LIB, resulting in the plotting of the curve depicting the variation of K_LIB over time, as presented in Figure 5.

K_LIB = (dP/dt)V^1/3

(2)

where dP/dt represents the rate of increase in chamber pressure over time during the LIB eruption process, while V refers to the volume of the sealed chamber.

The process of the battery transitioning from heating to the ejection of thermal runaway products spans a relatively long duration. In comparison, the ejection process itself occurs within a short time frame. The K_LIB curve in Figure 6 primarily focuses on analyzing the ejection process. In the calculation procedure, the maximum value of K_LIB is determined using Formula (3).

K_LIB,max = (dP/dt)_maxV^1/3

(3)

where K_LIB,max refers to the maximum value of K_LIB and (dP/dt)_max refers to the maximum rate of pressure increase in the chamber.

To quantitatively analyze the venting process of lithium-ion batteries, this study defines several characteristic times [19].

• The time of venting initiation (t_e) is defined as the time corresponding to the rapid increase in K_LIB.
• The end time of high-speed venting (t_u) is defined as the time corresponding to the maximum value of K_LIB.
• The end time of fast venting (t_f) is defined as the time at which K_LIB transitions from positive to negative.
• The end time of slow venting (t_s) is defined as the time at which K_LIB returns to the initial fluctuation state before venting.

Based on the aforementioned definitions, the ejection process of LIBs can be divided into three stages: the ultra-high-speed ejection stage, the fast ejection stage, and the slow ejection stage [16]. The summary of each typical time interval is presented in

Table 2. Timing of the eruption phases.

	Du (s)	Df (s)	Ds (s)	De (s)
1st	12.8	12.0	35.1	59.9
2nd	13.8	10.1	32.0	55.9
3rd	11.1	7.7	29.1	47.9

.

• The duration of the ultra-high-speed venting stage (D_u) is the time interval from venting initiation (t_e) to the end of ultra-high-speed venting (t_u);
• The duration of the fast venting stage (D_f) is the time interval from the end of ultra-high-speed venting (t_u) to the end of fast venting (t_f);
• The duration of the slow venting stage (D_s) is the time interval from the end of fast venting (t_f) to the end of slow venting (t_s);
• The total venting duration (D_e) is the time interval from venting initiation (t_e) to the end of slow venting (t_s).

4.2. Mass Loss Rate and Gas Production

In general, for fire studies, the combustion rate of lithium-ion batteries is primarily influenced by the rate at which heat is released from the TR reactions within the battery and the mass ejection rate into the environment. The combustion process is highly complex, as even slight variations in the oxygen mass concentration and combustion reaction rate can result in different combustion rates, affecting the mass loss rate and gas generation [24].

Due to the instability of temperature and pressure within the experimental chamber, precise measurement of gas volume can only be achieved under stable conditions. In this study, the gas generation was calculated under stable environmental conditions. During the thermal runaway process, once the gas pressure reaches the release pressure of the safety valve, it opens and releases internal gases, electrolytes, and active materials, resulting in battery mass loss. The residual mass of the battery was weighed in this study, and the mass loss rate was calculated using Equation (4) [19], while the gas generation was determined using Equations (5) and (6). The mass loss rate and steady-state normalized gas production (unit: mmol·Wh⁻¹) are presented in

Table 3. Thermal runaway analysis.

	Mass Loss Rate	Gas Production (mmol·Wh−1)	KLIB,max (kPa·m·s−1)
1st	37.78%	27.32	2.4
2nd	42.81%	29.03	2.3
3rd	35.80%	24.50	1.9

.

K = m_e/m_r × 100%

(4)

Within this context, K represents the mass loss rate, m_e denotes the initial mass of the battery, and, m_r represents the remaining mass of the battery after TR.

PV = nRT

(5)

n = (P₂V₂)/(RT₂) − n₀

(6)

In the context of the aforementioned variables, n represents gas production, P₂ denotes the real-time chamber pressure after TR, V₂ represents the volume of the experimental chamber, R represents the ideal gas constant, T₂ signifies the stable post-venting chamber temperature, and n₀ represents the initial chamber gas volume.

Reference [25] showcases 76 experimental research papers on thermal runaway of lithium-ion batteries from 2000 to 2021. In this study, we have referenced two experimental research papers that share the same cathode material and shape. In one study [26], the normalized gas generation rate was reported as 30.09 mmol·Wh⁻¹, while in another paper [27], it was reported as 28.16 mmol·Wh⁻¹. Our study yields similar results in terms of normalized gas generation rate, indicating that the maximum capacity of the battery has minimal influence on the normalized gas generation rate.

5. Conclusions

This study conducted in situ thermal runaway experiments on 50Ah lithium-ion batteries in an inert environment, specifically focusing on their thermal runaway characteristics at 100% state of charge (SOC). The experimental methodology involved the analysis of the thermal runaway phenomenon and its associated properties:

During thermal runaway, the lithium-ion battery exhibits rapid and significant emissions of smoke within a short period of time. The pressure curve reveals two distinct ejection events, with the first event (corresponding to the instantaneous opening of the safety valve) exhibiting a peak expulsion velocity. The velocity of smoke emission during the opening of the safety valve is approximately 40 m/s, while the maximum value of TE is observed during the second ejection event. These findings serve as valuable guidance for thermal hazard early warning systems.

During the extremely short duration of thermal runaway, TE reaches its maximum value (T_E,max = 340.6 ± 42.0 °C), while the experimental chamber pressure exhibits a rapid increase (K_LIB,max = 2.2 ± 0.3 (kPa·m·s⁻¹)). Notably, T_S,max surpasses T_E,max by approximately 150 °C. This observation emphasizes the importance of considering both the surface temperature of the battery and the temperature of the ejection stream when designing Battery Thermal Management Systems (BTMS), as they contribute to the thermal load imposed on the battery pack.

During the extremely short duration of thermal runaway, TE reaches its maximum value (T_E,max = 340.6 ± 42.0 °C), while the experimental chamber pressure exhibits a rapid increase (K_LIB,max = 2.2 ± 0.3 (kPa·m·s⁻¹)). Notably, T_S,max surpasses T_E,max by approximately 150 °C. This observation emphasizes the importance of considering both the surface temperature of the battery and the temperature of the ejection stream when designing Battery Thermal Management Systems (BTMS), as they contribute to the thermal load imposed on the battery pack.

The findings of this study provide valuable guidance for thermal hazard warning systems, encompassing factors such as the battery surface temperature, battery safety valve nozzle temperature, experimental chamber pressure, and battery gas evolution rate. The method for the quantitative analysis of the thermal runaway eruption of LIBs can provide the start time and duration of the eruption of LIBs, which offers further guidance for thermal runaway early warning systems and fire suppression strategies. These results provide important insights and guidance for the development of Battery Thermal Management Systems (BTMS).