Research on Thermal Runaway and Propagation Suppression of Energy Storage Batteries Based on Active Energy Dissipation Control Strategy of BMS

Abstract

With the increasing popularity of battery energy storage technology, safety issues have become increasingly important. The battery management system (BMS) is a key device for ensuring the safety of lithium-ion battery systems. While the BMS can effectively prevent faults such as external overheating, overload, or deep discharge, it cannot completely eliminate the possibility of internal short-circuit (ISC) faults—these faults may be caused by multiple factors, such as manufacturing defects. Therefore, reliable ISC detection or mitigation strategies need to be designed within the BMS to reduce the consequences of such faults. This study focuses on the critical role of the BMS in responding to thermal runaway (TR) and thermal propagation (TP) events caused by ISC faults and proposes an active energy-dissipation BMS control strategy. This strategy is compared with existing battery current interrupt device (CID) protection and threshold-type BMS protection schemes. A coupled electro-thermal simulation model was constructed based on thermal runaway test data of 280 Ah lithium iron phosphate batteries, and the proposed strategy was verified within this model. The proposed strategy can effectively suppress thermal propagation and thermal runaway in battery energy storage systems, providing a reference for the safety of battery energy storage systems (BESS).

1. Introduction

The global growth in the demand for renewable energy has accelerated the development of energy storage technologies. Lithium-ion batteries have become the mainstream energy storage solution in power stations and electric vehicles due to their high energy density and high efficiency [1,2]. Although lithium-ion batteries have been widely used, safety hazards pose a major challenge to their reliable operation. Thermal runaway (TR) is characterized by uncontrollable self-heating, posing a serious threat to personnel safety and the integrity of infrastructure [3]. Although the probability of TR in a single battery cell is statistically low, the high energy density of modern battery modules implies that the failure of a single cell may trigger catastrophic thermal propagation (TP) throughout the entire system [4,5]. Experimental analysis of battery packs shows that a closed space accelerates the temperature rise and hinders heat dissipation. Compared with an open system, this significantly amplifies the risk of TP [6]. Industry statistics show that since 2011, there have been over 90 large-scale stationary battery fires worldwide, highlighting the increased risk of systemic failures caused by high energy density [7]. Therefore, formulating strong strategies to curb TR and prevent its spread is crucial for the sustainable development of battery energy storage technology.

Multiple studies have delved deeply into the fundamental mechanisms of this thermal instability. Although the industry generally believes that internal short circuit (ISC) is a precursor to battery TR. Ren et al. found that the contribution of ISC to heat generation was negligible [8]. The main source of heat is the intense exothermic reaction between the negative electrode and the electrolyte. Comparative studies have shown that although the lithium iron phosphate (LFP) chemical system has superior thermal stability compared to the nickel cobalt manganese (NCM) system, electrolyte decomposition and the release of hazardous gases still occur under misuse conditions such as overcharging [9]. Wang et al. confirmed that the TR peak temperature of the 280 Ah LFP battery is lower than that of the NCM battery, but it releases a large amount of flammable gases (such as H₂, CO, and CH₄) during the exhaust process, posing an explosion risk in a confined space [10]. Zhao et al. pointed out that SOC is a key factor influencing reaction kinetics and gas toxicity. They found that when the carbon deposit level increased (≥75%), the reaction path shifted towards incomplete combustion, leading to an increase in the release of toxic carbon monoxide (CO) and hydrocarbons and expanding the combustible range of the emitted gas [11]. Ye et al. confirmed that under adiabatic overcharging conditions, side reactions contributed 80% of the total reaction heat, indicating that TR may occur within two minutes after the voltage reaches the inflection point, and immediate cooling must be carried out to prevent failure [12].

To combat these risks, modern safety strategies emphasize passive thermal management and physical isolation of materials. Previous studies have explored the physical barriers to heat conduction between battery cells [13]. Chen et al. demonstrated that phase change materials (PCMs) containing flame retardants can absorb excessive heat and delay TP [14]. Liu et al. optimized the thermal barrier through an anisotropic heat conduction structure to slow down the heat diffusion inside the module [15]. For active cooling, advanced two-phase heat transfer technologies such as immersion boiling and spray cooling have been proposed, which can effectively meet the heat dissipation needs of high-power charging [16]. Fire extinguishing strategies for lithium-ion battery fires (such as adding low-conductivity water mist in the form of a dosage form) can suppress the jet flame within 10 s and reduce the generation of CO by 56.6%. However, long-term application may lead to a short circuit risk for undamaged batteries [17]. Li et al. further quantified the energy flow during the fire spread process in the LFP module [18].

Despite the progress made in physical mitigation strategies, existing research has significant limitations. Most studies regard battery cell fault analysis and system-level protection as independent fields, ignoring the bidirectional interaction between battery cells and BMS. The traditional BMS safety logic mainly responds passively based on thresholds, only monitoring the three key parameters of voltage, current, and temperature. The circuit is cut off when these parameters exceed the safety limits. Although this measure can prevent the continuous inflow of external energy, it cannot regulate the energy accumulated inside the damaged battery.

Research has shown that SOC is the most crucial parameter affecting the intensity of energy release. Reducing the SOC to 50% or below can shorten the TP interval by more than 70%, or even completely prevent the propagation [11,18]. Scholars Li et al. found that even if the BMS functions normally, self-ignition may still occur due to battery consistency issues and the accumulation of heat and electrolyte during propagation [19]. As Zhang et al. found in a closed battery system, the decline in heat dissipation capacity triggers a reverse environmental heating phenomenon, and environmental heat accelerates the fault process, which is not affected by circuit disconnection [6]. These findings indicate that the current energy storage systems lack effective protective measures to suppress TR.

This study explored the crucial role of the BMS in BESS and proposed a BMS control framework based on active energy dissipation. This method reduces the SOC of adjacent TR cells through passive balancing, which is different from that of static physical barriers. A coupling simulation model of a 1P52S 280 Ah lithium iron phosphate battery and BMS control was developed in the research. This platform can accurately simulate the TR behavior caused by ISC under different working conditions, and conduct comparative analysis on three schemes: battery CID, traditional threshold BMS, and the active energy dissipation BMS protection strategy.

2. Mathematical Modelling

This section describes the process of establishing a simulation model that reflects the TR and TP characteristics of the energy storage battery modules. The simulation results of TR and TP occurrence of energy storage batteries under different operating conditions were obtained in this simulation model.

2.1. Energy Storage Battery Modeling

To study the thermal runaway characteristics of energy storage battery modules in simulations, it is first necessary to establish a physical simulation model that can reflect the thermal effects of lithium-ion batteries, including the electrical, thermal, and thermal runaway characteristics of the battery. In this study, a first-order RC circuit was used to construct the battery simulation model, with thermal characteristics represented by setting thermal resistance. The thermal runaway characteristics are more complex and were set using real experimental data. The established electro-thermal coupling model of the energy storage battery is shown in Figure 1.

The battery terminal voltage $U\left(t\right)$ is defined as the algebraic sum of the open circuit voltage (OCV), the voltage drop across the internal resistance of the battery, and the polarization voltage. According to Kirchhoff’s Voltage Law (KVL), the terminal voltage $U\left(t\right)$ equation can be expressed as:

$$U(t)=U_{\mathrm{oc}}(SOC,T)-I(t)R_0(SOC,T)-U_1(t)$$

(1)

Among them, $SOC$ is the state of charge, $T$ is the battery temperature, $I\left(t\right)$ is the load current (defined as positive for discharge). $U_1\left(t\right)$ represent the polarization voltages of RC parallel branches, and their evolution follows a first-order differential equation:

$${\displaystyle \frac{d{U}_1}{dt}}=-{\displaystyle \frac{1}{R_1{C}_1}}{U}_1+{\displaystyle \frac{1}{C_1}}I(t)$$

(2)

In this formula. All electrical parameters ($U_{\mathrm{o}\mathrm{c}}$, $R_0$, $R_1$, $C_1$) are parameterized as multidimensional nonlinear functions of SOC, T, and current direction, and are interpolated in real time through lookup tables.

The thermal model is established based on the law of energy conservation and is strongly coupled with the electrical model in a bidirectional manner, wherein the temperature field determines the electrical parameters, and the current and polarization voltage determine the heat generation rate. Using the lumped thermal mass method, the battery temperature evolution equation is expressed as follows:

$$C_{\mathrm{th}}{\displaystyle \frac{d{T}_i}{dt}}=P_{\mathrm{gen}}-P_{\mathrm{amb}}-P_{\mathrm{cond}}=P_{\mathrm{gen}}-{\displaystyle \frac{T_i-T_{\mathrm{amb}}}{R_{\mathrm{amb}}}}-{\displaystyle \sum _{j\in \{i-1,i+1\}}\frac{T_i-T_j}{R_{\mathrm{cond}}}}$$

(3)

$C_{\mathrm{t}\mathrm{h}}$ is the equivalent heat capacity, and $P_{\mathrm{a}\mathrm{m}\mathrm{b}}$ and $P_{\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}}$ are the heat dissipation between the battery and the air and the heat transfer between batteries. Since the heat dissipation coefficient is related to the battery position, the two sides are different from the middle. $i$ represents the serial number of a battery within a battery pack. The battery pack studied in this article consists of 13 pieces of 280 Ah LFP batteries, so the value of $i$ ranges from 1 to 13. $T_{\mathrm{a}\mathrm{m}\mathrm{b}}$ is the ambient temperature, $R_{\mathrm{a}\mathrm{m}\mathrm{b}}$ is the thermal resistance between the battery and the air, and $R_{\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}}$ is the thermal resistance between batteries.

The core heat generation term $P_{\mathrm{g}\mathrm{e}\mathrm{n}}$ is based on the extended Bernardi equation, including the Joule heat contributions of the RC circuit:

$$P_{\mathrm{gen}}=I^2{R}_0+{\displaystyle \frac{U_1^2}{R_1}}+P_{\mathrm{fault}}$$

(4)

In Equation (4), the first two terms represent the irreversible Joule heat generated by the ohmic resistance and the polarization process, respectively, and the third term, $P_{\mathrm{f}\mathrm{a}\mathrm{u}\mathrm{l}\mathrm{t}}$ is the additional heat source triggered by a fault. To evaluate battery safety, the model integrates a thermal runaway triggering mechanism based on reaction kinetics. When the local temperature exceeds the critical threshold $T_{\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{g}\mathrm{g}\mathrm{e}\mathrm{r}}$ or an internal short circuit is detected, the exothermic side reaction model is activated. The runaway reaction uses an empirical model established with real experimental data.

The above content explains the equations established in this study that can reflect the thermal characteristic simulation model of a real 280 Ah LFP battery, with the values of the key parameters listed in

Table 1. Key Parameters of Battery Cell.

Parameter Category	Symbol	Typical Value/Range	Unit
Thermal Basics	$C_{\mathrm{t}\mathrm{h}}$	5716.5	J/K
	$R_{\mathrm{a}\mathrm{m}\mathrm{b}}$(i = 2 ~ 12)	6.209	W/K
	$R_{\mathrm{a}\mathrm{m}\mathrm{b}}$(i = 1 or 13)	1.000	W/K
	$R_{\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}}$	0.1249	W/K
	$T_{\mathrm{a}\mathrm{m}\mathrm{b}}$	298.15	K
Electrical Core	$R_0$	$R_0(SOC,T)$	$\mathrm{\Omega }$
	$R_{\mathrm{1,2}}$	$R_1(SOC,T)$	$\mathrm{m}\mathrm{\Omega }$
	${\tau }_{\mathrm{1,2}}$	${\tau }_1(SOC,T)$	s
	$U_{\mathrm{o}\mathrm{c}}$	$U_{\mathrm{o}\mathrm{c}}(SOC,T)$	V
Fault Kinetics	$T_{\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{g}\mathrm{g}\mathrm{e}\mathrm{r}}$	368 (SOC = 100%)	K

.

This study uses MATLAB 2025b software to investigate thermal runaway and thermal propagation phenomena in energy storage batteries. Using the Battery Builder application, a battery pack simulation model including thermal effects was constructed. In the software, a battery pack composed of a set of two series-connected 1P52S battery modules was designed. The individual cells in the constructed battery modules were modeled according to the physical equations described above, and real 280 Ah LFP battery thermal runaway data were used to simulate the thermal runaway characteristics, which are described in detail in Section 2.2.

2.2. 280 Ah LFP Energy Storage Battery Simulation Model

To truly simulate the TR process of a 280 Ah lithium iron phosphate (LFP) battery cell in the simulation model, the TR test data of the 280 Ah LFP battery in reference [20] were adopted. This enables the extraction of heat release rate curves and their corresponding TR trigger temperatures for different SOC states. Thermal runaway tests were conducted on the 280 Ah LFP battery established in Section 2.1 using an Accelerating Rate Calorimeter (ARC) simulation model (as shown in Figure 2a), and the results were compared with the temperature curves measured in experiments reported in the reference [20] (as shown in Figure 2b). The results indicated that the heating process and peak temperature point remained consistent. This demonstrates that this article, through this simulation method, can simulate the TR of large-scale energy storage battery modules based on small-scale test data of TR from a single cell.

Notably, in Figure 2b, the ARC method was used for testing to allow a comparison with the experimental data of thermal runaway of the 280 Ah LFP battery referenced in this study. This is consistent with the experimental method referenced. The battery was heated slowly by a heating plate, and at approximately 3000 s, when the battery reached the thermal runaway trigger temperature, a violent exothermic reaction commenced. When the temperature reached its peak, it indicated that the internal energy release of the battery was complete, and a slow cooling process began. This is different from the rapid reaction process triggered by an internal short circuit, witch will be further explained in Section 2.3.

The trigger temperature range of the 280 Ah lithium iron phosphate battery at 100% and 50%SOC states is derived from the data in reference [20]. The TR trigger temperature of lithium iron phosphate batteries is negatively correlated with the SOC [21,22]. Note that the TR temperature of the battery cell is affected by multiple factors, including cell differences and environmental conditions. Therefore, in actual scenarios, the trigger temperature is not a fixed value but fluctuates within a specific range. To facilitate the simulation study, specific values must be selected during the calculation. The trigger temperature adopted in this study was determined to be within a reasonable range (see

Table 2. TR Triggering Temperatures for Different SOC Batteries.

SOC/%	TR Trigger Temperature/°C
100	95.0
80	126.4
60	158.9
40	192.6
20	240.0

).

2.3. Thermal Runaway Behavior Simulation and CID Protection

This study focuses on the battery module assembly with the 1P52S configuration, which consists of 52 series-connected 280 Ah LFP battery cells. In Section 2.1, we established a battery simulation model with thermal effects. In Section 2.2, we referenced the thermal runaway experimental data of a 280 Ah LFP battery and configured the simulation accordingly, which provided experimental data to support the simulation study. Based on this, a Simulink simulation model for studying the thermal runaway behavior of energy storage battery modules and battery CID protection was established, as shown in Figure 3.

The TR and TP failures occurring within a module composed of 13 independent battery cells (labeled cell1 to cell13) were studied and analyzed. Introduce an ISC fault as the initial fault source in cell7. The ISC fault first triggered TR in cell7 and then led to TR in the remaining 12 batteries within the same parallel module. An internal short-circuit fault was introduced in cell7, which served as the source of the initiating failure. The ISC triggered TR in cell7, subsequently causing TR in the remaining 12 cells within the same parallel assembly.

Since the cell7 battery where the TR occurred was located at the center of the module, with symmetrical cells on both sides, Figure 4 shows the temperature curves for cell1–7. The ISC fault in cell7 occurred at 500 s, thereby initiating an increase in the temperature of cell7. Subsequent to a period of 58 s, the temperature of cell7 increased to 95 °C (368 K), thereby reaching the self-reaction trigger temperature for TR. This development signified the commencement of a process that was subsequently characterized by an uncontrollable escalation.

The heat released during the TR of cell7 was distributed through the thermal resistance of the adjacent cells. After 332 s, cell6 and 8 achieved the designated TR-trigger temperatures. Subsequently, TR was triggered in the remaining cells at intervals of 338, 275, 271, 270, and 244 s. From the t TR of the first battery cell (cell7) to its TP throughout the entire module, the process took 1731 s. The simulated TP rate was aligned with the order of magnitude reported in previous studies.

Figure 5 shows the TR process in cell7. The ISC fault triggered a violent exothermic reaction in cell7 at 500 s. When the battery temperature reaches the TR trigger threshold of 95 °C at 558 s, it marks the initiation of an irreversible TR reaction. Subsequently, the sharp increase in internal temperature and pressure activated the CID, cutting off the external circuit at 576 s. As a result, the entire battery module enters an open circuit state, preventing subsequent charging and discharging processes.

Most existing studies have focused on the battery TR and TP behavior under open-circuit conditions. The open-circuit state can be considered as a charging/discharging rate of 0C. For example, for a 280 Ah battery, 1.0C indicates a charging or discharging current of 280 A. This study improves upon existing simulation methods and successfully establishes a TR and heat propagation simulation model for batteries during the charging and discharging processes, with a particular focus on the accelerated propagation mechanisms of TR during charging.

In addition to the open-circuit condition, two charging conditions were also considered: standard rate charging (0.5C) and maximum rate charging (1.0C). For all three operating conditions, the protection of the BMS was excluded, thus enabling the independent separation and analysis of the TR and TP characteristics of the battery module itself. The detailed data, such as the TR time, TP time, and TP time interval cells of each battery cell under the three operating conditions, are summarized in

Table 3. TR and TP time Statistics Table (SOC = 100%, battery CID protection, three charging rates).

Cell No.	TR Time/s			TP Time/s			TP Time Interval/s
	0C	0.5C	1.0C	0C	0.5C	1.0C	0C	0.5C	1.0C
Cell1	2289	2285	2274	1731	1727	1716	Δ = 244	Δ = 244	Δ = 244
Cell2	2045	2041	2030	1487	1483	1472	Δ = 270	Δ = 270	Δ = 268
Cell3	1775	1771	1762	1217	1213	1204	Δ = 271	Δ = 271	Δ = 270
Cell4	1503	1500	1492	945	942	934	Δ = 275	Δ = 274	Δ = 273
Cell5	1228	1226	1219	670	668	661	Δ = 338	Δ = 337	Δ = 334
Cell6	890	889	885	332	331	327	Δ = 332	Δ = 331	Δ = 327
Cell7	558	558	558	0	0	0	Δ = 0	Δ = 0	Δ = 0
Cell8	890	889	885	332	331	327	Δ = 332	Δ = 331	Δ = 327
Cell9	1228	1226	1219	670	668	661	Δ = 338	Δ = 337	Δ = 334
Cell10	1503	1500	1492	945	942	934	Δ = 275	Δ = 274	Δ = 273
Cell11	1775	1771	1762	1217	1213	1204	Δ = 271	Δ = 271	Δ = 270
Cell12	2045	2041	2030	1487	1483	1472	Δ = 270	Δ = 270	Δ = 268
Cell13	2289	2285	2274	1731	1727	1716	Δ = 244	Δ = 244	Δ = 244

.

The table presents the TR times, TP times (referenced from the TR onset of cell7), and TP time intervals (referenced from the TR onset of the preceding cell) during the TR and TP processes for individual cells within the energy storage battery module under the three distinct operating conditions. Increased charging rates resulted in an earlier TR onset, shorter TP times, and diminished trigger intervals. This suggests that the injection of external energy expedites the TR and TP processes within the energy storage battery module to a certain extent.

3. Research on BMS Control Strategies

3.1. Traditional Threshold BMS Control Strategy

After adding the BMS, the BMS monitored the voltage, current, and temperature parameters of the energy storage battery modules. When an ISC occurred, the current of cell7 battery cells increased, whereas the voltage decreased to a lower range. As the TR progressed, the temperature gradually increased. The BMS established in the simulation model can monitor these behaviors, promptly detect faults, and halt charging and discharging to mitigate or delay the risks of TR and TP caused by the ISC. However, TR caused by ISC will still continue, and traditional threshold-type BMS only cuts off external energy input, but TP may still occur.

Following an ISC and TR in cell7, the BMS protection system required a certain amount of time to activate, owing to sampling frequency limitations under actual conditions, rather than an instant response. Sampling was completed after 1 s, overvoltage and overcurrent were detected after 3 s, overtemperature was detected after 5 s, and a system fault was confirmed after 8 s. The sampling frequency was set at 1 Hz. Confirmation of the voltage, current, and temperature parameters requires 2 s of stable readings to avoid false alarms from transient erroneous signals. The 5 s fault confirmation period further enhances the accuracy of the fault diagnosis system. Although the actual parameters may vary, the underlying principle remains consistent.

Following the simulation model analysis in

Table 4. TP time Statistics Table (SOC = 100%, Comparison of CID and Threshold BMS Protection).

Operating Condition	Start Time/s	End Time/s	Duration/s	Time Advance/s
0C Charging Rate/Battery CID	558	2289	1731	0
0.5C Charging Rate/Battery CID	558	2285	1727	4
1.0C Charging Rate/Battery CID	558	2274	1716	15
0.5C Charging Rate/Threshold BMS	558	2288	1730	1
1.0C Charging Rate/Threshold BMS	558	2287	1729	2

, it was found that under open-circuit conditions, the battery module experienced the longest TR propagation duration (1731 s) owing to the absence of external energy injection. When the battery is under charge/discharge conditions, higher currents result in greater energy injection during the module TR, thereby accelerating the overall TR and TP processes.

However, when the BMS functions normally, it interrupts the charge/discharge current within 8 s after fault detection, thereby reducing the energy injection. Compared with the scenarios of battery CID protection, a healthy BMS extended the TR propagation time of the 280 Ah LFP battery module under charging conditions. At a 0.5C charging rate with BMS protection, the propagation duration increased from 1727 s (CID protection) to 1730 s, representing a 3 s extension. At a 1.0C charging rate with BMS protection, the propagation duration increased from 1716 s (CID protection) to 1729 s, representing a 13 s extension. This demonstrates that the BMS can mitigate TR and TP in energy storage battery modules to a certain extent by promptly interrupting external energy input.

3.2. Active Energy Dissipation BMS Control Strategy

The above analysis demonstrates that a BMS is critical for ensuring the safe operation of energy storage systems and reducing failure risks. The transition from battery CID protection to BMS control and protection has enhanced the safety of energy storage battery modules to a certain extent. However, to further enhance the capability of the BMS to suppress and control TR and heat propagation in battery modules, this study investigated the mechanisms of battery TR and proposed an active energy dissipation BMS control strategy.

Existing research indicates that the risks of TR and TP failures in battery modules decrease at low SOC values. The BMS should comprehensively analyze the TR and TP risks within the battery modules. If the TR risk is detected in any individual cell within the module, it should be capable of suppressing the TR and TP processes by reducing the SOC of the other cells. This strategy configures an energy-feedback electronic load on the outside of the energy storage battery pack, which actively releases the energy of the battery pack through this device when needed. Because the heat generated by the energy dissipation device is not produced by the battery itself, when placing the device, consideration should be given to whether the heat it generates will affect the battery. The advantage of placing the device outside the energy storage battery pack is that it avoids the heat generated by the device during the discharge process from affecting the temperature of the battery itself. This discharge process lowers the SOC, increases the TR temperature threshold, and reduces the risk of TR. However, the discharge process itself causes the battery temperature to rise, which counteracts the reduction in TR risk. Consequently, the relationship between these two factors must be analyzed to determine the optimal discharge current and duration to achieve the best suppression effect. The establishment of control strategies comprises four components: simulation model development, simulation modeling, parameter prediction, and strategy calculation, as shown in Figure 6.

The suppression strategy illustrated in Figure 6 was formulated based on the correlation between the trigger temperature curve and the predicted temperature curve of the battery pack. Specifically, prior to the implementation of the active energy dissipation strategy, the original temperature reached the trigger temperature, resulting in the TR. Upon the application of the active energy dissipation strategy, the battery did not undergo TR, provided that the predicted temperature curve did not intersect with the predicted trigger temperature curve, thereby effectively suppressing the TP. This approach necessitates the prediction of the temperature increase, SOC, and trigger temperature during battery charging and discharging processes.

The derivation process and physical significance of the three formulas are explained below. First, as shown in Figure 7a,b, the temperature increase and SOC reduction in the battery under different charging rates obtained through the established simulation model are presented. Figure 7c,d show the fitting of the simulation data using a quadratic function for the temperature increase rate and a linear function for the SOC change rate. These fitting formulas facilitate the flexible application of the BMS under various operating conditions.

Therefore, Equations (5) and (6) are obtained. Equation (5) provides an integral model for the temperature increase rate of a battery under discharge current:

$$\Delta T={{\displaystyle \int }}_{t_1}^{t_2}{\mathrm{K}}_1{I}^{2}dt$$

(5)

In Equation (5), $∆T$ represents the temperature rise of the battery due to thermal effect under different discharge currents. $t_1$ and $t_2$ respectively represent the start time and the end time when the active discharge suppression strategy is initiated. $I$ is the magnitude of the discharge current. ${\mathrm{K}}_1$ is a comprehensive coefficient that measures the temperature rise caused by the current thermal effect during the discharge process of a battery cell with a certain specific heat capacity. The result of the ${\mathrm{K}}_1$ simulation fitting is 2.49 × 10⁻⁷, with units of $\mathrm{K}·{\mathrm{A}}^2·{\mathrm{s}}^{-1}$.

Equation (6) provides an integral model for the SOC reduction rate under the discharge current:

$$\Delta SOC={{\displaystyle \int }}_{t1}^{t2}{\mathrm{K}}_{2}Idt$$

(6)

In Equation (6), $∆SOC$ represents the change in SOC under different discharge currents (negative during discharge), while the three parameters $t_1$, $t_2$, and $I$ have the same functional meanings as in Equation (5). ${\mathrm{K}}_2$ is also a comprehensive coefficient measuring the rate of $SOC$ decrease at a given discharge current. The fitting result is −9.92 × 10⁻⁵, with units of ${\mathrm{A}}^{-1}·{\mathrm{s}}^{-1}$.

Equation (7) was obtained by interpolating the data in Table 2. The formula for determining the TR trigger temperature at various SOC is as follows:

$$T_{\mathrm{trig}}=f(SOC)$$

(7)

Equation (7) has no specific physical meaning; it represents the temperature values at which irreversible TR occurs in the simulated 280 Ah lithium iron phosphate battery, which was studied between 20% and 100% SOC. These values were fitted using a quadratic function ${y=ax}^2+bx+c$ ($a$ = 5.221 × 10⁻³, $b$ = −2.3690, $c$ = 281.09).

After obtaining the predictive model for key parameters, active energy dissipation can be carried out based on the model. The following explains how to implement active energy dissipation. Within a battery pack, once any individual battery cell experiences an internal short circuit, the BMS can detect it within a short period. At this time, the temperature of the faulty battery is in the rising stage and has not completely lost control. The entire battery pack can then be connected to an energy-feedback type electronic load to actively release energy. The SOC of the batteries adjacent to the faulty battery will decrease, and their thermal runaway trigger temperature will increase. When the SOC drops to a relatively low value, the heat conducted from the faulty battery to nearby cells will be insufficient to trigger thermal runaway, thus suppressing thermal propagation. The following will illustrate this point through specific cases, including the determination of the discharge current magnitude and discharge time.

It is significant to note that battery packs at 100% SOC exhibit the highest energy capacity; consequently, they are also at the greatest risk of TR and TP occurrence. Therefore, this study uses TR and TP at 100% SOC as examples to illustrate the proposed BMS control strategy based on active energy dissipation. The subsequent illustration demonstrates the suppression of TR faults in a battery pack at 100% SOC. This strategy is initiated by the BMS upon detecting a TR in a specific battery cell, such as cell7.

Cell7 experienced TR due to an ISC that could not be suppressed, rendering it unsuitable for this treatment strategy. Therefore, it is necessary to analyze cell6 and 8, which are the battery cells immediately adjacent to cell7 and pose the highest risk of TR. As shown in Figure 8a, starting from the identification of cell7 as defective at 508 s, cell6 and 8 underwent a discharge process at 280 A for a duration of 550 s. The thermal trigger temperature threshold curve and the predicted temperature curve did not intersect, indicating that the battery temperature never reached the thermal runaway trigger temperature. In the simulation model, thermal runaway indeed did not occur. The simulation results confirm that this method effectively suppressed TR in cell6 and cell8, thereby preventing the spread of TR in the battery module.

However, in the three strategies illustrated in Figure 8b–d, owing to the low discharge currents or brief discharge durations, intersection points existed, failing to suppress the TR in cell6 and cell8. cell5 and cell9 and their adjacent cells did not inherently carry a risk of TR occurrence. The susceptibility of these cells was contingent on the presence of neighboring cell6 and cell8. Thus, it can be concluded that the prevention of TR in cell6 and cell8 resulted in effective suppression of TR and TP throughout the entire battery module.

4. Discussion

This study first established a simulation model of a 280 Ah LFP energy storage battery with a 1P52S structure. The parameters of the simulation model are set with reference to the ARC TR experimental data of the 280 Ah lithium iron phosphate battery. The ARC simulation results of the battery model are consistent with the experiment. By studying the TR trigger temperature, it was found that the TR temperature changes at different SOC values. A low TR trigger temperature will increase the risk of TR and TP in the energy storage battery module, while a higher TR trigger temperature can reduce this risk. Therefore, this study proposes an active energy dissipation BMS protection strategy. The proposed strategy was studied and analyzed, along with the battery CID and threshold BMS protection strategies, to investigate the TR and TP processes within the energy storage battery module when a single battery experiences TR due to ISC faults. The suppression capabilities of the three BMS protection schemes for TR and TP were also compared.

Firstly, the direct analysis of the data results indicates that under the 1.0C charging current condition, after TR occurred for 500 s, four parameters were compared: the number of TR batteries, TP time, average TP time interval, and external current injection time. This analysis examined the TR and TP processes under three protection conditions, as well as the inhibitory effects on TR and TP. The summary results are shown in

Table 5. Comparison of Key Parameters for Three Protection Strategies.

Parameter	Battery CID Protection	Traditional Threshold BMS Protection	Active Energy Dissipation BMS Protection
Number of TR batteries	13 cells	13 cells	1 cell
TP time (cell7-cell1)	1716 s	1729 s	No propagation
Average TP time interval	286 s	288.2 s	No propagation
External current injection time	76 s	8 s	8 s
Active energy dissipation time	-	-	550 s
Electric energy variation (E)	+7.66 × 104 J	+8.06 × 103 J	−5.46 × 105 J
Electric charge variation (Q)	+5.911 Ah	+0.622 Ah	−42.156 Ah
SOC variation	-	-	−15.3%
TR trigger temperature variation	-	-	+22.9 °C

. It can be observed that neither the battery CID protection nor the traditional threshold protection scheme has been able to prevent the TR from the short circuit in the cell7 from spreading throughout the entire energy storage battery module. Although the active energy dissipation scheme failed to suppress TR of the cell7, it successfully prevented its diffusion in the whole battery module. Compared with the battery-free CID protection and the traditional threshold protection scheme, the traditional threshold system extends the TP time by 13 s and increases the average TP time by 2.2 s. This is mainly due to the reduction in the external current injection time by 68 s. This mechanism effectively restricts the external energy input after the failure of the energy storage battery module, thereby partially delaying the progress of TR and TP.

Secondly, regarding the internal energy release capacity of the battery, both the CID protection of the battery and the traditional threshold type BMS protection strategy can prevent further injection of external energy after an ISC in battery cell7 leads to TR. Among them, CID protection is achieved through internal interrupts, while the BMS actively cuts off the main circuit breaker. However, neither of these two methods can reduce the internal energy. The active energy dissipation scheme improves the traditional threshold method by adjusting the fault state strategy after blocking the injection of external energy. This scheme adopts the method shown in Figure 8 to calculate the amplitude and duration of the discharge current. Passive balancing is achieved by paralleling discharge resistors at the battery terminals at the same time. This discharge process is not affected by the internal CID of the battery circuit, enabling the cell7 to continuously discharge to other battery cells even after CID protection. The active energy dissipation BMS protection strategy lasts for 280 A discharge for 550 s under these conditions. The energy of the non-faulty battery decreases by 5.46 × 10⁵ J, and the SOC drops by 15.3%. The TR trigger temperature rose from 95°C to 117.9°C, thereby suppressing the TR.

Based on the above results, it can be further discussed that when the TR phenomenon occurs in a single cell within an energy storage battery module, the cells directly adjacent to it are at the greatest risk of experiencing TR. If these adjacent cells do not undergo TR and their temperatures are maintained within a low range, TR will not propagate to the other cells. This indicates that if BMS can promptly identify battery cells with a high risk of TR and control the energy or temperature of the adjacent battery cells, it is possible to prevent the spread of TR within the battery module.

The method proposed in Section 3.2 can determine the specific discharge parameters for implementing active energy dissipation suppression strategies for batteries at risk of TR. When the discharge current is too low or the discharge duration is too short, the battery remains in a high-energy state, and the TR trigger temperature is relatively low. Excessively high current or overly long discharge duration can generate a large amount of heat, which may instead trigger TR. Therefore, the discharge current and duration in the suppression strategy must operate within the specified maximum and minimum values. The TR trigger temperature fluctuates under actual working conditions. This strategy should prioritize maintaining the discharge parameters near the midpoint of the range rather than approaching the boundary value of the range.

5. Conclusions

The main objective of this study is to investigate the behavior of energy storage battery modules integrated with BMS during TR and TP events, and to explore methods to suppress TR and TP by optimizing the BMS control strategy. Then a simulation model of the 1P52S 280 Ah LFP battery was established. This model can operate under charge and discharge conditions and reflect the electrothermal characteristics. Finally, an innovative TR and TP suppression strategy based on active energy dissipation BMS control and parameter determination method was proposed, and its performance was compared with two traditional battery TR protection strategies. The conclusions are as follows:

(1)

The trigger temperature of battery TR varies with SOC. When the SOC is 100%, the trigger temperature is below 100 °C. However, when the SOC drops to 50%, the trigger temperature exceeds 150 °C, thereby increasing the probability of TR occurrence and related risks. Conversely, a lower SOC can reduce the risks of TR and TP in battery modules.

(2)

BMS is crucial in alleviating TR and TP within energy storage battery modules. When a single battery undergoes TR at a charging rate of 1.0C, its average TP time is 2.2 s longer than that of the battery CID protection. This indicates that when a single cell triggers a TR due to ISC, a fully functional and normally operating BMS can to some extent delay the propagation of TR within the module.

(3)

The BMS control strategy based on active energy dissipation can suppress TP through active discharge. The SOC was reduced by 15.3%, the energy was decreased by 5.46 × 10⁵ J, and the trigger temperature was increased by 22.9 °C. This method was verified through the simulation model developed in this study.