Lithium battery thermal runaway gas production experimental device and testing method

Abstract

This application provides an experimental apparatus and testing method for thermal runaway gas generation in lithium batteries, relating to the field of lithium battery safety performance testing technology. The apparatus includes: an adjustable-capacity sealed container, the internal volume of which can be adjusted via removable filling material, with a light-transmitting window and sampling interface on the side wall; a reduction-sealed sphere, connected to the container via a pipeline, equipped with a heating module, a temperature probe, and a gas injection interface; a gas buffer tank, connected to the container's sampling interface via a sampling pipeline with a control valve, and connected to the reduction-sealed sphere via an outlet pipeline, used for temporary storage and gas delivery; a data acquisition system, including a Raman spectroscopy analysis device, a temperature sensor, and a pressure sensor; and a control and processing unit. This apparatus can synchronously and with high precision monitor the temperature, pressure, and gas component concentration during the thermal runaway process of lithium batteries, and supports real-time sampling and subsequent explosion verification, providing an efficient and reliable experimental platform for lithium battery safety assessment.

Description

Technical Field

[0001] This application relates to the field of lithium battery safety performance testing technology, and in particular to a lithium battery thermal runaway gas generation experimental device and testing method. Background Technology

[0002] With the widespread application of lithium batteries in the new energy field, combustion and explosion accidents caused by thermal runaway are frequent. Accurately obtaining multi-physics parameters (temperature, pressure, gas concentration) and combustion and explosion characteristics during thermal runaway is a core prerequisite for optimizing battery safety design. Existing technologies have the following key shortcomings: 1) Limitations in monitoring dimensions: For example, the in-situ testing device and method for battery charging and discharging gas generation based on a Raman probe disclosed in patent publication number CN117288740A only achieves in-situ gas detection through a Raman probe and does not collect pressure and temperature data simultaneously; the combustible gas collection method and device in lithium battery testing disclosed in patent publication number CN116202830A adopts segmented gas collection, which cannot capture the dynamic changes in concentration in real time, resulting in the lack of multi-parameter coupling analysis.

[0003] 2) Limited scenario simulation: Existing testing devices, such as the optical probe of a closed-space lithium battery thermal runaway gas real-time spectral monitoring system disclosed in patent publication number CN119827478A, have a fixed volume and cannot simulate thermal runaway environments of different spatial scales, such as electric vehicle battery packs (50-100L) and energy storage containers (150-200L). The test scenarios deviate greatly from actual applications.

[0004] 3) Insufficient reduction capability: For example, a quantitative assessment method for thermal runaway of lithium batteries disclosed in patent publication number CN119644160A can only assess the intensity of thermal runaway. Existing technologies mostly only reduce gas components and concentrations, and do not achieve the reconstruction of combustion and explosion scenarios in coordination of temperature, pressure and concentration. Furthermore, it is impossible to sample and verify in real time during the test process, and it is impossible to trace the explosion characteristics of gases at different stages.

[0005] 4) Data loop failure: Due to the lack of a complete system of real-time monitoring → real-time sampling → software simulation → physical reconstruction → model correction, the simulation results generally have an error of more than 15% compared with the real scene. For example, the battery charging and discharging gas generation in-situ test device and test method based on Raman probe disclosed in patent publication number CN117288740A has core formulas that are mostly single parameter correlations, ignoring multi-field coupling effects. The predicted value deviates from the measured value by more than 99%, which is completely divorced from engineering reality.

[0006] Therefore, there is an urgent need to develop an experimental device with adjustable volume, multi-parameter synchronous monitoring, real-time sampling verification, scene restoration and high-precision model correction capabilities to fill the gap in existing technology. Summary of the Invention

[0007] This application provides an experimental apparatus and testing method for thermal runaway gas generation in lithium batteries. The core objective is to achieve high-precision synchronous monitoring of temperature, pressure, and gas component concentration during thermal runaway (data timestamp synchronization error ≤1ms). It also supports real-time sampling, temporary storage, and subsequent explosion verification of the gas within the container during testing to trace gas characteristics at different stages. Through an adjustable capacity design, it can simulate thermal runaway environments with spatial scales ranging from 50 to 200L, improving the realism of the testing scenario and ensuring compatibility with various battery types, including ternary lithium and lithium iron phosphate. Ultimately, by constructing a complete closed-loop system of "monitoring-sampling-simulation-reduction-correction," the prediction error of combustion and explosion characteristics is controlled within 5%, thereby meeting the precise testing needs of different application scenarios such as electric vehicles and energy storage systems.

[0008] In a first aspect, this application provides an experimental apparatus for thermal runaway gas generation in lithium batteries, comprising: An adjustable-volume sealed container, the internal volume of which is adjusted by a removable filler, has a light-transmitting window and a sampling interface on its side wall; The reducing sealing sphere is connected to the adjustable-capacity sealing container via a pipeline. The reducing sealing sphere is equipped with a heating module, a temperature probe, and a gas injection interface. A gas buffer tank is connected to the sampling interface of the adjustable-volume sealed container via a sampling pipeline with a control valve, and to the reducing sealed sphere via an outlet pipeline, for temporarily storing and transporting the collected gas. The data acquisition system includes a Raman spectroscopy analyzer for detecting the concentration of gas components, a temperature sensor for detecting temperature, a pressure sensor for detecting pressure, and a data acquisition card. The control processing unit is configured to control the data acquisition system to perform synchronous data acquisition, control the sampling operation of the gas buffer tank, and run a multi-field coupled simulation model to predict the peak combustion and explosion pressure based on the acquired data.

[0009] In one possible design, the multi-field coupling simulation model includes: The gas generation source term model is used to calculate the total number of moles of gas and the total number of moles of combustible gas in the container; A gas mixing and spatial distribution model is used to introduce a complete mixing coefficient to correct the effective combustible mole number, and a volume correction function is used to adapt to the effects of different volumes. The combustion kinetics model employs a first-order global reaction mechanism, calculates the combustion progress and final combustion fraction based on the Arrhenius law, and calculates the total energy released during combustion. Thermodynamic and pressure response models are used to calculate the equilibrium temperature after combustion based on energy conservation, and to calculate the real-time pressure and peak explosion pressure according to the ideal gas law. The engineered closed-loop prediction module predicts peak pressure based on a physical quantity-driven closed-loop formula and includes a residual correction term.

[0010] In one possible design, in the gas-generating source term model, the total number of moles of gas in the container at time t is... Estimate using the following formula: in, For real-time initial pressure measurement inside the container, For real-time initial temperature measurement, The effective volume of the container. This is the universal gas constant; Total number of moles of combustible gas in the container at time t Calculated using the following formula: in, It is a collection of combustible gas components; For the first i Real-time mole fraction of a gas.

[0011] In one possible design, a perfect mixing coefficient is introduced into the gas mixing and spatial distribution model. The total number of moles of combustible gas in the container is corrected to obtain the effective number of combustible moles. : in, This represents the total number of moles of combustible gas in the container. The following continuously differentiable volume correction function is used. To adapt to the impact of different volumes on pressure response: in, The reference volume used for experimental calibration. The volumetric influence factor, This refers to the effective volume of the container.

[0012] In one possible design, in the combustion dynamics model, the combustion progress variable The dynamic evolution equation is: in, For combustion progress, For time, The rate constant is a temperature-dependent reaction and follows Arrhenius's law: in, Pre-exponential factor, The activation energy of the reaction. For real-time temperature, This is the universal gas constant. It is an exponential function with the natural constant as its base; Final combustion score Calculated using the following formula: in, This refers to the combustion termination time; Total chemical energy released by combustion Calculated using the following formula: in, For the first The molar calorific value of a combustible gas For the first i The number of moles of a type of combustible gas.

[0013] In one possible design, the equilibrium temperature after gas combustion is [value missing] in the thermodynamic and pressure response model. Calculated using the following energy conservation formula: in, This represents the total number of moles of gas inside the container. The molar specific heat capacity at constant volume of the gas mixture. The initial temperature. This refers to the heat loss between the system and the external environment during combustion. The total chemical energy released by combustion; The heat exchange loss between the system and the outside environment during the combustion process Calculated using the convective heat transfer formula: in, The convective heat transfer coefficient is... This refers to the heat exchange area of ​​the container's inner wall. This refers to the temperature of the container wall. Real-time pressure inside the container during combustion Calculated using the following formula: in, Let t be the total number of moles of gas in the container at time t. This is the universal gas constant. For real-time temperature, The effective volume of the container; Peak explosion pressure The maximum pressure value during the monitoring period: Where max represents the maximum value.

[0014] In one possible design, the physical quantity-driven closed-loop pressure prediction formula in the engineered closed-loop prediction module is: in, This is the predicted pressure value. The adiabatic index of the gas mixture is . , For isobaric molar specific heat capacity, This refers to the molar specific heat capacity at constant volume. For combustion efficiency; The average molar calorific value of the combustible component. , For the first i mole fraction of the combustible gas For the first i The molar calorific value of a combustible gas; Initial pressure; The initial temperature; This represents the total number of moles of gas in the container. The effective volume of the container; This is the residual correction term; The residual correction term Generated by a lightweight residual neural network, its expression is: in, It is a residual neural network. This is a set of mole fractions for each gaseous component. For time; The training objective of the lightweight residual neural network is to minimize the predicted value of the physical model. Compared with the measured pressure value The deviation is given by the objective function: in, These are network parameters.

[0015] In one possible design, the multi-field coupled simulation model also includes an online Bayesian adaptive calibration mechanism for iteratively correcting the key parameter set of the model, with the iterative formula being: in, This is the measured pressure value. These are the predicted values ​​from the physical model. Including the coefficient of complete mixing Combustion efficiency Volume Influence Factor Pre-exponential factors Activation energy of reaction At least one of them, The Kalman gain matrix is ​​adaptively calculated based on the real-time error statistics. This is the set of key parameters after calibration.

[0016] Secondly, this application provides a method for experimentally testing the thermal runaway gas generation of a lithium battery using a device as described in the first aspect and its various possible designs, characterized by comprising the following steps: S1: Adjust the volume of the adjustable-capacity sealed container by installing or removing filler, install the lithium battery sample, conduct a sealing test, and connect the gas buffer tank; S2: Set data acquisition parameters, sampling trigger conditions and sampling quantity, and calibrate each sensor; S3: Trigger thermal runaway and simultaneously collect temperature, pressure and gas concentration data; control the gas buffer tank to sample and temporarily store gas according to the triggering conditions; S4: Input the collected data into the multi-field coupled simulation model to generate combustion and explosion characteristic curves and predict pressure peaks; S5: Inject the gas temporarily stored in the gas buffer tank into the reduced sealed sphere to reproduce the environmental parameters at the sampling time and conduct an explosion verification experiment. Based on the verification results, correct the parameters of the multi-field coupling simulation model. S6: Generate a test report containing the original data, sampling records, simulation and experimental comparison results, and the corrected model.

[0017] In one possible design, the explosion verification experiment is conducted in at least three parallel sets; the model parameters are corrected using a combination of least squares fitting and online Bayesian updating, with the objective function being: in, The set of parameters to be fitted, including the coefficients of a perfect mixture. Combustion efficiency Volume Influence Factor Pre-exponential factors Activation energy of reaction At least one of them; For the first k The measured pressure value of this experiment; Based on the set of parameters to be fitted The k Predicted pressure values ​​for this experiment.

[0018] The experimental apparatus and testing method for lithium battery thermal runaway gas generation provided in this application have at least the following beneficial effects: 1) This application integrates Raman spectroscopy, distributed temperature sensors and pressure sensors, and combines them with a high sampling rate data acquisition system to achieve synchronous real-time monitoring of gas component concentration, temperature and pressure during thermal runaway. The data timestamp synchronization error is ≤1ms, which significantly improves the consistency and reliability of the data and provides a precise data foundation for in-depth analysis of multi-physics coupling mechanisms.

[0019] 2) The 200L main container adopts a removable filling design, and the effective volume can be continuously adjusted between 50L and 200L. It can accurately simulate closed space environments of different scales, from electric vehicle battery packs to energy storage containers, which greatly improves the coverage and realism of test scenarios and overcomes the limitations of existing devices with fixed volume and single scenario.

[0020] 3) This application, by adding a 5L gas buffer tank and a matching rapid sampling valve, supports manual or automatic sampling at any point during the thermal runaway process, and temporarily stores the gas sample completely. Subsequently, the environmental parameters (temperature, pressure, concentration) at the time of sampling can be accurately reproduced in the 5L reconstructed sealed sphere for explosion verification experiments. This function makes it possible to trace and analyze the gas explosion characteristics at different stages of thermal runaway, a key capability not possessed by existing technologies.

[0021] 4) By establishing a multi-field coupled physical model integrating gas production, mixing, combustion, and thermodynamics, and introducing a physics-data hybrid residual compensation neural network and an online Bayesian adaptive calibration mechanism, a complete technical closed loop was established, encompassing real-time monitoring, real-time sampling, software simulation, physical reconstruction, and model correction. This system reduced the prediction error of peak combustion and explosion pressure from over 15% using traditional methods to less than 5%, and improved the model's prediction robustness across battery types and container volumes by over 85%, significantly enhancing the accuracy of safety assessments and its engineering guidance value.

[0022] 5) The entire device is controlled by integrated control software, which realizes full-process automated management of parameter setting, data acquisition, process monitoring, automatic sampling, verification triggering and report generation, greatly reducing the complexity of operation. A single person can complete the entire test process and improve test efficiency. Attached Figure Description

[0023] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0024] Figure 1 A structural diagram of a lithium battery thermal runaway gas generation experimental device provided in this application embodiment; Figure 2 A flowchart of an experimental test method for thermal runaway gas generation in a lithium battery, provided as an embodiment of this application.

[0025] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this application to those skilled in the art through reference to specific embodiments. Detailed Implementation

[0026] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0027] The collection, storage, use, processing, transmission, provision, and disclosure of financial data or user data involved in the technical solution of this application all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0028] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0029] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0030] Example 1: This application provides an experimental apparatus for thermal runaway gas generation in lithium batteries, such as... Figure 1As shown, the lithium battery thermal runaway gas generation experimental device includes an adjustable-capacity sealed container 100, a reduction-sealed sphere 200, a gas buffer tank 300, a data acquisition system 400, and a control and processing unit 500. The internal volume of the adjustable-capacity sealed container 100 is adjusted by a removable filling material. The side wall of the adjustable-capacity sealed container 100 is provided with a light-transmitting window and a sampling interface. The reduction-sealed sphere 200 is connected to the adjustable-capacity sealed container 100 via a pipeline and is equipped with a heating module, a temperature probe, and a gas injection interface. The gas buffer tank 300 is connected to the adjustable-capacity sealed container 100 via a control system 500. The valve's sampling line is connected to the sampling interface of the adjustable-capacity sealed container 100 and to the reducing sealed sphere via the gas outlet line, for temporary storage and transport of the collected gas; the data acquisition system 400 includes a Raman spectroscopy analysis device for detecting gas component concentration, a temperature sensor for detecting temperature, a pressure sensor for detecting pressure, and a data acquisition card; the control processing unit 500 is configured to control the data acquisition system to perform synchronous data acquisition, control the sampling operation of the gas buffer tank, and run a multi-field coupled simulation model to predict the peak combustion and explosion pressure based on the acquired data.

[0031] In some embodiments, the adjustable-capacity sealed container 100 has a volume of 200L, and its main body is made of 316L stainless steel in one piece, with a design pressure resistance ≥5MPa. Eight M16 bolt fixing holes (50mm spacing, 25mm depth, 6H thread precision) are evenly distributed around its inner wall. Three quartz glass windows (100mm diameter, ≥15mm thickness, ≥95% light transmittance) are formed along the height direction (1 / 4, 1 / 2, 3 / 4) on the side wall of the adjustable-capacity sealed container 100. These windows are sealed to a tenon-and-mortise flange (5mm depth) using fluororubber U-shaped sealing rings (section: inner diameter 100mm × outer diameter 120mm × height 10mm), with a sealing pressure ≥0.8MPa. One to five sets of honeycomb alumina fillers (specifications: ...) can be installed inside the adjustable-capacity sealed container 100. 50mm×100mm or The adjustable-capacity sealed container 100 (100mm×200mm) has mounting ears with matching bolt holes on the outer wall of the filler, which can be detached and fixed, and the effective volume can be adjusted to 50-200L; the lower part of the side wall (at 1 / 4 height) of the adjustable-capacity sealed container 100 is provided with a DN10 sampling interface, equipped with a quick connector and a one-way valve to prevent gas leakage.

[0032] In some embodiments, the volume of the reduction sealing sphere 200 is 5L, its main body is made of stainless steel in one piece, it is designed to withstand a pressure of ≥20MPa, has an outer diameter of 220mm and a wall thickness of 15mm; the top of the reduction sealing sphere 200 is provided with a DN15 gas injection port (equipped with a one-way valve, including...) The device includes a 6mm×50mm stainless steel needle and a DN10 pressure regulating valve (connected to a 0-10MPa precision pressure gauge, accuracy 0.02MPa). A new DN10 sampling verification interface is added (connected to a 5L gas buffer tank). The bottom of the reduction sealing sphere 200 has a built-in annular micro heating module (power 500W, temperature control range -50~800℃, accuracy ±0.5℃), wrapped with an Al2O3 ceramic insulation layer on the outside, and a PT1000 temperature feedback probe (accuracy ±0.1℃) in the middle of the inner wall. It can form a closed-loop temperature control with the control software configured in the control processing unit 500. The reduction sealing sphere 200 is connected to a 200L container (adjustable volume sealing container 100) through a DN25 stainless steel pipeline (wall thickness 3mm), and a solenoid valve (response time ≤0.5s) is installed on the pipeline.

[0033] In some embodiments, the gas buffer tank 300 has a volume of 5L, and its main body is made of 316L stainless steel in one piece. It is designed to withstand a pressure of ≥10MPa and is equipped with a pressure monitoring gauge (range 0-10MPa, accuracy 0.1%FS). The top of the gas buffer tank 300 is equipped with a DN10 inlet port (connected to the sampling port of a 200L container via a DN10 stainless steel sampling pipeline) and a DN10 outlet port (connected to the sampling verification port of a 5L reduction sealing sphere via a pipeline). The control valve of the gas buffer tank 300 is an electric sampling valve on the sampling pipeline (response time ≤0.3s, supports manual / automatic control). The outlet of the buffer tank is equipped with a shut-off valve to ensure the gas storage is sealed. The outer wall of the gas buffer tank 300 is wrapped with a 5mm ceramic insulation layer to prevent sudden temperature changes of the stored gas from affecting its characteristics.

[0034] In some embodiments, the data acquisition system 400 includes a detection room and an acquisition system, wherein the detection components are shown in Table 1.

[0035] Table 1 Parameters and Installation Locations of Detection Components

[0036] The core components of the data acquisition system are a multi-channel data acquisition card (sampling frequency ≥1000Hz, 16-bit resolution), a fiber optic transmission module (transmission rate 1000Mbps), and an industrial host computer (CPU i7-12700K, 32GB memory). Temperature / pressure sensors and a pressure gauge in the buffer tank are connected to the analog interface of the acquisition card via shielded cables with a temperature resistance of ≥200℃. The Raman spectroscopy device is connected to the host computer's Ethernet interface via the fiber optic module. The electric sampling valve is connected to the digital output interface of the acquisition card via a control cable. The data timestamp synchronization error is ≤1ms, and the sampling action is forcibly associated with the corresponding T / P / C parameters in the storage. In some embodiments, the control processing unit 500 is configured with a real-time monitoring module, a sampling control module, and a multi-field coupled simulation model. The real-time monitoring module's functions include visually displaying temperature (layered curves), pressure (real-time values ​​and dynamic curves for a 200L container + buffer tank), gas concentration (bar charts and trend lines for CO, H2, CH4, etc.), and sampling status (sampling in progress / temporary storage / verification in progress). The real-time monitoring module's data storage supports CSV / SQL format switching, allows setting storage intervals from 10ms to 1s, and automatically marks sampling data with timestamps and corresponding T / P / C parameters.

[0037] The control functions that the sampling control module can implement include: 1) Sampling trigger: Supports manual triggering (click the "Sampling" button on the host computer) or automatic triggering (set the sampling time point / pressure threshold / temperature threshold, such as automatically sampling when the pressure rises to 1MPa). 2) Sampling parameter settings: The single sampling volume can be set (100mL-1L), and the electric sampling valve automatically controls the opening and closing time according to the sampling volume; 3) Temporary storage management: Displays the pressure and remaining volume of the buffer tank, supports multiple sampling and stacking of a single tank (up to 5 times, total sampling volume ≤ 4L) or clearing and resetting; 4) Verification Control: Click the “Explosion Verification” button to automatically inject the gas in the buffer tank into the 5L restored sealed sphere and start the verification process.

[0038] The multi-field coupling simulation model is one of the core components of this application, used to predict the peak explosion pressure of lithium-ion battery thermal runaway gas generation in a closed or semi-closed space. The model decomposes the real process into five interpretable and computable core sub-modules and an adaptive calibration system. Specifically, the multi-field coupling simulation model includes: The gas generation source term model is used to calculate the total number of moles of gas and the total number of moles of combustible gas in the container; A gas mixing and spatial distribution model is used to introduce a complete mixing coefficient to correct the effective combustible mole number, and a volume correction function is used to adapt to the effects of different volumes. The combustion kinetics model employs a first-order global reaction mechanism, calculates the combustion progress and final combustion fraction based on the Arrhenius law, and calculates the total energy released during combustion. Thermodynamic and pressure response models are used to calculate the equilibrium temperature after combustion based on energy conservation, and to calculate the real-time pressure and peak explosion pressure according to the ideal gas law. The engineered closed-loop prediction module predicts peak pressure based on a physical quantity-driven closed-loop formula and includes a residual correction term.

[0039] In this embodiment, the multi-field coupled simulation model is equipped with a physical-data hybrid residual compensation and an online Bayesian calibration mechanism. With mole number conservation and energy conservation as the core, it achieves accurate prediction of pressure peaks across volumes and operating conditions, solving the technical problems of poor generalization and high error of traditional models.

[0040] In some embodiments, the gas generation source term model obtains the real-time mole fraction of each gas component in the container based on raw monitoring data such as Raman spectra and pressure-temperature collected during the thermal runaway process. Total number of moles of gas in the container Estimated from the ideal gas law: in, The initial pressure (Pa) inside the container is measured in real time. For real-time initial temperature measurement (K). The effective volume of the container ( ), The universal gas constant ( The set of combustible gas components is defined as follows: The total number of moles of combustible species in the container for: in, For the first i Real-time mole fraction of a gas.

[0041] In some embodiments, to accurately characterize the degree of non-uniform mixing of gases inside a closed container, a complete mixing coefficient is introduced into the gas mixing and spatial distribution model. Its value range is The specific values ​​were obtained by inversion from the gas concentration profile data measured by the three-point Raman optical path. Indicates complete mixing. (Indicates complete layering).

[0042] Effective combustible moles corrected based on the perfect mixing coefficient for: To address the impact of different volumes on gas mixing and pressure response, a continuously differentiable volume correction function is employed. To avoid the discontinuity of traditional piecewise functions, its expression is: in, The reference volume for experimental calibration ( ), The volumetric influence factor was determined from multi-volume operating condition experimental data. This refers to the effective volume of the container.

[0043] In some embodiments, the combustion kinetics model employs a first-order global reaction mechanism to characterize the combustion process of combustible gases and defines a combustion progress variable. ( Indicates that it was not burned. (Indicating complete combustion), its kinetic evolution equation is: in, The rate constant is a temperature-dependent reaction and follows Arrhenius's law: in, Pre-exponential factor ( ), The activation energy of the reaction ( ), The real-time temperature (K) is given. It is an exponential function with the natural constant as its base. Integrating over the combustion progress variable yields the final combustion fraction: in, The combustion termination time is 10 seconds.

[0044] Total chemical energy released after combustion Calculated from the number of moles and molar calorific value of the combustible components: in, For the first The molar calorific value of a combustible gas ( ), For the first i The number of moles (mol) of a type of combustible gas.

[0045] In some embodiments, the equilibrium temperature after gas combustion in the thermodynamic and pressure response model Derived from the law of conservation of energy, the initial internal energy of the system, the energy released by combustion, and the heat transfer losses need to be considered. Its expression is: in, This represents the total number of moles of gas inside the container. The molar specific heat capacity at constant volume of a gas mixture ( ), The initial temperature (K) is given. The heat loss (J) between the system and the surrounding environment during combustion can be calculated using the convective heat transfer formula: in, The convective heat transfer coefficient ( ), The heat exchange area of ​​the inner wall of the container ( ), The container wall temperature (K); Accurate compensation can also be achieved through data-driven residual learning models. The real-time pressure inside the container during combustion is calculated using the ideal gas law: Define peak explosion pressure The maximum pressure value during the monitoring period: Where max represents the maximum value.

[0046] In some embodiments, to adapt to the real-time computing needs of embedded systems and online control systems, the engineered closed-loop prediction module applies the physical quantity-driven closed-loop pressure prediction formula derived in this application, avoiding the limitations of traditional empirical formulas. Its expression is: in, This is the predicted pressure value. The adiabatic index of the gas mixture is . , For isobaric molar specific heat capacity, This refers to the molar specific heat capacity at constant volume. Combustion efficiency is the ratio of actual energy released to theoretical energy released. The average molar calorific value of the combustible component ( ), , For the first i mole fraction of the combustible gas For the first i The molar calorific value of a combustible gas; Initial pressure; The initial temperature; This represents the total number of moles of gas in the container. The effective volume of the container; This is the residual correction term, generated in real time by the data-driven compensation model.

[0047] The above closed-loop pressure prediction formula is dimensionally self-consistent and does not require the introduction of a magnitude correction factor, enabling millisecond-level pressure peak prediction.

[0048] In some embodiments, to address unmodelable errors in combustion dynamics modeling such as turbulence disturbances, local mixing inhomogeneities, and complex heat transfer, this invention constructs a lightweight residual neural network (ResNet) to accurately compensate for physical model errors. The expression for the residual correction term is: in, It is a residual neural network. This is a set of mole fractions for each gaseous component. For time; the training objective of this residual network is to minimize the deviation between the physical model's predictions and the measured values, i.e.: in, For network parameters, This is the measured pressure value. These are the predicted values ​​from the physical model. The model employs an offline pre-training + online fine-tuning mode: in the offline stage, pre-training is completed using experimental data from multiple operating conditions; during online operation, network parameters are dynamically adjusted based on real-time sampling verification data to achieve adaptive fusion of the physical model and the data model, reducing the model's basic error by more than 60%.

[0049] In some embodiments, considering the system differences in container volume, structure, material, and cell type, this invention employs a Bayesian update algorithm to perform online calibration of key model parameters. The parameter iteration formula is as follows: in, ,Include , , , , At least one of them, The Kalman gain matrix is ​​adaptively calculated based on the real-time error statistics. This is the set of key parameters after calibration.

[0050] This calibration mechanism enables a closed-loop iteration of sampling test → 5L volume restoration verification → parameter correction, which improves the model's prediction robustness by more than 85% in cross-cell and cross-container scenarios, solving the technical problem of poor adaptability of traditional models to operating conditions.

[0051] Example 2: This application provides a method for testing the thermal runaway gas generation of a lithium battery using the device described in Example 1. Figure 2 As shown, the test method includes the following steps S1-S6.

[0052] S1: Adjust the volume of the adjustable-capacity sealed container by installing or removing filler, install the lithium battery sample, conduct a sealing test, and connect the gas buffer tank.

[0053] S2: Set data acquisition parameters, sampling trigger conditions and sampling quantity, and calibrate each sensor.

[0054] S3: Triggers thermal runaway and simultaneously collects temperature, pressure and gas concentration data. Based on the triggering conditions, it controls the gas buffer tank to sample and temporarily store the gas.

[0055] S4: Input the collected data into the multi-field coupled simulation model to generate combustion and explosion characteristic curves and predict pressure peaks.

[0056] S5: Inject the gas temporarily stored in the gas buffer tank into the reduced sealed sphere to reproduce the environmental parameters at the sampling time and conduct an explosion verification experiment. Based on the verification results, correct the parameters of the multi-field coupling simulation model.

[0057] S6: Generate a test report containing the original data, sampling records, simulation and experimental comparison results, and the corrected model.

[0058] In one specific embodiment, the specific operation content and key parameters / requirements of the above steps S1-S6 are shown in Table 2.

[0059] Table 2 Core Operation Procedures for Lithium-ion Battery Thermal Runaway Gas Generation Experimental Test Method

[0060] Example 3: The embodiments in this application are experimental verification cases based on Embodiment 2, in order to fully illustrate the feasibility and progressiveness of this application.

[0061] In this embodiment, the device assembly details are as follows: 200L adjustable volume sealed container: inner diameter 500mm, height 1019mm, top equipped with a 300mm quick-opening hatch (manual locking mechanism), bolt fixing holes are evenly distributed along the inner wall circumference (spaced 50mm), sampling interface is located at 1 / 4 height of the side wall, 255mm from the bottom; 5L gas buffer tank: outer diameter 220mm, wall thickness 12mm, insulation layer made of ceramic fiber, electric sampling valve model is two-position three-way solenoid valve, working pressure 0-10MPa, response time 0.2s; 5L Restored Sealed Sphere: The newly added sampling and verification interface is located in the middle of the side of the sphere and is connected to the outlet pipeline of the buffer tank via a quick connector, making disassembly and reassembly convenient; Raman spectroscopy device: fiber optic length 3m (temperature resistance ≥80℃), collimating lens fixed by bracket, coaxiality error between the lens and the center of the light transmission window ≤2mm.

[0062] Based on the assembled device, verification is carried out through the following steps.

[0063] Step 1: Determine the basic experimental conditions.

[0064] Effective volume of container: ; Initial ambient temperature (measured): ; Initial pressure of the container (actual measurement): ; The mole fractions of gas components (Raman spectral inversion, example) are shown in Table 3.

[0065] Table 3. Mole fraction of gas components

[0066] Step 2: Model calculation in steps.

[0067] Step 2.1: Calculate the total number of moles of gas in the container.

[0068] According to the ideal gas law, the formula for calculating the total number of moles is: Substitute parameters ( Perform an example calculation: .

[0069] Step 2.2: Calculate the total number of moles of combustible gas.

[0070] Define the set of combustible components The total number of moles of combustible gas is: Substitute the data into the calculation: .

[0071] Step 2.3: Correction of effective combustible molar number.

[0072] The perfect mixing coefficient was obtained by inversion of the concentration profile of the three-point Raman optical path. The effective number of combustible moles is: ;Calculation yields: .

[0073] Step 2.4: Calculate combustion kinetic parameters and final combustion fraction.

[0074] The first-order global reaction kinetic equation is used, and its reaction rate constant follows the Arrhenius law. Key kinetic parameters, including the pre-exponential factor, are obtained through experimental fitting. ,activation energy .

[0075] Integrating the combustion progress variable yields the final combustion score: Calculated .

[0076] Step 2.5: Calculate the total energy released by combustion.

[0077] The standard molar calorific value of each combustible component is shown in Table 4 below: Table 4 Standard Molar Calorific Value of Each Combustible Component

[0078] The average molar calorific value of the combustible component is: The formula for calculating the total energy released by combustion is: Substitute the data into the calculation: .

[0079] Step 2.6: Calculate the equilibrium temperature after combustion.

[0080] According to the law of conservation of energy, the temperature equation considering heat transfer losses is: Take the molar specific heat capacity of the gas at constant volume Heat exchange loss Substituting the data, we get: .

[0081] Step 2.7: Predicting peak pressure.

[0082] The final peak pressure is calculated using the ideal gas law: Substituting the parameters, we get: Step 3: Compare and verify the predicted values ​​with the measured values.

[0083] In the 5L container experiment, the measured peak pressure value was... Then the relative error of the model prediction is: The prediction error is less than ±5%, which meets the accuracy requirements for engineering safety early warning and explosion intensity assessment.

[0084] The key parameters of the model are calibrated using a combination of least squares fitting and online Bayesian update. The fitting objective function is: Among them, the set of parameters to be fitted .

[0085] After completing the offline least squares fitting, and combining the 5L restoration verification experimental data, the parameters were iteratively optimized online using the Bayesian update formula. in, The Kalman gain matrix is ​​adaptively generated from the real-time error statistics.

[0086] The optimal values ​​of the core parameters of the model were obtained by fitting three sets of thermal runaway experimental data at different pressure levels, as shown in Table 5 below.

[0087] Table 5 Optimal values ​​for the core parameters of the model

[0088] Step 4: Multi-condition error analysis.

[0089] Three typical thermal runaway experimental conditions—low pressure, medium pressure, and high pressure—were selected to verify the model's prediction accuracy. The error analysis results are shown in Table 6 below.

[0090] Table 6 Error Analysis Results

[0091] Error statistics show that the average prediction error for the three operating conditions is 2.93%, which is significantly better than the traditional empirical coupling model (10%–25% prediction error). Moreover, the error for all operating conditions is controlled within ±5%, which can be directly applied to engineering scenarios such as lithium battery thermal runaway explosion pressure prediction, safety relief device design, and energy storage compartment fire prevention.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. An experimental apparatus for thermal runaway gas generation in lithium batteries, characterized in that, include: An adjustable-volume sealed container, the internal volume of which is adjusted by a removable filler, has a light-transmitting window and a sampling interface on its side wall; The reducing sealing sphere is connected to the adjustable-capacity sealing container via a pipeline. The reducing sealing sphere is equipped with a heating module, a temperature probe, and a gas injection interface. A gas buffer tank is connected to the sampling interface of the adjustable-volume sealed container via a sampling pipeline with a control valve, and to the reducing sealed sphere via an outlet pipeline, for temporarily storing and transporting the collected gas. The data acquisition system includes a Raman spectroscopy analyzer for detecting the concentration of gas components, a temperature sensor for detecting temperature, a pressure sensor for detecting pressure, and a data acquisition card. The control processing unit is configured to control the data acquisition system to perform synchronous data acquisition, control the sampling operation of the gas buffer tank, and run a multi-field coupled simulation model to predict the peak combustion and explosion pressure based on the acquired data.

2. The lithium battery thermal runaway gas generation experimental apparatus according to claim 1, characterized in that, The multi-field coupling simulation model includes: The gas generation source term model is used to calculate the total number of moles of gas and the total number of moles of combustible gas in the container; A gas mixing and spatial distribution model is used to introduce a complete mixing coefficient to correct the effective combustible mole number, and a volume correction function is used to adapt to the effects of different volumes. The combustion kinetics model employs a first-order global reaction mechanism, calculates the combustion progress and final combustion fraction based on the Arrhenius law, and calculates the total energy released during combustion. Thermodynamic and pressure response models are used to calculate the equilibrium temperature after combustion based on energy conservation, and to calculate the real-time pressure and peak explosion pressure according to the ideal gas law. The engineered closed-loop prediction module predicts peak pressure based on a physical quantity-driven closed-loop formula and includes a residual correction term.

3. The lithium battery thermal runaway gas generation experimental apparatus according to claim 2, characterized in that, In the gas-generating source term model, the total number of moles of gas in the container at time t is... Estimate using the following formula: in, For real-time initial pressure measurement inside the container, For real-time initial temperature measurement, The effective volume of the container. This is the universal gas constant; Total number of moles of combustible gas in the container at time t Calculated using the following formula: in, It is a collection of combustible gas components; For the first i Real-time mole fraction of a gas.

4. The lithium battery thermal runaway gas generation experimental apparatus according to claim 2, characterized in that, In the gas mixing and spatial distribution model, a perfect mixing coefficient is introduced. The total number of moles of combustible gas in the container is corrected to obtain the effective number of combustible moles. : in, This represents the total number of moles of combustible gas in the container. The following continuously differentiable volume correction function is used. To adapt to the impact of different volumes on pressure response: in, The reference volume used for experimental calibration. The volumetric influence factor, This refers to the effective volume of the container.

5. The lithium battery thermal runaway gas generation experimental apparatus according to claim 2, characterized in that, In the combustion dynamics model, the combustion progress variable The dynamic evolution equation is: in, For combustion progress, For time, The rate constant is a temperature-dependent reaction and follows Arrhenius's law: in, Pre-exponential factor, The activation energy of the reaction. For real-time temperature, This is the universal gas constant. It is an exponential function with the natural constant as its base; Final combustion score Calculated using the following formula: in, This refers to the combustion termination time; Total chemical energy released by combustion Calculated using the following formula: in, For the first The molar calorific value of a combustible gas For the first i The number of moles of a type of combustible gas.

6. The lithium battery thermal runaway gas generation experimental apparatus according to claim 2, characterized in that, In the aforementioned thermodynamic and pressure response model, the equilibrium temperature after gas combustion Calculated using the following energy conservation formula: in, This represents the total number of moles of gas inside the container. The molar specific heat capacity at constant volume of the gas mixture. The initial temperature. This refers to the heat loss between the system and the external environment during combustion. The total chemical energy released by combustion; The heat exchange loss between the system and the outside environment during the combustion process Calculated using the convective heat transfer formula: in, The convective heat transfer coefficient is... This refers to the heat exchange area of ​​the container's inner wall. This refers to the temperature of the container wall. Real-time pressure inside the container during combustion Calculated using the following formula: in, Let t be the total number of moles of gas in the container at time t. This is the universal gas constant. For real-time temperature, The effective volume of the container; Peak explosion pressure The maximum pressure value during the monitoring period: Where max represents the maximum value.

7. The lithium battery thermal runaway gas generation experimental apparatus according to claim 2, characterized in that, In the engineered closed-loop prediction module, the physical quantity-driven closed-loop pressure prediction formula is: in, This is the predicted pressure value. The adiabatic index of the gas mixture is . , The molar specific heat capacity at constant pressure, This refers to the molar specific heat capacity at constant volume. For combustion efficiency; The average molar calorific value of the combustible component. , For the first i mole fraction of the combustible gas For the first i The molar calorific value of a combustible gas; Initial pressure; The initial temperature; This represents the total number of moles of gas in the container. The effective volume of the container; This is the residual correction term; The residual correction term Generated by a lightweight residual neural network, its expression is: in, It is a residual neural network. It is a set of mole fractions of each gas component. For time; The training objective of the lightweight residual neural network is to minimize the predicted value of the physical model. Compared with the measured pressure value The deviation is given by the objective function: in, These are network parameters.

8. The experimental apparatus for thermal runaway gas generation in lithium batteries according to any one of claims 1 to 7, characterized in that, The multi-field coupled simulation model also includes an online Bayesian adaptive calibration mechanism for iteratively correcting the key parameter set of the model. The iterative formula is as follows: in, This is the measured pressure value. These are the predicted values ​​from the physical model. Including the coefficient of complete mixing Combustion efficiency Volume Influence Factor Pre-exponential factors Activation energy of reaction At least one of them, The Kalman gain matrix is ​​adaptively calculated based on the real-time error statistics. This is the set of key parameters after calibration.

9. A method for experimentally testing the thermal runaway gas generation of a lithium battery using the device described in any one of claims 1 to 8, characterized in that, Includes the following steps: S1: Adjust the volume of the adjustable-capacity sealed container by installing or removing filler, install the lithium battery sample, conduct a sealing test, and connect the gas buffer tank; S2: Set data acquisition parameters, sampling trigger conditions and sampling quantity, and calibrate each sensor; S3: Trigger thermal runaway and simultaneously collect temperature, pressure and gas concentration data; control the gas buffer tank to sample and temporarily store gas according to the triggering conditions; S4: Input the collected data into the multi-field coupled simulation model to generate combustion and explosion characteristic curves and predict pressure peaks; S5: Inject the gas temporarily stored in the gas buffer tank into the reduced sealed sphere to reproduce the environmental parameters at the sampling time and conduct an explosion verification experiment. Based on the verification results, correct the parameters of the multi-field coupling simulation model. S6: Generate a test report containing the original data, sampling records, simulation and experimental comparison results, and the corrected model.

10. The experimental test method for thermal runaway gas generation in lithium batteries according to claim 9, characterized in that, The explosion verification experiment was conducted in at least three parallel sets; the model parameters were corrected using a combination of least squares fitting and online Bayesian updating, with the objective function being: in, The set of parameters to be fitted, including the coefficients of a perfect mixture. Combustion efficiency Volume Influence Factor Pre-exponential factors Activation energy of reaction At least one of them; For the first k The measured pressure value of this experiment; Based on the set of parameters to be fitted The k Predicted pressure values ​​for this experiment.

Images