Fire prediction and risk assessment method for new energy vehicles based on Fluent multi-physical field coupling

Abstract

This invention discloses a method for predicting and assessing fire risks in new energy vehicles based on Fluent multiphysics coupling, relating to the fields of fire science and public safety technology. The method includes: obtaining multi-mode heat flux distribution on the surfaces of adjacent vehicles from the fire source vehicle using multiphysics coupling numerical simulation; calculating the heat accumulation of adjacent vehicles and determining ignition and predicting ignition time using a dynamic correction model; conducting ignition risk classification assessment; converting the ignited adjacent vehicle into a dynamic fire source, and extrapolating the multi-vehicle chain spread process based on a time-series extrapolation algorithm and a dynamic fire source iteration mechanism; and outputting the full-cycle fire prediction results and risk assessment report. This invention balances the accuracy of ignition determination with the efficiency of chain spread prediction, solving the problem that high-precision simulation cannot meet the needs of real-time decision-making.

Description

Technical Field

[0001] This invention belongs to the field of fire science and public safety technology, specifically relating to a method for predicting and assessing fire risks in new energy vehicles based on Fluent multiphysics coupling. Background Technology

[0002] The ignition process of a neighboring vehicle in a car fire is a complex multi-physics coupled process involving multiple modes of heat transfer, including thermal radiation, convection, and conduction, as well as chemical reactions related to material pyrolysis and combustion. Currently, numerical simulation of fires is the primary technique for studying the spread of car fires, with commonly used software including FDS and Fluent. However, existing technologies suffer from the following core technical bottlenecks: First, the ignition criteria are too simplistic. Current technologies typically use a single critical heat flux criterion to determine whether a neighboring vehicle has been ignited, such as simply setting the threshold at 10... As an ignition threshold, the current method falls short. However, the actual ignition process is a complex heat accumulation process, with significant differences in the time required for ignition at different heat flux levels. Furthermore, the critical ignition parameters for different automotive components vary significantly, making a single criterion insufficient to accurately reflect the actual ignition risk. Second, there is a lack of multi-factor coupling correction. Existing technologies fail to adequately consider various factors influencing the ignition process. For example, the breakage of car windows under thermal effects significantly enhances convective heat transfer, the jet fire generated by thermal runaway of electric vehicle batteries produces directional high-intensity thermal radiation, and environmental factors such as wind speed and temperature also significantly impact the heat transfer process. The absence of these factors leads to significant deviations between the prediction results of existing methods and actual conditions. Third, there is insufficient consideration of vehicle model differences. Existing technologies fail to distinguish between the fire behaviors of gasoline-powered vehicles and electric vehicles. Electric vehicle battery thermal runaway fires differ significantly from traditional gasoline vehicle fires in terms of heat release rate, flame morphology, and thermal radiation intensity. A general judgment method cannot accurately assess the ignition risk under different vehicle model combinations. Fourth, the computational cost of full-domain simulation is extremely high. Existing technologies, if used to recreate the complete spread of a car fire and the entire process of multi-vehicle chain combustion using full-process transient numerical simulation, require extremely high computing power and take a very long time to calculate, which cannot meet the timeliness requirements of rapid early warning and real-time decision-making in fire emergencies. At the same time, existing technologies can only realize the judgment of single adjacent vehicle ignition and the prediction of ignition time, and cannot quickly extrapolate the subsequent fire trend and the multi-vehicle chain spread pattern based on short-term ignition simulation results, making it difficult to support full-process fire emergency decision-making.

[0003] Therefore, there is an urgent need to develop a method for judging the ignition and predicting the rapid spread of automobile fires that balances computational accuracy and efficiency. This method would combine short-term, high-precision numerical simulation of ignition with rapid extrapolation of time sequence in the later stages to solve the technical problems of insufficient accuracy, lack of spread prediction, and inability to balance computational power and timeliness in existing technologies. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a method for predicting and assessing fire risks in new energy vehicles based on Fluent multiphysics coupling, thereby achieving high-precision and high-efficiency dynamic simulation of the fire chain spread process within the monitoring area.

[0005] This invention provides a method for predicting and assessing the risk of fires in new energy vehicles based on Fluent multiphysics coupling, comprising the following steps:

[0006] Based on multiphysics field coupled numerical simulation, the multimode heat flux distribution of the fire source vehicle on the surface of the adjacent vehicle is obtained. The multimode heat flux distribution includes heat radiation flux, heat convection flux and heat conduction flux.

[0007] Based on the multi-mode heat flux distribution, the heat accumulation of the neighboring vehicle is calculated using a dynamic correction model, and the heat accumulation is used to determine whether the neighboring vehicle has been ignited and to predict the ignition time.

[0008] Based on the predicted ignition time, the adjacent vehicle is assessed for ignition risk classification.

[0009] The neighboring vehicles that will be ignited are transformed into dynamic fire sources, and the multi-vehicle chain spread process is deduced based on the time-series extrapolation algorithm and the dynamic fire source iteration mechanism;

[0010] Output fire full-cycle prediction results and risk assessment reports.

[0011] Before obtaining the multi-mode heat flux distribution of the fire source vehicle on the surface of the adjacent vehicle, the method further includes:

[0012] Extract the 3D geometric models of the fire source vehicle and adjacent vehicles, and model the key ignition components separately;

[0013] An unstructured hybrid mesh is used to divide the computational domain. The mesh is refined in the core area of ​​flame development and the vehicle surface area, while the mesh size gradually increases in the outer area of ​​the flame.

[0014] All mesh elements are inspected to ensure that the following mesh quality requirements are met: mesh skewness is no greater than 0.85, with more than 95% of the meshes having a skewness of less than 0.7, and the mesh skewness in critical areas being less than 0.6; mesh aspect ratio is no greater than 5, and the aspect ratio of the boundary layer mesh is no more than 10; orthogonality quality is no less than 0.3; there are 0 negative volume meshes.

[0015] Mesh independence verification was carried out: three mesh schemes with different densities—coarse, medium, and fine mesh—were generated, and steady-state thermal radiation calculations were performed under the same conditions. The average heat flux and maximum temperature of the adjacent vehicle surface were extracted as verification indicators. When the mesh was refined from medium to fine mesh, if the change in the verification indicators was less than 5%, the medium mesh was determined as the final calculation mesh scheme.

[0016] This preferred approach ensures the reliability and repeatability of numerical simulation results through rigorous mesh quality control and independence verification processes, eliminating the impact of mesh density on the calculation results.

[0017] Before obtaining the multi-mode heat flux distribution of the fire source vehicle on the surface of adjacent vehicles, the method also includes setting differentiated heat release rate curves for different vehicle models:

[0018] The ignition source model for gasoline-powered vehicles includes the growth stage, the stable combustion stage, and the decay stage.

[0019] The fire source model for ternary lithium battery electric vehicles includes the preheating stage, thermal runaway stage, jet fire stage, stable combustion stage, and decay stage.

[0020] The fire source model for lithium iron phosphate battery electric vehicles includes the preheating stage, thermal runaway stage, stable combustion stage, and decay stage.

[0021] The parameters of the heat release rate curves for different vehicle models can be determined based on test data, standard fire curves, or historical accident data, and then loaded into the calculation model as dynamic heat sources.

[0022] The heat release rate curve is imported as a volumetric heat source term using Fluent's DEFINE_PROFILE and DEFINE_SOURCE macros.

[0023] For different vehicle components, corresponding material thermophysical properties and ignition critical parameters are set, including density. Specific heat capacity thermal conductivity Critical ignition heat flux Critical ignition time The critical ignition heat flux of the tire is 16.5. The critical ignition time is 300 seconds, and the critical ignition heat flux of the bumper is 17.5 kilowatts. The critical ignition time is 280s, and the critical ignition heat flux of the interior materials is 10.0. The critical ignition time is 200s.

[0024] This preferred solution enables differentiated characterization of fire behavior in three types of vehicles: gasoline vehicles, ternary lithium battery electric vehicles, and lithium iron phosphate battery electric vehicles. It overcomes the shortcomings of existing technologies that make general judgments and cannot distinguish the ignition risk of different vehicle types.

[0025] The multiphysics coupled numerical simulation uses the following physical model and solver configuration:

[0026] Call RNG k- The turbulence model simulates the turbulent flow of fire smoke, with buoyancy effect correction enabled;

[0027] The DO radiation model is used to calculate the heat radiation transfer during a fire, supporting the radiation transmission effect of semi-transparent media such as glass.

[0028] The combustion model is invoked to simulate the chemical reaction process of fuel combustion;

[0029] By calling the conjugate heat transfer model and using coupled wall boundary conditions, the convective and radiative heat transfer between the fluid and the solid wall, as well as the heat conduction process inside the solid, are automatically solved.

[0030] A pressure-based transient solver was used, and the PISO algorithm was selected for the pressure-velocity coupling algorithm. All variables were in second-order precision schemes, with the gradient term using the nodal basis Green-Gaussian gradient algorithm, the pressure term using the PRESTO! scheme, and the momentum, turbulent kinetic energy, turbulent dissipation rate, component transport, and energy equations all using second-order upwind schemes.

[0031] An adaptive time step is used, with a recommended time step size between 0.0001 and 0.1 s. The convergence criterion adopts a dual standard combining the residual criterion and the monitoring point criterion. The residual convergence threshold for the continuity equation, momentum equation, turbulent kinetic energy equation, turbulent dissipation rate equation, and component transport equation is set to 1e-3, while the residual convergence threshold for the energy equation and radiation equation is set to 1e-6. The calculation of a time step is considered converged when the change amplitude of the temperature monitoring point is less than 0.1 K for 10 consecutive time steps.

[0032] Boundary condition settings include: a pressure inlet boundary is used for gas inflow, with a total pressure of 101325 Pa, an initial temperature of 300 K, and a turbulence intensity of 5%; a pressure outlet boundary is used for gas outflow, with a static pressure of 101325 Pa, a return temperature of 300 K, and the return component set to standard air; a no-slip wall boundary is used for solid boundaries, with an initial temperature of 300 K and wall emissivity set according to material properties; the wall surface in contact with the fluid is automatically identified as a coupled wall surface, and the energy conservation calculation of the fluid-solid interface is automatically completed by the conjugate heat transfer model; the combustion zone of the fire source vehicle is added to the energy equation and component transport equation in the form of a volumetric heat source or a surface heat source through a UDF macro;

[0033] The Fluent DEFINE_EXECUTE_AT_END macro outputs the temperature, thermal radiation flux, thermal convection flux, and thermal conduction flux of each grid cell on the surface of the adjacent vehicle in real time, with a time interval of 1 second.

[0034] This preferred scheme achieves accurate simulation of turbulent flow, heat radiation transfer, combustion chemical reaction and fluid-structure interaction heat transfer processes in fire scenarios through the coordinated configuration of multi-physics coupling models, providing high-fidelity heat transfer benchmark data for subsequent ignition judgment.

[0035] The step of calculating the heat accumulation of the neighboring vehicle based on the multi-mode heat flux distribution and combined with the dynamic correction model, and determining whether the neighboring vehicle has been ignited and predicting the ignition time based on the heat accumulation, includes:

[0036] Calculate the combined heat flux of each grid cell on the surface of the adjacent vehicle. The expression is:

[0037]

[0038] in, The thermal radiation flux is obtained by solving the DO radiation model, and the calculation formula is:

[0039]

[0040] in The intensity of incident radiation on the wall. For wall emissivity, The Stefan-Boltzmann constant is... The wall temperature;

[0041] For heat convection flux, given by RNG k- The turbulence model, combined with the standard wall function, yields the following calculation formula:

[0042]

[0043] in The convective heat transfer coefficient is... The near-wall flue gas temperature; the convective heat flux of the specified surface element is directly extracted using the Fluent built-in macro F_FLUX_CONV;

[0044] For heat conduction flux;

[0045] Based on the theory of thermal accumulation, the thermal accumulation of each grid cell is calculated. The expression is:

[0046]

[0047] in, Heat loss flux per unit area mainly includes heat dissipation to the environment and heat conduction within the material;

[0048] For different components, determine their critical heat accumulation threshold. The expression is:

[0049]

[0050] in, This is the critical ignition heat flux for this component. This is the ignition time of the component at the critical heat flux.

[0051] When the heat accumulation of a certain grid cell When the location of the component corresponding to the grid cell is determined to be ignited, and when any location of any component of a neighboring vehicle is determined to be ignited, the entire neighboring vehicle is determined to be ignited, the corresponding time is determined. That is, predicting the ignition time. .

[0052] This preferred scheme constructs a three-dimensional comprehensive ignition criterion of "heat flux-heat accumulation-component differentiation". Based on the principle of energy conservation, it quantifies the net energy absorbed by the material through time integration, which can accurately characterize the time difference required for the material to reach the ignition state under different heat flux levels, and avoids the problem of misjudging the ignition risk based solely on the instantaneous peak value.

[0053] The dynamic correction model includes a glass breakage correction model, a jet fire correction model, and an environmental correction model:

[0054] The glass breakage correction model includes: calculating the temperature distribution on the inner and outer surfaces of the vehicle window glass in real time using Fluent's solid temperature field calculation results; and determining when the highest temperature on the glass surface reaches the critical breakage temperature. At that time, it was determined that the glass had broken; before the glass broke, the convective heat transfer coefficient of the car window glass was... The natural convection heat transfer coefficient is adopted, and its value is [value missing]. After the glass breaks, the convective heat transfer coefficient of the adjacent vehicle surface is dynamically modified using UDF, expressed as:

[0055]

[0056] in, The convection enhancement factor for glass breakage is 1.5-2.5;

[0057] The jet fire correction model includes: when the fire source vehicle is a ternary lithium battery electric vehicle and is in the jet fire stage, the directional thermal enhancement effect of the jet fire on the surface of the adjacent vehicle is calculated based on the jet fire direction, jet fire duration and relative position relationship of the adjacent vehicle.

[0058] Establish the correction factor for directional thermal radiation of jet fire The thermal radiation received by the surface of adjacent vehicles is dynamically corrected to characterize the enhancing effect of thermal runaway jet fire on the local thermal exposure intensity. The expression is as follows:

[0059]

[0060] in, The maximum thermal radiation enhancement coefficient; This is the distance from the center of the jet fire to the surface of the adjacent vehicle, in meters (m). The characteristic attenuation distance; The angle between the direction of the jet fire and the normal to the surface of the adjacent vehicle is expressed in rad; the corrected thermal flux is:

[0061]

[0062] The environmental correction model includes: obtaining environmental wind speed. Ambient temperature and ambient humidity Parameters; adjust flame tilt angle according to ambient wind speed. The expression is:

[0063]

[0064] in, The vertical upward velocity of the flame is denoted as , and its value is . Correct the initial temperature of neighboring vehicles based on ambient temperature. And correct the temperature difference for heat transfer. Correct the air's radiation absorption coefficient based on ambient humidity. The expression is:

[0065]

[0066] in, The radiation absorption coefficient of dry air is taken as a value of .

[0067] This preferred scheme overcomes the shortcomings of traditional static boundary conditions in failing to reflect the sudden changes in heat transfer coefficient and heat source morphology during fire evolution by introducing three dynamic correction mechanisms, making the calculation of heat flux more consistent with the actual physical scenario.

[0068] The assessment of the ignition risk level of the adjacent vehicle includes:

[0069] Based on the predicted ignition time The risk of ignition from neighboring vehicles is divided into four levels:

[0070] Safety level: Ignition time is greater than 60 minutes or it will not be ignited, and no special emergency measures are required;

[0071] Alert Level: Ignition time greater than 30 minutes but not more than 60 minutes, requires enhanced monitoring and emergency preparedness;

[0072] Hazard level: If the ignition time is greater than 10 minutes but not more than 30 minutes, personnel must be evacuated immediately and preparations made for firefighting.

[0073] Extremely dangerous level: Ignition time is no more than 10 minutes. All personnel must be evacuated immediately and the source of fire must be dealt with first.

[0074] Generates a visualized ignition risk cloud map, displaying the ignition time distribution at different locations on the surface of adjacent vehicles using different colors; outputs an ignition risk assessment report, including simulation parameter settings, heat flux distribution, heat accumulation process, ignition time prediction, risk level, and corresponding emergency response recommendations.

[0075] This preferred approach transforms quantitative forecasts into qualitative risk levels that emergency decision-makers can directly use, significantly shortening the decision-making chain from risk identification to emergency response.

[0076] The process of converting adjacent vehicles into dynamic fire sources and simulating the chain reaction of fire spread across multiple vehicles based on a time-series extrapolation algorithm and a dynamic fire source iteration mechanism includes:

[0077] When any adjacent vehicle is identified as ignited, the chain spread prediction process is automatically triggered, which dynamically transforms the ignited vehicle into a secondary fire source. Based on its vehicle type, ignition location and ignition time, a unique dynamic combustion sequence is generated, and the volume heat source item is updated in real time through the UDF interface to achieve multi-fire source coupling loading.

[0078] The vehicle is designed with a 5-15 minute transition period from local component ignition to full vehicle combustion, while inheriting the different combustion characteristics of different vehicle models: after ignition, the fuel vehicle exhibits the standard oil pool fire combustion characteristics; after ignition, the ternary lithium battery electric vehicle automatically triggers the jet fire model, and the jet direction is determined according to the battery pack installation position and the direction of thermal runaway pressure relief; after ignition, the lithium iron phosphate battery electric vehicle maintains a slow and stable combustion characteristic.

[0079] The fire source failure mechanism is set according to the standard combustion time of each vehicle type: 20-30 minutes for fuel vehicles, 30-40 minutes for ternary lithium battery electric vehicles, and 40-60 minutes for lithium iron phosphate battery electric vehicles. After the time is reached, the heat release rate decays to 0 according to an exponential law, and the fire source is marked as extinguished, thus completing the dynamic iteration of the fire source throughout its entire cycle.

[0080] This preferred scheme constructs a dynamic iterative mechanism for the entire fire source cycle, automatically handling the generation, development, and extinguishing of fire sources. This enables the time-series extrapolation process to adaptively reflect the dynamic changes in fire load, ensuring the authenticity of secondary fire source behavior in chain spread simulations.

[0081] The method for extrapolating the multi-vehicle cascading propagation process based on the time-series extrapolation algorithm also includes:

[0082] Based on the updated list of multiple fire sources, instead of performing full-process transient CFD iterative simulation, we rely on the fire evolution patterns obtained from previous short-term simulations, combined with a time-series evolution extrapolation algorithm, and introduce a multi-fire source heat flux superposition mechanism and adaptive correction strategy to quickly extrapolate the entire fire spread process.

[0083] The total heat flux of any grid cell on the surface of an adjacent vehicle is the sum of the heat radiation, heat convection and heat conduction fluxes of all active fire sources.

[0084] The grid adaptive dynamic control mechanism is enabled. When a new fire source is generated, the grid around its 3m radius is densified by 50%, and the grid in the area where the fire source is extinguished is automatically coarsened.

[0085] A dynamic computational domain pruning strategy is adopted, with the effective computational domain extending 5m outward from the outer envelope of all active fire sources as the center, and areas with no impact are eliminated.

[0086] Set iteration termination criteria, and stop simulation when any of the following conditions are met: 1) No new ignition vehicles are added in the computational domain and all active fire sources are completely extinguished; 2) The simulation duration reaches the maximum emergency decision window of 120 minutes; 3) The fire spreads to the boundary of the computational domain and there is no continued spread trend.

[0087] This preferred scheme replaces the time-consuming flow field iteration solution with a multi-heat source heat flux superposition mechanism, and concentrates limited computing power on key areas with active fires through grid adaptive control and computational domain pruning, thereby achieving an order-of-magnitude improvement in computational efficiency while ensuring prediction accuracy.

[0088] The simulation of multi-vehicle cascading fire spread also includes constructing a full-cycle fire evolution database, extracting dynamic characteristics of the entire fire area in real time every second, including:

[0089] Chain ignition timing characteristics: record the ignition time, peak heat release rate, and complete extinguishing time for each vehicle to generate a multi-vehicle chain ignition timeline and fire spread path map;

[0090] Thermal field spread characteristics: Real-time output of global temperature field and thermal radiation flux field cloud map, and statistical analysis of high-temperature danger areas ( ), ignition hazard heat radiation area ( The real-time area and outward expansion rate of ).

[0091] Flue gas diffusion characteristics: Combining turbulence models and component transport equations, the flue gas spread trajectory, settling height, spatial visibility distribution, and CO2 concentrations are deduced. The toxic gas concentration field can accurately predict the time required to seal off evacuation routes and safety exits with smoke.

[0092] Key area early warning features: Mark key facilities and areas such as fire lanes, power distribution rooms, equipment rooms, and evacuation assembly points to predict the arrival time of fire and smoke.

[0093] This optimal solution, through multi-dimensional feature extraction, achieves a leap from judging a single ignition to predicting the entire fire cycle, providing a comprehensive quantitative scientific basis for fire emergency decision-making.

[0094] This invention discloses a method for predicting and assessing fire risks in new energy vehicles based on Fluent multiphysics coupling. First, by constructing a comprehensive ignition criterion of "multi-mode heat flux coupling + dynamic correction + thermal accumulation integral," it overcomes the shortcomings of traditional single critical heat flux criterion which ignores thermal inertia and dynamic environmental factors. This method can accurately capture the ignition timing under complex conditions such as glass breakage and directional spraying, significantly improving the accuracy of risk identification. Second, through the refined configuration of vehicle-specific heat release rate curves and material thermal property parameters, it achieves differentiated characterization of fire behavior between fuel vehicles and different types of electric vehicles. Third, it innovatively constructs a segmented prediction strategy of "short-term high-precision numerical simulation in the early stage + rapid extrapolation in the later stage." After obtaining reliable initial thermal field characteristics, it replaces the entire transient CFD iteration with a multi-heat source superposition and grid adaptive control mechanism. While ensuring the physical rationality of chain propagation prediction, it significantly reduces computational resource consumption and effectively solves the industry problem that high-precision numerical simulation cannot meet the timeliness requirements of emergency decision-making. Finally, by constructing a fire cycle evolution database, rapid prediction and dynamic early warning of the entire process of fire trend, multi-vehicle chain spread, smoke diffusion and regional risk evolution were achieved, providing scientific numerical calculation support for fire protection design of parking garages, fire risk assessment and fire emergency decision-making. Attached Figure Description

[0095] Figure 1 This is a flowchart of the method for predicting and assessing the chain ignition of automobile fires based on multi-physics field coupling, according to an embodiment of the present invention.

[0096] Figure 2 This is a schematic diagram of the layout of multiple vehicles arranged in an enclosed space and the vehicle catching fire in the center, according to an embodiment of the present invention. Detailed Implementation

[0097] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0098] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0099] like Figure 1As shown, this embodiment provides a method for predicting the cascading ignition of car fires based on multiphysics coupling, applicable to enclosed or semi-enclosed locations where cars congregate, such as car carriers, underground parking lots, urban tunnels, and multi-level parking garages. This method solves the technical problems of high computational consumption and low accuracy of single threshold criteria in traditional full-process numerical simulation by constructing a segmented prediction architecture that combines short-term, high-precision numerical simulation in the early stage with rapid extrapolation in the later stage. Specifically, the method includes the following steps.

[0100] Based on multi-physics coupled numerical simulation, the multi-mode heat flux distribution of the fire source vehicle on the surface of the adjacent vehicle is obtained. The multi-mode heat flux distribution includes heat radiation flux, heat convection flux and heat conduction flux.

[0101] In this embodiment, the multiphysics coupled numerical simulation is implemented based on the Fluent platform. By calling the synergistic coupling of turbulence, radiation, combustion, and conjugate heat transfer models, transient numerical simulations of the development process of a car fire within the computational domain are performed, and the temperature field, thermal radiation flux, thermal convection flux, and thermal conduction flux distribution of adjacent vehicle surfaces are obtained in real time at different times. The fire source vehicle refers to the vehicle on fire, and adjacent vehicles refer to vehicles around the fire source vehicle that may be ignited. Figure 2 As shown, in a typical enclosed space scenario, multiple cars are parked in a row, with the car in the center being the source of the fire, and the surrounding cars being neighboring cars. The multi-mode heat flux distribution encompasses the spatial distribution and time series data of the three heat transfer mechanisms—thermal radiation, thermal convection, and thermal conduction—on the surfaces of the neighboring cars.

[0102] Based on the multi-mode heat flux distribution, combined with the dynamic correction model, the heat accumulation of neighboring vehicles is calculated, and the heat accumulation is used to determine whether the neighboring vehicles are ignited and to predict the ignition time.

[0103] In this embodiment, the dynamic correction model is used to compensate for the drastic impact of sudden changes in physical state on the heat transfer process during fire evolution, including enhanced convection due to glass breakage, enhanced radiation due to battery jet fire, and the influence of environmental factors on heat transfer. The heat accumulation is calculated by subtracting heat loss from the material's output to the environment through time integration, yielding the net energy actually absorbed by the material and used for heating, pyrolysis, or phase change. Compared to traditional instantaneous heat flux threshold judgment, heat accumulation considers the material's thermal inertia and heat absorption history, effectively distinguishing the differentiated contributions of short-term high thermal shock and long-term low-temperature baking to ignition risk.

[0104] Based on the predicted ignition time, the ignition risk of adjacent vehicles is assessed.

[0105] In this embodiment, the ignition risk classification assessment transforms the quantitative prediction results into qualitative risk levels that can be directly used by emergency decision-makers. The system classifies each neighboring vehicle into different risk levels according to a preset time threshold range based on the predicted ignition time of each vehicle, and automatically generates a complete assessment report that includes a visualized risk cloud map and graded disposal recommendations.

[0106] The neighboring vehicles that will be ignited are transformed into dynamic fire sources, and the multi-vehicle chain spread process is deduced based on the time-series extrapolation algorithm and the dynamic fire source iteration mechanism.

[0107] In this embodiment, the temporal extrapolation algorithm is essentially a rapid evolutionary calculation method based on physical laws. Its core logic lies in: after halting the time-consuming full-process transient CFD iterative simulation, utilizing the baseline heat transfer laws obtained from previous short-term simulations, combined with the real-time heat release parameters of the dynamic fire source, it rapidly calculates the comprehensive heat flux distribution on the surfaces of adjacent vehicles at the next moment through the principle of superimposing heat flux from multiple heat sources. The dynamic fire source iteration mechanism automatically handles the entire process of fire source generation, development, and extinguishing, realizing the dynamic extrapolation of multi-vehicle chain spread.

[0108] Output fire full-cycle prediction results and risk assessment reports.

[0109] In this embodiment, the prediction results cover the ignition sequence of multiple vehicles, the thermal field spread path, the characteristics of smoke diffusion, and the risk classification results, providing a comprehensive quantitative scientific basis for fire emergency decision-making.

[0110] The 3D geometric models of the fire source vehicle and adjacent vehicles are extracted, and key ignition components such as the body, tires, windows, bumpers, and interior are modeled separately. The main geometric features affecting heat transfer are retained, while non-critical details are reasonably simplified. An unstructured hybrid mesh is used to divide the computational region. The mesh is refined in the core area of ​​flame development and the vehicle surface area to ensure that the calculation does not diverge due to mesh issues. The mesh size in the outer area of ​​the flame gradually increases to balance computational accuracy and efficiency.

[0111] Fluent's built-in mesh quality inspection tool was used to perform a comprehensive inspection of all mesh cells to ensure that the following requirements were met:

[0112] Grid skewness: Skewness is a core indicator for measuring the degree of mesh deformation. A skewness greater than 0.85 will significantly increase the error in solving discrete equations, and may even cause computational divergence. This method requires... The above mesh skewness is less than 0.7, and the mesh skewness in critical areas is less than 0.6.

[0113] Grid aspect ratio: High aspect ratio meshes can cause severe numerical dissipation in flow fields with large gradients. This method requires that the aspect ratio of all meshes be less than 5, and the aspect ratio of the boundary layer meshes not exceed 10.

[0114] Orthogonal Quality: Orthogonal quality reflects the orthogonality of mesh elements. The closer the value is to 1, the better the quality. Mesh elements with a value below 0.3 will lead to instability in conjugate heat transfer calculations.

[0115] Negative volume grids: 0. Negative volume grids will directly cause the calculation to terminate. After the grid is generated, a comprehensive check must be performed and all negative volume elements must be repaired.

[0116] To eliminate the influence of mesh density on the calculation results, mesh independence verification must be performed. The process is as follows: Generate three mesh schemes with different densities: coarse, medium, and fine. Under the same boundary conditions and model parameters, steady-state thermal radiation calculations are performed using each of the three mesh schemes, and the average heat flux and maximum temperature of adjacent vehicle surfaces are extracted as verification indicators. When the mesh is refined from medium to fine, if the change in the verification indicators is less than... If the required grid density is met, the medium-sized grid is considered to meet the computational accuracy requirements and is selected as the final computational grid scheme. This verification process ensures that the computational results are not affected by grid density, and possess repeatability and scientific validity.

[0117] Different heat release rate (HRR) curves are set for different vehicle models, and the curves are imported as volumetric heat source terms using Fluent's DEFINE_PROFILE and DEFINE_SOURCE macros:

[0118] The ignition source model for gasoline-powered vehicles includes the growth stage, the stable combustion stage, and the decay stage.

[0119] The fire source model for ternary lithium battery electric vehicles includes the preheating stage, thermal runaway stage, jet fire stage, stable combustion stage, and decay stage.

[0120] The fire source model for lithium iron phosphate battery electric vehicles includes the preheating stage, thermal runaway stage, stable combustion stage, and decay stage.

[0121] The parameters of the heat release rate curves for different vehicle models can be determined based on test data, standard fire curves, or historical accident data, and then loaded into the calculation model as dynamic heat sources.

[0122] For different components of the vehicle, corresponding material thermophysical properties and ignition critical parameters are set, including density. Specific heat capacity thermal conductivity Critical ignition heat flux Critical ignition time The specific parameter values ​​are shown in the table below:

[0123] Component Name Density (kg / m³) Specific heat capacity (J / (kg·K)) Thermal conductivity (W / (m·K)) Critical heat flux (kW / m²) Critical ignition time (s) tire 1100 1700 0.14 16.5 300 bumper 1050 1500 0.20 17.5 280 Interior materials 900 1400 0.12 10.0 200 Car window glass 2500 840 0.80 - -

[0124] Call RNG k- The turbulence model simulates the turbulent flow process of fire smoke, and with buoyancy effect correction enabled, the model can accurately simulate strong swirling and separated flows, and is suitable for calculating complex flow fields in automobile fires.

[0125] The heat radiation transfer during a fire is calculated using the Discrete Ordinates (DO) radiation model, with angle discretization employed. The pixels can accurately simulate the directional thermal radiation effect of flames and high-temperature smoke on adjacent vehicles, while also supporting the radiation transmission effect of semi-transparent media such as glass.

[0126] The combustion model is used to simulate the chemical reaction process of fuel combustion. This model can take into account the interaction between turbulence and chemical reaction, and is suitable for simulating diffusion combustion such as oil pool fire and jet fire. It also supports the pyrolysis combustion of solid materials.

[0127] By calling the Conjugate Heat Transfer (CHT) model and employing coupled wall boundary conditions, the system automatically solves for the convective and radiative heat transfer between the fluid and the solid wall, as well as the heat conduction process inside the solid, thus achieving a coupled solution for the temperature fields of the fluid and the solid.

[0128] Using Fluent's surface monitoring function and the UDF's DEFINE_EXECUTE_AT_END macro, the temperature, thermal radiation flux, thermal convection flux, and thermal conduction flux of each grid cell on the surface of the adjacent vehicle are output in real time at 1-second intervals.

[0129] For gas inflow boundaries such as ventilation openings and natural ventilation points, a pressure inlet boundary is uniformly adopted, with the total pressure set at standard atmospheric pressure (101325 Pa), the initial temperature at ambient temperature (300 K), and the turbulence intensity taken as... The hydraulic diameter is calculated based on the equivalent diameter of the ventilation opening. If simulating a forced air supply scenario, the pressure inlet can be replaced with a velocity inlet, and the corresponding ventilation velocity can be directly input. For gas outlet boundaries such as smoke exhaust vents and pressure relief vents, a pressure outlet boundary is used, with the static pressure set to standard atmospheric pressure (101325 Pa), the return temperature to 300 K, and the return composition set to standard air (oxygen mass fraction 0.233, nitrogen mass fraction 0.767). If simulating negative pressure smoke exhaust conditions, the static pressure can be adjusted to -50 Pa to -100 Pa to match the actual working pressure of the smoke exhaust system. For solid boundaries such as walls, floors, and vehicle exteriors in the computational domain, no-slip wall boundaries are used, with an initial temperature of 300K. Wall emissivity is set according to material properties: 0.85 for concrete floors and building walls, and 0.8 to 0.9 for vehicle components such as body panels, tires, and bumpers. Notably, walls in contact with fluids are automatically identified as coupled walls by Fluent, eliminating the need for manual temperature or heat flow boundary settings; the conjugate heat transfer model automatically performs energy conservation calculations at the fluid-solid interface. For the combustion zone of the fire source vehicle, no special boundary conditions are required. Heat release and mass exchange are directly added to the energy and component transport equations via UDF macros as volumetric or surface heat sources, enabling dynamic loading of the fire development process.

[0130] A pressure-based transient solver is adopted as the core solution framework. This solver is adapted to the physical characteristics of low-speed incompressible flow in automotive fires and can efficiently handle complex transient processes such as buoyancy-driven and combustion-exothermic processes. The pressure-velocity coupled algorithm uses the PISO algorithm, which, through step-by-step solution with pressure correction, exhibits superior stability and convergence speed in transient flow simulations, and is particularly suitable for calculating fire flow fields with strong buoyancy effects and combustion reactions. Regarding the discretization scheme, all variables are presented in a second-order precision scheme to ensure computational accuracy. The gradient term employs the nodal basis Green-Gaussian gradient algorithm, which has higher computational accuracy on unstructured grids. The pressure term uses the PRESTO! scheme, which accurately captures the drastic pressure changes in high-buoyancy flows. The momentum, turbulent kinetic energy, turbulent dissipation rate, component transport, and energy equations are all presented in a second-order upwind scheme to effectively reduce numerical dissipation. The radiation equations are also presented in a second-order upwind scheme to ensure the computational accuracy of directional thermal radiation transfer. The time step is uniformly set to 1 second, consistent with the output interval of subsequent adjacent vehicle surface thermal data. This time step accurately captures the transient changes in fire heat flux at the second level while also considering computational efficiency. If residual oscillations or divergences occur during the calculation, the time step can be temporarily reduced to 0.5 seconds, and then gradually restored to 1 second after the flow field stabilizes. The convergence criterion adopts a dual standard combining residual criteria and monitoring point criteria. The residual convergence threshold for the continuity equation, momentum equation, turbulent kinetic energy equation, turbulent dissipation rate equation, and component transport equation is set to 1e-3, while the residual convergence threshold for the energy equation and radiation equation is set to 1e-6. This is because the accuracy of energy and radiation calculations directly determines the accuracy of heat flux and temperature field, requiring higher convergence. At the same time, temperature monitoring points are set at key locations on the fire source vehicle center and adjacent vehicle surfaces. When the temperature change at the monitoring point is less than 0.1 K for 10 consecutive time steps, the calculation for that time step is considered converged.

[0131] Calculate the combined heat flux of each grid cell on the surface of the adjacent vehicle. The expression is:

[0132]

[0133] in, For thermal radiation flux, For heat convection flux, This represents heat conduction flux, with units of 1 and 2. .

[0134] thermal radiation flux This is the net heat flux density transferred from flames and high-temperature smoke to the surface of adjacent vehicles through thermal radiation, obtained by the DO radiation model. Its physical meaning is the difference between the radiant energy received by a unit area of ​​the wall per unit time and the radiant energy emitted by the wall itself. The calculation formula is:

[0135]

[0136] in The intensity of incident radiation on the wall. For wall emissivity, The Stefan-Boltzmann constant is... This refers to the wall temperature.

[0137] Heat convection flux is the heat flux density transferred between high-temperature flue gas and the surface of an adjacent vehicle through convective heat transfer, expressed as RNG k- The calculation is obtained by combining the turbulence model with the standard wall function. The calculation formula is as follows:

[0138]

[0139] in The convective heat transfer coefficient is calculated from the wall function based on the near-wall turbulent velocity gradient and temperature gradient. This represents the near-wall flue gas temperature. In Fluent, the convective heat flux of a specified surface element can be directly extracted using the built-in macro F_FLUX_CONV(f, t), with units of t. A positive value indicates that the flue gas transfers heat to the wall, while a negative value indicates that the wall dissipates heat from the flue gas.

[0140] Based on the theory of thermal accumulation, the thermal accumulation of each grid cell is calculated. The expression is:

[0141]

[0142] in, Heat loss flux per unit area, mainly including heat dissipation to the environment and heat conduction within the material, is measured in units of... .

[0143] For different components, determine their critical heat accumulation threshold. The expression is:

[0144]

[0145] in, This is the critical ignition heat flux for this component. This refers to the ignition time of the component at the critical heat flux; when the heat accumulation of a certain grid cell... When the location of the component corresponding to the grid cell is determined to be ignited, and when any location of any component of a neighboring vehicle is determined to be ignited, the entire neighboring vehicle is determined to be ignited, the corresponding time is determined. That is, predicting the ignition time. .

[0146] Glass breakage correction model: Based on Fluent's solid temperature field calculations, the temperature distribution on the inner and outer surfaces of the vehicle window glass is calculated in real time. When the highest surface temperature of the glass reaches the critical breakage temperature... At that time, it was determined that the glass had broken. Before the glass broke, the convective heat transfer coefficient of the car window glass was... The natural convection heat transfer coefficient is adopted, and its value is [value missing]. After the glass breaks, the convective heat transfer coefficient of the adjacent vehicle surface is dynamically modified using a UDF (User-Defined Function), expressed as:

[0147]

[0148] In the formula: for The rate of heat release or heat flux at any given time. This is the initial heat release rate or the baseline heat flux; The convection enhancement coefficient for glass breakage is set to 1.5-2.5 to simulate the enhanced heat transfer effect of high-temperature external hot airflow entering the vehicle after breakage.

[0149] Jet fire correction model: When the fire source vehicle is a ternary lithium battery electric vehicle and is in the jet fire stage, calculate the directional thermal radiation enhancement coefficient of the jet fire on adjacent vehicles. The expression is:

[0150]

[0151] in, The maximum thermal radiation enhancement coefficient; This is the distance from the center of the jet fire to the surface of the adjacent vehicle, in meters (m). The characteristic attenuation distance; The angle between the direction of the jet fire and the normal to the surface of the adjacent vehicle is expressed in rad.

[0152] The thermal radiation flux on the surface of the adjacent vehicle is corrected using UDF. for:

[0153]

[0154] Environmental correction model: Obtaining ambient wind speed Ambient temperature and ambient humidity Parameters. The flame tilt angle is adjusted based on ambient wind speed. The expression is:

[0155]

[0156] in, The vertical upward velocity of the flame is denoted as , and its value is . .

[0157] Correct the initial temperature of neighboring vehicles based on ambient temperature. And correct the temperature difference for heat transfer. The air's radiation absorption coefficient is corrected based on ambient humidity. The expression is:

[0158]

[0159] in, The radiation absorption coefficient of dry air is taken as a value of .

[0160] Based on the predicted ignition time The risk of ignition from neighboring vehicles is divided into four levels:

[0161] Security level: It may not ignite for several minutes and no special emergency measures are required.

[0162] Alert level: 30 minutes Minutes are needed; monitoring should be strengthened and emergency preparedness should be made.

[0163] Danger level: 10 minutes Within minutes, personnel must be evacuated immediately and preparations made for firefighting.

[0164] Extremely dangerous level: Within minutes, all personnel must be evacuated immediately, and the source of the fire must be dealt with first.

[0165] Generates a visualized ignition risk cloud map, using different colors to display the ignition time distribution at different locations on the surfaces of adjacent vehicles, intuitively presenting high-risk areas. Outputs an ignition risk assessment report, including simulation parameter settings (such as fire source vehicle type, vehicle layout, material properties, etc.), heat flux distribution (spatiotemporal distribution of radiation, convection, and conduction flux on the surfaces of each adjacent vehicle), heat accumulation process (evolution curve of heat accumulation of key components over time), ignition time prediction (predicted ignition time and ignition location for each adjacent vehicle), risk level, and corresponding emergency response recommendations.

[0166] Once it is determined that any adjacent vehicle has ignited, the chain spread prediction process is automatically triggered, which dynamically transforms the ignited vehicle into a new secondary fire source. Based on its vehicle type, ignition location, and ignition time, a unique dynamic combustion sequence is generated, and the volumetric heat source item is updated in real time through the UDF interface to achieve multi-fire source coupling loading.

[0167] The vehicle is designed with a 5-15 minute transition period from localized ignition to full-scale combustion, while also inheriting the differentiated combustion characteristics of different vehicle models: fuel vehicles exhibit standard oil pool fire combustion characteristics after ignition; ternary lithium battery electric vehicles automatically trigger a jet fire model after ignition, with the jet direction determined based on the battery pack installation location and the direction of thermal runaway pressure relief; and lithium iron phosphate battery electric vehicles maintain slow and stable combustion characteristics after ignition.

[0168] The fire source failure mechanism is set according to the standard combustion time of each vehicle type: 20-30 minutes for fuel vehicles, 30-40 minutes for ternary lithium battery electric vehicles, and 40-60 minutes for lithium iron phosphate battery electric vehicles. After the time is reached, the heat release rate decays to 0 according to an exponential law, and the fire source is marked as extinguished, thus completing the dynamic iteration of the fire source throughout its entire cycle.

[0169] Based on the updated list of multiple fire sources, instead of performing full-process transient CFD iterative simulation, the system relies on the fire evolution patterns, thermal field change rates, and fire growth characteristics obtained from previous short-term simulations. Combined with a time-series evolution extrapolation algorithm, it introduces a multi-fire source heat flux superposition mechanism and an adaptive correction strategy to quickly extrapolate the entire fire spread process, thus avoiding the drawbacks of high computing power and long time consumption in full-process numerical simulation.

[0170] The total heat flux of any grid cell on the surface of an adjacent vehicle is the sum of the heat radiation, heat convection, and heat conduction fluxes of all active fire sources.

[0171] The grid adaptive dynamic adjustment mechanism is enabled. When a new fire source is generated, the grid around it within 3m is densified by 50%, and the grid in the area where the fire source is extinguished is automatically coarsened, thereby improving the calculation efficiency by 30%-40% while ensuring the calculation accuracy.

[0172] A dynamic computational domain pruning strategy is adopted, with the outer envelope of all active fire sources as the center and an effective computational domain extending outward by 5m, eliminating unaffected areas and reducing the amount of invalid computation.

[0173] Set iteration termination criteria, and stop simulation when any of the following conditions are met: 1) No new ignition vehicles are added in the computational domain and all active fire sources are completely extinguished; 2) The simulation duration reaches the maximum emergency decision window of 120 minutes; 3) The fire spreads to the boundary of the computational domain and there is no continued spread trend.

[0174] During the iterative calculation process, dynamic features of the entire fire domain are extracted in real time every second to construct a database of the full life cycle evolution of the fire. The core extracted content includes:

[0175] Chain ignition timing characteristics: record the ignition time, peak heat release rate, and complete extinguishing time for each vehicle to generate a multi-vehicle chain ignition timeline and fire spread path map.

[0176] Thermal field spread characteristics: Real-time output of global temperature field and thermal radiation flux field cloud map, and statistical analysis of high-temperature danger areas ( ), ignition hazard heat radiation area ( The real-time area and outward expansion rate of ).

[0177] Flue gas diffusion characteristics: Combining turbulence models and component transport equations, the flue gas spread trajectory, settling height, spatial visibility distribution, and CO2 concentrations are deduced. The toxic gas concentration field can accurately predict the time required to seal off evacuation routes and safety exits with smoke.

[0178] Key area early warning features: Mark key facilities and areas such as fire lanes, power distribution rooms, equipment rooms, and evacuation assembly points within the site, predict the arrival time of fire and smoke, and achieve early warning.

Claims

1. A method for predicting and assessing the risk of fires in new energy vehicles based on Fluent multiphysics coupling, characterized in that, Includes the following steps: Based on multiphysics field coupled numerical simulation, the multimode heat flux distribution of the fire source vehicle on the surface of the adjacent vehicle is obtained. The multimode heat flux distribution includes heat radiation flux, heat convection flux and heat conduction flux. Based on the multi-mode heat flux distribution, the heat accumulation of the neighboring vehicle is calculated using a dynamic correction model, and the heat accumulation is used to determine whether the neighboring vehicle has been ignited and to predict the ignition time. Based on the predicted ignition time, the adjacent vehicle is assessed for ignition risk classification. The neighboring vehicles that will be ignited are transformed into dynamic fire sources, and the multi-vehicle chain spread process is deduced based on the time-series extrapolation algorithm and the dynamic fire source iteration mechanism; Output fire full-cycle prediction results and risk assessment reports.

2. The new energy vehicle fire prediction and risk assessment method based on Fluent multi-physical field coupling according to claim 1, characterized in that, Before obtaining the multi-mode heat flux distribution of the fire source vehicle on the surface of the adjacent vehicle, the method further includes: Extract the 3D geometric models of the fire source vehicle and adjacent vehicles, and model the ignition components separately; An unstructured hybrid mesh is used to divide the computational domain. The mesh is refined in the core area of ​​flame development and the vehicle surface area, while the mesh size gradually increases in the outer area of ​​the flame. All mesh elements are inspected to ensure that the following mesh quality requirements are met: mesh skewness is no greater than 0.85, with more than 95% of the meshes having a skewness of less than 0.7, and the mesh skewness in the preset area being less than 0.6; mesh aspect ratio is no greater than 5, and the aspect ratio of the boundary layer mesh is no more than 10; orthogonality quality is no less than 0.3; there are 0 negative volume meshes. Mesh independence verification was carried out: three mesh schemes with increasing mesh density (coarse, medium, and fine) were generated, and steady-state thermal radiation calculations were performed under the same conditions. The average heat flux and maximum temperature of adjacent vehicle surfaces were extracted as verification indicators. When the mesh was refined from medium to fine, if the change in the verification indicators was less than 5%, the medium mesh was determined as the final calculation mesh scheme. 3.The new energy vehicle fire prediction and risk assessment method based on Fluent multi-physical field coupling according to claim 1, characterized in that, Before obtaining the multi-mode heat flux distribution of the fire source vehicle on the surface of adjacent vehicles, the method also includes setting differentiated heat release rate curves for different vehicle models: The ignition source model for gasoline-powered vehicles includes the growth stage, the stable combustion stage, and the decay stage. The fire source model for ternary lithium battery electric vehicles includes the preheating stage, thermal runaway stage, jet fire stage, stable combustion stage, and decay stage. The fire source model for lithium iron phosphate battery electric vehicles includes the preheating stage, thermal runaway stage, stable combustion stage, and decay stage. The parameters of the heat release rate curves for different vehicle models are determined based on test data, standard fire curves, or historical accident data, and are loaded into the calculation model as dynamic heat sources. The heat release rate curve is imported as a volumetric heat source term using Fluent's DEFINE_PROFILE and DEFINE_SOURCE macros. For different vehicle components, corresponding material thermophysical properties and ignition critical parameters are set, including density. Specific heat capacity Thermal conductivity Critical ignition heat flux and critical ignition time The critical ignition heat flux of the tire is Critical ignition time is The critical ignition heat flux of the bumper is Critical ignition time is The critical ignition heat flux of the interior materials is Critical ignition time is .

4. The method for predicting and assessing the risk of fires in new energy vehicles based on Fluent multiphysics coupling according to claim 1, characterized in that, The multiphysics coupled numerical simulation uses the following physical model and solver configuration: Call RNG k- The turbulent flow process of fire smoke is simulated by a turbulent model, and the buoyancy effect is corrected. The DO radiation model is used to calculate the heat radiation transfer during a fire, supporting the radiation transmission effect of semi-transparent media including glass. The combustion model is invoked to simulate the chemical reaction process of fuel combustion; By calling the conjugate heat transfer model and using coupled wall boundary conditions, the convective and radiative heat transfer between the fluid and the solid wall, as well as the heat conduction process inside the solid, are automatically solved. A pressure-based transient solver was used, and the PISO algorithm was selected for the pressure-velocity coupling algorithm. All variables were in second-order precision schemes, with the gradient term using the nodal basis Green-Gaussian gradient algorithm, the pressure term using the PRESTO! scheme, and the momentum, turbulent kinetic energy, turbulent dissipation rate, component transport, and energy equations all using second-order upwind schemes. An adaptive time step is used, ranging from 0.0001 to 0.1 s. The convergence criterion adopts a dual standard combining the residual criterion and the monitoring point criterion. The residual convergence threshold for the continuity equation, momentum equation, turbulent kinetic energy equation, turbulent dissipation rate equation, and component transport equation is set to 1e-3, and the residual convergence threshold for the energy equation and radiation equation is set to 1e-6. The calculation of a time step is considered to have converged when the change amplitude of the temperature monitoring point is less than 0.1 K for 10 consecutive time steps. Boundary condition settings include: a pressure inlet boundary is used for gas inflow, with a total pressure of 101325 Pa, an initial temperature of 300 K, and a turbulence intensity of 5%; a pressure outlet boundary is used for gas outflow, with a static pressure of 101325 Pa, a return temperature of 300 K, and the return component set to standard air; a no-slip wall boundary is used for solid boundaries, with an initial temperature of 300 K and wall emissivity set according to material properties; the wall surface in contact with the fluid is automatically identified as a coupled wall surface, and the energy conservation calculation of the fluid-solid interface is automatically completed by the conjugate heat transfer model; the combustion zone of the fire source vehicle is added to the energy equation and component transport equation in the form of a volumetric heat source or a surface heat source through a UDF macro; The Fluent DEFINE_EXECUTE_AT_END macro outputs the temperature, thermal radiation flux, thermal convection flux, and thermal conduction flux of each grid cell on the surface of the adjacent vehicle in real time, with a time interval of 1 second.

5. The new energy vehicle fire prediction and risk assessment method based on Fluent multi-physical field coupling according to claim 1, characterized in that, The step of calculating the heat accumulation of the neighboring vehicle based on the multi-mode heat flux distribution and combined with the dynamic correction model, and determining whether the neighboring vehicle has been ignited and predicting the ignition time based on the heat accumulation, includes: calculating the integrated heat flux for each grid cell of the adjacent vehicle surface , the expression is: where, is the heat flux, which is solved by the DO radiation model with the following formula: in The intensity of incident radiation on the wall. For wall emissivity, The Stefan-Boltzmann constant is... The wall temperature; For the heat convection flux, the RNG k- The turbulent model is combined with the standard wall function to calculate, and the calculation formula is: in The convective heat transfer coefficient is... The near-wall flue gas temperature; the convective heat flux of the specified surface element is directly extracted using the Fluent built-in macro F_FLUX_CONV; is the heat conduction flux; Based on the heat accumulation theory, the heat accumulation of each grid unit is calculated , and the expression is: in, Heat loss flux per unit area mainly includes heat dissipation to the environment and heat conduction within the material; determining a critical heat build-up threshold value for different components , the expression being wherein, is the critical heat flux of the component, is the ignition time of the component at the critical heat flux; When the heat accumulation of a certain grid cell When the corresponding component location in the grid cell is determined to be ignited, and when any location of any component in a neighboring vehicle is determined to be ignited, the entire neighboring vehicle is determined to be ignited, the corresponding time is determined to be... That is, predicting the ignition time. .

6. The new energy vehicle fire prediction and risk assessment method based on Fluent multi-physical field coupling according to claim 5, characterized in that, The dynamic correction model includes a glass breakage correction model, a jet fire correction model, and an environmental correction model: The glass breakage correction model includes: calculating the temperature distribution on the inner and outer surfaces of the vehicle window glass in real time using Fluent's solid temperature field calculation results; and determining when the highest temperature on the glass surface reaches the critical breakage temperature. At that time, it was determined that the glass had broken; before the glass broke, the convective heat transfer coefficient of the car window glass was... The natural convection heat transfer coefficient is adopted, and its value is [value missing]. After the glass breaks, the convective heat transfer coefficient of the adjacent vehicle surface is dynamically modified using UDF, expressed as: wherein, is the glass breakage convection enhancement factor, having a value of 1.5-2.5; The jet fire correction model includes: when the fire source vehicle is a ternary lithium battery electric vehicle and is in the jet fire stage, determining the jet fire direction using the DPM model. jet speed and jet heat release rate ; Calculate the directional thermal radiation enhancement coefficient of the jet fire on adjacent vehicles. The expression is: in, The maximum thermal radiation enhancement coefficient is 3-5. This is the distance from the center of the jet fire to the surface of the adjacent vehicle, in meters (m). The characteristic attenuation distance is 2-3m. The angle between the direction of the jet fire and the normal to the surface of the adjacent vehicle is expressed in rad; the corrected thermal flux is: The environment correction model comprises: acquiring environment wind speed , environment temperature and environment humidity parameters; correcting the flame tilt angle according to the environment wind speed , and the expression is: in, The vertical upward velocity of the flame is denoted as , and its value is . Correct the initial temperature of neighboring vehicles based on ambient temperature. And correct the temperature difference for heat transfer. Correct the air's radiation absorption coefficient based on ambient humidity. The expression is: wherein is the radiation absorption coefficient for dry air, taken as .

7. The new energy vehicle fire prediction and risk assessment method based on Fluent multi-physical field coupling according to claim 1, characterized in that, The assessment of the ignition risk level of the adjacent vehicle includes: According to the predicted ignition time The ignition risk of the adjacent vehicle is classified into four levels: Safety level: Ignition time is greater than 60 minutes or it will not be ignited, and no special emergency measures are required; Alert Level: Ignition time greater than 30 minutes but not more than 60 minutes, requires enhanced monitoring and emergency preparedness; Hazard level: If the ignition time is greater than 10 minutes but not more than 30 minutes, personnel must be evacuated immediately and preparations made for firefighting. Extremely dangerous level: Ignition time is no more than 10 minutes. All personnel must be evacuated immediately and the source of fire must be dealt with first. Generates a visualized ignition risk cloud map, displaying the ignition time distribution at different locations on the surface of adjacent vehicles using different colors; outputs an ignition risk assessment report, including simulation parameter settings, heat flux distribution, heat accumulation process, ignition time prediction, risk level, and corresponding emergency response recommendations. 8.The new energy vehicle fire prediction and risk assessment method based on Fluent multi-physical field coupling according to claim 1, characterized in that, The process of converting adjacent vehicles into dynamic fire sources and simulating the chain reaction of fire spread across multiple vehicles based on a time-series extrapolation algorithm and a dynamic fire source iteration mechanism includes: When any adjacent vehicle is identified as ignited, the chain spread prediction process is automatically triggered, which dynamically transforms the ignited vehicle into a secondary fire source. Based on its vehicle type, ignition location and ignition time, a unique dynamic combustion sequence is generated, and the volume heat source item is updated in real time through the UDF interface to achieve multi-fire source coupling loading. The vehicle is designed with a 5-15 minute transition period from local component ignition to full vehicle combustion, while inheriting the different combustion characteristics of different vehicle models: after ignition, the fuel vehicle exhibits the standard oil pool fire combustion characteristics; after ignition, the ternary lithium battery electric vehicle automatically triggers the jet fire model, and the jet direction is determined according to the battery pack installation position and the direction of thermal runaway pressure relief; after ignition, the lithium iron phosphate battery electric vehicle maintains a slow and stable combustion characteristic. The fire source failure mechanism is set according to the standard combustion time of each vehicle type: 20-30 minutes for fuel vehicles, 30-40 minutes for ternary lithium battery electric vehicles, and 40-60 minutes for lithium iron phosphate battery electric vehicles. After the time is reached, the heat release rate decays to 0 according to an exponential law, and the fire source is marked as extinguished, thus completing the dynamic iteration of the fire source throughout its entire cycle.

9. The method for predicting and assessing the risk of fires in new energy vehicles based on Fluent multiphysics coupling according to claim 1, characterized in that, The method for extrapolating the multi-vehicle cascading propagation process based on the time-series extrapolation algorithm also includes: Based on the updated list of multiple fire sources, instead of performing full-process transient CFD iterative simulation, we rely on the fire evolution patterns obtained from previous short-term simulations, combined with a time-series evolution extrapolation algorithm, and introduce a multi-fire source heat flux superposition mechanism and adaptive correction strategy to quickly extrapolate the entire fire spread process. The total heat flux of any grid cell on the surface of an adjacent vehicle is the sum of the heat radiation, heat convection and heat conduction fluxes of all active fire sources. The grid adaptive dynamic control mechanism is enabled. When a new fire source is generated, the grid density within a 3m radius around it is increased by 50%. When the fire source is extinguished, the grid density within a 3m radius is automatically restored to the density before the increase. A dynamic computational domain pruning strategy is adopted, with the effective computational domain extending 5m outward from the outer envelope of all active fire sources as the center, and areas with no impact are eliminated. Set iteration termination criteria, and stop simulation when any of the following conditions are met: 1) No new ignition vehicles are added in the computational domain and all active fire sources are completely extinguished; 2) The simulation duration reaches the maximum emergency decision window of 120 minutes; 3) The fire spreads to the boundary of the computational domain and there is no continued spread trend.

10. The method for predicting and assessing fire risks in new energy vehicles based on Fluent multiphysics coupling according to claim 9, characterized in that, The simulation of multi-vehicle cascading fire spread also includes constructing a full-cycle fire evolution database, extracting dynamic characteristics of the entire fire area in real time every second, including: Chain ignition timing characteristics: record the ignition time, peak heat release rate, and complete extinguishing time for each vehicle to generate a multi-vehicle chain ignition timeline and fire spread path map; Thermal field spreading characteristics: real-time output of global temperature field, thermal radiation flux field cloud map, statistics of high temperature dangerous area ( ), real-time area and outward expansion rate of ignition dangerous thermal radiation area ( ); Smoke diffusion characteristics: combined with the turbulence model and component transport equation, the smoke spread trajectory, settling height, spatial visibility distribution and CO, toxic gas concentration field, predict the smoke sealing time of evacuation channel and safety exit; Key area early warning features: Mark facilities including fire lanes, power distribution rooms, equipment rooms, and evacuation assembly points to predict the arrival time of fire and smoke.

Images