Full-scale experiment and CFD simulation on smoke movement and smoke control in a metro tunnel with one opening portal

Three full-scale model experiments were conducted in a unidirectional tube, which is a part of a metro tunnel with one end connected to an underground metro station and the other end opened to outside in Chongqing, PR China. Three fire HRRs, 1.35 MW, 3 MW and 3.8 MW were produced by pool fires with different oil pan sizes in the experiments. Temperature distributions under the tunnel ceiling along the longitudinal direction were measured. At the same time, CFD simulations were conducted under the same boundary conditions with the experiments by FDS 5.5. In addition, more FDS simulation cases were conducted after the FDS simulation results agreed with the experimental results. The simulation results show that the smoke temperature and the decay rate of the temperature distribution under the tunnel ceiling along the longitudinal direction increase as HRR increases. The smoke exhausts effectively from the tunnel under mechanical ventilation system, whether the emergency vent is activated as a smoke exhaust or an air supply vent. The operation mode of the mechanical ventilation system depends on the evacuation route.     

Simulating longitudinal ventilation flows in long tunnels: Comparison of full CFD and multi-scale modelling approaches in FDS6

The accurate computational modelling of airflows in transport tunnels is needed for regulations compliance, pollution and fire safety studies but remains a challenge for long domains because the computational time increases dramatically. We simulate air flows using the open-source code FDS 6.1.1 developed by NIST, USA. This work contains two parts. First we validate FDS6’s capability for predicting the flow conditions in the tunnel by comparing the predictions against on-site measurements in the Dartford Tunnel, London, UK, which is 1200 m long and 8.5 m in diameter. The comparison includes the average velocity and the profile downstream of an active jet fan up to 120 m. Secondly, we study the performance of the multi-scale modelling approach by splitting the tunnel into CFD domain and a one-dimensional domain using the FDS HVAC (Heating, Ventilation and Air Conditioning) feature. The work shows the average velocity predicted by FDS6 using both the full CFD and multi-scale approaches is within the experimental uncertainty of the measurements. Although the results showed the prediction of the downstream velocity profile near the jet fan falls outside the on-site measurements, the predictions at 80 m and beyond are accurate. Our results also show multi-scale modelling in FDS6 is as accurate as full CFD but up to 2.2 times faster and that computational savings increase with the length of the tunnel. This work sets the foundation for the next step in complexity with fire dynamics introduced to the tunnel.     

Fire Dynamics Simulator (Version 4.0) Simulation for Tunnel Fire Scenarios with Forced,Transient, Longitudinal Ventilation Flows

In this study, a series of sensitivity analyses were conducted to evaluate a computational fluid dynamic (CFD) model, Fire Dynamics Simulator (FDS) version 4.0, for tunnel fire simulations. A tunnel fire test with a fire size on the order of a 100 MW with forced, time-varying longitudinal ventilation was chosen from the Memorial Tunnel Ventilation Test Program (MTVTP) after considering recent tunnel fire accidents and the use of CFD models in practice. A careful study of grid size and parameters used in the Large Eddy Simulation (LES) turbulence model—turbulent Prandtl number, turbulent Schmidt number, and Smagorinsky constant—was conducted. More detailed analyses were performed to refine the smoke layer prediction of FDS, especially on backflow (i.e., a reversed smoke flow near the ceiling). Also, energy conservation was checked for this scenario in FDS. A simple guideline is given for smoke layer simulations using FDS for similar tunnel fire scenarios.     

Experimental and Numerical Study of the Interaction Between Water Mist and Fire in an Intermediate Test Tunnel

The paper deals with interaction between water mist and hot gases in a longitudinally ventilated tunnel. The work aims at understanding the interaction of mist, smoke and ventilation.The study is based on one intermediate tunnel test and an extensive use of the computational code Fire Dynamics Simulator (FDS, NIST). The approach consists first of reconstructing the test with the CFD code by defining the relevant numerical parameters to accurately model the involved water mist system. Then, it consists of handling from the local data the complicated flows generated by the water mist flooding on the one hand and by fire and ventilation on the other hand. The last stage consists in quantifying each mechanism involved in interaction between water mist and hot gases. There are three main results in this study. Firstly, the CFD code prediction is also evaluated in this configuration, with and without water mist. Before the mist system activation, the agreement is satisfactory for gas temperatures and heat flux. After the activation time, the CFD code predicts well the thermal environment and in particular its stratification. Secondly, water mist plays a strong thermal role since in the test studied, roughly half of the heat released by fire is absorbed by water droplets. Thirdly, heat transfer from gaseous phase to droplets is the main mechanism involved (73%). The remaining heat absorbed by droplets results from tunnel surface cooling which represents (9%) and radiative attenuation (18%).     

New approach to estimate temperatures in pre-flashover fires: Lumped heat case

This paper presents a model for estimating temperatures in pre-flashover fires where the fire enclosure boundaries are assumed to have lumped heat capacity. That is, thermal inertia is concentrated to one layer with uniform temperature and insulating materials are considered purely by their heat transfer resistance. The model yields a good understanding of the heat balance in a fire enclosure and was used to predict temperatures in insulated and non-insulated steel-bounded enclosures. Comparisons were made with full scale experiments and with other predictive methods, including CFD modeling with FDS and the so called MQH relationship. Input parameter values to the model were then taken from well-known literature and the heat release rates were provided from the experiments. The fire temperature predictions of the model matched very well with experimental data. So did the FDS predictions while the original MQH relationship gave unrealistic results for the problems studied. Major benefits of using the model in comparison with CFD modeling are its readiness and simplicity as well as the negligible computation times needed. An Excel application of the presented pre-flashover fire model is available on request from the author.     

High-resolution CFD simulations for forced convective heat transfer coefficients at the facade of a low-rise building

High-resolution 3D steady RANS CFD simulations of forced convective heat transfer at the facades of a low-rise cubic (10 × 10 × 10 m³) building are performed to determine convective heat transfer coefficients (CHTC). The focus is on the windward facade. CFD validation is performed based on wind tunnel measurements of velocity and heat transfer for reduced-scale cubic models. The CFD simulations employ a high-resolution grid with, for the 10 m cubic building, cell centres at a minimum distance of 160 μm from the building surface to resolve the entire boundary layer, including the viscous sublayer and the buffer layer, which dominate the convective surface resistance. The results show that: (1) the wind flow around the building results in highly varying CHTC values across the windward facade; (2) standard and non-equilibrium wall functions are not suitable for CHTC calculation, necessitating either low-Reynolds number modelling or specially-adapted wall functions; (3) at every facade position, the CHTC is a power-law function of the mean wind speed; (4) the CHTC distribution at the windward facade is relatively insensitive to wind direction variations in the 0–67.5° angle range; (5) the CHTC shows a stronger spatial correlation with the turbulent kinetic energy than with the mean wind speed across the facade; and (6) the CHTC distribution across the windward facade is quite similar to the distribution of wind-driven rain (WDR), with both parameters reaching high levels near the top edge of the facade. This suggests that also the convective moisture transfer coefficient will be higher at this location and that the facade parts that receive most WDR might also experience a higher drying rate.     

Assessment of Fire Dynamics Simulator for Heat Flux and Flame Heights Predictions from Fires in SBI Tests

This paper presents an experimental and numerical study of heat flux and flame heights from fires generated in single burning item (SBI) tests. Thin steel plate probes were developed, as an inexpensive and reliable alternative to heat flux gauges, to measure the surface heat flux, whilst flame heights were determined by analyzing the instantaneous images extracted from the videos of the experiments by a CCD camera. Experimental results obtained at different heat release rates were subsequently used to assess the accuracy of the computational fluid dynamics (CFD) code, Fire dynamics simulator (FDS, V4.07). Simulation results indicated that though predicting reasonably flame heights FDS underpredicts significantly the surface heat flux at higher heat release rates. Consequently, a sensitivity study of the parameters used in the radiation and soot models in FDS was conducted.     

Development of a Gaussian glass breakage model within a fire field model

A glass breakage model has been implemented within the existing FDS fire field model. The field model allows the prediction of radiative and convective heating to solid-phase objects. This is utilised by filtering, within the code, the objects to be modelled as glass, and then linking a one-dimensional heat transfer across those objects to a Gaussian spread of breaking temperatures across panes within the compartment being modelled. The gas- and solid-phase heat transfer methods are described along with the implemented glass breakage model. A case study is then presented involving the modelling of a large-scale compartment fire in a building with a high degree of glazing. A key factor in the severity of the atmospheric conditions, and the variations in temperature prediction, comes as a result of the different breaking patterns of glass around the compartment. The Gaussian glass breakage model emphasises this, and equivalently high local temperatures are not predicted when the typically adopted method of solid-object removal is used, or when the assumption of an ‘open’ compartment condition, where all the glazing is removed, is modelled. Changing ventilation patterns from a Gaussian breakage model are thus shown to produce markedly different gas temperature predictions than other methods. The new glass brakeage model presented in this paper can be implemented into any existing CFD fire field model and is not exclusive to FDS.     

Three-dimensional nonlinear analysis for the coupled problem of the heat transfer of the surrounding rock and the heat convection between the air and the surrounding rock in cold-region tunnel

According to the basic theories of heat transfer, geocryology and fluid mechanics, taking the coupled problem of the heat transfer of the rock surrounding the tunnel and the heat convective between the air in the tunnel and the rock surrounding the tunnel into account, three-dimensional calculating model of the coupled problem are presented. The finite element formulae of this problem are obtained by Galerkin’s method, and the computer program of the finite element is compiled. Using the program, three-dimensional nonlinear analyses for the coupled problem of the heat transfer of the rock surrounding the tunnel and the heat convective between the air in the tunnel and the rock surrounding Fenghuo mountain tunnel on the Qinghai–Tibet Railway are made. The agreement between the calculated results and the in-situ observed data is seen to be very good. The calculated results illustrate that the freezing–thawing situation of the rock surrounding the tunnel can correctly be predicted even if the air temperature distribution along the tunnel is unknown. In thus way, the large cost of in-situ observation for the air temperature in the tunnel can be saved.     

A Study of the Critical Velocity of Smoke Bifurcation Flow in Tunnel with Longitudinal Ventilation

Small longitudinal velocity cannot prevent backlayering in tunnel fire, while excessive longitudinal velocity will destroy stratification of smoke layer and lead to bifurcation flow. As smoke bifurcation flow proceeds, the longitudinal flow is divided into two streams and flow along both sidewalls of the tunnel ceiling. The critical velocity of bifurcation flow is the minimum value at which bifurcation flow starts to occur. To investigate the critical velocity of bifurcation flow, experiments and CFD simulations were conducted. Experiment was carried out in a reduced-scale tunnel, which is 8 m long, 1 m wide and 0.5 m high. The numerical research was performed using FDS. In simulation, the computational region of a tunnel is 200 m long, 10 m wide. The heat release rate (1 MW to 6 MW) and the height (4 m to 8 m) is changed in the 30 simulation scenarios. Theoretical analysis showed that the dimensionless critical velocity of bifurcation flow only depends on the dimensionless heat release rates, and a mathematical equation is proposed. The reduced-scale experiments indicated that the critical velocity of bifurcation flow is 1.48 times that of critical velocity for preventing backlayering, and the coefficient is in agreement with CFD simulation.     

Verification of the accuracy of CFD simulations in small-scale tunnel and atrium fire configurations

In preparation for the use of computational fluid dynamics (CFD) simulation results as ‘numerical experiments’ in fire research, the agreement with experimental data for two different small-scale set-ups is discussed. The first configuration concerns the position of smoke-free height in case of fire with vertical ventilation in an atrium. The second set-up deals with the critical velocity for smoke backlayering in case of fire in a horizontally ventilated tunnel. An N-percent rule is introduced for the determination of the presence of smoke in the simulation results, based on the local temperature rise. The CFD package FDS is used for the numerical simulations. The paper does not scrutinize the detailed accuracy of the results, as this is hardly possible with any state-of-the-art experimental data at hand. Rather, the global accuracy is discussed with current numerical implementation and models in FDS, considering continuous evolution over different version releases with time. The agreement between the experiments and numerical simulations is very promising. Even when quantitative agreement with experimental data is not perfect, the trends are very well reproduced in the simulations. While much additional work is required, both in CFD as in ‘real’ experiments, the results are encouraging for the potential of state-of-the-art CFD to be used as numerical experiments.     

Numerical simulation of fire in a tunnel: Comparative study of CFAST and CFX predictions

The mathematical modeling of fire growth and smoke movement in any enclosure is a formidable task. Two types of deterministic models are in vogue, zone models and field models (popularly known as CFD technique). CFAST is a popular zone model used for modeling of fires in enclosures. Likewise, CFX is a general purpose CFD code used for various purposes including modeling of fires. In the present paper, a tunnel of length 150 m having a rectangular cross-section of 80 m² has been considered for analyzing the temperature and velocity profiles generated by fire, placed at a distance of 20 m from one end of portal, by both CFAST and CFX. The simulation by CFAST has been carried out by dividing the tunnel into 1, 2, 5, 8, 10, 12 and 15 compartments of equal size, where these compartments are joined by openings or vents having same cross-section as that of the tunnel. In case of tunnel divided into 15 compartments the fire source position lies at the position of vent; CFAST predicted very high temperatures. The simulations have also been carried out by dividing tunnel into unequal sized compartments such that position of fire was at the center of the compartment. It was found that for accuracy of results, location of fire source inside compartment is an important factor. Computational difficulty was experienced when tunnel was divided into more than 15 compartments. In this paper, a comparative study of temperatures predicted by CFAST and CFX has been done. The CFX and CFAST predictions show that smoke temperature changes with a pattern roughly similar to that of heat release rate. The temperature profiles at selected positions cannot be predicted by CFAST unlike CFX. The detailed features like flame tilt, flow field can only be observed from CFX predictions.     

Comparison of FDS Prediction of Smoke Movement in a 10-Storey Building with Experimental Data

In this study, the Fire Dynamics Simulator (FDS), a computational fluid dynamics (CFD) model developed by National Institute of Standards and Technology (NIST) is used to simulate fire tests conducted at the National Research Council of Canada (CNRC). These tests were conducted in an experimental 10-storey tower to generate realistic smoke movement data. A full size FDS model of the tower was developed to predict smoke movement from fires that originate on the second floor. Three propane fire tests were modelled, and predictions of O₂, CO₂ concentrations and temperature on each floor are compared with the experimental data. This paper provides details of the tests, and the numerical modelling, and discusses the comparisons between the model results and the experiments. The 10-storey experimental tower was designed to simulate the centre core of high-rise buildings. It includes a compartment and corridor on each floor, a stair shaft, elevator shaft and service shafts. Three propane fire tests were conducted in 2006 and 2007 to study smoke movement through the stair shaft to the upper floors of the building. The fire was set in the compartment of the 2nd floor. Thermocouples and gas analyzers were placed on each floor to measure temperature and O₂, CO₂ and CO concentrations. Comparisons in the fire compartment and floor of fire show that the FDS model gives a good prediction of temperature and O₂ and CO₂ concentrations. In the stair shaft and upper floors there are some small differences which are due to the effect of heat transfer to the stairs that was not considered in the model. Overall the study demonstrates that FDS is capable of modelling fire development and smoke movement in a high rise building for well ventilated fires.     

CFD and experimental studies of room fire growth on wall lining materials

CFD simulation and experimental tests have been carried out to study the room corner fire growth on combustible wall-lining materials. In the CFD simulation, the turbulent mass and heat transfer, and combustion were considered. The discrete transfer (DT) method was employed to calculate the radiation with an absorptivity and emissivity model employed to predict the radiation property of combustion products including soot, CO₂ and H₂O, which are usually the primary radiating species in the combustion of hydrocarbon fuels. The temperature of the solid boundary was determined by numerical solution of the heat conduction equation. A simple and practical pyrolysis model was developed to describe the response of the solid fuel. This pyrolysis model was first tested against the Cone Calorimeter data for both charring and non-charring materials under different irradiance levels and then coupled to CFD calculations. Both full and one-third scale room corner fire growths on particle board were modelled with CFD. The calculation was tested with various numbers of rays and grid sizes, showing that the present choice gives practically grid- and ray number-independent predictions. The heat release rate, wall surface temperature, char depth, gas temperature and radiation flux are compared with experimental measurements. The results are reasonable and the comparison between prediction and experiment is fairly good and promising.     

Dynamic thermal building analysis with CFD – modelling radiation

In an attempt to reduce the high computational effort required for dynamic thermal simulation of buildings using computational fluid dynamics (CFD) the authors have recently developed an adaptive freeze-flow method (i.e. freezing of flow equations over variable time periods). This article documents the work that has been carried out to predict the surface heat transfer in dynamic thermal building processes using CFD with particular focus on radiation. The Monte Carlo (MC) and discrete transfer (DT) radiation models were investigated and results compared with analytical solutions. The DT model has shown good performance whereas an unrealistic radiation distribution on the surfaces was observed when using the MC model. A further investigation of the DT model for the cooling of a solid wall has shown that the adaptive freeze-flow method is an efficient and accurate means of conducting dynamic thermal CFD simulations which involve radiation. Finally, application of the technique to a more realistic space comprising an uneven distribution of solar gain showed very good results when compared with a zonal dynamic thermal simulation program.     

Numerical studies on heat release rate in a room fire burning wood and liquid fuel

Heat release rates of burning gasoline and wood fires in a room were studied by computational fluid dynamics (CFD). Version 5.5.3 of the software Fire Dynamics Simulator (FDS), which is the latest one available, was selected as the CFD simulation tool. Predicted results were compared with two sets of reported data from full-scale burning tests. In the two sets of experiments, the scenarios were set at gasoline pool fire and wood chipboard fire with gasoline respectively. The input heating rate of gasoline pool fire based on experimental measurements was used in the first set of experiments. Three scenarios G1, G2 and G3 with different grid systems were simulated by CFD. The grid system of scenario G2 gave more accurate prediction, which was then used to study the second set of experiments on wood chipboard with gasoline. The combustion model in FDS was used in wood chipboard fire induced by gasoline pool. The wood chipboard was allowed to burn by itself using the pyrolysis model in FDS. The effects of the boundary conditions on free openings for the same set of experiments were studied by three scenarios SOB1, SOB2 and SOB3. Boundary condition SOB2 gave more reliable prediction among the three boundary conditions. Two other scenarios on the effect of moisture content of wood were also studied. The predicted HRR curve was found to agree better with experiment in using SOB2.     

城市V形坡隧道火灾最高温度数值模拟

采用FDS 数值模拟方法,对V 形坡隧道火灾时烟气运动特性及隧道纵向中心线上温度分布情况进行研究,并提出不同火灾位置时顶板最大温升参数经验预测模型。结果表明,火源位于变坡点右侧120 m 时,隧道纵向中心线峰值温度点向下游偏移,偏移距离随坡度的增加而增加,隧道顶板最高温度随坡度的增加而减小。通过推导无量纲火源位置与变坡点距离不同时的最大温升参数预测模型得出,无量纲最大温升参数随无量纲火源位置的增大而增大、与无量纲热释放速率的0.8 次幂成正比,且与隧道坡度呈非线性非单调关系。     

A Multi-Grid Model for Evacuation Coupling with the Effects of Fire Products

In the study, influences of fire products on pedestrians are introduced into a multi-grid evacuation model, in which the space is discretized into small grids with the size of 0.1 m × 0.1 m and each pedestrian occupies 5 × 5 grid sites. The fire products affect two walking parameters of pedestrians: the desired movement direction and the step frequency. The data of fire products are obtained from the simulation results of the Fire Dynamics Simulator (FDS), a well-founded computational fluid dynamic (CFD) program, developed by the National Institute of Standards and Technology (NIST). With this model, we investigated the routes of pedestrians in fires, and the evacuation times in scenarios with different fire intensities, pre-movement times or door widths. The results indicate that pedestrians will avoid moving towards the fire source, small fire may make egress process faster while large fire may reduce evacuation efficiency significantly, the movement time increases rapidly with the increasing pre-movement time, and the door width plays a more important role for evacuation in fire than normal condition. Furthermore, for the evacuation from a hall with two exits our model can reproduce the inefficient use of exits and predict the optimal condition that results in least evacuation time.     

A new model for determining neutral-plane position in shaft space of a building under fire situation

In this study, a new model is proposed for predicting the location of the neutral plane inside the shaft space of a building under fire situation. The shaft space of the building is divided into two zones, i.e., fire zone and inner space, in the model. The temperature is assumed uniform in each zone. To validate the proposed model, a parametric numerical study is also carried out by using a computational fluid dynamics (CFD) code, i.e., FDS code. The comparisons between the proposed model and the CFD model are processed and the agreements between the two approaches are satisfied. It is found that the location of the neutral plane is above the mid-height of the building and the simulation results are consistent with the proposed model approximately. The ratio of the neutral plane height to ceiling height depends on the ventilation condition of the fire room greatly, which is found varying between 0.50 and 0.62.     

Studies on buoyancy driven two-directional smoke flow layering length with combination of point extraction and longitudinal ventilation in tunnel fires

This paper investigates the buoyancy-driven smoke flow layering length (both upstream and downstream) beneath the ceiling with combination of point extraction and longitudinal ventilation in tunnel fires. A theoretical model is developed based on previous back-laying model with only longitudinal ventilation, with modified actual heat release rate, as well as modified upstream and downstream opposing longitudinal air flow velocities by the induced flow velocity due to point extraction. Experiments are carried out in a reduced scale model tunnel with dimensionless of 72 m×1.5 m×1.3 m. A LPG porous gas burner is used as fire source. The smoke flow layering length both upstream and downstream are identified based on temperature profiles measured along the ceiling, for different experiment conditions. CFD simulations with FDS are also performed for the same scenarios. Results show that with combination of point extraction and longitudinal ventilation, the smoke flow layering length is not symmetric where it is longer downstream than that upstream. The upstream smoke layering length decreases, while the downstream layering length increases with increase in longitudinal ventilation velocity; and they both decrease with increase in point extraction velocity. The predictions by the proposed theoretical model agree well with the measurements and simulation results.