Smoke simulation and emergency evacuation simulation under indoor fire scenario

According to relevant data, when a fire occurs in a large public building, human life and property will be greatly threatened. In order to solve the problem of personnel evacuation under fire conditions, this paper starts with ant colony algorithm and introduces heuristic function, node selection strategy, and pheromone update to improve ant colony algorithm, so as to provide a scientific basis for personnel evacuation after fire. Based on BIM technology, smoke simulation technology, and evacuation simulation technology, this paper studies the evacuation of people in fire scenarios. Simulation of fire smoke and personnel evacuation is carried out on the BIM model of the shopping mall. Using Pyrosim software to simulate the spread of smoke in the shopping mall after the fire, the smoke temperature, fire visibility and CO concentration of each node are obtained.


Introduction
Pyrosim is based on FDS (Fire Dynamic Simulation) development and creation, visual interface makes its unique characteristics, greatly convenient user operation [1].Pyrosim software mainly includes three aspects.The first is software model building, which can be directly built or imported through format conversion.The modeling method used in this article is to import the DFX format into Pyrosim to complete model building [2].Pyrosim software mainly includes three aspects, the first is the software model building.The second is FDS, because Pyrosim is developed on the basis of FDS, so it also includes the calculation function of FDS, which can predict the smoke concentration, the temperature of the test point, and the concentration of various toxic and harmful gases when a fire occurs [3].The third is Smokeview, which is mainly used to demonstrate the animation of smoke spread after the simulation is completed.BIM construction technology development is a technology that has developed rapidly in the fields related to construction engineering at home and abroad in recent years [4].It has been widely used in the field of modern construction engineering construction [5].As an important carrier of building digital information system integration, building information model (BIM) runs through all key stages of the life cycle of the construction industry and guides the basic design, construction, production and operation process management of the main structure of the building [6].

Path Optimization of Improved Ant Colony Algorithm
As a kind of heuristic algorithm, the ant colony algorithm has strong robustness and adaptability in dealing with path optimization problems, and is suitable for dealing with non-convex nonlinear problems of the NP-hard class.The general ant colony algorithm suffers from the problems of easily falling into local optimal solutions, large and time-consuming computation, and premature convergence [7].In order to effectively compute the global optimal solution, this paper improves the node selection method, the heuristic function and the pheromone update method in the algorithm to address these problems exhibited by the ant colony algorithm [8].The improved algorithm is improved in terms of solution accuracy, quality, and convergence.

Effect of Visibility on Evacuation Speed
By analyzing the experimental data, we found that the light reduction coefficient s K is related to the evacuation speed of the personnel, and the two are negatively correlated.

Effect of harmful gases on evacuation speed
We set a concentration of 0.25% CO as the threshold value for personnel incapacitation.
Where  is the concentration of CO and t is the length of time that evacuees are exposed to CO.

The effect of smoke temperature on the evacuation speed of people
Substitute the above data into the formula, after simplifying the effect of smoke temperature on the evacuation speed of personnel is calculated as: where there are three addresses, you should insert numbered superscripts 1, 2 and 3 to link surnames to addresses and then insert footnotes 4 and 5.Note that the first footnote in the main text will now be number 6.

Update of pheromone volatile factor 
A dynamic adaptive approach is adopted for  , which is updated during each iteration to ensure a balance between the global search capability and the convergence speed.Its update formula is as follows: where  is the constant used to control the decay rate of  , and min  is the minimum value of  , which can be adjusted artificially as needed.

Project Introduction
This mall is a three-story frame structure with a building area of 11575 m 2 .The floor area of each floor is 3858 m 2 , the floor height is 15m, and the height of each floor is 5m.There are 6 exits on the first floor connected to the outside world, 5 staircases, and 2 elevators leading to the second and third floors.The mall is equipped with food area, clothing area, furniture area, household department store, etc., with complete facilities and large customer flow.The structure of the mall is designed to have a service life of 50 years, with a waterproof grade of three and a fire resistance grade of two.

Select template
According to the information of each floor of the mall given in the drawing, open Revit 2016 to draw the 3D model of the mall.Firstly, draw the elevation and axis network in order according to the information on the drawing, secondly, determine the position of the foundation, wall, column and beam according to the position of the horizontal axis and vertical axis.Finally, windows and stairs are drawn .According to the characteristics of the building and the subsequent requirements for simulation, the structural part of the model is drawn according to the order of walls, columns, beams and floor slabs, and then the architectural parts such as doors, windows, curtain walls and canopies are drawn for the model, and finally the 3D model of the shopping mall is shown in Figure 1 and 2.

Simulation of fire smoke in shopping malls
Since the evacuation simulation model built by Pathfinder can only simulate the path and evacuation time of people, it cannot restore the real situation of the fire scene by parameter setting.Therefore, smoke simulation software needs to be added to restore the fire scene fire trend through the setting of the fire scenario [10].The tolerance limit time of evacuees is inferred through the collation and analysis of key data, and this time is used as one of the criteria to determine whether people can escape safely.

Division of the grid
Pyrosim's calculations are more accurate thanks to the grid division.If the building is irregularly shaped, the system will automatically generate a regular grid.If a part of the building is not meshed, FDS will not report an error when running, but the dynamic simulation process in the unmeshed part of the building will not be displayed in Smokeview.To achieve simulation accuracy, it is best to match the grid to the building size.The grid is set using a large vortex simulation and the fineness of the computational grid is directly related to the computational accuracy and computational resource usage.Therefore, the designed grid space size is 94m*61m*19m, and the grid size should be 1.0m*1.0m*1.0maccording to the feature size calculation formula given in the Pyrosim instruction manual.The number of meshes is 108946.The smaller the mesh size, the higher the accuracy of the calculation; the larger the mesh size, the lower the accuracy of the calculation.

Fire scene setting
There are many factors to consider when setting up a fire scenario, so when designing a fire scenario, the specific location of the fire, the amount of combustible material at the time of the fire, the distribution of people, and the structural characteristics of the building are usually taken into account.
When setting up a fire scenario, the most fire-prone location should be set as the location of the fire start point according to the actual situation of the mall.And in the process of fire development there are many uncontrollable factors, should be prepared for the worst case of multiple emergencies.With the development of the fire, the various stairways of the mall will also be filled with smoke, and the smoke will quickly spread to the second and third floors under the addition of high temperature, so the fire risk is higher and the setting is more reasonable.

Fire simulation initial conditions setting
The mall flue gas model built after steps 1-4 is shown in Figure 3.

Analysis of Pyrosim Simulation Results
According to the relevant data show that temperature, CO concentration, smoke visibility is the main factor affecting the evacuation of personnel, so these three factors pose a threat to the lives of personnel at the moment, as the life safety standards.That is, the evacuation process smoke temperature, CO concentration, smoke visibility in these conditions personnel can safely evacuate: (1) smoke temperature: the fire when the temperature is less than 120 ℃ below to ensure the smooth passage of personnel, greater than this temperature will be the evacuation of personnel repulsive force.
(2) when CO volume percentage reaches 0.25%, personnel will not have sufficient oxygen and fainting and other symptoms, resulting in the normal evacuation of personnel, so CO concentration should be less than in 0.25% to ensure the smooth passage of personnel; (3) in the fire after the building structure and area will affect the visibility of the smoke, in order to allow personnel to identify the location and see the evacuation path after the fire, so The time point to reach the critical temperature of 120℃ is T1, the time point to reach the CO concentration of 0.25% is T2, and the time point to reduce the visibility of smoke to 10m is T3.To ensure the safety of personnel, the tolerance limit of personnel should be taken as T=min{T1,T2,T3}.After setting the requirements, Pyrosim is run and 3D Smokeview is obtained.the spread of smoke with time and the distribution of fire temperature field are shown in Figure 4.    From the above figure, we can see that when T=90.2s, the fire smoke starts to spread to the lobby; when T=155.9s, the fire smoke starts to spread to the 2nd floor; when T=401.1s, the fire smoke starts to spread to the 3rd floor; when T=600s, the fire smoke has filled the whole building.
After simulation by Pyrosim software, the CO concentration values, visibility values and smoke temperature values of several key nodes at different time periods were obtained by setting temperature detectors, smoke detectors, thermocouples, etc. See Table 1-3.The data in the tables will be used in the improved ant colony algorithm.From the data analysis, it can be concluded that due to the large area of the mall and the complete ventilation system, the heat does not gather too much in one place.By simulating the temperature and circulation of airflow in the mall, it can be concluded that the variation of each node at different time points is small, and the temperature of other nodes, except for the above nodes, is basically kept within the acceptable range for human body.In terms of smoke, the mall has a large space and a good smoke exhaust system, and the exterior is mostly walls, floor-to-ceiling windows and more vents.Therefore, except for the above nodes, the CO values of the other nodes in the fire scenario are less than 0.10%, which has less impact on the passage.Due to the large and spacious arrangement of the mall, there is a good natural and artificial smoke exhaust system.The fire smoke simulation shows that the overall visibility is good, except for some nodes that are close to the fire source, all nodes are kept above 10m.Through the simulation process, it can be seen that the indoor smoke temperature is gradually rising with time, and when it reaches 582s, the temperature of some nodes rises to 120℃, T1=582s; at 545s, the CO concentration reaches 0.25%, T2=545s; at 588s, the visibility of smoke decreases to less than 10m, T3=588s.When only one of the three influencing factors reaches limit, the difficulty of indoor personnel to escape is great, so in order to ensure the safety of personnel life, personnel tolerance limit should be set as T=min{T1,T2,T3 } = min {582s,545s,588s} =545s.

Model Generation
In this paper, we use Pathfinder to build a fire evacuation simulation model of a shopping mall, and use BIM technology to highly restore the information of the building and complete the scenario setting for evacuation, so that people can carry out evacuation simulation in the mall.The model of the mall had already been built through Revit, and the next task was to import the model into Pathfinder software to complete the construction of evacuation scenarios.As with Pyrosim, the import of the BIM model also requires a format change, and the BIM model is imported into Pathfinder in DXF format.Compared with IFC 2×2 standard, IFC 2×3 is similar to providing a container to map Revit components to IFC one by one, so that more Revit components can be self-adapted in Pathfinder.To make it more convenient and intuitive to see the simulated evacuation process, we need to use the settings in Pathfinder to change the style of the BIM model after importing the model.The final model obtained is shown in Figure 8.

Determination of shopping mall evacuation parameters
After the evacuation simulation scenario is built, it is necessary to refer to the relevant regulations to decide the number of evacuees, evacuee type, walking speed, body width, evacuation mode, etc.The setting of evacuation parameters in Pathfinder will have a direct effect on the credibility of the simulation.

Number of evacuees
The number of evacuees is one of the basic parameters of evacuation simulation in Pathfinder.Accurate calculation of the number of people in the building can ensure the process of evacuation simulation is more scientific and representative.According to Table 5.5.21-2 of GB50016-2014 "Building Design Fire Code", the number of evacuees in a store should be multiplied by the floor area of each floor of the business hall to determine the density of personnel.The data of personnel density is shown in Table 4, and the calculation of evacuation personnel is shown in Table 5.

Personnel type and personnel parameters
The setting of personnel parameters include personnel type, personnel speed and body width.According to the actual situation observed in the mall, it is found that most of the people entering the mall are adult men and adult women, and some adults will lead is elderly or children, together with the mall.Therefore, set the type of people in the mall for four types of adult men, adult women, children, and the elderly.In order to match the actual observation, set adult men and women each 40%, children 10%, and the elderly 10%, set out in Pathfinder the shoulder width of various personnel steps, etc. as Figure 9.
When setting the walking speed of personnel, it is necessary to take into account the impact of sudden events on people's hearts, and other factors such as the temperature of the fire scene and the blockage of personnel can limit the walking speed of personnel after a fire.Therefore, the study set the walking speed to 0.8m/s for the elderly, 0.9m/s for children, 1.35m/s for adult men, and 1.15m/s for adult women.

SFPE Movement Mode
The SFPE movement mode follows the rules in the Fire Protection Design Manual, where individuals are independent of the radius of the cylinder and can penetrate each other, the speed of people walking is determined by the density of the room, and individuals will ignore the time factor and automatically choose the shortest path to escape.Therefore, the group in SFPE movement mode will pass along the inner side of the stairs in order when encountering stairs.

Steering movement mode.
The group in the Steering movement mode is able to complete the evacuation simulation by planning the path, controlling the flow and collision handling mechanisms.Under the premise of determining the behavior of individuals, setting thresholds for the distance to the group and the distance to the nearest escape exit, a new path will be generated after calculation and comparison.Individuals will not penetrate each other in any scenario and keep a certain distance.At the same moment, the same path planned between individuals will collide and form a report in the Run Simulation dialog box.In order to make the simulation results more realistic, the mall evacuation simulation will choose the steering motion mode.The simulation model with parameters set is run for a trial period, and the movement time of evacuating people in the Steering mode is 544.5s.

Analysis of Evacuation Results
Through analysis: exit passage rate of the whole mall is the highest in the first 60s, and the exit passage rate reaches 5 per/s at the highest, and decreases slowly after 60s.In the process of evacuation, the entire mall has the highest exit passage rate in the first 60s, and the highest exit passage rate reaches 5 per/s, which slowly decreases after 60s, the main reason for this phenomenon is: the crowd on the first floor evacuates out quickly.And the crowd on the upper floors took more time to pass the stairs, so it caused the lower evacuation rate after 60s.(3) Exit 1, 3 and 4 will be 0 at 100s, which means that the design of these three exits is defective and cannot be fully utilized in case of fire, and needs further improvement; Therefore, it is recommended to increase the width of stairways and evacuation stairways reasonably in the design.

Conclusion
This chapter describes the process of constructing a fire smoke model and a fire evacuation simulation model by using BIM models, and analyzing the data obtained to determine the tolerance limits, evacuation routes, and evacuation times of personnel.First, the smoke spread in the mall after the fire was simulated by pyrosim, and the tolerance limit time of the personnel was obtained by analyzing the smoke temperature, smoke visibility, and CO concentration data.This time is used as a criterion for the safe evacuation of people and as a basis for comparing the results of intelligent algorithms.Finally, the mall model is imported into Pathfinder for evacuation simulation, and the evacuation time is used as a criterion to judge the success of the intelligent optimization algorithm.The design flaws are also analyzed based on the evacuation time and exit passage rate, and suggestions are made to rectify them.

Figure 1 .
Figure 1.3D model of the mall.Figure2.Rendered mall model.

Figure 2 .
Figure 1.3D model of the mall.Figure2.Rendered mall model.

Figure 4 .
Figure 4.When T=90.2s, the fire smoke began to spread to the hall.

Figure 5 .
Figure 5.When T=155.9s, the fire smoke began to spread to the 2nd floor.

Figure 6 .
Figure 6.When T=401.1s, the fire smoke began to spread to the 3rd floor.

Figure 7 .
Figure 7.When T=600s, the fire smoke has filled the whole building.

Figure 12 .
Figure 12.Evacuation time and personnel passing rate.
(1) according to Figure 12(a), as the evacuation time increases, the number of people staying indoors becomes smaller and smaller, and the number of people evacuated to safe places increases, and the speed of evacuation also gradually decreases, reflecting the principle of "fast is slow"; (2) by analyzing the time change of people Figure 12(b), during the evacuation process, the ICCSS-2023 Journal of Physics: Conference Series 2613 (2023) 012022 IOP Publishing doi:10.1088/1742-6596/2613/1/01202210

Table 5 .
Calculation of Evacuated Persons.

Table 6 .
Shopping mall personnel characteristics and distribution of each floor.