APP下载

隧道内车辆运输储罐泄漏源机器人仿生感知算法研究

2015-03-19ZhiqiZHAOJiandongFANG

机床与液压 2015年6期
关键词:储罐运输隧道

Zhi-qi ZHAO,Jian-dong FANG

(Inner Mongolia University of Technology,Hohhot010080,China)

1 In troduction

With the fast development of industrial processes,the demand of hazardous gases gets increased in industry and related industries as a raw material or production ofmaterials,and the pressure of the transported dangerous good gets increased in order to transport more goods at one time[1-2].Accidental release of dangerous gases phenomenon often occurs in the course of transport due to the affected of long distance,humanity,the weather and other uncertainties during the transportation.Due to the feathers of toxic,flammable,explosive and so on,once the dangerous gas leakage occurs,it not only will cause serious damage to property,but alsomake a serious threat to the safety of personnel[3].

Especially hazardous gases leakage occurred in the tunnel and explosion and burning,the losses could become incalculable.Investigate its reason,relative to the open road,the space in the tunnel is relatively small and closed,once the hazardous gases get fired,the temperature,smoke density and concentration of harmful gases inside the tunnel will get increased sharply in a short time,it could easily cause continuous combustion of vehicles or explosive chain reaction,and the life of passengers will get a great threat[4].Meanwhile,as an important transportation hub,tunnel firewill cause traffic disruption,long road congestion and big economic losses.Therefore,how to timely and accurately determine the location of the gas leakage source plays an important role to reduce the property damage and personnel casualties when the hazardous gases leakage occurred,in practical terms,due to the hazardousmaterials have the characteristics of toxic,flammable and explosive,artificially accomplish the above tasks are highly dangerous.Therefore,the robot operation under hazardous conditions has gradually become a hot research field of robotics,since it could replace humans to complete the assigned task under inappropriate or hazardous conditions.

This paper proposed a hazardous gas detection system for mobile robot,bionic robot olfactory information is used to position the dangerous gas leakage source and combine the active olfaction positioning algorithm and physical and mathematicalmodels of gas diffusion which is based on Fluent to study and locate the source of the leakage occurred in the tunnel,and restore the real situation of the tunnel,it could provide a reference for the practical applications[5].

2 Fluent sim u lation of gas d iffusion of vehicle transport tank leaks in the tunnel

Fluent software is used to simulate the distribution of the gas leakage in the tunnel.

By repeatedly solving a large number ofmass conservation equation and momentum conservation equations,the fluent constitute the basic framework to solving flow problems.

The mathematical equation ofmass conservation is as follows:

Where,t is the time,ρis the fluid density;u,v,w are the components of velocities in the x,y,z direction,respectively.

The mathematical expression ofmomentum conservation equations are as follows:

Where,u is the dynamic viscosity;p is the pressure,Su,Sv,Sw are the generalized source terms for momentum equations.

2.1 Select the turbulencemodel

There exist many models of turbulence such as Spalart-Allmaras,Reynolds stress,k-ε and large eddy simulation[5].

k-εmodel iswidely used because it has the characteristics of processing the speed and length of the turbulent fluctuation simultaneously.k-εmodel could be divided into standard,overlapping technology and achievable models.Standard k-ε model is applicable to the initial iteration,design selection and parameter studies.Overlapping k-ε model is applicable for the complex shear flow which involves rapid strain,moderate vortex and local conversion;Achievable k-εmodel has better accuracy than that of overlapping k- ε model.According to the practical requirements,the standard k- ε model is adopted in this paper.

The transport equations of standard k-εmodel are as follows:

Where,Gkis produced by the turbulent kinetic energy which is caused by the average velocity gradient and Gbis produced by the turbulent kinetic which is caused by the buoyancy influence.YMis the effect of compressible turbulent fluctuation expansion to the total dissipation law.C1ε,C2ε,C3εare the empirical constants.The default values for these parameters in FLUENT are as follows:C1ε=1.44,C2ε=1.92,C3ε=0.09;

2.2 Establish the geometricmodel and set the boundary condition

Since it is a non-steady-state process about the diffusion processwhen the natural gas leakage occurs and the gas concentration,turbulence intensity will get changed as the time during the diffusion process.Assume the fluid is homogeneous flow(i.e.,no phase transition)when the dispersion process ismodeled.

Before the gas diffusion simulation by using Fluent,the pretreatmentwith Gambit is necessary[6].Simulate the real environment inside the tunnel.The tunnel is north to south and the leakagewas occurred at a distance of9meters to the inlet.The drain port radius is 10 cm and the inlet velocity is 10 m/s.The structure of themodel is shown in Fig.1.

Fig.1 The calcu late area

Generate themesh of 10×50 in XY plane by Gambit according to computational domain size in Fig.1.The upper boundary is the inlet of tunnel and the boundary condition is set to velocity-inlet.The left and right borders are the walls of tunnel and the boundary condition is set toWall.The lower boundary condition is exit of tunnel and the boundary condition is set to pressure-outlet.The boundary condition of gas leakage port is set to velocity-inlet.The mesh model generated by the Gambit is shown in Fig.2:

Fig.2 The m eshing figure

2.3 Numerical simulation of tank leak in the tunnel

Since natural gas is a kind ofmulticomponent gas and it will mix with air during the leakage process,the species transport model is used to simulate the process of natural gas leakage.The simulation of leakage for natural gas is considered in this paper and there is no chemical reaction,therefore the chemical reaction model will not be activated[7].During the definition process of boundary condition,the temperature of gas leakage port is300 K,set the componentof h2 s is 0.029 and ch4 is0.971.The rate of gas leakage is 2.5m/s,the turbulence intensity is10 and the hydraulic radius is 0.2.The temperature of inlet is 300 K,the component of h2s and ch4 remains 0.The velocity of inlet is 10 m/s,the turbulence intensity is 10 and the hydraulic radius is 5.The total pressure of the tunnel exit is0,the turbulence intensity is10 and the hydraulic radius is 5.

The effect of different leakage time for natural gas leak diffusion process is presented in the following section.

Natural gas could become a flammable gas when it ismixed with air after the leakage.Then,serious accidents such as fire or explosion could be caused when the fire is encountered,and the length of time has a direct impact on the severity of the accident.The space is relatively closed in the tunnel;long time leak of nature gas could cause a sharp increase in the concentration of gas,and itmight get exploded when the concentration of gas reaches the explosive limit,the consequences could be disastrous.Therefore,to accurately obtain the relationships between the concentration of the gas and the leakage of time could help people take the appropriate first aidmethodswhen the accident occurred and prevent the casualties.

The diffusion clouds of ch4 at different diffusion time are shown in Table 1 and Table 2.

Table 1 The diffusion cloud o f ch4

As shown in Table 1 and Table 2,the velocity of inlet is10m/s,the rate of leakage is2.5m/s.At the beginning of gas leakage,the gas diffusion concentration distribution ismore continuity and regularity,and it has the similar diffusion state with Gaussian model and the plume spread toward to the downwind.As the time goes by,affected by the turbulence in the space field,the turbulence becomes apparent and the leakage gas diffusion sharply throughout the tunnel,the concentration of gas is gradually increased.After 20 seconds,the plume has filled out the tunnel.

Table 2 The diffusion cloud of ch4

3 Robot odor localization based on evolutionary gradient algorithm

3.1 Active olfaction positioning algorithm

This paper presents an active olfaction positioning technology used in mobile robots.Assignmentmethod is used to search plume randomly in the design and a search algorithm is used to track the plume,the algorithm is called evolutionary gradient search algorithm.Fig.3 is the flow chart of the algorithm.

Fig.3 The flow chart o f evo lutionary gradient search algorithm

Mobile robot[8]is controlled by evolutionary gradient search algorithm,it is also called gradient search algorithm.The initial position of themobile robot is O and the concentration is C(O)at this location.Compare the concentration between C(O)and the eight points around O,the eight points are right in front of point A,right point of B,right rear point of C,front point of D,rear point of E,left front point of F,left point of G and the left rear point of H.After comparing the above-mentioned concentration,the robotmoves to the pointwhere ithas the highest concentration,and this cyclewill get continued.Assume that the initial position of themobile robot is O point,the coordinates are(tr,tc),the concentration of the point C(O)is C(tr,tc).First,the mobile robot moves a distance OA and reaches point A,the concentration of point A is C(A),if C(A)>C(O),put the value of C(A)assigned to C(O);If C(A)<(O),to maintain C(O)=C(tr,tc).Then the mobile robot moves a distance AB and reaches point B,the concentration of point B is C(B),If C(B)>C(O),put the value of C(B)assigned to C(O);If C(B)<(O),to maintain C(O)=C(tr,tc),accordance with the above procedure,successively compare the concentration between point of A and point of C,D,E,F,G,H,make the higher concentration of point assigned to C(O).After completion of the comparison,assuming that the point of G has the highest concentration,the next step will use the point of G as the search starting point of O,and compare the concentration between G and the 8 point between G,make the higher concentration of point values assigned to C(O)and this cycle continues.By using this method of assignment to search the plume randomly and in accordancewith the gradient search algorithm to track the evolution of plume,it could guarantee the mobile robot move to the place which has the higher concentration of plume.Therefore,the trajectory of the mobile robot will become increasingly close to the odor source.When Robot cannot find the point where have higher concentration than its own in the surrounding eight points,this shows that the robot reaches the vicinity of the odor source and the robotwill stop moving.

3.2 Comparison and analysis of the simulation results

Completed plume simulation data is exported from the Fluent and combine with simulated robot in Matlab.Simulation data of gas concentration at different time periods are stored as a file in ASCII format by Fluent and these datawill be imported into Matlab,so the two-dimensional simulation of the tunnel plume model are copied to the Matlab.Simulation robot will go through the ASCII file in Matlab at different positions and search gas concentration value of corresponding position follows the active olfaction positioning technology based on evolutionary gradient algorithm.Therefore,the robot could move in the plume model,and search and locate the position of odor source.Simulate the leakage source location from the initial position and the diffusion time.

1)The robot has the same initial position and the plume has different diffusion time.

As shown in Table 3 and Table 4,the concentration of the gas is presented at different time with different colors.The red area has the nearest distance from the source of the leakage and it has the highest concentration value,the concern of yellow-green area is lower than the red area and the dark blue area has the lowest concentration.The trajectories of robot are shown in red dotted lines.From the Table 2,under the condition of the robot has the same initial position and the plume has different diffusion time,the plume has the different diffusion distance in the X axis and Y axis.Initially,the plume shows a uniform and circular diffusion,with the time goes by plume diffusionmainly for Y-axis direction and diffusion of X axis is very small However the robot in four figures are all start moving from the point of(2.8,2)and finally navigate to the maximum concentration point(1.7,6),this proves that the robot could successfully find the odor source.The simulation results show that different diffusion time might affect the diffusion distance of plume in the X-axis and Y-axis,but it does not affect the robot to locate the odor source under the condition of the same starting point.

Table 3 Concentration o f the gas at different tim e

2)The robot has the different initial position and the plume has same diffusion time.Unified the diffusion time is 15 s.

From Table 5 and Table 6,under the condition that the robot has the different initial position and plume has same diffusion time,the robot will successfully target to the point ofmaximum concentration,and the point is(1.7,6).The above simulation results show that the starting point of the robot does not affect the type of robot for positioning the odor source at the same concentration field.

Table 4 Concentration o f the gas at different tim e

Table 5 Diffirent initial position

Table 6 Diffirent initial position

4 Conclusions

This paper established amathematicalmodel by using Fluent for the situation of vehicles transporting natural gas leakage in the tunnel.The numerical sim-ulations have been conducted for the actual phenomena and the influences of time on the plume diffusion are analyzed based on thismodel.Based on the evolutionary gradient algorithm,the robot active olfaction positioning algorithm is proposed.Verify the validity and correctness of this algorithm to the leakage source location in the condition of time-varying gaseous fluid based on the discussions of the initial position of the robotand the length of time to the gas leak.Robot positioning technology for leakage sources in the tunnel could provide a theoretical basis to the dangerous gas leakage accident occurred in the tunnel.Also,once the gas leakage occurred inside the tunnel,to promptly cut off the source of leakage and enhance ventilation to prevent the concentration of gas sharply increases in a local area.

Acknow ledgem ents

This paper is supported by National Natural Science Foundation of Inner Mongolia Autonomous Region(No.2014MS0619)and China Science and Technology Project of Inner Mongolia(No.20120304).

[1]Xuhai Pan,Chengjun Jiang.Statistical analysis of significant spills and accident patterns research[J].Chemical Industry and Engineering,2002,3(19):248-249.

[2]Xiuhai Pan,JunchengJiang.Risk analysis of hazard tank sudden spill for real-time environment[J].Journal of Environmental Sciences,2004,24(3):539-540.

[3]Jie Zhao,Sijing Ren,ChongxinCui.Dangerous operation robotmanipulator control system design based on ARM[J].Industrial Instrumentation & Automation,2013,3:111-112.

[4]Chao Wang.Diffusion of atmospheric accidental release of dangerous chemicals[D].Dalian:The master’s degree thesis of Dalian Maritime University,2008:3-5.

[5]Hongjun Zhu,Yuanhua Lin.FLUENT 12 fluid analysis and engineering simulation[M].Beijing:Tsinghua University Press,2011,88-91.

[6]Weixing Xu,Shouqi Yuan.Submersiblemixers stirring fluency analysis based on FLUENT[J].Mechanical Design and Manufacturing,2011,9:155-157.

[7]Zonghua Cheng,JianchunFan,Ming Du,et al.Uncontrolled blowout toxic gas diffusion simulation based on FLUENT[J].China Safety Science and Technology,2013,9:24.

[8]Lino Marques,AnibalT.de Almeida.Electronic Nose-Based Odour Source Localization[J].IEEE,2000:36-40.

猜你喜欢

储罐运输隧道
大型LNG储罐设计计算关键技术
大型LNG储罐珍珠岩在线填充技术实践
基于地震响应分析的大型LNG全容式储罐储罐基础方案设计
预见2019:隧道的微光
神奇的泥巴山隧道
黑乎乎的隧道好可怕
一生清廉
—— 储罐
受阻——快递运输“快”不起来
比甩挂更高效,交换箱渐成运输“新宠”
LED隧道照明节能改造探讨