重气体在障碍物附近扩散的数值仿真

1、Numerical Simulation on Heavy Gas Dispersion near ObstaclesFENG Zhihua1, 2, NIE Baisheng1 & LI Xiangchun1(1 Institute of Resource and Safety Engineering, China University of Mining &Technology, Beijing 100083, China;2 College of Mining Engineering, Taiyuan University of Science &Technolo

2、gy, Taiyuan 030024, Shanxi, China)Abstract: Storage of huge amount of dangerously heavy gas is usually involved in industrial storage and transportation processes. Many species of the hazardous gas are inflammable, explosive, and poisonous. Once the gaseous material leakage occurs, great hazardous i

3、mpact could happen on the environment and residents nearby，so studying on hazardous gas dispersion is very important. In this paper, a trial of liquid propane dispersion near an obstacle was simulated using the commercial software FLUENT. Basic species transport equations were described in detail. T

4、he standard model, implicit segregated solver, gravity item, under-relaxation factors, and boundary conditions were rightly set. Meshing gas dispersion dimension and the simulation results were discussed. The main results are: (1) FLUENT can be used to simulate the heavy gas dispersion in flat and o

5、bstacle terrains, the values of simulation and corresponding experiment data are much approximated. (2)There are some differences between the simulation values and the corresponding experiment data; this is mainly because of the uncertain factors of dispersion itself and the discrepancy between the

6、characteristics of numerical computation and the dispersion process. (3) Better simulating results of heavy gas dispersion near obstacles should be based on specific scenarios, characteristics of flow, and appropriately fine meshes. Keywords: FLUENT; heavy gas; dense gas; simulation; dispersion1 Int

7、roductionThe accidental release of hazardous gas has generally occurred during its process of transport or in storage. Although the frequency of the accident is low, the serious results make it important in the safety evaluation. In fact, many risk managements of the hazardous gas release are uncert

8、ain. For example, the sites where hazardous dense gas will release and its possibility, the toxic effects on the environment and human, the possible population range that may influence, and the effective way to mitigate losses. Investigation of hazardous dense gas dispersion plays an important role

9、in risk evaluation 1. In connection with density difference, the hazardous gas can be classed into the buoyant, the neutral and the heavy. Among them, heavy gas is the most hazardous to human, so it is very important to study on heavy gas dispersion in safety field. Up to the present, researchers ha

10、ve only formed a primary understanding of heavy gas dispersion in flat terrain and made some dispersion mathematical models. For gas dispersion in complexity topography such as slope, near obstacles, and in rough terrain, it is still not to be comprehended completely.With the rapid development of co

11、mputer, to simulate heavy gas dispersion in complex topography using CFD (Computational Fluid Dynamics) becomes a new study field: numerical simulation can not only save much money and time, reduce the risk of the poisoning, but also can make a systematic study on the parameters which affects the di

12、spersion. For example, when investigating the dispersion effects of slope, any interesting value of angles can be taken at will in simulation; yet in field experiment, the the local terrain is confined by specific slope.This paper simulated a trial of liquid propane release near an obstacle using FL

13、UENT, and discussed its applicability in heavy gas dispersion field further.2 A Release Trial of Liquid Propane near an ObstacleThe project Research on Instantaneous and Continuous Dense Gas Release or Major Technological Hazards (MTH project BA 1988-1991) was sponsored by the Commission of the Euro

14、pean Communities. The overall objective was to improve the understanding of accidental dense gas releases. This paper introduced simply one field trial in the series of liquid propane continuous release with momentum focus on the effect of obstacles 4. 2.1 Process Descriptions of the TrialLiquid pro

15、pane was released from a pressurized storage at near-ambient temperature. The source exit was a nozzle producing a momentum plum, and the source strength was about 3 kg/s. The obstacle was a 2-m high wall perpendicular to the wind direction at a distance about 48 m down stream of the sources. In ord

16、er to quantify the obstacle effect, the wall was removed in the middle of the trial. The selected trial was conducted with wind speeds on the order of 2 m/s, humidity conditions. The released conditions are summarized in table 1. Further details are in reference 4.Table 1 Overview of the selected tr

17、ialTrialEEC55Exit pressure p0 /Pa106Exit temperature /13.3Release rate /(kg×s-1)3.0Jet momentum Fjet /N208Effective molar weight M* /(kg×kmol-1)0.099Wind direction relative to ideal /(°)-12Wind speed at 6-m height /(m×s-1)3.2Air temperature Tair /9.9Relative air humidity /%992.2

18、Concentration StatisticsTable 2 shows concentration statistics at the back of wall measured with and without obstacle, ground concentration was extrapolated from known values at higher level. In table 2, Cb and Ca are mole concentrations with and without obstacle.Table 2 Mean mole concentration at b

19、ack of the wallHeight/m0124With obstacle, Cb/%0.70.951.150.42Without obstacle, Ca/%1.71.160.560.183 Introduction of FLUENT and Its Solution Mode in Heavy Gas Dispersion3.1 Introduction of FLUENTFLUENT is a computer program for modeling fluid flow and heat transfer in complex geometries. It is writte

20、n in C computer language. FLUENT provides complete mesh flexibility, solving flow problems with unstructured meshes that can be generated about complex geometries with relative ease 5. The FLUENT CFD package consists of several tools for defining a flow problem, setting boundary and initial conditio

21、ns, and solving the set of complex equations for conservations of momentum, mass, energy, and chemical species. The governing equations are discretized on a curvilinear grid to permit computations over irregular geometries and are solved using a control volume based finite difference method. Discret

22、e velocities and pressures are stored in a non-staggered grid, and interpolation is realized by using a first-order, second-order, power-law scheme or optionally by higher-order upwind scheme. The mesh types include 2D triangular/quadrilateral, 3D tetrahedral/hexahedral/pyramid/wedges, and mixed (hy

23、brid) meshes. FLUENT is one of the better software in simulation heavy gas dispersion, especially in rugged terrain. In the following part, model of species transport in FLUENT is used to predict the local mass fraction of heavy gas in a large dimension with obstacle. Equations, solver models, bound

24、ary conditions, and controls of solutions about species transport are presented in the following subsections. 3.2 Species ModelThe Species Transport model simulates the mixing and transport of different chemical species through the solution of a convection-diffusion equation for the ith species.3.2.

25、1 Species transport equationThe species transport equation is 6: (1)Where, Ri is the net rate of production of species i by chemical reaction (in heavy gas dispersion it equals 0), and Si is the rate of creation by addition from the dispersed phase plus any user-defined sources. An equation of this

26、form will be solved for N-1species where N is the total number of fluid phase chemical species present in the system. To minimize numerical error, the Nth species should be selected as that species with the overall largest mass fraction, such as air in the heavy gas dispersion.3.2.2 Mass diffusion i

27、n turbulent flows In turbulent flows, FLUENT computes the mass diffusion in the following form: (2)Where Sct is the turbulent Schmidt number, t is the turbulent viscosity, and Di,m is the turbulent diffusivity. The default Sct is 0.7.3.2.3 Treatment of species transport in the energy equationFor man

28、y multi-component mixing flows, the transport of enthalpy due to species diffusion: (3)Can have a significant effect on the enthalpy field and should not be neglected. In particular, when the Lewis number: (4)for any species is far from unity, neglecting this term can lead to significant errors. FLU

29、ENT includes this term by default. In equation (4), k is the thermal conductivity. 3.3 Turbulent Models and Setting Boundary ConditionsFLUENT provides six differential models6,7: Spalart-Allmaras model, the model, model, Reynolds Stress Equation Model (RSM), Algebraic Stress equation Model (ASM), an

30、d Large Eddy Simulation (LES). At present, no a general model can be used in every flow problem. So it is very important to select the appropriate turbulent models and setting the boundary conditions rightly for the heavy gas dispersion near obstacle.3.3.1 Turbulence model in heavy gas dispersionThe

31、 right model should be selected basing on the characteristics of flow and the computer sources. The distinctions of heavy gas dispersion from engineering fluid are: (1) The dispersion dimension is very large: from hundreds of meters to kilometers; (2) Dispersion speed is low. The physical properties

32、 of flow are almost unvaried in the dispersion. According to the characteristics and complexity of gas dispersion, the standard model is selected to simulate heavy gas dispersion. The treatment of near-wall adopts the normal wall function7. 3.3.2 Solver in heavy gas dispersionFLUENT provides two typ

33、es of solvers: the segregated solver and couple solver. Both of them are suitable for flow of uncompressible or compressible with high speed. However, it is disposed to couple solver for compressible flow with high speed.For relatively slow flow of heavy gas dispersion, the implicit segregated solve

34、r is selected 3.3.3 Setting the operating conditions and boundary conditions Because heavy gas is taken as incompressible ideal gas, and dispersed in an open dimension, the operating pressure is set to the normal-pressure value. Gravity plays an important role in gas dispersion direction which is ve

35、ry dangerous to residential areas, so the buoyancy forces in the simulation of heavy gas dispersion must be input in the operating conditions. Boundary conditions specify the flow and thermal variables on the boundaries of the physical model while initial conditions specify the distribution physical

36、 variables in dispersion dimension at the beginning of process. They are, therefore, critical components of FLUENT simulations. Setting of boundary conditions includes the source strength, surface roughness, the wind speed as a function of height, temperature etc. For the liquid propane, release is

37、steady in the dispersion process that it is unnecessary to specify the initial conditions.3.3.4 Setting under-relaxation factorsThe segregated solver uses under-relaxation to control the update of computed variables at each iteration. This means that all equations solved using the segregated solver

38、will have under-relaxation factors associated with them. In FLUENT, the default under-relaxation parameters for all variables are set to values that are near optimal for the largest possible number of cases. These values are suitable for heavy gas dispersion also.4 Meshing the Dispersion Dimension a

39、nd Discussing the Simulation Results4.1 Meshing the Dispersion DimensionBecause the obstacle in trial can not be moved during the simulation, FLUENT simulation will be divided into two steps: (1) setting the releasing scenario with obstacle, and simulating the concentration distribution near the obs

40、tacle; (2) setting the releasing scenario without obstacle, and simulating the concentration distribution in flat terrain under the same boundary conditions. The scheme of meshing the dispersion dimension is as followings:(1) Without obstacle, see Fig. 1. The sizes of dispersion dimension are: Lengt

41、h X equals 80 m; width Y equals 100 m, height Z equals 10 m. The length of 3D tetrahedral side equals 1 m, and the origin of Descartes coordinates locates at the left and down corner in fig.1. Fig.2 shows the position of air inlet, heavy gas inlet, and outflow. The air inlet face is on the left side

42、 of dispersion dimension and marked by green-yellow, the direction of air velocity parallels to ground. The face of outflow is on the right side and marked by baby blue, paralleling to the air inlet face. The radius of liquid propane jet is 0.06m, exit of the jet is located at (2, 50, 0), and the di

43、rection of the jet parallels to ground. (2) With obstacle, see Fig. 3. The meshing of dispersion is the same as that without obstacle. The place of obstacle is at (50, 50, 0), and the height of obstacle is 2 meters.From table 1, we know that the effective molar weight of liquid propane is about 99.

44、In FLUENT simulation, we take C7H16 as the substitute whose molar weight is about 100, so the concentration of C7H16 obtained directly from FLUENT should be modified to get the real concentration of C3H8 by the following formula: , where, C is mole concentration.4.2 Results and Discussion The trial

45、of liquid propane is simulated using FLUENT and results are shown in Fig 4-9. Fig.4 is the distribution of C7H16 mole concentration at the ground without obstacle. In order to show the results clearly, Fig.5 is appended as the amplificatory distribution of C7H16 mole concentration. Fig.5 shows that

46、the mole concentration at ground decreases gradually alone with the downwind distance, no sudden concentration change at the centre line. Fig. 6 is the concentration distribution of C7H16 at different height at the centre line of the downwind (i.e. Y=50 m, Z=0, 1, 2, and 4 m, X=45-55 m) without obst

47、acle. It is shown in Fig.6 that concentration of C7H16 attenuates with ascendant height, and this tendency denotes the particular gravity effects in propane vapor dispersion. Fig.1 3D tetrahedral meshes of dimension without obstacle Fig.2 Position of air inlet and outflowFig.3 3D tetrahedral meshes

48、of dimension without obstacleFig.7 is distribution of C7H16 mole concentration at the ground with obstacle, and Fig.8 is the amplificatory C7H16 distribution. Fig.8 also shows that the mole concentration at ground attenuates alone with the downwind distance, but there is obviously breaking concentra

49、tion changes near the obstacle at the centre line, and this phenomenon shows that the obstacle modified the concentration distribution rear it. Fig. 9 reveals the mole concentration distribution of C7H16 at different height in the centre line of the downwind (i.e. Y=50 m, Z=0, 1, 2, and 4 m, X=45-55

50、 m) with obstacle. Fig. 9 shows that mole concentration of C7H16 is augmented with increasing height at the front of obstacle while the distribution of mole concentration becomes uniform at different level behind the obstacle. So the concentration in the wake of obstacle is strongly effected by it.

51、Data of experiment are compared with the results of simulation in Fig. 10 and Fig. 11. (Ceb and Csb denote the experimental and simulated data with obstacle; Cea and Csa denote the experimental and simulated data without obstacle). Fig.10 shows that the simulated concentration is smaller than that o

52、f experiment slightly before the obstacle is removed when the height is less than 4 meters while the opposite tendency presents at the height of 4 meters. Fig.11 shows the similarly simulated sequence to that in Fig.10 after the obstacle is removed: the simulated concentrations are less than that of

53、 experiment when the height is less than 2 meters, and an inverted trend at the level of 2 to 4 meters. It seems that the values of simulation are at a closer range in height than that of experiment. However, from an overall view, the results of simulation are much approximated the data of experimen

54、t.According to Britter9: in heavy gas dispersion, no matter what models among integral model, semi-experienced model, CFD model, SLAB model, and shallow model, it will be better if the difference between the simulation values and that of experiment is within 2 times. In FLUENT simulation, the ratios

55、 of simulation concentration to corresponding experiment data are all within 2 times, so the FLUENT is applicable software to simulate heavy gas dispersion near obstacles. This is much significative for the researchers of heavy gas dispersion: “They can break away from confines of repeating and comp

56、lex programs and focus on the principles of gas dispersion” 10. The better application of FLUENT in heavy gas dispersion paves a way to solve the practical problem.However, there is a certain difference between the simulation concentration and the corresponding experiment value. The differential is

57、caused mainly by (1) the uncertain factors of gas dispersion itself which greatly affect heavy concentration distribution. For example, the measured concentration distribution is influenced by the capacity of analyzing instruments, the air scale of turbulence, air stability, wind speed, terrain, and surface roughness. Tiny difference of above factors may induce various results. (2) The discrepancy between the characteristics of numerical computation and the dispersion process. CFD simulated the concentration distribution by way of solving the turbulence models based on mas