You are on page 1of 11

Journal of Wind Engineering

and Industrial Aerodynamics 74—76 (1998) 315—325

An evaluation of topographical effects on neutral


and heavy-gas dispersion with a CFD model
Jan Burman
Defence Research Establishment, Division of NBC Defence, S-901 82 Umea> , Sweden

Abstract

Three different scenarios are studied to estimate how different structures at the side of a road,
influence concentration levels. A solid fence gives the strongest effect on concentration. It forces
the mean flow to move vertically which promotes mixing in the wake behind the fence. A hedge
reduces the level of turbulence and the level of concentration becomes higher due to less
turbulent dispersion. A road-valley generates turbulence in an intermediate regime. It captures
a part of the plume in the valley which gives high concentrations locally. Atmospheric stability
increases the concentration levels at a distance from the topographical disturbance. A heavy
gasplume encountering a solid fence is broadened in front of the fence depending on the height
of the fence and internal stability in the plume. The relation between these parameters, wind
speed and the source rate will also affect the level of concentration downwind the
fence. ( 1998 Elsevier Science Ltd. All rights reserved.

Keywords: Topographical effects; Heavy gas; Dispersion; k—e

1. Introduction

Transport of industrial chemicals through urbanized areas is a potential threat


since accidents with releases of toxic gases can occur. This threat has to be minimized.
One way to achieve this is to make the level of concentrations after a hypothetical
accident as low as possible. This paper concerns how dispersion of a released gas is
influenced by topography or obstacles at the side of the road. The scenarios are
chosen in cooperation with the Swedish Rescue Services Agency (SRSA). In each case
only one parameter is changed to reveal the significance on the field of concentration.
All cases are calculated as steady-state cases. This will give the same result as a release
lasting more than approximately 4 min, in the range of 200 m at a wind speed of 2 m/s.
The study concerns the effect on dispersion from the surroundings and as the assumed
release time is not unrealistic [1], the steady-state assumption is taken to be satisfactory.
The wind direction is in all cases orthogonal to the road. The road is assumed to be
10 m wide. The obstacles considered here are a solid fence, a hedge and a road-valley.

0167-6105/98/$19.00 ( 1998 Elsevier Science Ltd. All rights reserved.


PII: S 0 1 6 7 - 6 1 0 5 ( 9 8 ) 0 0 0 2 8 - 2
316 J. Burman / J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 315–325

2. Model description

2.1. Basic equations

The Reynolds-averaged conservation equations for mass, momentum, energy (tem-


perature) and concentration are used to simulate the processes of interest. It is
assumed that the averaged ideal gas law can describe the state of the air as well as the
state of the released gas.

2.2. Description of density variation

The density is assumed to have a linear dependence on the concentration of the


released gas. The variation of density is described as

A A B B
R
oN "oN 1# !!1 C/C , (1)
! R 4
'
where o is the density of the mixture between the air and the released gas. The
subscripts g and a in Eq. (1) stand for released gas and the surrounding air, respective-
ly. C is the source concentration.
4
2.3. Turbulence description

The equation for momentum is closed by using the k—e model which is a first-order
closure using the eddy viscosity concept which relates the Reynolds stress term to the
gradient of the mean flow. The same concept is used for heat. Two separate equations
are used to solve for k and e, respectively. The turbulent kinetic energy is denoted by
k and the dissipation of turbulent kinetic energy is denoted by e [2].
Both heat and airborne particles can be assumed to follow the same physical
pattern during the turbulent dispersion process [3]. Therefore, the concentration
equation is closed in the same way as the equation for heat, and the same Prandtl
number is assumed to be applicable in both equations. The stratification is evaluated
through the gradient Richardson number Ri, which relates the buoyant production
term to the destruction terms found in the equation for kinetic energy [4]:
g Lo*/Lx
Ri"! i i . (2)
o*((LuN /Lz)2#(LvN /Lz)2)
In the definition of the gradient Richardson number the potential density is used
instead of the ordinary density [4]. When h"¹, then follows by definition o"o*
and Eq. (1) can be inserted. This gives the following relation when the potential
density is replaced by the potential temperature:
g Lh/Lx g DLC/Lx
Ri" i i ! i i . (3)
h((LuN /Lz)2#(LvN /Lz)2) (C /D#C)((LuN /Lz)2#(LvN /Lz)2)
.!9
D in Eq. (3) equals (R /R !1).
! '
J. Burman / J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 315–325 317

The Prandtl number for heat is defined as the ratio between the non-dimensional
/ functions for heat and momentum [3]. The / functions are given by Ref. [5]. The
Prandtl number for mass and heat is denoted by p . For the stable case an
hC
approximate expression for f is used, see e.g. Ref. [6]. P denotes the production of
4
turbulent kinetic energy due to shear and is given by Ref. [7] and P represents the
"
contribution from buoyancy effects. Generation of turbulent kinetic energy by buoy-
ancy effects are written as P "P #P . The thermal contribution is written as
" h #
g g l LhM
P " w@h@"! i 5 . (4)
h hM hM p Lx
hC i
The contribution from the released gas is modelled by comparing P with the equation
h
for the Richardson number, then the contribution from released gas becomes (D as in
Eq. (3))
gD l LC
P" i 5 . (5)
# (C /D#C) p Lx
.!9 hC i
The processes studied reside in the surface layer and therefore the Monin—Obukhov
theory is used to describe the boundary conditions. The surface is described by Ref.
[8].

2.4. Model evaluation

The effect of buoyancy is to reduce the dispersion and in Fig. 1, both the model and
the Monin—Obukhov theory show a damping of the vertical fluctuation. It is assumed

Fig. 1. Ratio of vertical fluctuations for stable and neutral stratification respectively. The left line is the
simulated fluctuations and the line to the right is the ratio prescribed by the Monin-Obukhov theory.
318 J. Burman / J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 315–325

Fig. 2. Decay of concentration downwind from the source. Squares stands for heavy-gas concentration
with a 1 : 4 slope. Stars stand for heavy-gas concentration with a flat surface. Solid line is the simulation
with flat surface and dashed line is the simulation with a slope.

that the Monin—Obukhov theory can describe the homogenous surface layer [9]. The
plots in Fig. 1 are similar only up to 10 m but the Obukhov length is only 30 m in this
case so it is assumed that the damping effect, that a released plume at the surface will
encounter, is well described.
Fig. 2 shows a comparison between a wind-tunnel experiment [10] and a simula-
tion. In the experiment the obstacle was a slope of 1 : 4 (&14°), up to an elevated
plateau. The simulation was scaled using suggestions in Ref. [10]. Downwind from the
source the decay of concentration follows x~1.5, which is a reasonable value in the
boundary layer [11]. The influence on the concentration from the 1 : 4-slope in the
simulation is similar to what seen in the experiment. The slope generates turbulence
and the dispersion process is stronger. Beyond the slope, the concentration curves
tend to converge, which is also seen in the experiment. The lower level of concentra-
tion is probably due to a difference in the boundary description. The simulation has
a higher level of turbulence. The effect of the 1 : 4-slope indicates that the model can
simulate the effects of topography on the dispersion process.

3. Results

3.1. Scenario 1, solid fence

The solid fence is situated 15 m downwind from the road and it forces the mean
flow to climb over the fence. This generates turbulence that increases the dispersion
J. Burman / J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 315–325 319

(see Fig. 3). The concentration downwind from the fence is approximately linearly
dependent on the inverse of the height of the fence. There is a dependence on the
ratio between the height of the plume in the absence of the fence, and the height of the
fence [12], but as this investigation only looks at a fence close to the road the main
effect is taken to be the fence height. A fence height of 4 m reduces the concentration
100 m downwind from the source by 50%, see Fig. 4. The concentration is normalised
with the value in the absence of a fence. In Fig. 4 the concentration from a heavy-gas
release is also plotted. It is clear that the heavy-gas cloud is more strongly affected by
the fence and relatively more diluted than the neutral plume, but in Fig. 5 it is obvious
that there is almost no differences in concentration when the plume has travelled
100 m.

Fig. 3. The distance a is 15 m.

Fig. 4. The lines represent distance from the source and the height of the fence is on the x-axis. The single
symbols stand for the same distances as the lined symbols but for heavy gas.
320 J. Burman / J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 315–325

Fig. 5. The same data as in Fig. 4 but now the lines represent a certain height of the fence and the distance
from the source is on the x-axis. The symbol X represents the case without a fence but with heavy gas. The
other symbols are as in Fig. 4.

Fig. 6. The distance a is 15 m.

3.2. Scenario 2, hedge

The simulated hedge consists of birch with a homogenous foliage area index of
0.286 m~1 except at the top and at the lowest level where the leaf area density is
assumed to be reduced by 50% (Fig. 6). The extra source terms due to the hedge
follow the work by Green [13]. There are source terms for the momentum equations
as well as the kinetic energy and dissipation equation. The effect is that the turbulence
is damped downwind the hedge up to 10 times the hedge height. The dissipation is
lower and the concentration higher than in the case without hedge. The effect is
consistent with wind tunnel experiments [14]. For a 4 m high hedge, a significant part
of the plume passes over the hedge and is dispersed at the same rate as in the case
without hedge (see Figs. 7 and 8).
J. Burman / J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 315–325 321

Fig. 7. The lines represent distance from the source and the height of the hedge is on the x-axis.

Fig. 8. The same data as in Fig. 7 but now the lines represent a certain height of the hedge and the distance
from the source is on the x-axis.
322 J. Burman / J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 315–325

Fig. 9. The distance a is 5 m.

Fig. 10. The lines represent distance from the source and the road-valley depth is on the x-axis. The single
squares represent the same distance as the squared line but for neutral stability. The same for diamond and
cross. The X stands for concentration at 30 m with neutral stability.

3.3. Scenario 3 road-valley

The previous scenarios had neutral stratification but here we also study stable
stratification. The temperature gradient is 0.01 K/m. The concentration at a distance
from the source is higher for the stable case but close to the road-valley the difference
is small. The dispersive effect from the road-valley is dominant up to approximately
100 m (Figs. 9—11).

4. Conclusions

Three different scenarios (fence, hedge and road-valley) are studied to estimate the
impact on the concentration level downwind from the source, from different structures
at the side of a road.
J. Burman / J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 315–325 323

Fig. 11. The same data as in Fig. 10 but now the lines represent a certain road-valley depth and the
distance from the source is on the x-axis. The symbol X represent the case without valley but with neutral
stability. The other symbols as in Fig. 10.

Scenario 1. A solid fence gives the strongest effect of all scenarios. The concentration
at the distance of a few fence heights, is approximately linearly dependent on the
inverse of the height when the atmospheric stability is neutral. A heavy-gas plume is
broadened in front of the fence before it climbs over. The broadening depends on the
internal stability in the plume, wind velocity, height of the fence and the volume
source. The dispersive effect from the fence on the heavy-gas plume is stronger than
the effect on a neutral gas plume.
Scenario 2. A hedge creates a low turbulent wake down-wind and this wake can
extend up to 10 times the hedge height. The turbulent dispersion is lower and thus the
concentration compared with the case without hedge, is higher.
Scenario 3. A road-valley generates turbulence but in an intermediate regime
compared with the other scenarios. For this scenario, a comparison is made between
stably and neutrally stratified atmosphere. In the stable case vertical fluctuations are
damped and therefore the level of concentration is higher but only at distances of the
order of 100 m or more. Closer to the road-valley the topographical effect dominates
the dispersion process. A main observation for this scenario is that between the sides
of the valley recirculation distributes high levels of concentration.
Fig. 12 shows the average of the kinetic energy in the plume weighted with the
concentration. The plume encounters different levels of turbulence depending on the
type of object and height/depth of object. The road-valley case has stably stratified
atmosphere and therefore also has lower level of kinetic energy initially. The local
324 J. Burman / J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 315–325

Fig. 12. The weighted average of kinetic energy within the plume for the three scenarios in the study.

stability within the plume damps the kinetic energy in the heavy gas case. For a hedge
high enough, the kinetic energy in the plume is lower than without any obstacle.
The study shows that a hazard range defined using a homogenous approach
regarding the flowfield, can be influenced by different topographical structures at the
side of a road. A high hedge can increase the level of concentration and thereby also
extending the hazard range. A solid fence though will increase the dispersion, reduce
the concentration and consequently also the hazard range.
Therefore to reduce the potential threat from a hypothetical accident it seems
plausible to use solid fences.

Acknowledgements

The author thanks the Swedish Rescue Services Agency (SRSA) for their support
and co-operation during this work and Erik Näslund and Per-Erik Johansson for
valuable discussions.

References
[1] J.O. Andersson, P. Wulff, Hot och Mota> tgärder vid Kemikalieolyckor, FOA report, E 40042-4.5 FOA
ABC-Skydd. Umea> , 1989.
[2] F. Pasquill, F.B. Smith, Atmospheric Diffusion, Ellis Horwood, Chichester, UK, 1983.
[3] W. Rodi, Turbulence models and their application in hydraulics — A state of the art review (IAHR),
University of Karlsruhe, 1978.
[4] A.S. Monin, A.M. Yaglom, Statistical fluid mechanics: Mechanics of Turbulence, MIT press, Cam-
bridge, MA, 1971.
[5] R.A. Dobbins, Atmospheric Motion and Air Pollution, Wiley, New York, 1979.
J. Burman / J. Wind Eng. Ind. Aerodyn. 74—76 (1998) 315–325 325

[6] U. Högström, Non dimensional wind and temperature profiles in the atmospheric surface layer:
a re-evaluation, Boundary-Layer Meteor. 42 (1988) 55—78.
[7] H.A. Panofsky, J.A. Dutton, Atmospheric Turbulence, Wiley, New York, 1984.
[8] B.E. Launder, D.B. Spalding, The numerical computation of turbulent flow, Comp. Meth. Appl.
Mech. Eng. 3 (1974) 269—289.
[9] R.E. Britter, W.H. Snyder, Fluid Modeling of Dense Gas Dispersion Over a Ramp, J. Hazard. Mater.
18 (1988) 37—67.
[10] U. Högström, Kompendium i Spridningsmeteorologi, Uppsala, 1981.
[11] R.E. Britter, J. McQuaid, Workbook on the Dispersion of Heavy Gases, HSE Contract Research
Report No. 17, 1988.
[12] S.R. Green, Modelling turbulent air flow in a stand of widely spaced trees, Phoenics J. 5 (3) (1992)
294—312.
[13] M.J. Judd, M.R. Raupach, J.J. Finnigan, A wind tunnel study of turbulent flow around single and
multiple windbreaks, Part 1: velocity fields, Boundary-Layer Meteorol. 80 (1996) 127—165.
[14] M.E. Davies, S. Singh, The phase II trials: A data set on the effect of obstructions, J. Hazard. Mater. 11
(1985) 301—323.

You might also like