Professional Documents
Culture Documents
0
b
x
, and of the
ordinate is the maximum velocity at that distance divided by the jet exhaust velocity
0
U
U
m
.
The plots show a reasonable agreement between the CFD results and the empirical equation.
The maximum velocity is slightly over-predicted by the CFD simulation and stays above
Rajaratnams empirical curve. The rate of decay is similar for both plots, with the difference
between their maximum velocities being almost constant.
The different behaviour in the upstream region may be easily explained. As reported in the
previous study 0, the empirical formula does not take into account the transitional region
between the initial and the main regions of the jet. The length of the potential core has to be
corrected according to Abramovichs law to include the transition phase 0, as explained in the
previous report 0. According to the CFD results, the length of the potential core extends to
about 7.5 diameters. Correcting the empirical results, it was found that its prediction of the
potential core extends to about 8 diameters. Thus, the CFD and empirical results are very close
near the jet exhaust.
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
6 EEC/SEE/2007/004
Recollecting the data used in the previous report and changing the scales to correct to the
parameters previously used, it is possible to compare the maximum velocity decays of the free
and wall jets, as shown in Figure 5.
Figure 5 Free and wall jet comparison for the maximum velocity decay
The plots show a substantial difference between the wall and free jet. First, the decay rate is
higher for the free jet than for the wall jet. The maximum velocity is still high at 60 diameters for
the wall jet, whereas the free jet is relatively low. Second, the length of the potential core is
shorter for the free jet than for the wall jet. The length of the potential core predicted by the CFD
simulation is about 5.5 diameters for the free jet and 7.5 diameters for the wall jet. Similar
results were presented by Curd 0, who compared free and wall jets in controlling air
contaminants.
The extension in the potential core length and the lower rate of decay for the wall jet as
compared to the free jet can only be a consequence of the presence of the wall, because all
other parameters are kept the same. Baydar & Ozmen 0 and ODonovan & Murray 0, while
studying impinging jets, also reported similar differences between confined and unconfined
configurations. They reported that confining the jet increases the length of the potential core,
and explained this behaviour by the wall limiting the entrainment of ambient air only on one
side. By restricting the entrainment, the rate of spreading decreases leading to an increase in
the potential core length for the wall jet.
2.2.2 Vertical Velocity Profile for Non-Buoyant Wall Jet
In the case of a free jet, it was shown that there exists a unique self-preserving flow that
represents the whole vertical velocity profile after the flow development region. The behaviour of
the wall jet is slightly different due to the presence of wall. Two types of flow can be
encountered in the wall jet, a boundary layer and a shear layer type of flow, each of them
exhibiting their respective self-similar profile.
In the literature, there is a disagreement regarding the point where the flow changes from the
boundary layer type to a shear layer type of flow. As mentioned earlier, in this report, Glauberts
demarcation of layers will be used. This refers to the point where the flow attained the maximum
velocity, which is also the boundary layer thickness. Many researchers adopted this definition,
such as Launder & Rodi 0. Kurka & Eskinazi 0, on the other hand, found that the best
separation point between the two layers is where the Reynolds shear stress vanishes, but they
also stated that this method of determination is impractical.
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
EEC/SEE/2007/004 7
While there is an agreement amongst researchers concerning the self-similarity of the outer
layer, the establishment of a self-preserving inner layer is unclear. Some, such as Schwarz &
Cosart 0, believe that the velocity profiles in both inner and outer layers showed self-similarity
characteristics. But others, like Wygnanski et al. 0 and Swamy & Bandyopadyay 0, partially
disagree with this statement, refuting the self-similarity in the inner layer.
Figure 6 represents the vertical velocity profile plots at different locations behind the wall.
Similar to Figure 4, the non-dimensional coordinate of the abscissa is the distance away from
the wall jet axis divided by the distance where the velocity reaches half the axial velocity
b
y
,
and of the ordinate is the ratio of the local velocity to the axial velocity
m
U
U
.
Figure 6 CFD vertical profile comparison at different distances behind the wall jet
Unlike the free jet, which represented a unique pattern throughout the control volume, the wall
jet presents some different characteristics. The profiles only start to become similar at about
0.5m behind the wall jet. This confirms the previous discussion concerning the length of the
potential core. The length taken to reach a similar state represents the portion where the flow
undergoes the potential core and intermediate regions. This statement correlates with the
findings of Wygnanski et al. 0 who showed that the self-similarity is attained at longer distances
from the nozzle for the wall jet than for the free jet.
When the profile reached a similarity pattern, the boundary layer region is very small compared
to the shear layer region. Townsend 0 found that this region occupies between 1/10 to 1/5 of the
total wall jet velocity profile. Further away from the wall, all plots converge to zero, showing the
end boundary of the mixing properties with the ambient fluid.
Glaubert 0 was the first to derive an analytical solution for the wall jet problem, based on the
Blasius empirical power friction formula and on using matching and integrating techniques.
His work offers a discussion on this problem which will not be repeated here. Other researchers
also presented empirical formulae to represent the self-similarity of both profiles, such as
Newman et al. 0 and Schwarz & Cosart 0, but all failed to do so because the inner and outer
layer have completely different self-similar profiles. This point will be discussed more thoroughly
in the next section.
The CFD results were also compared with the Forthmann experimental results given by
Rajaratnam 0, who used only data from 0.6m behind the jet to show the vertical velocity pattern
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
8 EEC/SEE/2007/004
at different distances downstream, confirming the previous discussion on the length of the
potential core. Figure 7 shows that the general velocity profile is well predicted by the CFD
simulation at different distances behind the jet. Slight differences are only found near the wall
and further away from the wall, and these will be explained separately.
Figure 7 CFD and experimental results for the vertical velocity profile
The differences near the wall come from the fact that the wall properties are unknown in
Forthmanns experiment, such as the surface roughness which is an important property
especially near the wall where its effects are significant 0.
The differences at the end of the shear layer may come from the fact that the experiment
instruments were quite basic. As explained in a previous report 0, Forthmanns experiment
failed to comply with three criteria for assessing experimental data. In addition to this, the
results were taken very near the jet exhaust where the flow is supposed to exhibit considerable
scatter at its edges, as in the results of the CFD simulation 0.
Similar to the free jet study, it is possible to analyse the spreading parameters affecting the flow
in the case of the wall jet. Figure 8 shows a comparison between the spreading parameters for
a free and a wall jet. The non-dimensional coordinate of the abscissa is the distance behind the
jet divided by the jet diameter
0
b
x
, and of the ordinate is the length where the velocity attained
half of the axial velocity divided by the jet diameter
0
b
b
. It was found in 0 that the plots are
linear and, as a consequence, can be characterised by a linear trend line in the form:
+ =
2
0
1
0
K
b
x
K
b
b
When 0
0
=
b
b
, the approximate position of the virtual point is given by
1
2
K
K
while
0
1
b
K
gives a
measure of the spreading rate the wall jet is subject to.
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
EEC/SEE/2007/004 9
Figure 8 Angle of spread comparison between free and wall jets
Considering first the position of the virtual point, its location is situated behind the actual origin
of the wall jet and the value
1
2
K
K
gives a magnitude of 9.75 diameters. This compares very
well with the findings of Rajaratnam 0, who predicted the location of the virtual origin to be
around 10 diameters behind the nozzle.
Although the location of the virtual origin is accurately predicted by the CFD results, the angle of
spread is slightly under predicted with a difference of about 11.5% below the value given by
Rajaratnam 0 ( ) x b 068 . 0 = . However, there are uncertainties regarding this parameter. Launder
& Rodi 0 found a growth rate in the order of 002 . 0 073 . 0 but Tachie et al. 0 found this value to
be much higher. This value can vary from 0.085 to 0.09 at low Reynolds numbers. The position
of the virtual origin is clearly different for the free and the wall jet, with the wall jet located further
behind the nozzle compared to the free jet.
The spreading angle is more pronounced in the case of the free jet than in the wall jet. Launder
& Rodi 0 predicted this behaviour and showed that the growth rate of the free jet is more than
30% above that of the wall jet. Similarly, Rajaratnam 0 stated that the wall jet growth is about
0.7 times that of the free jet. Tangemann et al. 0 explained this behaviour by highlighting the
walls influence as the main reason for the reduction in the spreading growth as compared with
the free jet. This confirms the wall effect on increasing the potential core length.
As in the previous report 0, the existence of streamwise vortices can be proved in the case of a
wall jet by both the effects of the flow encountering the surrounding irrotational fluid and the
presence of the wall. This is illustrated in Figure 9, which presents isolines for the streamwise
vorticity contours.
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
10 EEC/SEE/2007/004
Figure 9 Vorticity contour around z-axis with isolines (1/s)
The figure shows that there is no symmetrical pattern as in the free jet case, only the upper side
mixing with the ambient fluid resembles slightly the figure obtained for the free jet. The points
highlighted in the previous report are still valid:
The potential core is characterised by lower vortices inside it.
The boundary of the jet spreading can be clearly seen as a separation between
rotational and irrotational flows.
The vortices are found to be counter-rotating but unlike the free jet, the negative rotation around
the z-axis is due to the wall.
Townsend 0 explained that the presence of a solid boundary restricts the vortices created by
the shear layer to sizes which, in comparison to the free jet, are smaller thus causing a
reduction in the dissipating length scale. Figure 10 shows the vortices around the z-axis at
different distances behind the wall jet.
Figure 10 Z-Vorticity profile comparison at different distances behind the non-buoyant wall jet
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
EEC/SEE/2007/004 11
It can be seen that, near the wall, the magnitude of negative streamwise vortices is very high
compared to the positive vortices created by the shear layer. These negative vortices around
the z-axis remain very high at large distances behind the wall jet, whereas the effect of the
positive vortices decreases downstream. The findings of Aloysius et al. 0 on the structure of
contra-rotating vortices formation and their development, also discussed in 0, are still valid here.
Another interesting point concerns the location of the maximum positive vorticity amplitude;
which are shown to be going into the wall jet axis. The free jet, on the other hand, was found to
go away from the jet axis 0. It was explained in 0 that this behaviour is related to the widening of
the jet. This influence can clearly be seen here; this pushing down effect restricts the widening
of the jet or, in other words, the growth rate. The wall plays an important role in the creation of
high magnitude negative vortices, thus gradually influencing the positive vortices which, as a
consequence, reduce the jet growth rate.
The location of the point of maximum velocity was found to vary behind the wall jet. To show
this variation, Figure 11 represents the height of the maximum velocity at different distances
behind the wall jet. The coordinate of the abscissa is the distance behind the jet divided by the
jet diameter
0
b
x
, while that of the ordinate is obtained by subtracting the vertical distance
where the velocity is maximum by the jet radius ( ) 2 /
0
b y
m
.
The figure shows that the location of the maximum velocity starts below the centreline axis of
the jet but, at a distance of about 29 diameters, this position starts to rise. Overall, the position
of the maximum velocity is below the centreline axis whereas in the free jet the position was
found to be at the centreline axis. This clearly shows the important effect of the wall on the flow
pattern. The pushing down mechanism can be seen here and is due to the high magnitude of
negative vortices and decreasing positive vortices around the z-axis, which pushes down the
flow near the wall, confirming the reduction in the spread rate and the increase in the potential
core length when compared to the free jet.
Figure 11 Height of the plume centreline as a function of downwind distance
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
12 EEC/SEE/2007/004
2.3 Buoyant Wall Jets
This section aims to compare the results obtained in the previous non-buoyant configuration
with a similar buoyant flow. Such condition is possible by adding a temperature difference and
assigning a release of different species and density magnitude.
The control volume and the boundary conditions were kept the same, except on the jet exhaust
where the velocity magnitude was maintained at 35m/s but this time releasing a mass fraction of
NOx ( )
4
10 37 . 0
at a temperature of 690K. The simulation was run for the same time step as
the non-buoyant case.
2.3.1 Axial Velocity Profile for Buoyant Wall Jet
A comparison between the maximum velocity for the buoyant and non-buoyant wall jets can be
seen in Figure 12. The decay rate for the buoyant wall jet is greater than for its non-buoyant
counterpart. All plots share a very similar pattern near the jet exhaust, but far downwind
divergences occur. The decay of the buoyant jet follows the theoretical curve upstream but
completely diverge from it further behind. The non-buoyant wall jet, on the other hand, follows
the theoretical rate of decay as seen in Figure 4.
Figure 12 Maximum axial velocity decay profiles with and without buoyancy
The length of the potential core for the buoyant wall jet was found to be about 6.5 jet diameters,
which is about 1 diameter shorter than that of the non-buoyant wall jet. This compares well with
the free jet results of 0, where the length of the potential core was 4.9 diameters for the buoyant
free jet and 5.6 diameters for the non-buoyant free jet. This difference was attributed to the
buoyancy effect on the flow pattern. Appendix A shows a comparison between the results for
free and wall jets, with or without buoyancy, with their respective theoretical profiles.
Figure 13 presents the velocity contours for the buoyant wall jet, which can be compared with
Figure 2 for the non-buoyant wall jet. The upstream region is very comparable to the non-
buoyant simulation, as shown in Appendix B. The potential core region can be recognized by its
triangular shape right at the exit of the nozzle creating the potential core length discussed
previously. In the fully developed region, the profile has the same configuration: a boundary
layer profile created by the wall and a shear layer profile formed by the surrounding fluid.
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
EEC/SEE/2007/004 13
Figure 13 Velocity contours for buoyant wall jet
Unfortunately, the control volume of this simulation is not wide enough to show plume rise
results, but it will be shown later in the 3D section of the report that the plume will actually lift-off
and rise due to buoyancy.
2.3.2 Vertical Velocity Profile for Buoyant Wall Jet
The vertical velocity profiles at different distances behind the jet for the buoyant wall jet are
presented in Figure 14. All the curves have different patterns from the nozzle exit to about 0.5m
behind the jet, whereas after that distance they all display a similar profile. As discussed
previously, this is due to the flow still being in the potential core region or in the intermediate
region. After 0.6m behind the jet, the figure shows that the profiles are very similar throughout
the vertical distance.
Figure 14 Velocity profiles with buoyancy at different distances behind the wall jet
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
14 EEC/SEE/2007/004
Appendix C compares the CFD results for the buoyant and non-buoyant wall jets. The ordinate
was changed to a dimensional form to highlight the differences. The analysis can be done in
two different sections. The first section is the boundary layer; except for the 0.6m behind the jet
which shows a shorter length for the buoyant wall jet than the non-buoyant case, the other
sections show approximately the same boundary layer thickness.
Secondly, the shear layer behaves very differently for the buoyant and the non-buoyant cases.
Although the point where the velocity ratio reaches zero is shorter for the buoyant simulation
than the non-buoyant simulation, it can be seen that this position becomes closer to the non-
buoyant jet and has almost the same velocity ratio at 0.2m above the wall. This point was
discussed in the previous report 0 as a confirmation of buoyancy effects acting on the flow.
As in the case of the non-buoyant wall jet, the analysis of different locations where the velocity
reaches half the axial velocity can be made by plotting these at different distances behind the
wall jet.
Figure 15 shows a comparison between the spreading parameters for the buoyant and non-
buoyant simulations. Even though the upwind region is fairly similar for both simulations, the
downwind region is very different, with the buoyant simulation showing higher values of the
position where the velocity reached half of the maximum velocity. This happens after about 30
diameters behind the jet nozzle, and is due to the buoyancy effect on the flow. The same
behaviour has been reported in 0 and the relevant figure is presented in Appendix D for
comparison.
Figure 15 Spread b comparison for wall jet with and without buoyancy
As mentioned earlier on the report, the spread b as a function of distance is much lower for the
wall jet than for the free jet simulation without buoyancy. The addition of buoyancy does not
change this pattern; spread b is still higher for the free jet than for the wall jet at large
downwind distances.
Another interesting point concerns the location where the buoyant curve separates from the
non-buoyant curve, for free and wall jet simulations. This was found to be about 20 diameters
for the free jet and 30 diameters for the wall jet. This means that the wall jet takes longer to go
from a jet like to a plume like behaviour as compared with the free jet simulation, confirming the
potential core length discussion.
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
EEC/SEE/2007/004 15
Figure 16 presents a comparison of the height of the plume centreline for the buoyant and non-
buoyant simulations, at different positions behind the wall jet. The axes are kept the same as in
Figure 11.
Figure 16 Comparison of the height of plume centreline as a function of downwind distance
Figure 16 shows that both simulations follow the same pattern near the jet exhaust, whereas
further downwind differences clearly start to appear. The buoyant wall jet near the end of the
control volume exhibits a sudden rise of the height of the plume centreline, while in the non-
buoyant case this rise progresses at a lower rate. All locations of the plume centreline are below
the jet axis for the non-buoyant simulation while the buoyant simulation shows a rise above it
further downwind, confirming the lift-off of the plume due to buoyancy. This point will be
discussed in greater detail for three-dimensional simulations.
The reason why the height of the plume centreline is kept below the jet axis for such long
distance can be explained by the Coanda effect. Ramsdale & Tickle 0 explained this effect as
the tendency of the jets to attach themselves to nearby solid boundaries. Sharp & Vyas 0
added that the flow clings to the ground for large distances before buoyancy takes over and
causes the jet to lift and rise above the ground.
An analogy was made in the non-buoyant study between the roles of the vortices in the pushing
down effect of the flow near the ground. It is interesting to see whether the same observations
can be made for the buoyant wall jet simulation.
Figure 17 shows Z-Vorticity profiles at different distances downwind. Similar to the non-buoyant
case shown in Figure 10, the location of the maximum positive vorticity amplitude goes down
into the wall jet axis affecting the widening of the jet. This explains why the growth rate of the
buoyant free jet is higher than the buoyant wall jet. As a matter of fact, the high magnitude of
negative vortices, which are created due to the presence of the wall, gradually influences the
positive vortices created by the entrainment with the ambient fluid leading to a pushdown of the
flow. As a consequence, the length of the potential core is increased while, at the same time,
the growth rate is decreased.
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
16 EEC/SEE/2007/004
Figure 17 Z-Vorticity profile at different distances behind the buoyant wall jet
With the initial results and discussions from this two-dimensional study, it is now possible to
advance with confidence to three-dimensional simulations. A more realistic scenario can be
adopted, with a jet close to the ground and a co-flowing velocity to replicate the headwind flow.
3 THREE-DIMENSIONAL STUDY OF TURBULENT WALL JET
The simulation in this section involves a round jet of 0.93m diameter very close to a wall, in an
ideal co-flow headwind configuration. First an isothermal simulation is carried out; then, the
same geometry will be simulated with buoyancy effects. The CFM56-3C jet engine was selected
because it has already been studied in previous reports 0 and 0.
A transient simulation was conducted to monitor the progress of the flow throughout the control
volume. The geometry, mesh distribution and boundary conditions are discussed in the
following section.
3.1 Geometry and Boundary Conditions
The computational domain is rectangular, composed of 104 different volumes of different mesh
densities. This fine decomposition of the control volume was done to optimise the mesh
distribution. The near wall region including the jet and the first few 100 meters, where the
buoyancy effects are expected to be most significant, was set up as fine as possible. The
overall geometry is presented in Figure 18, and extends to
0
6 . 537 b ,
0
5 . 64 b and
0
86b in the x, y
and z directions, respectively, with
0
b the jet diameter. The total mesh density for this problem
comprises about 3,965,760 nodes.
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
EEC/SEE/2007/004 17
Figure 18 Geometry and mesh distribution of the 3D wall jet
The boundary conditions are as follows:
The jet exhaust is set up as a velocity inlet with velocity magnitude 80m/s, releasing hot
gases (690K) of NOx emitted at a mass fraction of
4
10 74 . 0
, as described in 0.
The faces adjacent to the exhaust are also defined as velocity inlets, but with a different
velocity of magnitude 2.5m/s to replicate the co-flow condition of a headwind. The co-
flow is adiabatically stratified across the domain with a temperature of 290K.
The bottom wall is setup as stationary with surface roughness of 0.003m, corresponding
to the concrete PCN 60 R/B/W/T used at Zurich airport. Werner and Wengles near wall
treatment was used to account for the boundary layer formation when the mesh is
coarse.
The external boundaries are defined as symmetry walls, as they are located far away
from the jet.
The face opposite to the jet is the outflow of the control volume.
The Navier-Stokes equations of fluid flow are solved by the Large Eddy Simulation (LES)
technique.
3.2 Mean Velocity Profile Comparison
3.2.1 Streamwise Direction Analysis
This section aims to compare the results of the co-flowing wall jet with the co-flowing free jet,
both with buoyancy, to assess the effects on the fluid mechanics when the wall is added.
Results of the two-dimensional simulations have shown that the decay rate for the free jet is
greater than the for wall jet, while the potential core is much longer for the wall jet than for the
free jet. These trends are confirmed in the three-dimensional simulations, as shown in Figure
19.
537.6b
0
64.5 b
0
86 b
0
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
18 EEC/SEE/2007/004
An analysis of the results for the first time step shows that there is a slight difference in the
length where the free and wall jet both reach 0
5 . 2
0
=
U
U
m
. The free jet simulation attained this
value faster than the wall jet simulation; on the other hand, the potential core is much shorter
and its decay rate is much greater for the free jet simulation than for the wall jet simulation. The
following time step also shows a similar behaviour upwind, but with a more elongated pattern
because the fluid is still progressing through time and has not yet attained a steady state
condition.
Figure 19 Maximum velocity decay comparison between free and wall jet
Figure 20 shows a comparison of the mean velocity profile for the simulation of the free jet (left)
and the wall jet (right), at different times.
The first time level (after 1s) shows a much further penetration for the wall jet than for the free
jet. The free jet still exhibits an almost symmetrical pattern, as in the case of a non-buoyant free
jet. The wall jet, on the other hand, shows a different pattern with the fluid rising at some
distance behind the jet exhaust.
Buoyancy effects can clearly be seen in the second time level (after 5s), when the free jet
velocity profile shows a deviation from the centreline axis. The wall jet has approximately the
same pattern as in the previous time step, with the flow rising much higher than for the free jet
simulation. The flow penetration is also much deeper for the wall jet than for the free jet
simulation, as stated earlier.
The third time level (after 10s) continues the previous trends, with greater flow penetration and
higher rise for the buoyant wall jet than for the buoyant free jet.
Another interesting point that can be observed for the wall jet is that, at large distances behind
the exhaust, the flow separates from the wall. Although not so apparent, this effect can already
be seen after 1s, at about 28m behind the exhaust, and is clearly seen after 10s at about 60m
behind the exhaust. This lift-off effect will be discussed in more detail in the next section.
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
EEC/SEE/2007/004 19
After 1 second
Free jet
Wall Jet
After 5 seconds
Free jet
Wall Jet
After 10 seconds
Free jet
Wall Jet
Figure 20 Mean velocity profile evolution through time for buoyant free and wall jets
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
20 EEC/SEE/2007/004
The streamwise velocity profile at different distances behind the exhaust for the buoyant wall jet
is shown in Figure 21. This figure is very comparable to Figure 14 in section 2.3.2. From the exit
of the nozzle to a distance of 7.5m, the wall jet is still in the potential core and intermediate
regions, whereas from this distance onwards the wall jet is in the fully developed region and
exhibits a self similarity profile.
Forthmanns experimental results are used to compare the self-similar profile. It can be seen
that all the plots after 7.5m behind the exhaust closely follow the experimental results, both in
the boundary layer and the free shear layer parts.
Figure 21 Streamwise velocity profiles at different distances behind the buoyant wall jet
It was previously argued that the profile of the free shear layer is comparable with the free jet
situation; Appendix E shows this comparison with Forthmanns experimental data for the wall jet
and the theoretical results of Tollmien, Goertler and Bradbury for the free jet. It can be seen that
they all have the same pattern; the CFD results follow very closely the experimental curve to
about
b
y
75 . 1 , whereas further away Tollmiens theoretical results are the closest to the CFD
results. A comparison between the co-flow and stagnant condition shows that the boundary
layer thickness is greater in a co-flowing situation. This is in line with the findings of Launder &
Rodi 0, who explained this phenomenon as a consequence of the presence of the external
stream weakening the relative strength of the free shear layer. This, in turn, has less impact on
the wall boundary layer, making it thicker.
A comparison between the spread b as a function of distance of the free and wall jets, at two
different time levels, is given in Appendix F. Both graphs show the same general pattern as for
the 2D simulations, section 2.2.2. The rate of spread is greater for the free jet than for the wall
jet. As mentioned earlier, several authors such as Launder & Rodi 0 had already obtained this
behaviour in a non-buoyant situation, and claimed that the growth rate of the free jet is more
than 30% higher that of the wall jet. As can be seen in Appendix F, this trend tends to grow as
the flow progresses through time. It is interesting to note that the growth rate is actually
increasing with time for the buoyant free jet simulation, whereas for the buoyant wall jet the
growth rate is actually decreasing. It will be shown later in the report that this behaviour is due
to the vortices generated by the presence of the wall; this is true for some distances behind the
exhaust but, at larger distances, the wall effects decrease and the buoyancy effects will take
over and lift jet above the ground.
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
EEC/SEE/2007/004 21
While studying the 2D wall jet, another influence of the wall was found on the maximum height
of the plume centreline. To understand it better in a 3D situation, Figure 22 gives a comparison
of the maximum height of the plume centreline as a function of downward distance. The
ordinate axis was changed to the non-dimensional form
m
y
b y
0
, with
m
y the maximum height of
both simulations.
Figure 22 3D comparison of height of plume centreline at different distances behind the exhaust,
after 10s
Figure 22 shows a different pattern for both simulations; the rise above the actual centreline
axis of the jet is much faster for the free jet than for the wall jet simulation. The buoyant wall jet
starts rising above the centreline at about 55 diameters; before this point, the plot never reaches
a positive value. As a matter of fact, it is either approaching the centreline value (0) or stays
negative, meaning the flow is going down. This behaviour was mentioned earlier in the 2D part
of the report, as the clinging phenomenon due to the Coanda effect.
Although in a steady-state condition, Sharp & Vyas 0 found a crude linear relation between the
relative distance
0
b
L
over which the flow stays attached to the ground and the densimetric
Froude number F :
c F
b
L
=
0
with 2 . 3 = c found experimentally and L the length of cling. For this simulation, at the time
when it reaches a steady-state condition, the length of cling should be 166m behind the
exhaust. At the time 10s, the length of cling in Figure 22 is about 55 diameters, which is in the
region of the plume lift-off shown in Figure 20.
A comprehensive literature review on the lift-off phenomenon was carried out by Ramsdale &
Tickle 0 in their study of ground-based buoyant clouds. One interesting parameter when
studying the lift-off of a buoyant gas is the Richardson number
p
L , given by:
ALAQS CFD Comparison of Buoyant Free and
Wall Turbulent Jets
22 EEC/SEE/2007/004
=
a
p
u
gH
L
2
with H the effective depth,