Download as pdf or txt
Download as pdf or txt
You are on page 1of 56

Turbulence Intensity

Turbulence intensity is defined as the ratio of standard deviation of fluctuating wind


velocity to the mean wind speed, and it represents the intensity of wind velocity
fluctuation.

From: Innovative Bridge Design Handbook, 2016

Related terms:

Energy Engineering, Turbulence, Boundary Layer, Turbines, Wind Tunnels, Wind


Turbines, Reynolds' Number, Freestream, Streamwise

View all Topics

Heat and Mass Transfer Across a Hol-


low Fiber Membrane Bundle
Li-Zhi Zhang, in Conjugate Heat and Mass Transfer in Heat Mass Exchanger Ducts,
2013

7.7.2 Turbulence for the air flow


Turbulence intensities (ua /ua) in the air flow for the in-line and the staggered
arrangements are shown in Figure 7.22a and b, respectively. As can be seen, the
turbulence intensities are higher where velocities are higher. It has been found that
vortexes appear behind the fibers. The turbulence intensities behind the fibers where
vortexes appear are higher than other regions. For both the in-line and the staggered
arrangements, the turbulence intensities are higher than 0.07, meaning that the
fluctuations of velocities for the air flow in the bundle are relatively large [13,14].
In other words, the turbulent behavior of the air flow across the bundle should be
taken into account. Therefore the turbulence model employed here is reasonable
and necessary to account for the disturbances from the fibers.
Figure 7.22. Contours of the turbulence intensities for the air stream, PL/(2ro) = PT-
/(2ro) = 2.0,  = 0.196,Rea = 300. (a) In-line arrangement; (b) staggered arrangement.

> Read full chapter

Experimental Investigation of Turbu-


lent Wake-Blade Interaction in Axial
Compressors
A. Sentker, W. Riess, in Engineering Turbulence Modelling and Experiments 4, 1999

3.2 Turbulence intensity distribution first stage


The turbulence intensity, or unresolved unsteadiness, which is calculated from the
mean square fluctuation components divided by the mean absolute velocity [3], gives
a good insight into the behaviour of the unsteady flow field travelling through the
stator.

Considering the turbulence intensity in E2 (fig.6) most striking is the high turbulence
in the wakes of the rotor blades with more than 5%. With a distance of 12° rotorposi-
tion they move from the right to the left side through the three dimensional diagram.
The wake of the IGV is also characterised by an increased turbulence intensity (3%).
It is situated spatially fixed at 0.5° to 3° measuring position. At the position where
the rotor and IGV wakes intersect the IGV wake is shifted, with the part on the
suction side of the rotor wake moving faster than that on the pressure side. Inside
the stator channel in E2b presented in fig.7 the rotor wake is still visible due to its
high turbulence intensity of 4.5%. The shape of the rotor wake is changing inside
the stator channel. The absolute velocity near the suction side of the channel is
higher so that the part of the rotor wake near to the suction side of the channel
is transported faster downstream. This results in a twisting of the rotor wake in the
three dimensional diagram. Near the pressure side of the stator channel occurs a
spatially fixed region with a higher turbulence intensity. It can be assumed that the
increased turbulence at this measuring position results from an IGV wake. Changing
the operating point of the compressor leads to other positions of this region of
increased turbulence inside the stator channel. This fact underlines, that the higher
turbulence near the suction side of the stator channel is not generated by the velocity
or pressure distribution inside the stator channel but originates from the IGV wake.

Figure 6. Turbulence intensity distribution in E2

Figure 7. Turbulence intensity distribution in E2b

Considering fig.8, where the turbulence intensity in E2d is shown, supports the
observation that an influence of the IGV wake on the unsteady flow field is still
detectable there. In a measuring range of –5.5° to –7.5° a spatially fixed region
with increased turbulence intensity is visible, which can be identified as the IGV
wake. Compared to the absolute velocity distribution the position of the IGV wake
has changed, because the highest turbulence occurs inside the wake and the max.
velocity in fig. 4 comes from the suction side of the IGV wake. The maximum
turbulence inside the rotor wake in E2d is less than in E2b with 4%.

Figure 8. Turbulence intensity distribution in E2d

In the measuring plane behind the stator, presented in fig.9, the wakes of the stator
with a turbulence intensity of about 10% are dominant. They are spatially fixed
at 3.5° and –10.5° measuring position. In contrast to fig.5 the diagonal regions
with an increased turbulence level are the wakes of the upstream rotor, transported
through the stator. The time distance between two adjacent rotor wakes is no longer
constant at 12° but has diminished. The wakes are shifted clockwise due to the
velocity gradient inside the stator pitch.

Figure 9. Turbulence intensity distribution in E3

Considering the position of the interaction between stator and rotor wake, a differ-
ence of 6° rotor position between the part of the rotor wake on the suction side
compared to the part on the pressure side of the stator wake can be determined.
Near the pressure side of the stator wake the curvature of the rotor wakes is strong.
The IGV wake, too, is weakly visible at –4.5° to –7.5° measuring position. The increase
of the turbulence intensity due to the IGV wake is not very strong behind the stator.
It has been nearly mixed out.

In fig. 10 the way of the IGV wake through the first stage is reconstructed schemat-
ically from the measured unsteady flow data. The extension in circumferential
direction in E2a and E2b could not be measured because the IGV wake is situated at
the border of the max. possible probe traverse. The IGV wake follows an imaginary
streamline on its way through the stator. Its circumferential extension grows from
E2 to E3, it follows the curvature of the stator blade.

Figure 10. IGV wake in first stage

> Read full chapter

LDA-Measurements of the Turbulence


in and Around a Venturi
R.F. Mudde, ... H.R.E. van Maanen, in Engineering Turbulence Modelling and
Experiments 6, 2005

Turbulence Intensities
The axial and tangential turbulence intensities, Iax and Itan, have been calculated via
the definition: with the local mean velocity. The intensities have been corrected for
noise contributions by inspecting the ACF. This correction is better for signals with
a high SNR, stressing the importance of high quality data. As the definition of the
intensities contains the local mean velocity, the values of the intensities go up when
approaching the wall. The axial turbulence intensities are shown in fig.(8).
Figure 8. Turbulence intensity profiles of the axial velocity

The turbulence intensity in the throat is much lower than 1D upstream of the Venturi
(where the turbulence intensity is equal to that of the pipe further upstream), in
agreement with the literature. In the diffusor, the intensity in the central part is
still low: the flow that comes out of the throat moves as a jet of low turbulence
intensity into the diffusor section. The intensity in the wall region has increased to
values much higher than that of a well developed turbulent pipe flow. The increase
continues further downstream. Just outside the diffusor the levels in the center are
around those of the turbulent flow in the pipe; in the wall region they have increased
further. Further downstream, the profile becomes flatter but is higher than that
of the turbulent flow. From the turbulence intensity it is obvious that the flow 5D
downstream is not in equilibrium: the turbulence intensity profile is flatter than for
turbulent flow in a fully developed pipe flow and its magnitude is roughly twice
as high, indicating that the dissipation is not in equilibrium with the production.
The tangential velocity fluctuations show a similar turbulence intensity development
along the Venturi.

> Read full chapter

Application of a lagrangian PDF


Method to Turbulent Gas/Particle Com-
bustion
M. Rose, ... M.G. Neuhaus, in Engineering Turbulence Modelling and Experiments
4, 1999

5.2 Results of Calculations


The effect of turbulence intensity on temperature and CO mass fraction in a
turbulent reacting gas/particle mixture was investigated. The upper part of Fig. 3
shows the distribution of temperature in the x/y-plane 20 ms after particles start to
devolatilize CO into the preheated ambient air.

Figure 3. Distribution of temperature (upper part, 300 K ≤ T ≤ 1950 K) and CO mass


fraction (lower part, 0 ≤ [CO] ≤ 0.4) in the x/z-plane for low (left) and high (right)
turbulence intensity 20 ms after particles start to devolatilize.

The left and right parts of the figure correspond to low and high turbulence
intensity, respectively. At this time, the mean characteristics of the flow field have
become stationary, i.e. spreading angle of the ”flame” and mean gas/particle-mix-
ture compositions at different cross-sections normal to the x-axis. Independent of
turbulence intensity, high chemical activity indicated by light colours in the T-picture
is restricted to the vicinity of CP-particles in the inflow section, x ≤ 1 cm. Further
downstream, turbulent transport influences significantly the evolution of the
temperature field. For lower turbulent mixing, the spreading angle of the flame front
in the flow direction is much lower than in the case of higher turbulence. In the
latter case, also the distance between isothermals in the z-direction is greater than
for lower turbulence intensity. This represents a much higher flame brush thickness.
In addition, the more erratic form of the isothermals for higher turbulence indicate
a much more wrinkled and corrugated shape of the flame.

Characteristic distributions of the CO mass fractions are shown in the lower part of
Fig. 3. In both parts of the figure, the amount of unburnt CO in the inflow region
is high, which is represented by light colours. For higher turbulence, the enhanced
transport in the z-direction leads to a much better mixing of CO with fresh air further
downstream, resulting in a much lower amount of CO at the right border of the
volume. In the case of lower turbulence intensity, devolatilized CO is not burned
completely and leaves the volume mostly concentrated near z = 0 cm.
A quantitative description of the results illustrated above is given in Figs. 4 to 6.
Figs. 4 shows the mean temperature and CO mass fraction along the x-axis. The
values were obtained by averaging in the z-direction between z = 0 cm and z = 0.5 cm,
i.e. the originally particle loaded flow part. The inflow region x ≤ 1 cm is characterised
by peaks in temperature and CO concentration up to 1400 K and 40 %, respectively.
The very high heat release results from homogeneous burning of devolatilized CO
and heterogeneous particle burning in the fresh surrounding air. A local minimum
in temperature and CO concentration follows the zone of individual particle burning
at about x = 1 cm and x = 0.5 cm, respectively. From about x = 1.5 cm the graphs of
temperature and CO mass fraction become more continuous, indicating that the
chemical activity is no longer concentrated to the vicinity of individual particles.
The combustion mode changes from diffusion/devolatilization controlled reaction
to partially premixed reaction in this region. In the case of high turbulence intensity,
the temperature remains nearly constant at a value of about 1400 K for x ≥ 1.5 cm,
whereas the temperature further increases up to 1800 K for less intensive turbulent
mixing. At the same time, the mass fraction of CO decreases down to less than 5%
at the exit, whereas for low turbulence more than 12% of the total mass is unburned
CO.

Figure 4. Temperature distributions (left) and CO mass fraction (right) along x-axis
averaged for 0 cm ≤ z ≤ 0.5 cm.

Figure 5. Temperature distributions (left) and CO mass fraction (right) along x-axis
averaged for 1.5 cm ≤ z ≤ 2 cm.
Figure 6. Profiles of temperature (left) and CO mass fraction (right) at the right end
of the flow volume averaged for 4.5 cm ≤ x ≤ 5 cm.

The effect of different turbulence intensities on the flow field becomes even more
significant when looking at the distribution of temperature and CO mass fraction
in the upper section of the flow field at z = 1.75 cm, see Fig. 5. In the case of high
turbulence intensity, the left part of the figure shows for x ≥ 1.5 cm a significant
continuous increase in temperature, resulting in an average outlet temperature
which is about 200 K higher than for lower turbulent mixing. The CO mass fraction
shown in the right part of Fig. 5, is always less than 1%, and has a maximum at
x = 3 cm in the case of high turbulence. This is exactly the position where the
turbulent flame brush influences the flow field, whereas in the case of low turbulence
intensity only very little CO reaches this upper flow section by diffusion and turbulent
transport.

The profiles of temperature and CO mass fraction in the vertical outlet region are
given in Fig. 6. The left part shows that lower turbulence intensity causes more
heat release in the originally particle loaded flow part z ≤ l cm and less one for
z ≥ 1 cm. compared to the case with high turbulence intensity. The corresponding
temperature graph shows a high gradient around z = 1 cm, connecting the lower
temperature region of about 800 K with a high temperature region of about 1900 K.
The temperature graph of the corresponding high turbulence case is more smooth
with less extreme values ranging between 1000 K and 1500 K. It becomes obvious
from the right part of Fig. 6, that the amount of unburned CO leaving the flow sec-
tion for z ≤ 0.5 cm is very high in the case of low turbulence intensity. Furthermore,
the temperature is also high in this flow part, but the expected high reactivity is
restricted by a lack of oxidizer.

> Read full chapter


Angelberger et.al. (2000), Fureby (2000)).Direct numerical simulation still is restrict-
ed to low Reynolds numbers or spatially small regions of interest (Ruith et.al. 2003).

> Read full chapter

Theoretical and numerical methods


Anthony F Molland, Stephen R Turnock, in Marine Rudders and Control Surfaces,
2007

6.3.6 Boundary conditions


In order to obtain correct solutions from the governing equations, it is necessary to
define the initial and/or boundary conditions for the dependent variables (U, V, W, P,
k, )T that describe the problem to be solved. The correct selection of these boundary
condition is fundamental to obtaining accurate flow solutions.

The use of unrealistic and badly posed boundary conditions can lead to spurious
and incorrect flow solutions, or more usually rapid solver divergence. The two most
frequently used linear boundary conditions are, the Dirichlet (value specified) and
Neumann conditions (gradient of value specified) [6.2]. Typical boundary conditions
used for rudder problems are:

6.3.6.1 Inlet
The inlet boundary condition is a form of Dirichlet boundary condition. On an inlet
boundary, the dependent variables of U, V, W, k, are prescribed. The pressure is not
set for incompressible flows, as it is extrapolated from downstream. The turbulence
quantities k and or equivalent for other turbulence models, are often difficult
to specify. If the computations are to be compared with experimental data, the inlet
turbulence quantities of k and should ideally be set according to measured values
found from experiment. When these measurements are not available the sensitivity
of the flow solution to the selection of k and parameters must be carried out.

If measurements of turbulence intensity do exist and the turbulent length scale of


the problem is known, then crude estimates of k and can be made. The turbulence
level or turbulence intensity within a flow is defined by

(6.46).

which reduces to
(6.47).

if all the Reynolds stresses are assumed equal and hence using Equation (6.44),

(6.48)

The specification of the rate of dissipation of kinetic energy is more difficult.


Estimates of can be made if measurements of the turbulent length scale l of the
problem exist [6.21, 6.24]

(6.49)

If the flow is known to be free of residual turbulence (which is unusual especially in


the case of flows in the marine environment or in wind tunnel experiments), k and
can be set to zero, the free-stream undisturbed condition. In practice, both k and
are set to small values, say 0.0001 to avoid solver convergence problems if k and
can turn negative. If the upstream inlet boundary is placed far enough upstream,
the choice of k and is less critical as they tend to dissipate to low values. Such a
choice can often come at too high a computational cost.

6.3.6.2 Wall
The wall boundary condition requires that the velocity on the wall satisfies the no-slip
condition. Also, on the wall k is zero and is nonzero. The k and turbulence models
can only be applied to regions that are fully turbulent and cannot be applied in
regions where viscous effects are dominant, such as those found in the laminar
sublayer. The wall function approach proposed by Launder and Spalding [6.27] can
be used to overcome this problem.

From experimental work, it is known that near-wall flows have a characteristic


multilayered structure within the boundary layer as shown in Figures 3.10 and 6.9.

Figure 6.9. Flow regimes for thin laminar-turbulent boundary layer

This consists of a laminar sublayer (viscous stress dominated) close to the wall, fol-
lowed by a buffer layer (viscous and turbulent stress of similar magnitude) and then
an outer turbulent core (turbulent stress dominated). Direct methods of resolving
the turbulent eddies within this boundary layer require extremely fine grids down
to the wall, through the laminar sublayer, which is computationally intensive and
very costly. However, most RANS codes can use turbulence models that employ wall
functions, based on the universal law of the wall [6.17]. The use of wall functions
avoids the need for fine grids in the laminar sublayer, by making use of empirical
fits within this region. When dealing with near-wall flows, positions and velocities
within the boundary layer are usually considered in nondimensional form, and y+ is
a local Reynolds number, with length scale in the direction perpendicular to the wall
yp as represented by:

(6.50)

and u+ a velocity based on the wall shear stress,

(6.51)

where up velocity at distance yp.

Specific flow structures within the boundary layer lie within strict bounds of y+, and
these are used in the formulation of wall functions. Positions within the boundary
layer, in which y+ ≤ 11.63, are regarded as laminar in structure and above 11.63 as
turbulent. It has been shown that within these two regions, two different functional
relationships exist between y+ and u+. These are shown as

(6.52)

(6.53)

for the laminar linear sublayer and for the turbulent log-law regions, respectively.

The constants in Equation (6.53) are determined from experiment. For hydraulically
smooth walls the Von Karman constant, = 0.4, and the log-layer constant, E =
9.793. Roughness can be simulated by increasing the value of E. The buffer layer
crossover value of 11.63, is found by finding the intersection of the linear laminar
sublayer profile (6.52); and the log-law turbulent profile (6.53).

The use of wall function turbulence models places specific requirements on the
meshes used in solving turbulent flow problems. When considering turbulent
near-wall flows, the most critical mesh parameter is the near-wall grid spacing. It
is of paramount importance, that near-wall grid spacing is selected in accordance
with the requirements of the wall function. In wall function turbulent calculations,
a y+ of 11.63 usually sets the lower limit for the distance of the first cell, to the
wall boundary, with the optimum near-wall position lying somewhere between y+
= 30 and no more than 500. Although the first cell spacing is critical for accurate
near-wall flow modelling, enough cells should also be placed through the whole
boundary layer to resolve the flow gradients. The use of wall functions, therefore,
poses a specific mesh independence problem for near-wall flows.

6.3.6.3 Mass flow outlet


The mass flow boundary condition is a form of the Neumann boundary condition.
Here the gradients of the dependent field variables (U, V, W, k, )T normal to the
boundary are initially set to zero and, later, (U, V, W)T is modified to have a constant
gradient to maintain global mass continuity. The pressure is extrapolated from
upstream. This boundary condition relies on the assumption that the flow is fully
developed when it reaches the outlet. Therefore, all outlet boundaries should be
placed far enough downstream to ensure that the fluid flow is fully developed, i.e.,
zero flow variable gradients in the flow direction. Positioning an outlet too close to an
area with a flow disturbance may result in solution errors, since the assumed outlet
condition of zero-flow gradient will not hold. Bluff body flows have areas of reversed
flow and these regions will violate the outlet boundary condition of outward flow if
the outlet is placed too close to the body. It is necessary in the case of incompressible
flows for global mass continuity to be maintained in order for the pressure correction
equation to be well posed. This means that the total flow out of the domain must
equal the total flow into the domain at all stages of the solution procedure.

6.3.6.4 Positioning of boundaries


It is important that all boundaries are positioned to ensure that they have no
demonstrable effect on the flow solution. In any CFD study, where high-accuracy
results are required, a sensitivity study should be carried out to demonstrate that the
interior flow solution is unaffected by the location of the boundaries. It must also be
remembered that it is not good practice to place boundaries at excessive distances
from the body. This wastes valuable computational resources, which could be put to
better use in resolving areas with high flow gradients.

> Read full chapter

A Computer-Controlled Wind Tunnel


A. NISHI, ... K. HIGUCHI, in Computational Wind Engineering 1, 1993

5. CONTROL OF TURBULENCE INTENSITY


At first, the turbulence intensity was controlled in four cases. The results are present-
ed in Fig.7 to Fig.10. In Fig.7 an example of input voltage distribution is given, and
the corresponding mean velocity profile is presented. A pair of turbulence intensities
in longitudinal and vertical directions, Iu and Iw, are shown in the same figure.

Fig. 7. Turbulence factors for the given input voltages (Case (a)).

Fig. 10. Turbulence factors for the given input voltages (Case (d)).

Almost the same mean velocity profile, except then just below the ceiling, was
measured for a quite different input voltage distribution, as shown in Fig.8. The
even numbered fans were driven at high voltages and the odd numbered ones were
stopped. Both of the larger intensities were obtained in longitudinal and vertical
directions. This means that an arbitrary intensity between this case and the former
case could be produced by providing an appropriate input voltage distribution
between these two cases.
Fig. 8. Turbulence factors for the given input voltage (Case (b)).

The third trial is presented in Fig.9. A periodical change in pulse input voltages with
a frequency of 1 Hz was used for all fans. Almost the same mean velocity profile
was obtained. The intensity in the longitudinal direction was three times larger than
in the vertical. Therefore, when different magnitudes of intensity are required, this
method can be employed.

Fig. 9. Turbulence factors for the given input voltages (Case (c)).

The fourth trial is shown in Fig.10. A periodical change in input voltages was applied
to the even numbered fans in different directions. Almost the same magnitude of
intensities as in case (b) was obtained.

These trials indicate that different magnitude of intensities can be produced by


controlling the input voltages.

> Read full chapter


LEADING EDGE FILM COOLING
HEAT TRANSFER INCLUDING THE
EFFECT OF MAINSTREAM TURBU-
LENCE
A.B. Mehendale, ... J.C. Han, in Transport Phenomena in Heat and Mass Transfer,
1992

5 RESULTS AND DISCUSSION


The local velocity and turbulence intensity distribution for both upstream turbulence
conditions are shown in Fig. 2. For the no grid case, the incident mainstream velocity
(U∞) at X/b = 20 is 10 m/s. A decrease in centerline velocity is observed due to
the approaching stagnation region; whereas, a gradual increase in right-side line
velocity is observed due to the blockage effect of the test model. Turbulence intensity
along the right-side line decays with distance. Along the centerline, the turbulence
intensity first decays; but as the flow approaches stagnation, there is an increase in
turbulence intensity due to a decrease in the local average mainstream velocity. For
a given upstream turbulence condition, the minimum value of the corresponding
turbulence intensity curve was chosen as the reference turbulence intensity for that
condition. Thus, the reference turbulence intensities for the no grid and passive
grid cases are 0.75% and 9.67%, respectively. Based on the method of Hancock
and Bradshaw (1983), the corresponding dissipation length scale (Leu) at the same
location for the passive grid is about 1.5 cm (Leu/D = 0.099).

FIGURE 2. Streamwise distribution of normalized mainstream velocity and stream-


wise turbulence intensity.

Effect of blowing ratio on span wise averaged heat transfer coefficient and film
effectiveness, for hole and slot injections and no grid case, is shown in Fig. 3. Also
shown for comparison are the no film heat transfer data for the no grid case from
Mick and Mayle (1988) and Mehendale et al. (1990). For both hole and slot injections,
an increase in blowing ratio causes an increase in heat transfer coefficient. This is
consistent with the fact that as the blowing ratio increases, the turbulence intensity
(the secondary-to-mainstream flow interaction) at the exit of the film holes/slots
increases. The heat transfer coefficient for 3-ℓ slots is lower over the entire leading
edge than for 4-d holes for corresponding blowing ratios. This is due to the effect of
film hole geometry. The blowing ratio of B = 0.8 shows the best film effectiveness for
both hole and slot cases over most of the test surface. B = 1.2 shows the worst film
effectiveness. This is due to penetration of the secondary flow into the mainstream.
In general, slots show better film effectiveness than holes for no grid low mainstream
turbulence condition.

FIGURE 3. Effect of blowing ratio on span wise averaged Nusselt number and film
effectiveness for 4-d hole and 3-ℓ slot injections for the no grid (Tu = 0.75%) case.

The effect of blowing ratio on spanwise averaged heat transfer coefficient and film
effectiveness, for hole and slot injections and passive grid case, is shown in Fig. 4. As
shown in Fig. 3 for the no grid case, an increase in blowing ratio causes an increase
in heat transfer coefficient for both hole and slot injections. The data are higher
than before since the mainstream turbulence intensity is much higher compared
to the no grid case. As before, 3-ℓ slots have lower heat transfer coefficient than
4-d holes. B = 0.8 produces the best film effectiveness. The spread in heat transfer
coefficient and film effectiveness due to different blowing ratios is reduced for
passive grid high mainstream turbulence condition.

FIGURE 4. Effect of blowing ratio on spanwise averaged Nusselt number and film
effectiveness for 4-d hole and 3-ℓ slot injections for the passive grid (Tu = 9.67%)
case.
Fig. 5 shows the effect of mainstream turbulence on spanwise averaged heat transfer
coefficient and film effectiveness for the case of hole and slot injections at the
blowing ratio of B = 0.4. An increase in mainstream turbulence causes an increase
in heat transfer coefficient on the leading edge in both cases of slot and hole
injection. Considerable increases in heat transfer coefficient are observed for the
case of film holes over film slots just downstream of film injection. This is due solely
to the effect of film hole geometry. For the lowest blowing ratio B = 0.4, slot injection
clearly provides better film coverage than hole injection between the first and second
rows of film injection. This effect is reduced after the second row of film injection.
For the case of slot injection, an increase in mainstream turbulence adversely affects
the film effectiveness. The effect is not so severe for the case of hole injection.

FIGURE 5. Effect of mainstream turbulence on spanwise averaged Nusselt number


and film effectiveness for 4-d hole and 3-ℓ slot injections for the blowing ratio of B
= 0.4.

For the intermediate blowing ratio B = 0.8, Fig. 6 shows the effect of mainstream
turbulence on spanwise averaged heat transfer coefficient and film effectiveness for
the case of hole and slot injections. An increase in mainstream turbulence produces
higher heat transfer coefficient for both injection cases over the entire leading
edge. The increases in heat transfer coefficient, due to mainstream turbulence,
are not as severe as for B = 0.4 in Fig. 5. The increases are very minimal except for
just downstream of film injection. This is because at B = 0.8, the secondary flow
turbulence at the exit of the film holes/slots is much higher than that for the case of
B = 0.4. For the same reason, the decreases in film effectiveness due to mainstream
turbulence are much less compared to B = 0.4 (see Fig. 5).
FIGURE 6. Effect of mainstream turbulence on spanwise averaged Nusselt number
and film effectiveness for 4-d hole and 3-ℓ slot injections for the blowing ratio of B
= 0.8.

It is important to study the overall effect of film cooling. A better cooling design
should be the one which provides a greater reduction in heat transfer (heat load) over
the no film cooling case. The no film cooling data from Mehendale et al. (1990) is
used for the purpose of heat load calculations. The parameter head load ratio (Mick
and Mayle, 1988) is defined as

(3)

where is the spanwise averaged heat flux without film cooling at any streamwise
location, is the spanwise averaged heat flux with film cooling at the same streamwise
location, n is the number of thermocouples in the row, ho is the heat transfer
coefficient without film cooling at the same streamwise location, and is the
overall cooling effectiveness given by = (Tw - T∞)/(Ts - T∞). Thus, if heat load ratio
is less than one, then film cooling provides a reduction in heat load as compared
to the no film cooling case and vice versa. The value of is usually 0.5-0.6 for gas
turbines. Using = 0.6, a comparison of overall performance due to film cooling is
made for the cases of both circular hole and slot injections.

Effect of blowing ratio on spanwise averaged heat load ratio, for hole and slot
injections and the no grid case, is shown in Fig. 7. For both cases of hole and slot
injections, the blowing ratio of B = 0.8 shows the best performance and B = 1.2 shows
the worst performance. This is due to an optimum film coverage for the intermediate
blowing ratio of B = 0.8. For both cases of hole and slot injection, except for the
highest blowing ratio of B = 1.2, film cooling is seen to reduce heat load ratio over
the entire test surface. In general, 3-ℓ slot shows better performance than 4-d hole.
FIGURE 7. Effect of blowing ratio on spanwise averaged heat load ratio for 4-d hole
and 3-ℓ slot injections for no grid (Tu = 0.75%) and passive grid (Tu = 9.67%) cases.

Fig. 7 also shows the effect of blowing ratio on spanwise averaged heat load ratio
for hole and slot injections and the passive grid case. It is observed that higher
mainstream turbulence causes a considerable reduction in heat load ratio at the
flow reattachment point for all blowing ratios. The heat load ratios for the passive
grid case are less than 1.0 over most of the test surface. Thus, it appears that the
mainstream turbulence breaks the penetration of secondary flow for the highest
blowing ratio of B = 1.2. For both cases of hole and slot injection, the intermediate
blowing ratio of B = 0.8 still provides the lowest heat load ratio. However, the heat
load ratio for the hole injection is slightly lower than that for the slot injection.

The liquid crystals technique, as in Hippensteele et al., 1981, was used for thermal
visualization. For all test cases, the secondary air temperature was maintained at
57°C. The results for 3-ℓ slot geometry are shown in Fig. 8. For the low turbulence
case, the blowing ratios of B = 0.4 and 0.8 provide more film coverage than B =
1.2. For the high turbulence case, the blowing ratio of B = 0.8 provides the best
film coverage. An increase in mainstream turbulence is seen to cause a reduction
in film coverage. This effect is most severe for the lowest blowing ratio of B = 0.4
and least severe for the highest blowing ratio of B = 1.2. These results are reasonably
consistent with data presented earlier. The 4-d hole geometry results were similar
except the coverage was more in the streamwise direction. The results are not shown
due to space constraints.
FIGURE 8. Comparison of film thermal visualizations for 3-ℓ slots at blowing ratios
of B = 0.4, 0.8, and 1.2 for no grid (Tu = 0.75%) and passive grid (Tu = 9.67%) cases.

> Read full chapter

Turbulent Structure in the Three-Di-


mensional Boundary Layer on a Swept
Wing
Motoyuki Itoh, Minoru Kobayashi, in Engineering Turbulence Modelling and Exper-
iments 4, 1999

3.3 Turbulence field


Figure 8 shows the turbulence intensities and . In the outer region (y + > 200), both
and increase with going downstream. This may be ascribable to the effect of adverse
pressure gradient (Nagano et al. [11]). It should be remarked that, in the near-wall
region, decreases appreciably with going downstream from x = 396 mm to x = 540-
 mm. This seems to be due to the effects of increased three-dimensionality. On the
other hand, increases even in the near-wall region with going downstream from
x = 396 mm to x = 540 mm, as seen in Fig. 8(b). These results are consistent with the
numerical prediction (Moin et al. [12]) studying the effects of three-dimensionality
on the turbulence statistics.

Figure 8. Distributions of turbulence intensities

The wall-normal turbulence intensity is shown in Fig. 9. As we used an X-wire


probe to obtain -values, the data in the near-wall region could not be obtained. The
increase of with going downstream seems to be due to the dominant effect of the
adverse pressure gradient [11].

Figure 9. Wall-normal component of turbulence intensity

Figure 10 shows the turbulent shear stress , where u* is the turbulent velocity
component in the local mean flow direction. It should be noted that, in the region
y/ 99 < 0.4, the value of decreases with going downstream from x = 396 mm to
x = 540 mm. This seems to be caused by the effect of increased three-dimensionality.
Figure 10. Turbulent shear stress

Although all the components of Reynolds shear stress were obtained by the tech-
nique of a rotated hot-wire, they will not be disussed here any more because the
main concern in the present work is to clarify the causes for decreasing structure
parameter, A1, with increasing three-dimensionality. The Reynolds stresses data
obtained in the present work are available in machine-readable form for use by
interested workers.

The distributions of the turbulent kinetic energy, , are shown in Fig. 11. The values
of near the wall decrease with going downstream from x = 396 mm to x = 540 mm,
which also seems to be due to the effect of increased three-dimensionality.

Figure 11. Distributions of turbulent kinetic energy

Figure 12 shows the distributions of structure parameter A1 (ratio of shear-stress


magnitude in the plane parallel to the wall to twice the turbulent kinetic energy). As
has been well known, the value of A1 decreases with increasing three-dimensionality.
In Fig. 12, A1 decreases from about 0.15 (at x = -108 mm) to 0.11 (at x = 540 mm) in
the central region of the boundary layer.

Figure 12. Distributions of structure parameter A1


Figures 13 (a) and (b) show the turbulent transport of turbulence energy component
and the turbulent shear stress . It is seen in Fig. 13(a) that with going downstream
the values of decrease near the wall and become negative at the stations x = 396 mm
and X = 540 mm. The negativevalues of indicate that the turbulence energy is
transported toward the wall. It will be known from Fig. 13(b) that the turbulent
transport of is also toward the wall in the same region as for . The wallward turbulent
transport has also be found in the turbulent boundary layers with adverse pressure
gradient ([11][13])

Figure 13. Distributions of turbulent transport

> Read full chapter

THE EFFECT OF WALL ROUGHNESS


ON AN OPEN CHANNEL BOUNDARY
LAYER
M.F. Tachie, ... R. Balachandar, in Engineering Turbulence Modelling and Experi-
ments 5, 2002

Turbulence Quantities
The distributions of the streamwise turbulence intensity on the three different
surfaces are shown in Figure 2a and 2b, using the friction velocity U and the outer
scaling (Ue) advocated by [GC], respectively. The use of inner scaling (Figure 2a)
shows that wall roughness reduces the turbulence level in the immediate vicinity of
the wall (i.e. y+ ≤ 30). Outside this region, the data obtained on the wire screen (WM)
are significantly larger than those of the smooth surface (SM). The deviation of SG
from SM in the region y+ > 30 is only minimal, i.e. within measurement uncertainty.
Figure 2b demonstrates that the sand grain data (SG) deviate from the smooth data
over the inner half of the boundary layer, but the deviation of the wire mesh data
(WM) from the smooth wall profile persist up to y/ ≈1. The differences are most
pronounced in the immediate vicinity of the wall where the peak values also occur.
Although the average roughness height of WM (i.e. the wire diameter) is only 50
percent of the nominal diameter of the sand grains (SG), WM shows higher level of
turbulence than SG over a significant portion of the flow. This is a clear evidence that
the extent to which surface roughness influence the turbulence structure depends
on the specific geometry of the roughness elements.

Figure 2a. Streamwise turbulence intensity in inner coordinates

Figure 2b. Streamwise turbulence intensity in outer coordinates

Figure 3a and 3b, respectively, show the wall-normal turbulence intensity for the
smooth and rough surfaces using U and Ue, which is the correct scaling according
to the theory of [GC]. Due to spatial resolution limitations, two-component data (e.g.
v and <uv>) could not be obtained in the region y ≤ 30 or y/ ≤ 0.02. Irrespective of the
scaling used, the figures suggest that surface roughness enhances the level of the
wall-normal turbulence intensity. It is also noted that the wall-normal component
of the velocity fluctuation is even more sensitive to the specific wall condition than
the streamwise component.
Figure 3a. Wall-normal turbulence intensity in inner coordinates

Figure 3b. Wall-normal turbulence intensity in outer coordinates

Figure 4 shows plots of the Reynolds shear stress on the three different surfaces
using both inner and outer scaling. Again, clear differences are observed in the three
profiles, irrespective of the scaling, with the peak value increasing markedly with
increasing roughness.

Figure 4a. Distribution of Reynolds shear stress in inner coordinates

We also examine the effect of surface roughness on the triple correlations. For
brevity, only the components, <v3> and <u2v> which represent turbulent transport
of <v2> and <u2>, respectively, in the wall-normal direction are shown in this paper.
These components are also the major contributors to the turbulent diffusion term in
the energy budget. The sum of <v3> and <u2v>, i.e. (<v3> + <u2v>) is shown in Figure
5. Following the analysis of [GC], the parameter U 2Ue is used to normalise the data.
It is apparent from Figure 5 that surface roughness increases the triple correlation
in the inner half of the boundary layer. The location of the outer (and higher) peaks
is also strongly influenced by surface roughness.

Figure 5. Roughness effects on triple correlation

> Read full chapter

ScienceDirect is Elsevier’s leading information solution for researchers.


Copyright © 2018 Elsevier B.V. or its licensors or contributors. ScienceDirect ® is a registered trademark of Elsevier B.V. Terms and conditions apply.
Turbulence Intensity
Turbulence intensity is defined as the ratio of standard deviation of fluctuating wind
velocity to the mean wind speed, and it represents the intensity of wind velocity
fluctuation.

From: Innovative Bridge Design Handbook, 2016

Related terms:

Energy Engineering, Turbulence, Boundary Layer, Turbines, Wind Tunnels, Wind


Turbines, Reynolds' Number, Freestream, Streamwise

View all Topics

Heat and Mass Transfer Across a Hol-


low Fiber Membrane Bundle
Li-Zhi Zhang, in Conjugate Heat and Mass Transfer in Heat Mass Exchanger Ducts,
2013

7.7.2 Turbulence for the air flow


Turbulence intensities (ua /ua) in the air flow for the in-line and the staggered
arrangements are shown in Figure 7.22a and b, respectively. As can be seen, the
turbulence intensities are higher where velocities are higher. It has been found that
vortexes appear behind the fibers. The turbulence intensities behind the fibers where
vortexes appear are higher than other regions. For both the in-line and the staggered
arrangements, the turbulence intensities are higher than 0.07, meaning that the
fluctuations of velocities for the air flow in the bundle are relatively large [13,14].
In other words, the turbulent behavior of the air flow across the bundle should be
taken into account. Therefore the turbulence model employed here is reasonable
and necessary to account for the disturbances from the fibers.
Figure 7.22. Contours of the turbulence intensities for the air stream, PL/(2ro) = PT-
/(2ro) = 2.0,  = 0.196,Rea = 300. (a) In-line arrangement; (b) staggered arrangement.

> Read full chapter

Experimental Investigation of Turbu-


lent Wake-Blade Interaction in Axial
Compressors
A. Sentker, W. Riess, in Engineering Turbulence Modelling and Experiments 4, 1999

3.2 Turbulence intensity distribution first stage


The turbulence intensity, or unresolved unsteadiness, which is calculated from the
mean square fluctuation components divided by the mean absolute velocity [3], gives
a good insight into the behaviour of the unsteady flow field travelling through the
stator.

Considering the turbulence intensity in E2 (fig.6) most striking is the high turbulence
in the wakes of the rotor blades with more than 5%. With a distance of 12° rotorposi-
tion they move from the right to the left side through the three dimensional diagram.
The wake of the IGV is also characterised by an increased turbulence intensity (3%).
It is situated spatially fixed at 0.5° to 3° measuring position. At the position where
the rotor and IGV wakes intersect the IGV wake is shifted, with the part on the
suction side of the rotor wake moving faster than that on the pressure side. Inside
the stator channel in E2b presented in fig.7 the rotor wake is still visible due to its
high turbulence intensity of 4.5%. The shape of the rotor wake is changing inside
the stator channel. The absolute velocity near the suction side of the channel is
higher so that the part of the rotor wake near to the suction side of the channel
is transported faster downstream. This results in a twisting of the rotor wake in the
three dimensional diagram. Near the pressure side of the stator channel occurs a
spatially fixed region with a higher turbulence intensity. It can be assumed that the
increased turbulence at this measuring position results from an IGV wake. Changing
the operating point of the compressor leads to other positions of this region of
increased turbulence inside the stator channel. This fact underlines, that the higher
turbulence near the suction side of the stator channel is not generated by the velocity
or pressure distribution inside the stator channel but originates from the IGV wake.

Figure 6. Turbulence intensity distribution in E2

Figure 7. Turbulence intensity distribution in E2b

Considering fig.8, where the turbulence intensity in E2d is shown, supports the
observation that an influence of the IGV wake on the unsteady flow field is still
detectable there. In a measuring range of –5.5° to –7.5° a spatially fixed region
with increased turbulence intensity is visible, which can be identified as the IGV
wake. Compared to the absolute velocity distribution the position of the IGV wake
has changed, because the highest turbulence occurs inside the wake and the max.
velocity in fig. 4 comes from the suction side of the IGV wake. The maximum
turbulence inside the rotor wake in E2d is less than in E2b with 4%.

Figure 8. Turbulence intensity distribution in E2d

In the measuring plane behind the stator, presented in fig.9, the wakes of the stator
with a turbulence intensity of about 10% are dominant. They are spatially fixed
at 3.5° and –10.5° measuring position. In contrast to fig.5 the diagonal regions
with an increased turbulence level are the wakes of the upstream rotor, transported
through the stator. The time distance between two adjacent rotor wakes is no longer
constant at 12° but has diminished. The wakes are shifted clockwise due to the
velocity gradient inside the stator pitch.

Figure 9. Turbulence intensity distribution in E3

Considering the position of the interaction between stator and rotor wake, a differ-
ence of 6° rotor position between the part of the rotor wake on the suction side
compared to the part on the pressure side of the stator wake can be determined.
Near the pressure side of the stator wake the curvature of the rotor wakes is strong.
The IGV wake, too, is weakly visible at –4.5° to –7.5° measuring position. The increase
of the turbulence intensity due to the IGV wake is not very strong behind the stator.
It has been nearly mixed out.

In fig. 10 the way of the IGV wake through the first stage is reconstructed schemat-
ically from the measured unsteady flow data. The extension in circumferential
direction in E2a and E2b could not be measured because the IGV wake is situated at
the border of the max. possible probe traverse. The IGV wake follows an imaginary
streamline on its way through the stator. Its circumferential extension grows from
E2 to E3, it follows the curvature of the stator blade.

Figure 10. IGV wake in first stage

> Read full chapter

LDA-Measurements of the Turbulence


in and Around a Venturi
R.F. Mudde, ... H.R.E. van Maanen, in Engineering Turbulence Modelling and
Experiments 6, 2005

Turbulence Intensities
The axial and tangential turbulence intensities, Iax and Itan, have been calculated via
the definition: with the local mean velocity. The intensities have been corrected for
noise contributions by inspecting the ACF. This correction is better for signals with
a high SNR, stressing the importance of high quality data. As the definition of the
intensities contains the local mean velocity, the values of the intensities go up when
approaching the wall. The axial turbulence intensities are shown in fig.(8).
Figure 8. Turbulence intensity profiles of the axial velocity

The turbulence intensity in the throat is much lower than 1D upstream of the Venturi
(where the turbulence intensity is equal to that of the pipe further upstream), in
agreement with the literature. In the diffusor, the intensity in the central part is
still low: the flow that comes out of the throat moves as a jet of low turbulence
intensity into the diffusor section. The intensity in the wall region has increased to
values much higher than that of a well developed turbulent pipe flow. The increase
continues further downstream. Just outside the diffusor the levels in the center are
around those of the turbulent flow in the pipe; in the wall region they have increased
further. Further downstream, the profile becomes flatter but is higher than that
of the turbulent flow. From the turbulence intensity it is obvious that the flow 5D
downstream is not in equilibrium: the turbulence intensity profile is flatter than for
turbulent flow in a fully developed pipe flow and its magnitude is roughly twice
as high, indicating that the dissipation is not in equilibrium with the production.
The tangential velocity fluctuations show a similar turbulence intensity development
along the Venturi.

> Read full chapter

Application of a lagrangian PDF


Method to Turbulent Gas/Particle Com-
bustion
M. Rose, ... M.G. Neuhaus, in Engineering Turbulence Modelling and Experiments
4, 1999

5.2 Results of Calculations


The effect of turbulence intensity on temperature and CO mass fraction in a
turbulent reacting gas/particle mixture was investigated. The upper part of Fig. 3
shows the distribution of temperature in the x/y-plane 20 ms after particles start to
devolatilize CO into the preheated ambient air.

Figure 3. Distribution of temperature (upper part, 300 K ≤ T ≤ 1950 K) and CO mass


fraction (lower part, 0 ≤ [CO] ≤ 0.4) in the x/z-plane for low (left) and high (right)
turbulence intensity 20 ms after particles start to devolatilize.

The left and right parts of the figure correspond to low and high turbulence
intensity, respectively. At this time, the mean characteristics of the flow field have
become stationary, i.e. spreading angle of the ”flame” and mean gas/particle-mix-
ture compositions at different cross-sections normal to the x-axis. Independent of
turbulence intensity, high chemical activity indicated by light colours in the T-picture
is restricted to the vicinity of CP-particles in the inflow section, x ≤ 1 cm. Further
downstream, turbulent transport influences significantly the evolution of the
temperature field. For lower turbulent mixing, the spreading angle of the flame front
in the flow direction is much lower than in the case of higher turbulence. In the
latter case, also the distance between isothermals in the z-direction is greater than
for lower turbulence intensity. This represents a much higher flame brush thickness.
In addition, the more erratic form of the isothermals for higher turbulence indicate
a much more wrinkled and corrugated shape of the flame.

Characteristic distributions of the CO mass fractions are shown in the lower part of
Fig. 3. In both parts of the figure, the amount of unburnt CO in the inflow region
is high, which is represented by light colours. For higher turbulence, the enhanced
transport in the z-direction leads to a much better mixing of CO with fresh air further
downstream, resulting in a much lower amount of CO at the right border of the
volume. In the case of lower turbulence intensity, devolatilized CO is not burned
completely and leaves the volume mostly concentrated near z = 0 cm.
A quantitative description of the results illustrated above is given in Figs. 4 to 6.
Figs. 4 shows the mean temperature and CO mass fraction along the x-axis. The
values were obtained by averaging in the z-direction between z = 0 cm and z = 0.5 cm,
i.e. the originally particle loaded flow part. The inflow region x ≤ 1 cm is characterised
by peaks in temperature and CO concentration up to 1400 K and 40 %, respectively.
The very high heat release results from homogeneous burning of devolatilized CO
and heterogeneous particle burning in the fresh surrounding air. A local minimum
in temperature and CO concentration follows the zone of individual particle burning
at about x = 1 cm and x = 0.5 cm, respectively. From about x = 1.5 cm the graphs of
temperature and CO mass fraction become more continuous, indicating that the
chemical activity is no longer concentrated to the vicinity of individual particles.
The combustion mode changes from diffusion/devolatilization controlled reaction
to partially premixed reaction in this region. In the case of high turbulence intensity,
the temperature remains nearly constant at a value of about 1400 K for x ≥ 1.5 cm,
whereas the temperature further increases up to 1800 K for less intensive turbulent
mixing. At the same time, the mass fraction of CO decreases down to less than 5%
at the exit, whereas for low turbulence more than 12% of the total mass is unburned
CO.

Figure 4. Temperature distributions (left) and CO mass fraction (right) along x-axis
averaged for 0 cm ≤ z ≤ 0.5 cm.

Figure 5. Temperature distributions (left) and CO mass fraction (right) along x-axis
averaged for 1.5 cm ≤ z ≤ 2 cm.
Figure 6. Profiles of temperature (left) and CO mass fraction (right) at the right end
of the flow volume averaged for 4.5 cm ≤ x ≤ 5 cm.

The effect of different turbulence intensities on the flow field becomes even more
significant when looking at the distribution of temperature and CO mass fraction
in the upper section of the flow field at z = 1.75 cm, see Fig. 5. In the case of high
turbulence intensity, the left part of the figure shows for x ≥ 1.5 cm a significant
continuous increase in temperature, resulting in an average outlet temperature
which is about 200 K higher than for lower turbulent mixing. The CO mass fraction
shown in the right part of Fig. 5, is always less than 1%, and has a maximum at
x = 3 cm in the case of high turbulence. This is exactly the position where the
turbulent flame brush influences the flow field, whereas in the case of low turbulence
intensity only very little CO reaches this upper flow section by diffusion and turbulent
transport.

The profiles of temperature and CO mass fraction in the vertical outlet region are
given in Fig. 6. The left part shows that lower turbulence intensity causes more
heat release in the originally particle loaded flow part z ≤ l cm and less one for
z ≥ 1 cm. compared to the case with high turbulence intensity. The corresponding
temperature graph shows a high gradient around z = 1 cm, connecting the lower
temperature region of about 800 K with a high temperature region of about 1900 K.
The temperature graph of the corresponding high turbulence case is more smooth
with less extreme values ranging between 1000 K and 1500 K. It becomes obvious
from the right part of Fig. 6, that the amount of unburned CO leaving the flow sec-
tion for z ≤ 0.5 cm is very high in the case of low turbulence intensity. Furthermore,
the temperature is also high in this flow part, but the expected high reactivity is
restricted by a lack of oxidizer.

> Read full chapter


Angelberger et.al. (2000), Fureby (2000)).Direct numerical simulation still is restrict-
ed to low Reynolds numbers or spatially small regions of interest (Ruith et.al. 2003).

> Read full chapter

Theoretical and numerical methods


Anthony F Molland, Stephen R Turnock, in Marine Rudders and Control Surfaces,
2007

6.3.6 Boundary conditions


In order to obtain correct solutions from the governing equations, it is necessary to
define the initial and/or boundary conditions for the dependent variables (U, V, W, P,
k, )T that describe the problem to be solved. The correct selection of these boundary
condition is fundamental to obtaining accurate flow solutions.

The use of unrealistic and badly posed boundary conditions can lead to spurious
and incorrect flow solutions, or more usually rapid solver divergence. The two most
frequently used linear boundary conditions are, the Dirichlet (value specified) and
Neumann conditions (gradient of value specified) [6.2]. Typical boundary conditions
used for rudder problems are:

6.3.6.1 Inlet
The inlet boundary condition is a form of Dirichlet boundary condition. On an inlet
boundary, the dependent variables of U, V, W, k, are prescribed. The pressure is not
set for incompressible flows, as it is extrapolated from downstream. The turbulence
quantities k and or equivalent for other turbulence models, are often difficult
to specify. If the computations are to be compared with experimental data, the inlet
turbulence quantities of k and should ideally be set according to measured values
found from experiment. When these measurements are not available the sensitivity
of the flow solution to the selection of k and parameters must be carried out.

If measurements of turbulence intensity do exist and the turbulent length scale of


the problem is known, then crude estimates of k and can be made. The turbulence
level or turbulence intensity within a flow is defined by

(6.46).

which reduces to
(6.47).

if all the Reynolds stresses are assumed equal and hence using Equation (6.44),

(6.48)

The specification of the rate of dissipation of kinetic energy is more difficult.


Estimates of can be made if measurements of the turbulent length scale l of the
problem exist [6.21, 6.24]

(6.49)

If the flow is known to be free of residual turbulence (which is unusual especially in


the case of flows in the marine environment or in wind tunnel experiments), k and
can be set to zero, the free-stream undisturbed condition. In practice, both k and
are set to small values, say 0.0001 to avoid solver convergence problems if k and
can turn negative. If the upstream inlet boundary is placed far enough upstream,
the choice of k and is less critical as they tend to dissipate to low values. Such a
choice can often come at too high a computational cost.

6.3.6.2 Wall
The wall boundary condition requires that the velocity on the wall satisfies the no-slip
condition. Also, on the wall k is zero and is nonzero. The k and turbulence models
can only be applied to regions that are fully turbulent and cannot be applied in
regions where viscous effects are dominant, such as those found in the laminar
sublayer. The wall function approach proposed by Launder and Spalding [6.27] can
be used to overcome this problem.

From experimental work, it is known that near-wall flows have a characteristic


multilayered structure within the boundary layer as shown in Figures 3.10 and 6.9.

Figure 6.9. Flow regimes for thin laminar-turbulent boundary layer

This consists of a laminar sublayer (viscous stress dominated) close to the wall, fol-
lowed by a buffer layer (viscous and turbulent stress of similar magnitude) and then
an outer turbulent core (turbulent stress dominated). Direct methods of resolving
the turbulent eddies within this boundary layer require extremely fine grids down
boundary layer to resolve the flow gradients. The use of wall functions, therefore,
poses a specific mesh independence problem for near-wall flows.

6.3.6.3 Mass flow outlet


The mass flow boundary condition is a form of the Neumann boundary condition.
Here the gradients of the dependent field variables (U, V, W, k, )T normal to the
boundary are initially set to zero and, later, (U, V, W)T is modified to have a constant
gradient to maintain global mass continuity. The pressure is extrapolated from
upstream. This boundary condition relies on the assumption that the flow is fully
developed when it reaches the outlet. Therefore, all outlet boundaries should be
placed far enough downstream to ensure that the fluid flow is fully developed, i.e.,
zero flow variable gradients in the flow direction. Positioning an outlet too close to an
area with a flow disturbance may result in solution errors, since the assumed outlet
condition of zero-flow gradient will not hold. Bluff body flows have areas of reversed
flow and these regions will violate the outlet boundary condition of outward flow if
the outlet is placed too close to the body. It is necessary in the case of incompressible
flows for global mass continuity to be maintained in order for the pressure correction
equation to be well posed. This means that the total flow out of the domain must
equal the total flow into the domain at all stages of the solution procedure.

6.3.6.4 Positioning of boundaries


It is important that all boundaries are positioned to ensure that they have no
demonstrable effect on the flow solution. In any CFD study, where high-accuracy
results are required, a sensitivity study should be carried out to demonstrate that the
interior flow solution is unaffected by the location of the boundaries. It must also be
remembered that it is not good practice to place boundaries at excessive distances
from the body. This wastes valuable computational resources, which could be put to
better use in resolving areas with high flow gradients.

> Read full chapter

A Computer-Controlled Wind Tunnel


A. NISHI, ... K. HIGUCHI, in Computational Wind Engineering 1, 1993

5. CONTROL OF TURBULENCE INTENSITY


At first, the turbulence intensity was controlled in four cases. The results are present-
ed in Fig.7 to Fig.10. In Fig.7 an example of input voltage distribution is given, and
the corresponding mean velocity profile is presented. A pair of turbulence intensities
in longitudinal and vertical directions, Iu and Iw, are shown in the same figure.

Fig. 7. Turbulence factors for the given input voltages (Case (a)).

Fig. 10. Turbulence factors for the given input voltages (Case (d)).

Almost the same mean velocity profile, except then just below the ceiling, was
measured for a quite different input voltage distribution, as shown in Fig.8. The
even numbered fans were driven at high voltages and the odd numbered ones were
stopped. Both of the larger intensities were obtained in longitudinal and vertical
directions. This means that an arbitrary intensity between this case and the former
case could be produced by providing an appropriate input voltage distribution
between these two cases.
Fig. 8. Turbulence factors for the given input voltage (Case (b)).

The third trial is presented in Fig.9. A periodical change in pulse input voltages with
a frequency of 1 Hz was used for all fans. Almost the same mean velocity profile
was obtained. The intensity in the longitudinal direction was three times larger than
in the vertical. Therefore, when different magnitudes of intensity are required, this
method can be employed.

Fig. 9. Turbulence factors for the given input voltages (Case (c)).

The fourth trial is shown in Fig.10. A periodical change in input voltages was applied
to the even numbered fans in different directions. Almost the same magnitude of
intensities as in case (b) was obtained.

These trials indicate that different magnitude of intensities can be produced by


controlling the input voltages.

> Read full chapter


LEADING EDGE FILM COOLING
HEAT TRANSFER INCLUDING THE
EFFECT OF MAINSTREAM TURBU-
LENCE
A.B. Mehendale, ... J.C. Han, in Transport Phenomena in Heat and Mass Transfer,
1992

5 RESULTS AND DISCUSSION


The local velocity and turbulence intensity distribution for both upstream turbulence
conditions are shown in Fig. 2. For the no grid case, the incident mainstream velocity
(U∞) at X/b = 20 is 10 m/s. A decrease in centerline velocity is observed due to
the approaching stagnation region; whereas, a gradual increase in right-side line
velocity is observed due to the blockage effect of the test model. Turbulence intensity
along the right-side line decays with distance. Along the centerline, the turbulence
intensity first decays; but as the flow approaches stagnation, there is an increase in
turbulence intensity due to a decrease in the local average mainstream velocity. For
a given upstream turbulence condition, the minimum value of the corresponding
turbulence intensity curve was chosen as the reference turbulence intensity for that
condition. Thus, the reference turbulence intensities for the no grid and passive
grid cases are 0.75% and 9.67%, respectively. Based on the method of Hancock
and Bradshaw (1983), the corresponding dissipation length scale (Leu) at the same
location for the passive grid is about 1.5 cm (Leu/D = 0.099).

FIGURE 2. Streamwise distribution of normalized mainstream velocity and stream-


wise turbulence intensity.

Effect of blowing ratio on span wise averaged heat transfer coefficient and film
effectiveness, for hole and slot injections and no grid case, is shown in Fig. 3. Also
shown for comparison are the no film heat transfer data for the no grid case from
Mick and Mayle (1988) and Mehendale et al. (1990). For both hole and slot injections,
an increase in blowing ratio causes an increase in heat transfer coefficient. This is
consistent with the fact that as the blowing ratio increases, the turbulence intensity
(the secondary-to-mainstream flow interaction) at the exit of the film holes/slots
increases. The heat transfer coefficient for 3-ℓ slots is lower over the entire leading
edge than for 4-d holes for corresponding blowing ratios. This is due to the effect of
film hole geometry. The blowing ratio of B = 0.8 shows the best film effectiveness for
both hole and slot cases over most of the test surface. B = 1.2 shows the worst film
effectiveness. This is due to penetration of the secondary flow into the mainstream.
In general, slots show better film effectiveness than holes for no grid low mainstream
turbulence condition.

FIGURE 3. Effect of blowing ratio on span wise averaged Nusselt number and film
effectiveness for 4-d hole and 3-ℓ slot injections for the no grid (Tu = 0.75%) case.

The effect of blowing ratio on spanwise averaged heat transfer coefficient and film
effectiveness, for hole and slot injections and passive grid case, is shown in Fig. 4. As
shown in Fig. 3 for the no grid case, an increase in blowing ratio causes an increase
in heat transfer coefficient for both hole and slot injections. The data are higher
than before since the mainstream turbulence intensity is much higher compared
to the no grid case. As before, 3-ℓ slots have lower heat transfer coefficient than
4-d holes. B = 0.8 produces the best film effectiveness. The spread in heat transfer
coefficient and film effectiveness due to different blowing ratios is reduced for
passive grid high mainstream turbulence condition.

FIGURE 4. Effect of blowing ratio on spanwise averaged Nusselt number and film
effectiveness for 4-d hole and 3-ℓ slot injections for the passive grid (Tu = 9.67%)
case.
Fig. 5 shows the effect of mainstream turbulence on spanwise averaged heat transfer
coefficient and film effectiveness for the case of hole and slot injections at the
blowing ratio of B = 0.4. An increase in mainstream turbulence causes an increase
in heat transfer coefficient on the leading edge in both cases of slot and hole
injection. Considerable increases in heat transfer coefficient are observed for the
case of film holes over film slots just downstream of film injection. This is due solely
to the effect of film hole geometry. For the lowest blowing ratio B = 0.4, slot injection
clearly provides better film coverage than hole injection between the first and second
rows of film injection. This effect is reduced after the second row of film injection.
For the case of slot injection, an increase in mainstream turbulence adversely affects
the film effectiveness. The effect is not so severe for the case of hole injection.

FIGURE 5. Effect of mainstream turbulence on spanwise averaged Nusselt number


and film effectiveness for 4-d hole and 3-ℓ slot injections for the blowing ratio of B
= 0.4.

For the intermediate blowing ratio B = 0.8, Fig. 6 shows the effect of mainstream
turbulence on spanwise averaged heat transfer coefficient and film effectiveness for
the case of hole and slot injections. An increase in mainstream turbulence produces
higher heat transfer coefficient for both injection cases over the entire leading
edge. The increases in heat transfer coefficient, due to mainstream turbulence,
are not as severe as for B = 0.4 in Fig. 5. The increases are very minimal except for
just downstream of film injection. This is because at B = 0.8, the secondary flow
turbulence at the exit of the film holes/slots is much higher than that for the case of
B = 0.4. For the same reason, the decreases in film effectiveness due to mainstream
turbulence are much less compared to B = 0.4 (see Fig. 5).
FIGURE 6. Effect of mainstream turbulence on spanwise averaged Nusselt number
and film effectiveness for 4-d hole and 3-ℓ slot injections for the blowing ratio of B
= 0.8.

It is important to study the overall effect of film cooling. A better cooling design
should be the one which provides a greater reduction in heat transfer (heat load) over
the no film cooling case. The no film cooling data from Mehendale et al. (1990) is
used for the purpose of heat load calculations. The parameter head load ratio (Mick
and Mayle, 1988) is defined as

(3)

where is the spanwise averaged heat flux without film cooling at any streamwise
location, is the spanwise averaged heat flux with film cooling at the same streamwise
location, n is the number of thermocouples in the row, ho is the heat transfer
coefficient without film cooling at the same streamwise location, and is the
overall cooling effectiveness given by = (Tw - T∞)/(Ts - T∞). Thus, if heat load ratio
is less than one, then film cooling provides a reduction in heat load as compared
to the no film cooling case and vice versa. The value of is usually 0.5-0.6 for gas
turbines. Using = 0.6, a comparison of overall performance due to film cooling is
made for the cases of both circular hole and slot injections.

Effect of blowing ratio on spanwise averaged heat load ratio, for hole and slot
injections and the no grid case, is shown in Fig. 7. For both cases of hole and slot
injections, the blowing ratio of B = 0.8 shows the best performance and B = 1.2 shows
the worst performance. This is due to an optimum film coverage for the intermediate
blowing ratio of B = 0.8. For both cases of hole and slot injection, except for the
highest blowing ratio of B = 1.2, film cooling is seen to reduce heat load ratio over
the entire test surface. In general, 3-ℓ slot shows better performance than 4-d hole.
FIGURE 7. Effect of blowing ratio on spanwise averaged heat load ratio for 4-d hole
and 3-ℓ slot injections for no grid (Tu = 0.75%) and passive grid (Tu = 9.67%) cases.

Fig. 7 also shows the effect of blowing ratio on spanwise averaged heat load ratio
for hole and slot injections and the passive grid case. It is observed that higher
mainstream turbulence causes a considerable reduction in heat load ratio at the
flow reattachment point for all blowing ratios. The heat load ratios for the passive
grid case are less than 1.0 over most of the test surface. Thus, it appears that the
mainstream turbulence breaks the penetration of secondary flow for the highest
blowing ratio of B = 1.2. For both cases of hole and slot injection, the intermediate
blowing ratio of B = 0.8 still provides the lowest heat load ratio. However, the heat
load ratio for the hole injection is slightly lower than that for the slot injection.

The liquid crystals technique, as in Hippensteele et al., 1981, was used for thermal
visualization. For all test cases, the secondary air temperature was maintained at
57°C. The results for 3-ℓ slot geometry are shown in Fig. 8. For the low turbulence
case, the blowing ratios of B = 0.4 and 0.8 provide more film coverage than B =
1.2. For the high turbulence case, the blowing ratio of B = 0.8 provides the best
film coverage. An increase in mainstream turbulence is seen to cause a reduction
in film coverage. This effect is most severe for the lowest blowing ratio of B = 0.4
and least severe for the highest blowing ratio of B = 1.2. These results are reasonably
consistent with data presented earlier. The 4-d hole geometry results were similar
except the coverage was more in the streamwise direction. The results are not shown
due to space constraints.
FIGURE 8. Comparison of film thermal visualizations for 3-ℓ slots at blowing ratios
of B = 0.4, 0.8, and 1.2 for no grid (Tu = 0.75%) and passive grid (Tu = 9.67%) cases.

> Read full chapter

Turbulent Structure in the Three-Di-


mensional Boundary Layer on a Swept
Wing
Motoyuki Itoh, Minoru Kobayashi, in Engineering Turbulence Modelling and Exper-
iments 4, 1999

3.3 Turbulence field


Figure 8 shows the turbulence intensities and . In the outer region (y + > 200), both
and increase with going downstream. This may be ascribable to the effect of adverse
pressure gradient (Nagano et al. [11]). It should be remarked that, in the near-wall
region, decreases appreciably with going downstream from x = 396 mm to x = 540-
 mm. This seems to be due to the effects of increased three-dimensionality. On the
other hand, increases even in the near-wall region with going downstream from
x = 396 mm to x = 540 mm, as seen in Fig. 8(b). These results are consistent with the
numerical prediction (Moin et al. [12]) studying the effects of three-dimensionality
on the turbulence statistics.

Figure 8. Distributions of turbulence intensities

The wall-normal turbulence intensity is shown in Fig. 9. As we used an X-wire


probe to obtain -values, the data in the near-wall region could not be obtained. The
increase of with going downstream seems to be due to the dominant effect of the
adverse pressure gradient [11].

Figure 9. Wall-normal component of turbulence intensity

Figure 10 shows the turbulent shear stress , where u* is the turbulent velocity
component in the local mean flow direction. It should be noted that, in the region
y/ 99 < 0.4, the value of decreases with going downstream from x = 396 mm to
x = 540 mm. This seems to be caused by the effect of increased three-dimensionality.
Figure 10. Turbulent shear stress

Although all the components of Reynolds shear stress were obtained by the tech-
nique of a rotated hot-wire, they will not be disussed here any more because the
main concern in the present work is to clarify the causes for decreasing structure
parameter, A1, with increasing three-dimensionality. The Reynolds stresses data
obtained in the present work are available in machine-readable form for use by
interested workers.

The distributions of the turbulent kinetic energy, , are shown in Fig. 11. The values
of near the wall decrease with going downstream from x = 396 mm to x = 540 mm,
which also seems to be due to the effect of increased three-dimensionality.

Figure 11. Distributions of turbulent kinetic energy

Figure 12 shows the distributions of structure parameter A1 (ratio of shear-stress


magnitude in the plane parallel to the wall to twice the turbulent kinetic energy). As
has been well known, the value of A1 decreases with increasing three-dimensionality.
In Fig. 12, A1 decreases from about 0.15 (at x = -108 mm) to 0.11 (at x = 540 mm) in
the central region of the boundary layer.

Figure 12. Distributions of structure parameter A1


Figures 13 (a) and (b) show the turbulent transport of turbulence energy component
and the turbulent shear stress . It is seen in Fig. 13(a) that with going downstream
the values of decrease near the wall and become negative at the stations x = 396 mm
and X = 540 mm. The negativevalues of indicate that the turbulence energy is
transported toward the wall. It will be known from Fig. 13(b) that the turbulent
transport of is also toward the wall in the same region as for . The wallward turbulent
transport has also be found in the turbulent boundary layers with adverse pressure
gradient ([11][13])

Figure 13. Distributions of turbulent transport

> Read full chapter

THE EFFECT OF WALL ROUGHNESS


ON AN OPEN CHANNEL BOUNDARY
LAYER
M.F. Tachie, ... R. Balachandar, in Engineering Turbulence Modelling and Experi-
ments 5, 2002

Turbulence Quantities
The distributions of the streamwise turbulence intensity on the three different
surfaces are shown in Figure 2a and 2b, using the friction velocity U and the outer
scaling (Ue) advocated by [GC], respectively. The use of inner scaling (Figure 2a)
shows that wall roughness reduces the turbulence level in the immediate vicinity of
the wall (i.e. y+ ≤ 30). Outside this region, the data obtained on the wire screen (WM)
are significantly larger than those of the smooth surface (SM). The deviation of SG
from SM in the region y+ > 30 is only minimal, i.e. within measurement uncertainty.
Figure 2b demonstrates that the sand grain data (SG) deviate from the smooth data
over the inner half of the boundary layer, but the deviation of the wire mesh data
(WM) from the smooth wall profile persist up to y/ ≈1. The differences are most
pronounced in the immediate vicinity of the wall where the peak values also occur.
Although the average roughness height of WM (i.e. the wire diameter) is only 50
percent of the nominal diameter of the sand grains (SG), WM shows higher level of
turbulence than SG over a significant portion of the flow. This is a clear evidence that
the extent to which surface roughness influence the turbulence structure depends
on the specific geometry of the roughness elements.

Figure 2a. Streamwise turbulence intensity in inner coordinates

Figure 2b. Streamwise turbulence intensity in outer coordinates

Figure 3a and 3b, respectively, show the wall-normal turbulence intensity for the
smooth and rough surfaces using U and Ue, which is the correct scaling according
to the theory of [GC]. Due to spatial resolution limitations, two-component data (e.g.
v and <uv>) could not be obtained in the region y ≤ 30 or y/ ≤ 0.02. Irrespective of the
scaling used, the figures suggest that surface roughness enhances the level of the
wall-normal turbulence intensity. It is also noted that the wall-normal component
of the velocity fluctuation is even more sensitive to the specific wall condition than
the streamwise component.
Figure 3a. Wall-normal turbulence intensity in inner coordinates

Figure 3b. Wall-normal turbulence intensity in outer coordinates

Figure 4 shows plots of the Reynolds shear stress on the three different surfaces
using both inner and outer scaling. Again, clear differences are observed in the three
profiles, irrespective of the scaling, with the peak value increasing markedly with
increasing roughness.

Figure 4a. Distribution of Reynolds shear stress in inner coordinates

We also examine the effect of surface roughness on the triple correlations. For
brevity, only the components, <v3> and <u2v> which represent turbulent transport
of <v2> and <u2>, respectively, in the wall-normal direction are shown in this paper.
These components are also the major contributors to the turbulent diffusion term in
the energy budget. The sum of <v3> and <u2v>, i.e. (<v3> + <u2v>) is shown in Figure
5. Following the analysis of [GC], the parameter U 2Ue is used to normalise the data.
It is apparent from Figure 5 that surface roughness increases the triple correlation
in the inner half of the boundary layer. The location of the outer (and higher) peaks
is also strongly influenced by surface roughness.

Figure 5. Roughness effects on triple correlation

> Read full chapter

ScienceDirect is Elsevier’s leading information solution for researchers.


Copyright © 2018 Elsevier B.V. or its licensors or contributors. ScienceDirect ® is a registered trademark of Elsevier B.V. Terms and conditions apply.

You might also like