Turbine Blade HT

Turbine Blade Heat Transfer 69
CHAPTER 6.0 TURBINE BLADE HEAT TRANSFER

In this chapter, unsteady heat transfer will be investigated from the solutions of the
quasi 2-D Navier-Stokes equations. Experimental data was obtained from the Virginia
Tech Transonic Wind Tunnel representing engine similar conditions. The temperature
conditions were actually less than those experienced in an actual turbine stage, but the
remaining flow physics were simulated in the wind tunnel experiment. The experimental
pressures and temperatures measured on the turbine blade surface were used to establish
the initial conditions for the numerical approach. All of the resulting calculated data was
used as input to a first order time response numerical model in an attempt to model the
finite time response of the heat flux gage. The resulting data was compared to the
experimental data. Similarities observed from this comparison justified the removal of the
gage time response effects from the calculated heat flux data. The blade surface heat
transfer effects during the shock/boundary layer interaction are then observed in the
absence of gage effects.
6.1 The Wind Tunnel Experiment
In this section, the wind tunnel experiment is briefly discussed. These conditions
were obtained from a previously performed experiment in the Virginia Tech Transonic
Blowdown Wind Tunnel facility. This facility will be briefly described along with the test
section and the blade location where heat transfer data is obtained.
Fig 6.1 The Virginia Tech Transonic Blowdown Wind Tunnel Facility
6.1.1 The Virginia Tech Transonic Blowdown Wind Tunnel
Unsteady heat transfer measurements were obtained from the Virginia Tech
Transonic Blowdown Wind Tunnel Facility, show in Figure 6.1. The air supply for this
tunnel is provided by a four stage reciprocating centrifugal compressor and stored in
tanks. Prior to a heated tunnel run, a fan circulates air across a 36kW heater and through a
13 inch I.D. pipe, containing 300lb of copper flow straighteners, referred to as the heat
exchanger. When a predetermined temperature on the surface of the heat exchanger is
reached, the heater and the fan are turned off. The high-pressure air from the storage tanks
becomes heated as it moves through the heat exchanger, and toward the wind tunnel test
section.
6.1.2 The Test Section
The test section is shown in Figure 6.2. Heated air moves towards the test section
containing a highly loaded cascade of 11 aluminum turbine blades. These turbine blades
are supported with two Plexiglas endwalls and two aluminum endblocks. Aluminum doors
hold the endwalls in place. The Plexiglas windows enable four blades to be visible. During
the beginning of a tunnel run, a shock tube is used to generate a shock, which propagates
into pressurized tubing at its endcap. This tubing is connected to a shock shaper (a
diverging channel) which is connected to the wind tunnel. This allows a shock of
predetermined pressure ratio to propagate into the test section, in a direction parallel to
the leading edge plane of the turbine blades. The inlet and exit Mach numbers to this
cascade 0.36 and 1.2, respectively. The turbine cascade turns the flow 120 and the
trailing edge Reynolds number is 1.1million. The Instrumented Blade Window can be seen
in Figure 6.3. This shows the blades that are visible from outside of the wind tunnel test
section. This window is magnified in Figure 6.3 to show the location of the
Instrumented
Blade
Window
Fig 6.2 The Virginia Tech Cascade Wind Tunnel Test Section
Fig 6.3 The Instrumented Blade Window
instrumented blades. The instrumented blade passage consists of five gage positions, three
on the suction side and two on the pressure side. Each of the five locations is equipped
with a Kulite pressure transducer and a Vatell heat flux microsensor. As stated in Chapter
2, the pressure and heat flux data showed simultaneous increases, suggesting that the
increases in heat transfer occurred during shock impact.
In order to test the approach discussed in this document, attention is given to gage
position 1. At the instant prior to shock impact, the pressure and temperature on the blade
surface is 13.5psig and ~300K. During an unheated tunnel run, the temperature of the
freestream is ~300K. During the heated tunnel run, the temperature of the freestream is
330K. From these conditions, the freestream density profiles can be calculated from the
ideal gas law. In addition, the freestream velocity (of 197m/s) can be calculated from the
isentropic relations. The wind tunnel data has provided all of the boundary layer edge flow
conditions needed to begin the calculations.
6.1.3 The Initial Conditions
The measured steady wind tunnel data are used as the initial conditions for the
numerical approach in this Chapter. The quasi 2-D Navier-Stokes equations with heat
transfer and viscous terms are the governing equations. The initial conditions are
comprised of the freestream conditions ( ) upstream of the shock, and the shock tube
relations applied downstream of the shock (induced). This is done in an effort to capture
the discontinuous change of the flow variables, which is modeled as a step function. The
freestream conditions represent the state of the flow just outside of the boundary layer,
measured in the wind tunnel experiment of Nix. Referring to Figure 1.4, the core (a.k.a.
freestream) flow region is between the boundary layers edge and the initial position of the
shock wave. The induced conditions represent the estimated state of the shock-altered
core flow conditions. Again referring to Figure 1.4, this would be the region above the
shock wave. The changes in the flow conditions across the shock are represented
numerically as a step function, which are calculated by the shock tube relations. These
relations are used to describe the flow variable magnitudes on each side of the shock
(initially). The conditions discussed here do not apply inside the boundary layer. The
freestream conditions are
n 0
=1.06985
kg
m
3
u
n
0
=197
m
s
v
n
0
0
P
n
0
= 0.9atm
T
n
0
=330K.
The pressure and temperature (0.9atm and 330K) listed above are obtained from
the Kulite pressure transducer and the HFM during the heated wind tunnel experiment.
They are used along with the ideal gas law to obtain the density of the freestream
(between the boundary layer edge and the shock wave). The isentropic relations are used
to determine the velocity (u) in the freestream. The velocity perpendicular to the blade
surface (v) at gage location 1 is 0. The above mentioned parameters apply to the initial
time step (n=0) for the region between the boundary layer edge and the shock wave.
Behind the shock wave, the flow conditions of the shock are obtained from the
shock tube relations. These relations are all dependent on the strength of the shock and the
ratios of specific heats (), and enable the magnitudes of the induced flow conditions to be
calculated. The strength of the shock is defined as the pressure ratio
P
P
= User defined.
This becomes the input for the relations defining the induced flow properties, which are
shown below. These induced flow properties are defined by

induced
n n
P
P
P
P
+
+
_
,
1
]
1
1
1
1
1
0 0
1
1
1
1
1
[6.1]
u
a P
P
P
P
induced
n
n

_
,
+
+

+
_
,
0
0
1
2
1
2
1
1
1
[6.2]
v
induced
n
0
0 [6.3]
T T
P
P
P
P
P
P
induced
n n
_
,
+
+
+
_
,
0 0
1
1
1
1
1
[6.4]
It should be noted that a RT
n n
0 0
is the speed of sound, as defined for an ideal gas.
It is also a parameter used for nondimensionalizing the governing equations (see Appendix
A). With these relations mentioned, the initial flowfield properties can be established
everywhere except inside the boundary layer. The initial flowfield parameters defining the
boundary layer will be described in the next section.
6.2 The Boundary Layer Profiles
Referring back to Figures 4.1 and 4.2, the numbering for the cell-centered node
points at the blade surface and at the far field boundary were illustrated. As stated in the
literature, ~10 grid points has been used in the past to spatially descritize the boundary
layer region. In this research, approximately 60 node points are used to occupy the region
known as the velocity boundary layer, shown in Figure 1.4. This increase not only
provides more spatial resolution in the boundary layer, but also defines a smaller y,
which reduces the magnitude of the truncation error. A total of 9000 node points are used,
and the region outside of the boundary layer contains 8940 node points. This section
discusses the initial conditions of the 60 node points defining the boundary layer.
The estimated velocity boundary layer thickness was obtained from previous
steady-state viscous calculations, using the boundary layer code BL2D. This estimated
velocity boundary layer thickness was ~0.10mm. Kays and Crawford
33
defined the ratio of
boundary layer thickness as

1
1026
1
3
. Pr
[6.5]
where the velocity boundary layer thickness () and the Prandtl number are known. Using
this relation, the thermal boundary layer thickness is calculated to be ~0.11mm. These
initial estimates provide a starting point for the boundary layer profiles.
The blade surface velocity (0 due to no-slip conditions) and temperature
(measured with HFM to be 300K) are held constant throughout the calculations. Knowing
the velocity and temperature at the wall and at the boundary layer edge, a third-order
parabola is used to quantify the velocity and temperature profiles. The third-order
parabola for both the velocity and thermal boundary layer profiles have the algebraic form
of
u
u
y y
_
,
3
2
1
2
3

[6.6]
T T
T T
y y
surface
surface
_
,
3
2
1
2
3

[6.7]
The parameter y corresponds to the spatial distance away from the wall. The previously
mentioned estimates of velocity boundary thickness () and the thermal boundary layer
thickness () are used to produce the velocity and temperature profiles near the blade
surface. Pressure and temperature are used along with the ideal gas law to produce the
boundary layer density profile. These initial conditions were applied to both an unheated
flow, and a heated flow.
For the unheated flow, the temperature, pressure, and density profiles were linear,
equaling the freestream value. The induced transverse velocity is zero and the streamwise
velocity profile is a third-order parabola, defined by Equation 6.6.
For the heated flow, pressure was linear, equaling the freestream value. The
induced velocity is zero. The streamwise velocity and temperature profiles were defined by
Equations 6.6 and 6.7. The temperature and pressure was then used (along with the ideal
gas law) to define the density profile.
The above-mentioned boundary layer profiles were applied to a shock-free
flowfield. The resulting calculated surface heat transfer was plotted for both unheated and
heated flows, and is shown in Figure 6.4. The purpose of this step was to obtain the most
accurate initial conditions for analyzing shock induced heat transfer. The third-order
parabolas used to define the velocity and temperature profiles are merely estimates. Once
these profiles are used to begin the calculations, the final profiles are determined from the
governing equations.
Figure 6.4 shows the results of the shock-free unheated and heated calculations. In
the unheated flow, the heat transfer shown is a result of viscous heating. At the start of the
calculations, the heat transfer at the surface slowly increases until reaching a maximum,
where it remains constant. This occurs at ~40 microseconds. This suggests that a non-zero
temperature gradient develops at the blade surface.
In the heated flow, the heat transfer seen is driven by the temperature difference
between the freestream (330K) and the surface (300K), and viscous heating. At the start
of the calculations, there is a drop in heat transfer during the first microsecond. This
suggests that the surface temperature gradient obtained from the solution of the governing
equations is substantially less than that defined by a third-order parabola. The solutions to
the governing equations quickly adjust the thermal boundary layer to reflect a more
accurate profile. After one microsecond, the heat transfer starts to increase again, due to
viscous heating. After 40 microseconds, the heat transfer level due to viscous heating has
become constant.
Figure 6.4 Surface Heating due to Viscous Heating in a Shock-Free Flowfield
(both unheated and heated)
The boundary layer profiles that exists after the viscous heating has become
constant will serve as the initial conditions when analyzing shock induced heat transfer.
The purpose of using these initial conditions is to isolate the viscous heating from the
shock-induced heat transfer calculations. This should eliminate the growing effect of
viscous heating during the shock/boundary layer interaction.
Another way to address this problem would be to make the computational window
very long and place the shock initial position a significant distance away from the wall. As
the shock approaches the boundary layer, enough time would pass (~40 microseconds) so
that the viscous heating would become constant before the shock reaches the boundary
layer. The problem with this is that long computational windows would result in long run
times and large files (due to the stability constraint, which requires small advances in
time). This approach was less desirable.
Performing calculations on a shock-free flowfield is ideal since the calculations are
performed quickly and the result is boundary layer profiles, which are computationally
fully-developed. The resulting boundary layer profiles after 40 microseconds of unheated
and heated shock-free calculations can be seen in Figures 6.5 and 6.6, respectively.
In Figure 6.5 (unheated), changes occur in the temperature and density profiles.
The temperature profile is no longer constant. While the surface and freestream
temperature remains 300K, the temperature inside the boundary layer actually increases by
2-3 K, due to the viscous dissipation of mechanical energy. The temperature and pressure
profiles are then used to produce a density profile, which is also no longer constant.
In Figure 6.6 (heated), changes occur in the temperature and density profiles. The
temperature profile, which started off as a third-order parabola, does not contain the same
high gradients. The thermal boundary layer thickness has increased to ~0.12mm. In
addition, the density distribution has changed since it is temperature dependent.
Figure 6.5 Shock-Free Unheated Boundary Layer Profiles after 40 sec
Figure 6.6 Shock-Free Heated Boundary Layer Profiles after 40 sec
In both cases, all other properties (pressure, and streamwise and transverse velocities) did
not experience any notable changes.
6.2.1 The Initial Conditions for Shock-Induced Heat Transfer Calculations
Figures 6.5 and 6.6 only shows a 0.12mm window adjacent to the blade surface.
This window will provide details of the shock/boundary layer interaction, which will be
discussed in future sections.
The pressure in the 0.12mm region adjacent to the surface has a value of P/P =1.
This indicates that the pressure throughout this region has the magnitude of the
freestream. This value is constant all the way to the shock front. At the shock front, the
pressure increases by a factor of 1.1, which is the pressure ratio across the shock.
Outside of the thermal boundary layer, the temperature is a constant of 300K (for
unheated flow) and 330K (for heated flow), all the way to the shock front. At the shock
front, the temperature increases by a factor of 1.026, which is the temperature ratio across
a shock of this strength.
The density in the freestream to the shock front is 1.1614kg/m
3
(for unheated
flow) and 1.07kg/m
3
(for heated flow), all the way to the shock front. At the shock front,
the density increases by a factor of 1.067, which is the density ratio across a shock of this
strength.
The streamwise velocity is constant outside of the boundary layer and throughout
the flowfield (Mach Number equals 0.52). The transverse y-velocity is zero from the blade
surface to the shock front. At the shock front, the y-velocity increases by a constant of
0.07, which is in the direction toward the blade surface.
The initial conditions in the boundary layer for the cases that were calculated are:
Stagnant Gas - No velocity (zero streamwise velocity) or thermal boundary
(zero temperature gradient),
Unheated Flow - Velocity boundary layer and small thermal boundary layer
(both obtained after 40 microseconds of shock-free calculations in an unheated
flow. See Figure 6.5.),
Heated Flow - Velocity and thermal boundary layer (both obtained after 40
microseconds of shock-free calculations in a heated flow. See Figure 6.6).
6.3 The Experimental Data
The experimental data used for analysis comes from unheated and heated tunnel
runs. Samples of these tunnel runs are shown in Figure 6.7. The time of shock arrival
occurs at approximately 0 seconds. Prior to this time are the heat flux levels at the blade
surface just before shock impact. The jump in heat transfer and its subsequent decay
occurs due to the impact and the propagation of the reflected shock away from the blade
surface, respectively. The time response effects of the gage will be simulated and applied
to the numerical solutions of the quasi 2-D Navier-Stokes equations. The resulting heat
flux curves will be compared to a 40 second time window (of the experimental data)
occurring immediately after the shock impact the blade surface.
6.4 Solutions to the Quasi 2-D Navier-Stokes Equations
The heat flux histories produced by the solutions to the governing equations are
shown together in Figure 6.8. The solutions shown are for stagnant air (no velocity or
thermal boundary layers), unheated (velocity boundary layer only) and heated (both
velocity and thermal boundary layers) flows. The stagnant air adjacent to the blade surface
has the lowest instantaneous heat transfer levels, while the heated flow has the highest.
The trends are similar, but the instantaneous levels prior to shock impact are different. In
addition, the instantaneous levels after shock impact are different.
The calculated solution for stagnant air is shown in Figure 6.8. The stagnant air
does not have any boundary layer (velocity or thermal). The temperature difference
Figure 6.7 Unheated and Heated Experimental Data at Gage Position Number 1
Figure 6.8 Stagnant, Unheated and Heated Numerical Data Corresponding to Gage
Position Number 1
between the blade and the air is 0 degrees C. As a result, heat transfer does not occur at
the blade surface prior to shock impact. The heat transfer seen at time 0 is strictly due to
shock heating. Since there are no boundary layers, there is no mechanism present to serve
as an insulating layer. This is not surprising for the stagnant flow since it is unheated.
The calculated solution for unheated flow is also shown in Figure 6.8. The
unheated flow has a velocity boundary layer, but a minimal thermal boundary layer. The
temperature difference between the blade and the freestream flow is 0 degrees C. At the
time of impact, low levels of heat transfer (~3,100W/m
2
) are produced prior to shock
impact. Since there is only a 2-3K temperature difference between the blade surface and
the freestream flow before the shock arrives, the heat transfer contributions must be due to
viscous heating in the velocity boundary layer. Viscous heating occurs from the interaction
of the adjacent layers of gas, which are moving at different speeds. The heat transfer seen
at time 0 is due to a combination of viscous heating and shock heating. After shock
impact, the velocity boundary layer is unchanged. However, an induced thermal boundary
layer profile has been added by the shock impact. Therefore, viscous and shock heating
are the only mechanisms significantly contributing to surface heat transfer after shock
impact.
Finally, the calculated solution for heated flow is shown in Figure 6.6. The heated
flow has a velocity and a thermal boundary layer. The initial temperature difference
between the blade and the freestream flow is 30 degrees C. Higher levels of heat transfer
(~13,500W/m
2
) are produced prior to shock impact. The elevated temperature of the
thermal boundary layer contains layers of gas adjacent to the blade surface that have a
higher thermal conductivity, resulting in a higher heat transfer level. In addition, viscous
heating is still occurring in the velocity boundary layer, prior to shock impact. Therefore,
the pre-shock heat transfer contributions come from viscous heating in the velocity
boundary layer and the elevated temperature of the thermal boundary layer. The heat
transfer seen at time 0 is due to a combination of the elevated temperature of the thermal
boundary layer, viscous heating and shock heating.
An interesting feature appears from the data in Figure 6.8. If the heat transfer level
prior to shock impact is taken off, then all of the curves collapse identically. This suggests
that the effects of shock wave impact are identical in stagnant air, as with unheated and
heated flows. The main differences in these heat transfer histories are due to the presence
of the boundary layer and the temperature difference that exists between the freestream
flow and the blade surface. Having observed this, an estimation of shock-induced heat
transfer along a blade surface can be assembled by first using the shock-induced heat
transfer in stagnant air. If the flow characteristics along the blade surface are known, then
the mean heat transfer level prior to shock impact can be determined. Their combination
results in an estimate of shock-induced heat transfer in an unheated flowfield. Performing
the same procedure and adding the heat transfer level due to a temperature difference,
results in an estimate of shock-induced heat transfer in a heated flowfield.
The flow physics in an unsteady flow is an interesting case since it contains similar
heat transfer mechanisms that would exist in an actual turbine stage. Typically, multiple
shocks would strike the turbine blade surface. In this analysis, only the single
shock/boundary layer interaction will be examined. In the next section, the results of a
computational animation generated by this approach will be presented. The details of the
changes in the flow variables (, u, v, P, and T) during the shock/boundary layer
interaction will be described. These changes contribute to the heat flux levels experienced
at the surface during shock impact.
6.4.1 Time Estimate of the Shock/Boundary Layer Interaction
The time when the shock wave enters and exits the boundary layer was estimated
prior to obtaining heat flux data at the surface. The time for shock impact was also
estimated prior to solving the governing equations. In Figure 6.9, the temperature
differences between the blade surface and the edge of the boundary layer are referenced
based on the time during shock impact with the blade surface.
As shown in Figure 6.9, the shock enters the boundary layer approximately 0.3
seconds before the shock impacts the blade surface. This is indicated by the first jump in
temperature, which represents the incident shock passing the edge of the boundary layer.
Recall that there is a temperature ratio of 1.026 across the shock, so the temperature
behind the shock will be a factor of 1.026 higher than the temperature in front of the
shock.
The shock impacts the blade surface and a reflected shock propagates away from
the blade surface. The reflected shock exits the boundary layer approximately 0.3
seconds after the shock impacts the blade surface. This is noted by the second jump in
temperature, which is a result of the temperature at the boundary layer edge increasing by
a factor of 1.026.
In Figure 6.9, the symbols along the time axis represent the instances in times that
make up the animation, which will be shown in later sections. Time histories for pressure
(P), temperature (T), density () and the x- and y-velocities (u and v respectively) will be
shown at roughly 0.2, and 0.1 seconds before shock impact, during shock impact, and
0.1, 0.2, and 0.3 seconds after shock impact. Shock impact was been determined by the
jump in heat flux as seen in Figure 6.10. The absolute time of impact was subtracted from
the data in this figure so that shock impact would occur at time zero. This is done just for
ease of presentation. This figure also contains the symbols corresponding to the instant in
time for which flow variable time histories will be shown. These symbols define the time
window when the shock/boundary layer interaction occurs. This is when the shock enters
the boundary layer, impacts the blade surface and the exits the boundary layer in the form
of a reflection. Figure 6.10 shows that there is not a dramatic effect in blade surface heat
Fig 6.9 Temperature Difference between the Boundary Layer Edge
and the Blade Surface
Figure 6.10 Blade Surface Heat Transfer Immediately Before and After Shock Impact
transfer when the shock enters the boundary. The immediate effect is visible only during
and after the shock impacts the blade surface.
6.4.2 Animation of the Shock/Boundary Layer Interaction
In the previous section, an estimate of when the shock is in the boundary layer was
determined. This time range was estimated to be 0.3 seconds before and after shock
impact. The numerical approach allows the shock to be frozen at various moments in
time so that the flow variables can be observed at a specific instant. Figures 6.11-6.16
show the state of the flow variables at various instances in time while the shock is inside
the boundary. Figure 6.17 shows the state of the flow variables after the reflected shock
has propagated away from the blade surface for approximately 10seconds. These
instantaneous moments in time help create a picture regarding the response of the
boundary layer. This picture is difficult to obtain experimentally due to the short duration
of the shock/boundary layer interaction.
The following figures that make up the animation utilizes a similar format as the
initial conditions, which were shown in Figure 6.4. The figures are basically a 0.12mm
window adjacent to the blade surface. The blade surface is located at y=0 and the distance
away from the blade surface is characterized by positive values of y. Each graph in Figures
6.11-6.17 are spatial distributions of pressure (P), temperature (T), density (), x-velocity
(u), and y-velocity (v), at a specific moment in time.
Figure 6.11 corresponds to the shock location 0.2 seconds before impacting the
blade surface. The velocity boundary layer occupies the space of 0 < y 0.1 mm, while the
thermal boundary layer occupies the space of 0 < y 0.11 mm. The pressure distribution
in the boundary layer is constant before the shock arrives. In Figure 6.11, the discontinuity
(representing the shock) can be seen at y 0.07 mm. The pressure is shown strictly as a
step function. The ratio across the step function is consistent with the shock strength. As
the shock moves toward y = 0 (the blade surface), the shock has the effect of increasing
the absolute pressure of every location it passes by a factor of 1.1.
The temperature profile before the shock is a third order parabola (as discussed
earlier). Before the shock enters the boundary layer, the temperature ranges from 300K (at
the blade surface) to 330K (in the freestream). The air temperature goes down while
approaching the blade surface. Once the shock enters the boundary layer, it raises the
absolute temperature in the boundary layer by a factor of 1.026 everywhere except at the
blade surface (y=0), where the temperature is held constant throughout the entire event.
This increase in temperature can be seen in Figure 6.11 at y 0.07 mm.
The density is related to the pressure (P) and the temperature (T) by the ideal gas
law
P=RT
or
=
P
RT
Prior to the shock entering the boundary layer, the pressure is constant and the
temperature decreases while moving toward the wall. This suggests that the density of the
constant pressure gas will increase while approaching the blade surface. This is due to the
decreasing temperature while approaching the blade surface. When the shock is at y
0.07 mm, the density behind the shock increases by a factor of 1.067. This is the density
ratio across a shock of this magnitude.
The x-velocity shows no visible change in the presence of the shock. This is
expected since the direction of the x-velocity (crossflow) is tangent with respect to the
shock motion. Normal shock theory indicates that there is no change in the tangential
velocity (crossflow) across a normal shock. The y-velocity is zero before the shock
appears in the flowfield. However, as the shock enters the boundary layer, it produces a
low velocity airflow in the direction of the shock motion. This low velocity flow is ~25
m/sec, which corresponds to a Mach number < 0.10. All of the above mentioned effects
produced behind the shock are called induced properties, because they are the result of the
shock wave interaction with the flow parameters.
The oscillation seen behind the shock wave is a numerical instability resulting from
the solution to the governing equations across a discontinuity. This is expected and is
similar in behavior to the Gibbs phenomena
34
. Attempts were made to reduce the
presence of the Gibbs phenomena during test problem number 2. These attempts were not
successful, unless excessive damping was used. The use of excessive damping will affect
the accuracy of the temperature profiles at the surface. Therefore, for the purpose of this
analysis, the oscillations behind the shock wave were tolerated.
Figure 6.12 corresponds to the shock location ~0.1 seconds before impact with
the blade surface. At this time, the shock is located at y 0.04 mm. The pressure shows
the step function has moved closer to y=0. Also the air pressure in the range 0.04 y
0.07 mm has increased due to the induced pressure resulting from the shock wave. The
temperature in this spatial range has also increased. The temperature still increases by a
factor of 1.026 (the temperature ratio across the shock), but the discontinuity is not as
large since the absolute magnitude of the temperature prior to the shock is decreasing.
Since this location is closer to the blade surface, the velocity gradients are higher. Higher
velocity gradients mean higher viscous stresses, which may play a role in the reduction of
Gibbs phenomena in the temperature profile.
Figure 6.13 is a numerical approximation of the shock impact. The shock has hit
the blade surface. The pressure is approximately equal to the shock strength. The
reflection has not yet developed, so the remainder of the boundary layer is at the
magnitude of the induced pressure. The temperature has experienced an increase due to
the shock-induced temperature behind the shock. Since the shock is at the blade surface,
the entire boundary layer temperature profile has been increased by the temperature ratio
across the shock (which is 1.026), except at the blade surface. This is due to the constant
temperature boundary condition, which was imposed at the blade surface. The density
profile has also experienced a uniform increase by a factor of 1.067, which is the density
ratio across the shock. The x-velocity does not experience a major change, but when
Figure 6.11 ~0.2 sec before shock impact: (a) pressure, (b) temperature, (c) density,
(d) x-velocity, and (e) y-velocity
Figure 6.12 ~0.1 sec before shock impact: (a) pressure, (b) temperature, (c) density,
examined closely, the boundary layer thickness has decreased. This is explained by the
increased pressure and density profiles once the shock has impacted the wall. The decrease
in boundary layer thickness is less than 1% and can not be seen in these figures. The y-
velocity shows that the induced velocity did not make it all the way to the blade surface
even though the shock did. If this induced velocity appeared at the blade surface, that
would suggest that the resultant flow direction is no longer parallel to the surface. Since
the y-velocity is in the direction towards the surface and the x-velocity is in the direction
tangent to the surface, then the resultant velocity vector at the surface would be in a
direction toward the surface. Since this is not a porous surface, this would not be
considered a realistic solution. As a result, the combination of the x- and y-velocities at the
wall produces a zero resultant velocity vector, which is consistent with the no-slip
condition. Gibbs phenomenon appears to be minimal since the shock is located at the
surface and is not moving at this time. Also, viscous stresses are high close to the surface
and may help reduce the effects of this phenomena due to the inherent viscosity of the
flow field.
Figure 6.14 shows that the reflection has developed and is moving away from the
blade surface. This figure corresponds to 0.1 seconds after shock impact. The reflected
shock is at y 0.03mm and has re-established the shock ratios (pressure, temperature, and
density) which are superimposed on the induced properties that existed prior to the
development of the reflected shock. Since the reflected shock has the same strength, it is
producing the same induced property characteristics as the incoming shock wave. This
means that the shock is propagating through an already induced flowfield. The absolute
pressure at y < 0.03mm is 1.21 while the absolute pressure at y > 0.03mm is 1.1,
maintaining a pressure ratio of 1.1 (which is the strength of the reflected shock). The
temperature exhibits similar behavior. The temperature ratio is still 1.026 across the shock
wave. The discontinuity in the temperature profile appears to grow, but the ratio is still the
same. The higher the absolute temperature, the larger the induced temperature
Figure 6.13 Shock Impact: (a) pressure, (b) temperature, (c) density,
magnitude. The temperature gradient is much higher once the reflection has been
produced because the temperature near the wall has been subjected to the induced
temperature twice. The region near the wall has increased by a factor of (temperature
ratio)
2
, while at the same time maintaining a constant temperature at the blade surface.
This contributes to the high temperature gradient, which results in a high magnitude of
heat transfer. At y 0.01mm, the temperature distribution appears linear in this thin region
adjacent to the boundary layer. It appears that as the reflected shock develops and begins
moving away from the blade surface, the temperature field is the last to transform to its
new state. At y 0.01mm, a change in curvature is observed in the temperature profile.
This is probably due to the fact that the rate of penetration of heat transfer does not occur
fast enough to keep up with the development of the shock-induced temperature profile.
The air away from the blade surface has heated up as a result of the inducing properties of
the shock. At the same time, the air very close to the wall is being cooled by the constant
temperature surface at y=0. The tendency of the air is to warm up, but the cool surface at
y=0 prohibits this. More than likely, this is what is seen by the change in curvature 0.01
mm away from the blade surface.
The steep temperature gradients very close to the blade surface also contribute to
the production of steep gradients in the density profile of Figure 6.14. The higher pressure
air is cooled at a fast rate. The ideal gas law suggests that this air should become denser at
a fast rate as the blade surface is approached. Again, the x-velocity does not appear to
undergo any major changes. However, the y-velocity produces an induced velocity that is
in the direction of the reflected shock motion. The incident and reflected shock strength is
the same. Therefore, the induced velocities produced from each are of the same magnitude
but in opposite directions. As the reflected shock moves away from the blade surface, the
shock effectively restores the y-velocity back to zero.
Figure 6.15 shows the location of the shock 0.2 seconds after impacting the blade
surface. The reflected shock is located at y 0.07mm. The pressure field directly adjacent
to the blade surface has settled to the elevated state. However, it is evident that
Figure 6.14 ~0.1 sec after shock impact: (a) pressure, (b) temperature, (c) density,
the shock is still inside the boundary layer. The pressure field is the quickest to reach its
elevated state near the blade surface. The temperature field still shows a change in
curvature close to the blade surface. However, this occurs at y 0.015mm, indicating that
as the change in curvature decreases, it moves farther away from the blade surface. The
change in curvature is due to the conflict of the heating of the twice-induced gas and the
cooling of the constant temperature blade. With increasing time, this change in curvature
becomes less distinct. The result is the decay of surface heat transfer as the reflected shock
propagates away from the blade.
The density shows the same change of curvature at y 0.015mm. As the blade
surface is approached, the temperature and density gradients increase. As the temperature
decreases quickly, the density increases quickly. The x-velocity does not display a visible
change and the y-velocity is restored to zero. The Gibbs phenomenon is attached to the
moving shock. This can be observed by looking at the location of the numerical
oscillations on the incident and reflected moving shocks. These oscillations move along
with the shock.
Figure 6.16 shows the shock location 0.3 seconds after impacting the blade
surface. The shock has just left the boundary layer region and is approximately 0.125 mm
away from the blade surface. The remaining effects of the Gibbs phenomena can be seen
although the shock is just outside of the viewing window. The pressure within the entire
boundary layer is at its elevated state. The temperature has increased uniformly from the
initial profile by a factor of 1.026
2
(temperature ratio squared). The blade temperature is
still 300K. The change in curvature of the temperature profile is still approximately
0.015mm away from the surface. This location appears to be the boundary between the
cooling effects of the blade and the heating effects of the induced temperature. When y
0.015 mm, the temperature profile remains a factor of the temperature ratio squared
higher than the initial temperature profile and is nonlinear. When y 0.015 mm, the slope
of the temperature profile decreases linearly with increasing time. With increasing time,
Figure 6.17 10.0 sec after shock impact: (a) pressure, (b) temperature, (c) density,
the slope of the temperature profile decays. The density profile remains a factor of the
density ratio squared higher than the initial density profile.
At y 0.015 mm, the change in curvature of the density begins to settle, producing
a smoother profile through this location. This suggests that as the energy transfers to the
surface (in the form of heat transfer), the change of the air density and temperature at y
0.015 mm begin to merge. The difference above and below this point equalizes, producing
a smooth profile. The x-velocity remains unchanged while the y-velocity has been restored
to zero. At y ~0.015mm, the change in curvature in the temperature and density profiles
suggests a distinct change in behavior. This change could be due to low velocity air
present when y < 0.015 mm, which may effect the rate of change of the air properties in
this region when compared to y > 0.015 mm.
Figure 6.17 represents the shock location 10 seconds after shock impact. At this
time, the shock is approximately 0.5 mm away from the blade surface. The pressure is
uniformly at an elevated state. The temperature profile resembles a smooth third order
parabola with a higher temperature difference between the surface and the boundary layer
edge. The change in curvature at y ~ 0.015 mm is no longer present. This is also true of
the density. Sufficient time has passed for the rate of change of air properties (specifically,
the temperature and density) in the low velocity sublayer (y 0.015 mm), to match the
rate of change of air properties outside of this sublayer (y 0.015 mm). The rate of
change of the new conditions inside of the boundary layer dictate the new mean heat
transfer level, which occurs well after the reflected shock has propagated away from the
blade surface.
6.4.3 Unsteady Surface Heat Transfer Verification
In the previous section, the shock/boundary layer interaction for a heated flow was
discussed. This animation was produced from the solutions of the governing equations. In
this section, the heat transfer occurring at the blade surface will be presented for both
unheated and heated flow cases. The numerically calculated heat transfer will be compared
to 50 seconds of the experimental data, shown in Figure 6.7.
The calculated heat transfer data are referred to as the raw calculated data. Unlike
the heat flux gage used in the experiment, these raw calculated data based on the
numerical simulation dont suffer from limited time response of the instrument. In the wind
tunnel measurements, a Vatell heat flux microsensor was used to measure the unsteady
heat transfer from a shock impacting a blade surface normally. Although this gage has a
frequency response of 100 kHz and a time response between 6-20 seconds, this is still
not fast enough to completely capture all of the details of the shock/boundary layer
interaction. This is due to the fact that the shock/boundary layer interaction occurs in such
a short span of time. In the previous section, the numerical animation suggests that the
shock enters and exits the boundary layer in less than 1 second. If the time response of
any instrumentation is greater than that, then some detail will be missed.
In previous work by Holmberg
35
, the gage was characterized as having a first order
time response. A first order time response model was satisfactorily used to simulate the
effects of a gage subject to a step change in heat transfer. The same time response model
will be used with the numerical data obtained from the solutions to the governing
equations. The model used is shown below
q q e
el raw
t
" "
mod

'
1

[6.8]
After this model is used, the results are compared to the experimental data. The data
resulting from Equation 6.8 will be referred to as the modeled data.
Figure 6.18 and 6.19 shows several different heat transfer traces. The raw
calculated data has the highest peak and is the result of the solutions to the governing
equations. The raw calculated data is used as input to the first order time response model
of Equation 6.8. This is done for three different values of time constant . The three values
(8, 12, and 16 seconds) were used since previous attempts to determine the heat
Figure 6.18 Unsteady Heat Transfer on Blade Surface (Unheated Flow): Raw calculated
data, modeled data using three time constants (8, 12, and 16 seconds), and
experimental data
Figure 6.19 Unsteady Heat Transfer on Blade Surface (Heated Flow): Raw calculated
data, modeled data using three time constants (8, 12, and 16 seconds), and
experimental data
flux gage time constant were unsuccessful. The results of this investigation
16
could only
suggest that the actual time constant was < 20 seconds. The symbols represent the
experimental measurements, which were previously shown in Figure 6.1.
In Figure 6.18 (unheated flow), the raw calculated data previously showed a low
mean heat transfer level, microseconds before shock impact. This level corresponded to
viscous heating in the velocity boundary layer. It was minimal (~3,100 W/m
2
), but
distinctly present. The time constant from the time-response model was varied, producing
a range of modeled heat flux histories. The experimental data is contained by the modeled
data. It is uncertain why the experimental data would suggests a varying time constant (by
comparison), but overall the comparison is considered qualitatively good.
In Figure 6.19 (heated flow), the raw calculated data previously also showed a
mean heat transfer level, microseconds before shock impact. Also, the mean heat transfer
level prior to shock impact is considerably higher (~13,500 W/m
2
). This is due to the 30
C temperature difference between the freestream and the blade surface in addition to the
viscous heating (~3,100 W/m
2
). The time constant from the time-response model was
again varied, producing a range of modeled heat flux histories. The heated modeled data
did not match the experimental data as well as the previous unheated case, but the
comparison is considered qualitatively fair.
In both of these figures, the calculated data exhibits similar behavior. However, the
rise time of the experimental data appears to be considerably slower. This could be due to
the limiting time response of the heat flux gage. Although this gage is a state-of-the-art
heat flux gage, the shock/boundary layer interaction occurs too fast to be measured
accurately with available instrumentation. Despite this, the modeled data in Figures 6.18
and 6.19 show reasonable qualitatively agreement with the experimental data.
Shock wave effects on unheated and heated boundary layer flow fields appear to
be similar to that seen in stagnant air. Elements such as viscous heating can be added to
the stagnant air shock-induced heat transfer history to obtain the shock-induced unheated
boundary layer heat transfer history. Similarly, viscous heating and temperature difference
(between blade and freestream) heat transfer can be added to the shock-induced stagnant
air heat transfer history to obtain the shock-induced heated boundary layer heat transfer.
The experimental heat flux histories in Figures 6.18 and 6.19 can be viewed as the
gages interpretation of the heat transfer phenomena during this unsteady event. The
experimental data is the result of the heat transfer phenomena and the gage limitations.
The first order response model has enabled the modeled data to simulate the effects of the
gages limiting time response, resulting in a good comparison between modeled and
experimental data.
To take advantage of the numerical solutions, the heat transfer phenomena can be
observed without the interference of gage limitations by simply looking at the raw
calculated data.
6.4.4 Unsteady Blade Heat Transfer Without Gage Effects
In the previous section, the modeled data compared reasonably well with the
experimental data. The first order time response model was a reasonable way to simulate
gage effects. This provides a certain level of confidence that the raw data actually
represents the heat transfer phenomena occurring during this unsteady event. In this
section, the total energy transfer while the shock is in the boundary layer will be observed.
The total energy transfer while the shock is in the boundary layer can be observed
in two parts. First, during the 0.3 seconds from the time the shock enters the boundary
layer to just before impact and second, and from shock impact to the remaining 0.3
seconds it takes for the reflected shock to exit the boundary layer. This is best estimated
by the following relations:
Q q dt
before

1
03
0 3
0
.
"
.
[6.9]
Q q dt
after
=

1
0 3
0
0 3
.
"
.
[6.10]
where Equation 6.9 is applied just prior to shock impact and Equations 6.10 is applied just
after shock. Both Equations 6.9 and 6.10 represent the time-averaged heat transfer rate at
the blade surface for a 0.3second period before and after shock impact. The results of
these equations can be seen in Table 6.1.
Table 6.1 Time-averaged Heat Transfer (0.3 second period)
Q
before
Eqn (6.9)
Q
after
Eqn (6.10)
Q
after
- Q
before
Q Q
Q
after before
before
Stagnant Flow 0 131,150 131,150 N/A

Unheated Flow 3,100 134,250 131,150 42
Heated Flow 13,500 144,700 131,200 10
Table 6.1 shows the results of integrating the instantaneous heat transfer (in Watts
per meters squared) during a time range while the shock is inside of the boundary layer.
The subscript before refers to the time range when the incident shock has just entered the
boundary layer, and up to the instant before shock impact. The subscript after refers to the
time range when shock impact occurs, and until the reflected shock has propagated to the
boundary layer edge. The ratio of the jump in energy transfer changes for the different
flow conditions, but the difference does not change by more than 0.1 %. While the surface
energy transfer is a function of shock strength, the increased temperature of the boundary
layer increases its thermal conductivity. This controls how much heat transfer occurs at
the surface.
In Table 6.2, the time-averaged heat transfer was calculated from shock impact,
until the reflected shock propagated away from the surface for 50seconds. Equation
6.11, which is defined as
Q q dt
after
=

1
50
0
50
" [6.11]
was used to represent the time-averaged heat transfer level 50second after shock impact.
It is compared to the mean heat transfer levels prior to shock impact. The ratios of the
time-averaged heat transfer levels are lower than seen in Table 6.1,
Table 6.2 Time-averaged Heat Transfer (50 second period)
Q
before
(mean level)
Q
time-av
Eqn (6.11)
Q
time-av
- Q
before
Q Q
Q
time ave before
before

Stagnant Flow 0 14,300 14,300 N/A
Unheated Flow 3,100 17,400 14,300 4.6
Heated Flow 13,500 28,100 14,600 1.1
which suggests that the effects of time-averaging over a larger time period effectively
reduces the magnitude of the thermal load. Great care needs to be taken when estimating a
time-average unsteady load to avoid under-estimating its magnitude. Although it is short
duration, it can be very significant.
6.5 Simplified Energy Equation
In Section 6.4.2, the numerical animation of the shock/boundary layer interaction
was discussed in detail. During the animation, the velocity boundary layer changed very
little. Also, the y-velocity (induced velocity) was very small. This suggests yet another
simplification that could be applied to the governing equations to produce unsteady heat
transfer data. In this section, this simplification will be tested and compared to the solution
to the governing equations.
As stated before, the x-velocity profile changed very little during the
shock/boundary layer interaction. Therefore, the x-velocity profile shown in Figure 6.4
was used and held constant. The y-velocity was set equal to zero. The initial conditions
were chosen so that the pressure P was equal to (1.1)
2
*P
atm
throughout, the density was
equal to (1.067)
2
*
atm
, and temperature T was equal to (1.026)
2
*T
atm
. These conditions
were used to solve the simplified energy equation, which is
( )
{ }

E
t y
u q
xy y
+ + " 0 [6.12]
This simplification to the fourth term in the vector equation was shown in Equations 3.4,
3.5, and 3.6. All terms that were coefficients of the y-velocity (v) were eliminated. The x-
velocity (u) is a constant. Also, the density field is not updated due to the absence of the
continuity equation. The animation suggests that these are worthwhile simplifications. The
solution to Equation 6.12 produces a heat transfer history that can be seen in Figure 6.20.
The solution to Equation 6.12 applies after shock impact only. This equation does not
contain enough of the flow physics to produce a propagating shock wave. However, it can
produce solutions that represent the transient response of the thermal boundary layer.
In Figure 6.20, the blade surface heat transfer from the solution of the simplified
energy equation is higher than that obtained from the quasi 2-D Navier-Stokes equations.
This could be due to several reasons. As the temperature near the blade surface decreases,
the density increases. The simplified energy equation does not account for the transient
decay of the magnitude of density directly. The pressure is constant, but the total energy is
coupled with the density. This creates difficulty using the ideal gas law. The continuity
equation is not included, so the density is not updated in time. Setting the y-velocity to
zero reduces the continuity equation to the trivial solution (0 = 0). Therefore, with
increasing time, the density does not show any changes from its initial profile.
When solving the quasi 2-D Navier-Stokes equations, the primitive variable (E) was used
to obtain temperature. The temperature was used along with the ideal gas law (P=RT) to
update pressure. The density history was available from the continuity equation. When
solving the simplified energy equation, the primitive variable (E) was used to obtain
temperature. However, pressure does not vary in this flowfield. The
Figure 6.20 Comparison of Simplified Energy Equation with Quasi 2-D N-S
Equations
ideal gas law is not required to solve Equation 6.11. As a result, the decaying nature of
density near the blade surface is not accounted for by the simplified energy equation. The
simplified energy equation shows significant error while the shock is inside the boundary
layer (-0.3 < time < 0.3seconds). This makes sense since the density varies considerably
during this time.
With increasing time, the real density begins to increase. However, the simplified
energy equation does not contain the flow physics required to simulate this. The density
profile time history is required to obtain a more accurate solution from this equation.
Using an estimate of the pre-shock pressure and temperature profiles, the pre-shock
density profile can be produced. This profile should be uniformly increased by a factor of
the density ratio squared to produce an estimate of the post-shock density profile. Using
the post-shock density profile along with the simplified energy equation can produce
reasonable heat flux time history estimates.

Turbine Blade HT

Uploaded by

Copyright:

Available Formats

Turbine Blade HT

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Turbine Blade HT

Uploaded by

Copyright:

Available Formats

Turbine Blade Heat Transfer 69

CHAPTER 6.0 TURBINE BLADE HEAT TRANSFER

Stagnant Flow 0 131,150 131,150 N/A

You might also like