Professional Documents
Culture Documents
Zhou2014 PDF
Zhou2014 PDF
1 Introduction requires a thick trailing edge (Zhou et al. [5]). In high pressure
turbines, the base pressure is much lower than the mixed-out static
The profile loss of a low pressure turbine blade is associated
pressure, so a thick trailing edge creates a large blockage loss,
with the blade surface boundary layers, the effects of the trailing
which is due to the mixing and flow diffusion downstream of the
edge, and flow mixing downstream of the trailing edge. Quite a
blade trailing edge.
few publications studied the boundary layers of high and ultra-
In low pressure turbines, the trailing edges are thinner than
high lift low pressure turbine blades, e.g., Zhang and Hodson [1]
those in high pressure turbines, and the trailing edge loss was
and Stieger and Hodson [2]. In these studies, the effects of Reyn-
found to be lower than the blade boundary layer loss. For a low
olds number and freestream turbulence intensity on the character-
pressure turbine blade with a Zweifel coefficient of about 0.9,
istics of the blade boundary layer, such as transition, were
Roberts [6] found that the trailing edge loss is approximately one
investigated. Hodson and Howell [3] reviewed the process of
third of the blade profile loss. Curtis et al. [7] studied two low
wake-induced boundary-layer transition and how this has enabled
pressure turbine blade profiles with equal loading. They found
successful development of ultrahigh-lift low-pressure turbines. In
that although the blade with the entirely laminar boundary layers
contrast, there are few open publications which specifically focus
had a lower boundary layer loss than the blade with a more turbu-
on the issue of the effects of the trailing edge on the profile loss of
lent boundary layer, the reduction in overall loss was much less
high or ultrahigh-lift low-pressure turbine blades.
than the reduction in the suction side boundary layer loss. This
In a high pressure turbine, the loss due to the blade trailing
highlights the importance of the effects of the blade trailing edge
edge could be higher than the boundary layer loss (Xu and Denton
and base pressure on loss.
[4]). This is because the cooling configuration of the blade
To predict the blade trailing edge loss, a control volume method
was proposed by Denton [8]. In this model the velocity (also the
1
Corresponding author. static pressure) was assumed to be uniform across the throat. If
Contributed by the International Gas Turbine Institute (IGTI) of ASME for
publication in the JOURNAL OF TURBOMACHINERY. Manuscript received July 6, 2013;
the trailing edge base pressure is lower than the mixed-out static
final manuscript received December 14, 2013; published online February 12, 2014. pressure, the trailing edge creates a loss. In general, the value of
Editor: Ronald Bunker. the base pressure must be obtained from empirical data. Hart et al.
Parameter T106C T2
velocities measured in the current study are within this range. The
Fig. 4 Changing the thickness of blade trailing edge
uncertainty of the static and stagnation pressure measured by the
Neptune probe is about 0.3 Pa. The error bar is used later in this
instead to complete the velocity distribution. The remainder of the paper to indicate the uncertainty of the loss measurement.
pressure field was then estimated by means of a simple linear fit
between the blade surface values and nearest available Neptune
4 Results and Discussions
probe pressure value. The error introduced by this procedure is
very small as the estimations apply to only 4% of the pitch. The 4.1 T106C Cascade
Neptune probe also measures the flow at plane ‘2’ in Fig. 2, which
located 30% Chord downstream of the trailing edge. 4.1.1 Blade Surface Static Pressure Coefficient. The surface
The trailing edge thickness was changed by attaching a thin static pressure coefficient is defined as
insert on the blade pressure side near the trailing edge, as shown
in Fig. 4. Pieces with different thickness were used to simulate P01 P
Cp ¼ (1)
different trailing edge thicknesses. The attached pieces were made P01 P2
using a rapid prototyping technique, so that the insert nicely fol-
lowed the profile of the blade pressure side. The insert was where P01 is the upstream stagnation pressure, P is the local static
blended into the pressure side surface about 20% chord upstream pressure and P2 is the averaged cascade exit static pressure.
of the trailing edge. The Trailing edge pieces were attached using Figure 5 shows the measured static pressure coefficient for the
glue and thin tape. It is noted that a slight change in the throat T106C blade with the standard blade trailing edge thickness
area is introduced by the inserts. The profile of the blade suction (1.9% s) and double trailing edge thickness (3.8% s). The abscissa
side surface remains unchanged to minimize the change of the is the fraction of the surface length. A separation bubble occurs on
suction side surface boundary layer, which is the main contributor the suction side surface at around 55% of the suction surface
to the profile loss. The loss due to the boundary layer on the blade length. The flow then reattaches at about 70% of the suction side
pressure side surface is small. During the experiment, the trailing surface length. The five taps located from 80% s to 100% s on the
edge thicknesses of the bottom four blades in Fig. 1 were set to be blade pressure surface are not plotted for the case with a trailing
the same. edge thickness of 3.8% s, because the insert covers them.
The base pressure is measured by a static pressure tapping As the trailing edge thickness increases, the throat area is
located on the blade trailing edge, which is indicated in Fig. 4. reduced, so the mass flow rate and the overall loading decrease.
The static pressure tapping is located on the center of the blade As a result, increasing the trailing edge thickness slightly reduces
trailing edge, when no piece was attached to thicken the trailing the peak Cp on the suction side. Nevertheless, this effect is small.
edge. When pieces were used on the blade pressure side surface to The base pressure coefficient is defined as
simulate thicker trailing edges, this static pressure tapping was
still used to measure the base pressure. With the attached piece, Pb PM
Cpb ¼ (2)
the trailing edge is closer to, but no longer truly circular. The mea- P01 PM
surement point will become closer to the suction side. Based on
the CFD simulation, the error in the base pressure measurement where Pb is the base pressure and PM is the mixed-out static
will introduce a maximum error of 62% in the magnitude of pro- pressure.
file loss coefficient, which is about 0.03. This results in an uncer- The measured static pressure distribution at the trailing plane of
tainty of 60.0006 in the profile loss, which is small. the cascade is normalized in the same way as the base pressure
coefficient and is presented in Fig. 6. In Fig. 6, a negative value
3 Uncertainty Analysis means that the static pressures at these locations are lower than
the mixed-out static pressure. For the T106C cascade, this area
A Pressure System Inc. (PSI) Pressure Scanner 9016 was used extends from 14% to 67% of the pitch. Near the blade surfaces,
to measure the mean pressure. The accuracy of the pressure read- the static pressure is higher than the mixed-out static pressure.
ings was validated by Druck DPI 520. The combined uncertainty This pressure difference accelerates the low velocity flow near the
of both the random and the systematic uncertainty were better trailing edge. This effect reduces the loss in the downstream mix-
than 0.2 Pa for the current measurement. The variation of exit ing process (Denton [8]).
Reynolds number in the experiments was less than 0.5%.
The Neptune probe was calibrated at Reynolds number corre- 4.1.2 Reduction of Loss. The mixed-out loss can be derived
sponding to velocities of 30 m/s and 15 m/s. The difference in the from the measured data either at the trailing edge plane or 30%
calibration maps of the two velocities is negligible. All of the chord downstream of the cascade in either case by using a
P01 P0;M Fig. 7 Comparison of the mixed-out loss (Yp) calculated from
Yp ¼ (3)
P01 PM measurements in the trailing edge plane and in a plane 0.3 C
downstream of the trailing edge
where P0;M and PM are mixed-out stagnation and static pressure.
To breakdown the profile loss, the loss due to the blade surface distribution at the trailing edge plane (e.g., Denton [8]) cannot be
boundary layer can be obtained in the following way. The entropy used in this case. The analytical method used in this paper con-
generation within the boundary layer can be derived based on the ducts constant area mixing calculation (Greitzer et al. [13]) based
hotwire measurement as (Schlichting [14]) on the flow data measured on the trailing edge plane, the base
ð TE pressure and the trailing edge thickness.
q
S_BL ¼ S_A dS ¼ ðU 3 dE ÞTE (4)
LE 2T 4.1.3 Profile Loss of Standard Trailing Edge. A comparison
between the mixed-out total pressure loss coefficient values calcu-
where S_A is the rate of entropy production per surface area, U is lated from the traverses at the trailing edge plane and the traverses
the outer velocity of the boundary layer and dE is the Energy performed at a plane 0.3 C downstream of the trailing edge is pre-
thickness of the blade surface boundary layer at the blade trailing sented in Fig. 7. The ratio of the standard trailing edge thickness
edge. to pitch is 1.9%s. The Reynolds numbers of low pressure turbine
Using the method proposed by Greitzer et al. [13], the entropy blades range from about 0:5 105 in the final stage at high alti-
generation can then be related to the loss coefficient. tude in small business jet applications to about 5 105 at sea level
takeoff in the first stage of the largest turbofans. (Hodson and
dE U3 P0;M PM Howell [3]); when the Reynolds number is higher than 2 105 ,
Yp;BL ¼ 3
(5)
s cosðaM Þ VM TE P01 PM the effects of the Reynolds number on the loss become small. So,
in the current study, measurements were made at six Reynolds
In this calculation, the parameters of the energy thickness of the numbers of 0:5 105 , 0:7 105 , 0:9 105 , 1:2 105 , 1:5 105
boundary and the boundary layer edge velocity at the trailing edge and 2:1 105 . The dashed line is based on the measurement at six
were all measured. Using this method, the loss due to the blade Reynolds numbers. The loss coefficient reduces with increasing
pressure side boundary layer and suction side boundary layer can the Reynolds number. The effects of the Reynolds number on the
be derived, respectively. loss are mainly related to the flow separation and transition on the
The mixed-out loss coefficient can be broken down into two blade suction side surface, and are beyond the scope of the current
terms: the loss of stagnation pressure in the blade passage paper. Interested readers can refer to Zhang and Hodson [1] and
upstream of the trailing edge and the loss downstream of the cas- Mahallati and Sjolander [15].
cade as For the steady flow condition, it was decided only to consider
the trailing edge measurements down to a Reynolds number of
P01 P0;TE P0;TE P0;M 120,000. For lower Reynolds numbers, where an open separation
Yp ¼ þ (6) increasingly occurs in steady flow conditions, the agreement
P01 PM P01 PM
between the measurements was not very good because of the
where P0;TE is the mass averaged stagnation pressure at the trail- reversed flow at the trailing edge, which could not be properly
ing edge plane. The first term on the right hand side is mainly due evaluated. However, excellent agreement for the mixed-out loss
to the boundary layers on the blade surface, the second term on coefficient was found over the entire Reynolds number range
the right hand side is the mixing loss, in which the base pressure under unsteady flow conditions because the flow is always
has an effect. The second term is the mixing loss downstream of attached at the trailing edge. This provides confidence in the anal-
the trailing edge plane. ysis that follows later in this paper.
Base pressure coefficient values reported in the literature are
P0;TE P0;M commonly based on far downstream or mixed-out reference val-
Yp;M ¼ (7)
P01 PM ues. Values reported in the literature range between 0.1 and
0.2 (e.g., 0.13 by Hart et al. [10]). The base pressure coeffi-
The static pressure distribution at the trailing edge plane is highly cient calculated using the measured mixed-out values is presented
nonuniform, so the methods suggest a uniform static pressure in Fig. 8. In contrast to typical values reported in the literature, the