Professional Documents
Culture Documents
com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
Gursharanjit Singh
GE Power, Rugby, UK
singh.gursharanjit1@gmail.com
ed
ite
d
The present study aims to investigate the interaction between a last-stage steam
turbine blade row and diffuser. This work is carried out using computational
(CFD) simulations of a generic last stage low pressure (LP) turbine and axialradial exhaust diffuser attached to it. In order to determine the validity of the
computational method, the CFD predictions are first compared with data
obtained from an experimental test facility. A computational study is then
performed for different design configurations of the diffuser and rotor casing
shapes. The study focuses on typical flow features such as effects of rotor tip
leakage flows and subsequent changes in the rotor-diffuser interactions. The
results suggest that the rotor casing shape influences the rotor work extraction
capability and yields significant improvements in the diffuser static pressure
recovery.
Andrew P. S. Wheeler
Department of Engineering
University of Cambridge, UK
Gurnam Singh
Co
py
GE Power, Rugby, UK
sc
rip
tN
ot
Keywords: Steam turbines, rotor, diffusers, pressure recovery, last stage blades
Ac
ce
pt
ed
Ma
nu
INTRODUCTION
The majority of the worlds power production is made
through steam turbine power plants using different natural
resources. Further improvement in efficiency of a steam turbine
is important to meet increasing power needs, controlling fuel
costs with less exploitation of the natural resources. Over the
past decades, the technology has been developed to use longer
last stage blades [1] in the steam turbines and these are now
contributing up to 20% of the total power produced in the steam
power plants [2]. This is self-evident from thermodynamics of
the Rankine cycle that large enthalpy drops occur at the lower
pressure side of the steam turbine. It also means that there will
be large entropy productions at the lower pressure (and
temperature) areas of the steam turbine [3].
Specific power output from a Low Pressure (LP) steam
turbine can be enhanced by increasing its internal efficiency or
raising the pressure ratio across its ends. Increase in efficiency
can be achieved by better designs changes [4]; exploiting a full
3D design [1,5], or reduction in the internal losses [3]. Increase
in pressure ratio across the turbine stage can be achieved by
connecting it to an exhaust diffuser that helps to recover the
leaving kinetic energy losses from the LP turbine. The exact
nature of flow inside a diffuser is still unknown to some extent.
A large number of diffuser mappings (channel, conical,
annular) have been studied [6] and compiled (for instance [7]).
However these mappings cannot be used directly for axialradial diffusers which are commonly used for steam-turbine
exhausts. One of the main reasons for this is the short axial
GTP-16-1174
Singh
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
SysPr
LSB
MP
/
tN
ot
Co
py
ed
ite
d
COMPUTATIONAL SET-UP
The flow solver used for the entire study is TBLOCK-11.9, a
3D RANS solver developed by Prof. John Denton and is widely
used in the industry. It is mainly derived from the time
marching methods as described in [23-26]. The spatial
differences are completely centered, using cell corner storage
and the time stepping is based on the 'Scree' algorithm [27].
The algorithm is comparatively fast and requires minimum
artificial viscosity. It employs a multi-block approach by
dividing the flow domain into various blocks of different sizes
and shapes while joining them with different patches (as
desired by the user).
The turbulence is modeled using either the Baldwin-Lomax
mixing length model or the Spalart-Allmaras (SA) model.
Depending on the near wall mesh resolution, wall functions can
be employed if the first grid-line away from the wall is outside
of the viscous sub-layer [28]. For this study, the first grid point
away from the wall was positioned within y+<5 for all viscous
surfaces. Also, for most of the blade surface the y+ values lie
between 2-3 and the corresponding y+ values fall between 1-3
on the diffuser casing side. With SA model, the solver gives an
option to use the mixing length model for the first 10 grid
points (streamwise direction) in the inlet block to generate a
realistic profile of turbulent viscosity in the endwall boundary
layer entering the rotor.
The flow domain was divided into the 14 blocks (9-rotor, 3diffuser and 2-tip gap if present). The rotor passage has 2M grid
points, while the diffuser contains 1.6M grid points. TBLOCK
being a structured mesh solver, an O-H mesh topology was
adopted for the rotor passage (the last stage blades are highly
staggered from hub to tip and a normal H-mesh is not suitable),
while the tip-gap and diffuser were modeled with an H-mesh.
Figure 1 shows the mesh used for the case with a tip-gap of 1%
span. Here, part A shows the mesh for the whole domain as
seen in the meridional plane. Part B depicts the mesh on the
rotor surfaces as viewed at an oblique angle from the top. Part
C gives the zoomed-in view of the mesh around rotor trailing
edges. Part D shows the computational mesh for the rotor tip
gap. Grid expansion ratios are applied such that smooth
transition of cell aspect ratios occurs across the computational
domain. A mesh independence study was performed
individually for each section i.e. rotor, tip gap and diffuser. To
achieve this, three separate meshes i.e. coarse, medium and fine
were created for each section and performance factors were
compared along with the spanwise distribution of flow field and
local pressure profiles along the diffuser casing. For singlepassage rotor calculations, with points per passage of 0.6M,
1.5M and 3.1M, the predicted rotor total-to-total efficiencies
were 93.893%, 93.931% and 94.059% respectively. Hence the
rotor total-total efficiency changes by 0.13% when the rotor
mesh is doubled from 1.5M (medium) to 3.1M cells (fine). For
ce
pt
ed
Ma
nu
sc
r
ip
done by [16] and [9] showed the coupling between the last
stage blade and exhaust with the diffuser and demonstrated the
scope of three-dimensional diffuser shape optimizations.
Fu et al [17] carried out numerical and experimental studies
of unsteady interactions between the LSB and exhaust hood
under different operating conditions, revealing strong
interations between its components. They demonstrated that the
stage performance is dependent on diffuser inlet pressure and
swirl distributions. In the case where rotor tip flows are
supersonic, such as for supersonic reversed cambered blades
[18], there is likely to also be a strong interaction between rotor
trailing-edge shocks and the downstream diffuser.
The diffuser exit conditions are set by the exhaust collector.
Finzel et al [19] presented in-depth experimental comparison
for different exhaust hood confirgurations, diffuser lengths, tip
leakage flows and with swirl. They found that the leakage flow
positively affects the diffuser performance by supressing
adverse pressure effects on the diffuser casing. Further, both
numerical and experimental studies for the gas turbine annular
diffuser have also been shown to benefit from tip clearance
flow, in terms of increasing diffuser pressure recovery [20-21].
Musch et al [22] found an optimum diffuser casing profile
with a tip leakage flow using a coupled boundary layer
calculations in a auto-optimiser, and later confirmed by 3DRANS runs. They paratmetrized the diffuser shape mainly as a
circular arc with variable diffuser exit radii. In the present
study, the focus is on the interaction of the blade tip flow field
and the diffuser casing. The diffusers presented for the most
part of this paper have small levels of separation on the casing.
The present study involves the investigation of flows within
an LP turbine diffuser attached to an upstream rotor, such that
the effects of the rotor flow (such as tip-leakage and trailingedge shocks) on the diffuser and overall turbine work output
can be explored. Initially, the computational method is
validated against rig data provided by GE Power. Then a
generic stage design, representative of a modern LP stage is
used to investigate computationally the rotor/diffuser
interaction. Different design configurations of the diffuser and
rotor casing shapes with a range of hade angles are analysed at
a constant swallowing capacity. The study focuses on the
effects of rotor tip leakage flows and trailing-edge shocks, and
subsequent changes in the stage-diffuser interactions. The
results suggest that the rotor casing shape influences its work
extraction capability. It can yield a better diffuser performance
by significantly improving the static pressure recovery.
Ac
NOMENCLATURE
Cp
Pressure coefficient (Pout - Pin)/ (Po,in - Pin)
Ys
Loss coefficient (Po,in - Po,out)/ (Po,in - Pin)
GTP-16-1174
Singh
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
DIFFUSER DESIGN
The datum geometry for the blade was provided by Denton
[29]. Using the datum shape, meshes were generated via
automated parametric software written in the MATLAB
specifically for this study. After generating the stage mesh, an
axisymmetric diffuser was generated using the blade exit
coordinates.
The casing profile was generated by four control points
(marked 1-4 in Figure 2) using a Bezier curve which can be
derived from the sum of Bernstein polynomials as shown
below:
py
ed
ite
d
Ma
nu
sc
r
ip
tN
ot
Co
Ac
ce
pt
ed
GTP-16-1174
Singh
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
section. The S/H value for this study was set at 0.8 with an area
ratio of 1.75. The circumferential width of the diffuser was set
to be equal to the rotor exit blade passage pitch.
sc
r
ip
tN
ot
Co
py
ed
ite
d
Ma
nu
Ac
ce
pt
ed
GTP-16-1174
Singh
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
Ac
ce
pt
ed
Ma
nu
sc
r
ip
tN
ot
Co
py
ed
ite
d
GTP-16-1174
Singh
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
RESULTS
The rotor-diffuser working geometry was generated as
described in previous sections and the diffuser area ratio was
set to 1.75, with L/H = 1.617 and S/H = 0.8 (refer to Figure 2).
Eight test cases (see Table 1) were investigated in order to
determine the effects of the rotor/diffuser interactions on
diffuser performance. The test cases with and without a mixing
plane (MP) interface between the rotor and diffuser blocks were
investigated; they allow us to determine the influence of
circumferential non-uniformity on the diffuser and
rotor/diffuser system. Also test cases were investigated with
and without a rotor tip gap for both with and without mixingplanes. The 'no tip gap' case was generated by extending blades
up to the rotor casing. These test cases were also configured to
run at two system pressure ratios of 3.92 and 4.40.
Figure 6 plots the rotor and system efficiencies for the eight
test cases described in Table 1. The system efficiency is based
on the total-to-static efficiency across the rotor-diffuser, where
the exit static pressure is determined from the area-average
static pressure at the diffuser exit. The rotor total-to-total
efficiency is determined based on a constant area mixed-out (in
circumferential and spanwise directions) average at rotor exit.
For the idealized case where an isentropic diffuser is able to
diffuse the flow to a stagnation condition at diffuser exit, the
system and rotor efficiencies would be equal. The difference
between rotor and system efficiencies is thus a measure of the
diffuser performance, while the rotor total-total efficiency is a
measure of the rotor performance alone, since the use of mixedout quantities accounts for any downstream loss which would
have been generated if the rotor was coupled to a constant area
duct.
tN
ot
Co
py
ed
ite
d
sc
r
ip
Ac
ce
pt
ed
Ma
nu
No
No
with
with
TipGap
TipGap TipGap
TipGap
MP
MP
A1
B1
C1
D1
SysPr = 3.92
A2
B2
C2
D2
SysPr = 4.40
Table 1. TESTS CASES FOR ROTOR/DIFFUSER
INTERACTIONS
GTP-16-1174
Singh
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
tN
ot
Co
py
ed
ite
d
ed
Ma
nu
sc
r
ip
pt
ce
Ac
GTP-16-1174
Singh
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
tN
ot
Co
py
ed
ite
d
Ac
ce
pt
ed
Ma
nu
sc
r
ip
GTP-16-1174
Singh
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Ma
nu
sc
r
ip
tN
ot
Co
py
ed
ite
d
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
ed
Ac
ce
pt
GTP-16-1174
Singh
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
position ensuring a fixed leakage area along the chord. All tests
cases were run with and without tip gap flow.
Figure 14 plots the diffuser performance factors for
different hade angles. It reveals that the diffuser Cp is highly
enhanced with an increase in the hade angle. Also, the total
pressure loss (Ys) is relatively unaffected by hade. The changes
in Cp are therefore a consequence of changes in diffuser
inviscid loading (1-). Figure 15 plots the pitchwise averaged
radial profiles of static and total pressure and stagnation
temperature from 60-100% span at the diffuser inlet, for the
cases with and without leakage flow. For simplicity, only three
hade angles are provided. The results show that the effect of
hade is to greatly reduce the static pressure in the upper portion
of the span, increasing the work-output in this region
significantly. For cases with a tip gap, the results show a large
reduction in the stagnation pressure and temperature in the
region of the leakage flow ~99% span with the rise in the hade
angle.
ed
Ma
nu
sc
r
ip
tN
ot
Co
py
ed
ite
d
ce
pt
Ac
GTP-16-1174
Singh
10
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
ip
tN
ot
Co
py
ed
ite
d
Ac
ce
pt
ed
Ma
nu
sc
r
GTP-16-1174
CONCLUSIONS
A study was performed of the flows within a last-stage
steam turbine rotor and coupled axial-radial diffuser. A CFD
solver was first validated against rig data before being used to
investigate various aspects of the rotor and diffuser flow field.
The paper focused on two main lines of investigation: the first
was the effects of circumferential non-uniformity and leakage
flows on the coupled rotor/diffuser performance; the second
area of study was the effect of rotor casing shape (namely hade
angle) on the coupled system performance. The study revealed
that the majority of the entropy rise across the diffuser is a
result of trailing-edge shocks and the mixing loss generated by
the rotor flow. For cases without a leakage flow, trailing-edge
shocks were found to greatly influence the unsteady pressure
field on the diffuser casing. The presence of a leakage flow
virtually eliminated this effect, by off-loading the rotor tip and
reducing the trailing-edge shock strength. In most cases, mixing
of flow features in the diffuser generated similar losses if mixed
upstream of the diffuser across a mixing-plane, or within the
diffuser.
Increasing hade angle of the rotor casing significantly
increased the diffuser pressure recovery and reduced tip leakage
flows; the combined effects led to an increase in overall system
performance of around 2% as hade was increased from 0 to
40deg.
Singh
11
http://tarjomebazar.com
09372121085 Telegram
026-33219077
py
ed
ite
d
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
ip
tN
ot
Co
Ac
ce
pt
ed
Ma
nu
sc
r
GTP-16-1174
Singh
12
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
py
ed
Co
ot
tN
Ac
ce
pt
ed
Ma
nu
sc
r
ip
REFERENCES
[1] Stuer H., Truckenmuller F., Borthwick D., Denton JD.
Aerodynamic Concept for Very Large Steam Turbine Last
Stages. GT2005-68746. ASME Turbo Expo 2005: Power
for Land, Sea and Air.
[2] McBean, I., Havakechian, S., & Masserey, P.-A. (2010).
The Development of Long Last Stage Steam Turbine
Blades. ASME Turbo Expo 2010: Power for Land, Sea,
and Air.
[3] Denton, J. D. (1993). The 1993 IGTI Scholar Lecture:
Loss Mechanisms in Turbomachines. Journal of
Turbomachinery.
[4] Haller, B. R.; Hesketh, J. A. (1993). Large Steam Turbine
Retrofitting to Improve Performance. Power-Gen Europe.
[5] Denton, JD and Xu, L (1999). The exploitation of 3D flow
in turbomachinery design. Proceedings of the Institution
of Mechanical Engineering, Part C: Journal of Mechanical
Engineering Science, 213. pp. 125-137. ISSN 0954-4062
[6] Sovran G., Klomp E. D., 1967, Experimentally
Determined Optimum Geometries for Rectilinear
Diffusers with Rectangular Conical or Annular Cross
Section, in Fluid Mechanics of Internal Flow (G. Sovran,
ed.), Elsevier
[7] Japikse, D., & Baines, N. C. (1998). Diffuser Design
Technology. Concepts Eti; 2 edition.
[8] Sakuma A., Matsuura T., Kodama H. Steam Turbine
Retrofits Technology for Improved Performance and
Reliability, Challenges of Power Engineering and
Environment, p251-257 Vol 1-2, 2007.
[9] Yoon, S., Stanislaus, F. J., Singh, G., & Claridge, M.
(2011). A Three-Dimensional Diffuser Design for the
Retrofit of a Low Pressure Turbine Using In-House
Exhaust Design System. Proceedings of ASME Turbo
Expo 2011: Power for Land, Sea and Air.
[10] Li Z., Li J., Yan X., Feng Z., Ohyama H., Zhang M.
Investigations on the flow pattern and aerodynamic
performance of last stage and exhaust hood for large
power steam turbines.
[11] Greitzer, E. M., Tan, C. S., & Graf, M. B. (2004). Internal
Flow - Concepts and Applications. Cambridge University
Press.
[12] Liu, J. J., Cui, Y., & Jiang, H. (2001). Investigation of
Flow in a Steam Turbine Exhaust Hood With / Without
Turbine Exit Conditions Simulated. Proceedings of ASME
Turbo Expo 2001.
ite
d
ACKNOWLEDGMENTS
The authors would like to thank Prof. John Denton for
providing his CFD solver along with the datum geometry. His
initial comments to start this study during a personal meeting
are deeply regarded. The comments and ideas from Dr. Brian
Haller and Dr. Sungho Yoon, GE Power, Rugby proved to be
the underline building of this paper. The funding for this project
was provided by QMUL, London. The solver validation data
was provided by GE Power, Rugby.
GTP-16-1174
Singh
13
http://tarjomebazar.com
09372121085 Telegram
026-33219077
Journal of Engineering for Gas Turbines and Power. Received May 09, 2016;
Accepted manuscript posted August 11, 2016. doi:10.1115/1.4034417
Copyright (c) 2016 by ASME
Ac
ce
pt
ed
Ma
nu
sc
r
ip
tN
ot
Co
py
ed
ite
d
GTP-16-1174
Singh
14