Professional Documents
Culture Documents
Numerical Investigation of Leakage and Power Loss For Different Seal Types in Turbine Stage Environment
Numerical Investigation of Leakage and Power Loss For Different Seal Types in Turbine Stage Environment
GT2010
July 14-18, 2010, Glasgow, UK
GT2010-22743
Outlet
Shroud
Inlet
Bristle pack
NUMERICAL APPROACH
All configurations of the baseline labyrinth seals and
improved honeycomb seals and brush seals were analyzed by
CFD to predict leakage rate at different seal clearance and (b) Efficiency
structure parameters. The flow field details were also discussed Fig.9 Computational convergence information
to obtain pressure distribution and sealing effects of seals. Porous Medium Model
Mesh Generation The bristle pack in a brush seal forms a porous medium. To
All blades and seals structure meshes were generated in estimate leakage and pressure load of brush seals, three
AUTOGRID5 and IGG of commercial software NUMECA. categories flow models are developed, cross flow analysis
Considering the axi-symmetric, only one periodic blade and through voids among bristles, BULK flow models, and porous
seal mesh was generated. Total mesh size for blades was medium models. The porous medium approach is basically
approximately 1 million nodes. Approximately 10000 nodes solving the Navier-Stokes equation with the additional flow
were specified with in one honeycomb and 14000 nodes in resistances due to friction between flow and bristles. For the
bristle pack domain. All blades together with honeycomb seals highly resistive porous media, this equation is simplified by
in shroud flow path and brush seals in diaphragm flow path neglecting the inertial and viscous terms, which yields a
mesh sizes were approximately 4 million. The y+ values near balance equation between pressure gradient and flow resistance
the solid were below 15. For accelerating the computation terms. This approach has been successfully applied for brush
convergence, multi-grid technology was used in all calculation seals [23-26]. Dogu’s analysis results [22,24] prove that the
Leakage Rate(kg/s)
[24], namely the Darcy law, the Ergun law and an Integral law. 0.9
0.8
In this paper, Darcy law is implemented to simulate the
0.7
flow in the bristle pack [25]. In the Darcy law, the pressure
0.6
drop is given by the following relation: 0.5
dp μ
− =
0.4
ui (1) 0.3
dxi K i 0.2
Efficiency
0.9425
Where the mean diameter of the inclusion d is given by 0.94
the following formula: 0.9375
d = 6/ A (3) 0.935
0.9325
Where, A is the ratio of the Volume of an inclusion to its 0.93
surface. For a spherical inclusion, the mean diameter is the blade stage blade stage+DS blade stage+SS blade
Leakage rate(kg/s)
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
labyrith seals brush seals
Seal types
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
-0.05 0 0.05 0.1 0.15 0.2 0.25 0.3
Clearance(mm)
0.943
0.94275
0.9425
0.94225
0.942
0.94175
0.9415
-0.05 0 0.05 0.1 0.15 0.2 0.25 0.3
Clearance(mm)
0.13
0.12
0.11 Fig. 21 Radial pressure distribution in bristle pack
0.1 For brush seals, the interior domain of bristle pack’ flow
0.09
0.08 condition is not the major problem concerned. However, the
0.07 flow field which close to the heading plate and backing plate is
0.06
0.05 usually chosen for analyzing sealing effect. The flow lines in
1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 these domains are presented in fig. 22. Stream flows into fence
Fence height(mm) height domain by the axial and radial pressure impulse, and
Fig. 19 Leakage rate of brush seals in different fence height pressure drop mainly occurs in fence height, pressure in upper
Flow Field region of bristle pack almost equals to the upstream pressure.
The axial pressure distribution in bristle pack is presented Comparing to the fence height, the axial velocity in the upper
in fig. 20. The different color lines represent the axial pressure region is lower. Under the backing plate, steam accelerates
distribution from the upstream side to the downstream side at because the flow area decreases. After flowing through the
different radial height. The results are reasonable similar with backing plate, the stream expands to a larger vortex in the
the Dogu’s [23] investigation. Pressure changes little in the cavity near the backing plate. This vortex could decrease the
upper region of bristle pack, and the downstream pressure equivalent flow area which helpful to leakage rate reduction.
almost equals to the upstream, pressure mainly loads on the
backing plate. In the fence height, the inertial momentum of
flow hitting to bristles face diffuses into the bristle pack
without any accumulation, unlike the upper region. Thus, axial
pressure drops mainly in the fence height, and this high
pressure drop may cause the bristle “blow-down” during
operation. Fig. 21 illustrates pressure distribution in radial
direction. In fence height region, the radial pressure is almost
uniform, then, rapidly reaches the upstream pressure close to
backing plate tip.
1.2
1
0.8
0.6
0.4
0.2
0
labyrith seal honeycomb seal honeycomb seal honeycomb seal
0.75 0.75 0.5 0.25
1.15
0.95
0.75
0.55
0.35
0.2 0.3 0.4 0.5 0.6 0.7 0.8
Clearance(mm)
(a) Leakage
0.944
0.943
Fig. 26 Two-dimension Flow lines in honeycomb seals
Fig. 27 illustrates static pressure distribution from the inlet
Efficiency
(b) Efficiency
Fig.24 Honeycomb seals’ leakage and stage efficiency
Flow field
Fig. 25 and fig. 26 show pressure distribution and flow
lines at honeycomb seals’ axial cutting plane. Steam flows into
the honeycombs, with the high turbulent effect, many vortexes
are formed. The each honeycomb is more like the teeth of the
labyrinth seals. However, every honeycomb is separated from
other honeycombs by the solid wall, which will slow down
flow velocity in circle direction, and thus reduce the leakage Fig. 27 Static pressure in the honeycomb seal
flow. (clearance=0.5mm)
In order to research the flow field in the honeycomb, one
honeycomb in the honeycomb band is chosen for analyzing
(fig. 28). Fig. 29 demonstrates the velocity vectors in the axial
cutting plane. A larger eddy occupies the full area of the
cutting plane. At the bottom of the honeycomb where close to
the clearance, the magnitude of velocity is up to 50m/s,
whereas at the center of the eddy, the magnitude of flow speed
is below 10m/s.
Fig. 30 shows the relative velocity vectors at different
radial height of honeycomb. At 10% radial height, the flow
ACKNOWLEDGMENTS
The authors would like to thank the Shanghai Power
Generation Equipment CO., LTD. for supporting this
investigation. Special thank goes to Mr. QI (from Beijing
(30%) (50%) Institute of Technology) for his contribution in this work..
REFERENCES
[1] Stoff, H., Incompressible Flow in a Labyrinth Seal,
Journal of Fluid Mechanics, 1980, 100: 817-829.