Proceedings of ASME Turbo Expo 2015: Turbine Technical Conference and Exposition

GT2015
June 15 – 19, 2015, Montréal, Canada

GT2015-44068

CAE BASED BRUSH SEAL CHARACTERIZATION FOR STIFFNESS AND STRESS

LEVELS

E. Tolga Duran Mahmut F. Aksit

SDM R&D Sabanci University
Istanbul, Turkey Istanbul, Turkey

Murat Ozmusul
SDM R&D
Istanbul, Turkey

ABSTRACT σVM = Von Mises stress

Brush seals are complex structures having variety of design
parameters, all of which affect the seal behavior under turbine INTRODUCTION
operating conditions. The complicated nature of the seal pack Improvements in steam and gas turbine technologies
and frictional interactions of rotor, backing plate and bristles constantly drive harsher conditions, which result in
result in nonlinear response of the brush seal to variances of thermodynamic cycles having higher pressure ratios and inlet
design parameters. This study presents CAE based temperature levels as well as increased leakage flows. The
characterization of brush seals, which aims to investigate the necessity of reducing the internal leakage flows and
main effects of several brush seal design parameters on brush thermodynamic cycle losses while increasing the turbine overall
seal stiffness and stress levels. Characterization work of this efficiency puts greater importance on improvements in sealing
study includes free-state rotor rub (unpressurized seal), steady technologies. Brush seals are fulfilling the leakage performance
state (pressure load without rotor interference) and pressurized- requirements and they can successfully compensate rotor
rotor interference conditions. interference at turbine transient conditions. Since the bristles
slide against the high-speed rotating shaft, friction and wear at
NOMENCLATURE the bristle tip contact becomes a major concern as it determines
BH = Free Bristle Height the life and efficiency of the seal as well as the rotor stability.
BP = Backing Plate Evaluating the bristle tip contact forces and the resulting stress
BPCR = Backing Plate Corner Radius levels under operating conditions is critical to optimize the seal
BTF = Bristle Tip Force performance and to achieve the bristle durability.
CAE = Computer Aided Engineering There are several analytical studies in the literature which
FE = Finite Element
analyses the stiffness and stress levels for brush seals. Flower [1]
FEA = Finite Element Analysis
FH = Fence Height considered a single bristle as a cantilever beam deforming under
FP = Front Plate rotor interference. Stango et al. provided a more complex
NR = Number of Bristle Rows formulation [2] to calculate bristle tip forces and displacements
VM = Von Mises under rotor rub conditions without pressure load. Zhao et al. [3]
PD = Downstream Pressure developed an analytical formulation for bristle deformation
PU = Upstream Pressure under pressurized conditions, but they do not evaluate bristle tip
R = Rotor radius forces under pressurized conditions.
SA = Axial spacing between bristles The bristle tip forces and stress levels strongly depend on
ST = Tangential spacing between bristles frictional effects. Inter-bristle, bristle pack backing plate and
ΔP = Pressure load rotor-bristle contacts play important role in determining the
μbp = Backing plate-bristle friction coefficient
bristle tip force levels and wear life of a seal. The complicated
μib = Inter-bristle friction coefficient
μrb = Rotor-bristle friction coefficient nature of contacts and frictions within the bristle pack do not lend
itself to full analytical formulation. Therefore, comprehensive

1 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 FEA is needed to account for inter-bristle interactions. There are amongst hexahedral elements, “C3D8I-8 node hexahedral
several studies in the literature where CAE based analyses were elements with incompatible modes [12]” is selected, which is
reported. Aksit [4] performed seal analyses by using space beam known to have good accuracy in bending problems. However,
elements at unpressurized/pressurized-static rotor interference care must be taken in FE modeling since the accuracy of the
conditions (non-rotating rotor). Crudgington et al. [5, 6, 7, 8] simulation strongly depends on the element quality and Jacobian
conducted brush seal FE analyses using Hexahedral and space criteria. A brush seal FE model constructed with C3D8I elements
beam elements at static-unpressurized rotor interference is given in Fig. 1. Each bristle has 16 nodes at the bristle
conditions and for a loaded seal. Guardino [9] and Lelli [10] were circumference, which assures the element quality for
used in-house developed software, which was based on simulations. Backing plate and rotor are modeled as rigid bodies
mathematical formulae, for simulating the bristle deflection at due to their larger stiffness in comparison with bristles. Backing
non-rotating rotor interference. Pekris et al. [11] conducted a plate corner radius is introduced to the FE model for more
DOE study for an idealized brush seal to examine the effect of realistic simulation of localized contact between bristles and
geometric factors on brush seal performance measures including backing plate at the fence height region. Cyclic boundary
bristle tip reaction force for zero tip deflection. condition has been defined in circumferential direction, which
In this study, FE methodology and in-house software has enables simulating brush seal under operating conditions. Inter-
been developed for performing stiffness and stress bristle, bristle-backing plate, rotor-bristle contacts and frictions
characterization of brush seals. The automatic input file have also been defined in the FE model.
generation by privately developed software enables preparing
brush seal input models that are ready to run in Abaqus
automatically. Two different FE methodologies have been
developed and correlated with the test data of custom design test
rig for bristle tip force measurements. Correlated CAE models
have been used for stiffness and stress characterization of brush
seals under rotor rub conditions for a loaded and unloaded seal,
and also under steady state operating conditions. This study
focuses on the main effects of selected design parameters on seal
stiffness and bristle stress levels, and further analyses are
required for including the cross effects.

COMPARISON OF DIFFERENT FE METHODOLOGIES

Selecting the correct element type has crucial importance in
achieving the FE simulations with high fidelity. The element
library of ABAQUS consists of variety of elements with hybrid
and modified formulations. Commonly used element families
are solid, shell, beam, membrane, truss and spring elements. In
the FE analyses of brush seals, using shell, membrane, truss
elements is not possible due to geometry and stress formulation
of bristles. Amongst continuum elements, penta elements are not
preferred due to their inaccuracy in stress calculations during
structural analyses. Similarly, first order tetrahedral elements are
not used since they act extremely rigid due to their formulations
and have shear locking problems. Second order tetrahedral
elements are known to be accurate in structural analyses, but
increased number of nodes and possible warpage problems
during bending of bristles avoids usage of these elements in
brush seal simulations. Therefore, the most appropriate element
types for brush seal analyses are Hexahedral and Beam type
elements. Figure 1: Brush seal CAE model with C3D8I elements – CAE
model details
Second order hexahedral elements are not preferred due to
the fact that formulation of second order elements may yield The circular cross-section and large L/d ratios of bristles
contact pressure discontinuities, which in turn will cause enable usage of beam type elements during simulations. B31 –
accuracy problems in BTF correlation. C3D8R [10] elements are space beam with circular cross-section is the most appropriate
8-node brick elements with reduced integration, and they are not element type for bristles that can simulate three dimensional
preferred in brush seal modeling because hour-glassing mode in deformation as well as stress state for the brush seal. The B31
bending will yield smaller stress levels than expected. Therefore, section and integration point layout is detailed in Fig. 2.

2 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 BPCR = 0.1mm

B31 bristles
d=0.1016mm
Default beam
SA NR = 15
integration in space

Figure 2: B31 space beam with circular cross-section, integration Bristles are constrained
points (3 points radially, 8 points circumferentially, 17 in total) at the pinch point.
[12]

Brush seal FE model has also been constructed by using the BACKING PLATE
B31 elements for bristles, as shown in Fig. 3. Unwrapped seal- ROTOR
rotor geometry has been used for this model, and cyclic boundary
conditions have been defined in circumferential direction. Inter-
bristle, bristle-rotor and backing-plate-bristle frictional contacts
have also been defined in this model. Rotor and backing plate
have been modeled as rigid bodies, and backing plate corner
radius has been introduced at FH region.
The pressure profile for a loaded seal is extracted from the 7th
order polynomial fit to experimental data (in seal radial
direction) of Bayley et al. [13] and Turner et al. [14]. Linear
pressure drop is used through the bristle pack in rotor axial
direction (Almost linear pressure drop at fence height region has Cyclic BC is defined in
been reported in the literature [15, 16]). Pressure profile for circumferential direction
brush seal FE models using C3D8I and B31 elements are given
in Fig. 4 and Fig. 5 respectively. ST
Bristles are modeled by using Haynes 25 material properties,
which is typically preferred in most of turbine sealing
applications due to their superior strength and satisfactory Backing plate corner
radius is introduced
ductility (up to 600oC). Cold worked Haynes 25 material with
in the CAE model
10% cold reduction has 725MPa tensile yield strength and
1070MPa ultimate tensile strength limits at room temperature
[17].
Analyses have been conducted by using two CAE models,
one of which has bristles with C3D8I elements and the other one
has B31 bristles. Three different cases have been simulated:
Free-state static rotor interference (unloaded seal), Pressurized
static rotor interference (loaded seal and non-rotating rotor rub) SA
and steady state conditions (pressure load only). Analyses results
are compared with each other as well as with the BTF
measurements on RTR-stiffness test rig given in Figures 6 and 7.
BACKING PLATE

Rotor-bristle friction coefficient of 0.35 used during analyses.

Inter-bristle and bristle pack-backing plate friction coefficients
ST
are taken as 0.25.

Staggered configuration for an unloaded seal

(Used in the FE models)
Figure 3: Brush seal CAE model with B31 elements – CAE model
details

3 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Pressure load profile

Figure 7: RTR-Rotary Test Rig, Details

Figure 4: Brush seal CAE model-C3D8I, Pressure load profile Free-state rotor interference simulations have been run up to
Bristles are constrained 0.6 mm, and BTF (Bristle Tip Force) levels are extracted and
at the pinch point. compared with the Rotary Test Rig stiffness measurements. As it
can be seen from the Fig. 8, results of both simulations show
good agreement with the test data. Please note that the upper
limit for all figures of this study are normalized by dividing the
respective forces by the maximum force measured in all of the
BACKING PLATE

Pressure load profile

tests.
Brush seal structural analyses using C3D8I and B31 elements
have also been conducted at 0.3MPa pressure load and up to 0.4
mm rotor rub. BTF results of those analyses and their
comparison with the RTR-BTF measurements are detailed in
Fig. 9. As it can be seen from Fig. 9, the real testing conditions
have been simulated successfully both in loading and unloading
FH steps with the brush seal model using C3D8I elements, which is
mainly the result of detailed contact modeling as well as the
ROTOR defined pressure load which creates additional bristle moment
profile of real life application. Pressure stiffening, BTF
stabilization and pressurized hysteresis can successfully be
Figure 5: Brush seal CAE model-B31, Pressure load profile simulated with CAE analyses with C3D8I elements. CAE
analyses with B31 elements also give consistent results with tests
in the loading step. However, the CAE model with B31 elements
cannot simulate the unloading step and pressurized hysteresis;
therefore BTF values obtained from B31 analyses deviate from
test results and CAE-C3D8I simulations while rotor is being
pulled back. The reason for zero BTF values in unloading step
of CAE-B31 elements is the applied equivalent line loads instead
of pressure, which result in bristle hung-up at any magnitude of
pressure load, even in very low magnitudes.
Steady state analyses have also been run by using two seal
models, and stress levels are compared with each other.
Simulations have been run under the pressure load of 0.3MPa.

Figure 6: RTR-Rotary Test Rig, Leakage and stiffness

measurement system

4 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 stress in the model is observed at the FH point of downstream
(backing plate) side bristles. The maximum VM stress level is
observed as 125 MPa at that region.

σVM-Row1-PP = 15MPa σVM-Row15-PP = 100MPa

Figure 8: Free-state BTF (normalized) comparison of RTR

measurements and CAE analysis with B31&C3D8I elements
(Rotor bristle friction, µrb=0.35)
ΔP = 0.3MPa
ΔR = 0

σVM-Row15-FH = 35MPa

σVM-Row1-FH = 125MPa
Undeformed
bristle pack

BACKING PLATE Leakage flow

PD PU
Figure 10: Pressure load without rotor interference – CAE-
C3D8I, VM stress profile at ΔP=0.3 MPa

VM stress magnitudes for the brush seal steady state

analyses-B31 elements are detailed in Fig. 11. Similar to C3D8I
model, the maximum VM stress is obtained at the region of
bristles where they have contact with the backing plate.
However, B31 elements approximate lower stress levels (99
Figure 9: Pressurized seal, BTF (normalized) comparison of RTR MPa) when compared with the C3D8I-brush seal model.
measurements and CAE analysis with B31&C3D8I elements
Displacement and stress comparison of C3D8I and B31 brush
(Rotor bristle friction, µrb=0.35)
VM stress magnitudes for the brush seal steady state seal FE analyses at 0.3 MPa pressure load is detailed in Tab. 1.
analyses-C3D8I elements are given in Fig. 10. The pressure As it can be seen from the table, C3D8I elements give higher
causes axial spacing to be closed and initiates the inter-bristle bristle stress levels at the fence height point. Since the localized
contact. The inter-bristle contact at the upstream side (front plate contact of bristle-backing plate can be better simulated with full
side) (last three rows) is localized at the region just above the solid elements (C3D8I), the FE model with C3D8I elements are
bristle tip, and extended up to FH point as moved in the axial expected to give more accurate stress data in comparison with
direction towards to backing plate. Taking the advantage of the the B31 type elements.
solid elements and small mesh size, the stress profile of the
loaded seal can be visualized clearly. The localized contact at the
backing plate corner causes increased contact pressure levels at
that region, where the pressure load of the seal pack is carried by
the backing plate bristles. As a result of this, the maximum VM

5 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 ΔP = 0.3MPa , ΔR = 0 Free State Rotor Interference Simulations
FE Model with B31 (space beam) elements Pre- Post- Output
RAM CPU Simulation
process process size
High stress is observed at
100-
the pinch point of front CAE-B31 8 [email protected] 1-2 days 5-10 min ~0.1 day ~150Kb
150Mb
plate side bristles (Row 15)
CAE- 16
40-45Gb 2-3 days 3-4 days ~0.2 day ~60Gb
C3D8I [email protected]
Steady State Simulations (Pressure load w/o rotor interference)
Pre- Post- Output
RAM CPU Simulation
process process size
150-
CAE-B31 8 [email protected] 2-3 days 5-10 min ~0.1 day ~200Kb
200Mb
PU CAE- 16
PD 50-60Gb 3-4 days 3-4 days ~0.2 day ~80Gb
C3D8I [email protected]
Maximum VM stress is Pressurized Rotor Interference Simulations
observed at the FH Pre- Post- Output
RAM CPU Simulation
region of backing plate process process size
side bristles 150-
CAE-B31 8 [email protected] 2-3 days 15-20 min ~0.1 day ~300Kb
200Mb
CAE- 16
50-60Gb 3-4 days 5-6 days ~0.2day ~120Gb
C3D8I [email protected]
Table 2: CAE methodologies for brush seal analyses – Hardware
and CPU requirements

Figure 11: Pressure load without rotor interference – CAE- Brush seal characterization study will be performed by
B31, VM stress profile at ΔP=0.3 MPa conducting approximately 60-70 analyses, which means pre-
processing periods up to 200 days and unnecessary burden. In
Pressure load without rotor interference for Test Seals order to shorten pre-processing periods, in-house software
Comparison of CAE Analyses and Analytical Studies (Brush Seal Analyzer) has been developed, which construct the
ΔP=0.3 MPa
ABAQUS input file automatically from seal parameters. The
CAE
software shortens the pre-processing time requirements to 1-2
CAE-C3D8I CAE-B31
Max. displacement magnitude 0.42 mm 0.39 mm minutes, which enables efficient FE modeling.
Initial pack thickness 1.68 mm 1.68 mm
Deformed pack thickness 1.34 mm 1.34 mm BRUSH SEAL DESIGN PARAMETERS FOR STIFFNESS
Decrease in pack thickness 25% 25% AND STRESS CHARACTERIZATION
Backing plate side FH point 125.00 MPa 99.20 MPa Brush seal characterization studies have been performed for
(downstream, 1st row) Pinch point ~15 MPa ~7 MPa seal design parameters of bristle diameter, d, free bristle height,
Table 1: Pressure load without rotor interference for Test Seals – BH and cant angle, θ. Effect of those parameters on BTF and
Comparison of CAE analyses and analytical studies stress levels have been examined in free-state rotor excursion,
steady state and pressurized-rotor interference conditions.
SELECTING THE APPROPRIATE FE METHODOLOGY Nominal values for design parameters and friction coefficients
Two different CAE methodologies have been developed are given in Tab. 3. While the effect of specific parameter on
within the content of this study, one of which uses C3D8I brush seal behavior is being analyzed, nominal values are used
elements for bristles, and the second methodology runs the FE for other parameters. Brush seal design parameters are also
model having B31 elements. CAE analyses and correlation visualized in Figures 12 and 13.
studies show that B31 elements show good agreement with BTF
measurements, yet they have shorter simulation times when Brush Seal Design Parameters – Nominal Values
compared with CAE-C3D8I simulations. The hardware Bristle diameter, d [mm] 0.10
requirements and CPU requirements for two different models are Free bristle height, BH [mm] 12.50
summarized in Tab. 2. Fence height, FH [mm] 1.35
Cant angle, θ [deg] 45
As summarized in tabulated values, CAE-C3D8I simulations Bristle density, η [per mm] 80
take 5-6 days while CAE-B31 analyses are completed less than Number of bristle rows, NR 12
an hour. In addition to that, the output file sizes on disk for CAE- Rotor-bristle friction coefficient, µrb 0.30
Inter-bristle and bristle-backing plate friction
C3D8I analyses are at the levels of 50-60GB. On the contrary 0.30
coefficient, µib= µbp
CAE-B31 “.odb” file sizes are as small as several kilobytes.
Other Parameters
Considering that the characterization study requires variety of Bristle elastic modulus, E [MPa] 225000
analyses, and bearing in mind that CAE-B31 simulations show Bristle poisson’s ratio, υ 0.3
good agreement with tests, CAE-B31 FE model is used for brush Backing plate corner radius [mm] 0.10
seal stiffness and friction characterization. Pressure load, ∆P [MPa] 0.30
Table 3: CAE based brush seal characterization – Nominal values
for design parameters

6 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 experience, and larger diameters are usually preferred for
applications that operate under high pressure loads. In order to
improve the pressure load capacity of the brush seal, multi-stage
and multi-layer configurations have been tried [18, 19, 20, 21,
22, 23]. However, due to nonlinear pressure drop between seal
stages for multi-stage applications and introduced bristle locking
phenomenon in multi-layer seal applications, the improvement
in pressure load capacities are limited. Maximum pressure load
capacity for a brush seal is reported as 27 bar (Dinc et al. [24]).
In this section, the effect of bristle diameter on BTF and
stress levels under transient and steady state conditions will be
analyzed through correlated CAE models. Since the axial and
tangential spacing parameters change with the bristle diameter,
the number of bristle rows also changes (if other parameters such
as bristle density, cant angle etc. are kept constant).

Unpressurized rotor interference-BTF. The effect of bristle

diameter on BTF at unpressurized-nonrotating rotor interference
conditions (free-state rotor rub) has been examined by running
the cases detailed in Tab. 4.
Figure 12: Typical brush seal layout
Effect of Bristle Diameter, ΔP=0
Case 1 Case 2 Case 3 Case 4 Case 5
Bristle diameter, d [mm] 0.09 0.10 0.11 0.12 0.13
Number of bristle rows, NR 11 12 13 15 16
PU Max. rotor interference, ∆Rmax [mm] 0.80 0.80 0.80 0.80 0.80
Pinch PS: Nominal values given in Table 3 have been used for remaining parameters.
L
point Table 4: Effect of bristle diameter – Cases for unpressurized
nonrotating rotor interference conditions
PD
BH
Free-state BTF (BTF under unpressurized rotor rub) change
FH
with bristle diameter is given in Fig. 14 at different rotor
interference levels. Second moment of area for the bristle cross-
CANT ANGLE, θ
R section is a 4th order function of the bristle diameter, therefore
Seal inner circle normal required tip force for bending the bristles is higher for larger
SA bristle diameters. Increase in free-state BTF with bristle diameter
is more pronounced at high interference levels.
BACKING PLATE

ST

DETAIL B
Staggered configuration for an unloaded seal
Figure 13: Typical brush seal design parameters

EFFECT OF BRISTLE DIAMETER

Bristle diameter is one of the most important design
parameters, which directly affects BTF and bristle stress levels
at pressurized/unpressurized rotor interference conditions as
well as steady state case. Typical industrial application for Figure 14: Unpressurized rotor interference, - CAE w/ B31
selecting the appropriate bristle diameter strongly depends on the elements, Effect of bristle diameter on BTF (normalized)

7 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Steady state (ΔP=0.3 MPa)–VM Stress. The effect of bristle Pressurized rotor interference-BTF. The effect of bristle
diameter on bristle stress levels for a loaded seal (pressure load diameter on BTF and bristle stress levels under pressurized-
without rotor interference) has been examined by running the nonrotating rotor rub conditions (combined loading) has been
cases detailed in Tab. 5. Stress levels at the most critical section, examined by running the cases detailed in Tab. 6.
which is the FH point of backing plate side bristles (where
downstream side bristles touching the backing plate corner, Effect of Bristle Diameter, ΔP=0.3 MPa
Figures 10 and 11), are examined during characterization study. Case 1 Case 2 Case 3 Case 4 Case 5
Bristle diameter, d [mm] 0.09 0.10 0.11 0.13 0.14
Number of bristle rows, NR 11 12 13 16 17
Effect of Bristle Diameter, ΔP=0.3 MPa
Max. rotor interference, ∆Rmax [mm] 0.35 0.35 0.35 0.35 0.35
Case 1 Case 2 Case 3 Case 4 Case 5
PS: Nominal values given in Table 3 have been used for remaining parameters.
Bristle diameter, d [mm] 0.09 0.10 0.11 0.13 0.14
Table 6: Effect of bristle diameter – Cases for combined loading,
Number of bristle rows, NR 11 12 13 16 17
Max. rotor interference, ∆Rmax [mm] 0 0 0 0 0
∆P=0.3 MPa, ∆Rmax=0.35 mm
PS: Nominal values given in Table 3 have been used for remaining parameters.
Table 5: Effect of bristle diameter – Cases for steady state

Figure 16: Change of BTF (normalized) with bristle diameter,

Pressurized-nonrotating rotor interference-loading step
Figure 15: Change of backing plate side FH point-VM stress levels
with bristle diameter, steady state conditions at ∆P=0.3 MPa Combined loading BTF change with bristle diameter is detailed
in Fig. 16 for ∆P=0.3 MPa. Increase in bristle diameter does not
VM stress levels at the FH point of back plate side bristles alter pressurized-BTF at interference levels of 0.1 mm. However,
and their change with bristle diameter is detailed in Fig. 15 for for higher rotor incursion levels, the inter-bristle interactions and
steady state conditions at ∆P=0.3 MPa. Up to bristle diameter pressure stiffening effect becomes more pronounced with larger
value of 0.11 mm, the maximum VM stress decreases as values of bristle diameter, and pressurized - BTF magnitudes
expected since the cross-sectional inertia increases. However, increase with bristle diameter. It should be noted that the increase
steady state VM stress levels do not decrease continuously with in BTF is more severe for bristle diameter levels above 0.11 mm
increasing bristle diameter, and stabilize after a certain level of (for the selected seal parameters).
bristle diameter. The stabilization point also depends on other
seal parameters, and for the selected nominal values and pressure EFFECT OF FREE BRISTLE HEIGHT, BH
load, this stabilization point for steady state stress level is bristle Free bristle height, “BH”, is the radial distance between the
diameter of 0.11 mm. The reason for the steady state VM stress pinch point and the bristle tip, and it affects both BTF and bristle
stabilization with bristle diameter is the complex contact- stress levels under steady state operating conditions as well as at
frictional inter-bristle and bristle-backing plate interactions. transients. Selecting free bristle height is relevant to bristle
When the pressure load is applied, the overhung portion of the diameter since obtaining appropriate L/d ratios is important in
downstream side bristles bend under the backing plate in the flow determining seal stiffness and wear at operating conditions.
direction, while the portion above FH region deforms in the Although selecting appropriate L/d ratios is the common
reverse direction. This deformation profile is mainly the reason methodology used in the industry, which is mostly based on
for localized contact and high stress peaks at backing plate corner experience on free-state stiffness calculations, CAE based
radius (FH point). For the selected nominal design parameters characterization is required due to the fact that inter-bristle and
and pressure load of 0.3 MPa, the effect of contact localization bristle-backing plate interactions are very effective in
and bristle diameter on FH point stress levels balances itself for
the bristle diameter of 0.11 mm.

8 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 determining stress levels and BTF magnitudes for a pressurized Steady state (ΔP=0.3 MPa)–VM Stress. The effect of BH on
seal. bristle stress levels at steady state conditions (pressure load
without rotor interference) has been examined by running the
Unpressurized rotor interference-BTF. The effect of free cases that have already been detailed in Tab. 6.
bristle height on seal response at unpressurized-nonrotating rotor
interference conditions (free-state rotor rub) has been examined
by running the cases detailed in Tab. 7.

Effect of Free Bristle Height

Case 1 Case 2 Case 3 Case 4 Case 5 Case 6
Free bristle height, BH [mm] 10 11 12 13 14 15
PS: Nominal values given in Table 3 have been used for remaining parameters.
Table 7: Effect of free bristle height – Cases for unpressurized
nonrotating rotor interference conditions

Free-state BTF change with free bristle height is given in Fig.

17 for different rotor incursion levels. Higher BH values result
in longer bristles with high L/d ratios, and causes BTF
magnitudes to decrease. Free-state BTF change with BH is
nonlinear, and decrease in the BTF with bristle free height is
more pronounced at large interference levels.

Figure 19: Backing plate side bristle deformation profiles at

different BH cases, steady state conditions at ∆P=0.3 MPa

Figure 17: Change of BTF (normalized) with free bristle height, As detailed in the previous pages, FH point is critical for
Unpressurized-nonrotating rotor interference-loading step steady state conditions since pressure load causes localized
bristle contact at backing plate corner radius. Therefore, steady
state characterization studies mainly focus on stress magnitudes
at that critical section. VM stress levels at FH point of backing
plate side bristles and their change with bristle free height is
detailed in Fig. 18 for steady state conditions at ∆P=0.3 MPa. As
it can be seen from the figure, VM stress level decreases as BH
takes larger values, which can be better explained by examining
the downstream side deformation profile (Fig. 19). As detailed
in the previous chapters with advanced CAE analyses, FH point
stress peaks under pressure load are mainly due to localized
contact induced by different bending characteristics of overhung
portion (aft bending) and bristle portion above FH (forward
bending). Keeping the FH same, the bristle length above the FH
regions become longer with BH, and forward bending as well as
Figure 18: Change of backing plate side FH point-VM stress levels the aft bending decreases, which in turn cause less severe
with free bristle height, steady state conditions at ∆P=0.3 MPa localization of contact. As a result of this, difference between aft
and forward bending decreases for larger BH values, which in

9 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 turn result in reduction in contact localization and therefore VM angle on pressurized/unpressurized conditions has been analyzed
stress levels under pressure load. Please note that this case might by using correlated CAE-models. Keeping the other parameters
be different for different nominal values for the seal or at constant, the number of bristle rows is affected by the cant angle,
different pressure levels, which requires the inspection of cross- which is due to change in tangential spacing between bristles.
effects.
Unpressurized rotor interference-BTF. Effect of cant angle on
Pressurized rotor interference-BTF. The effect of bristle free seal response at unpressurized-nonrotating rotor interference
height on BTF and bristle stress levels at pressurized-nonrotating conditions (free-state rotor rub) has been examined by running
rotor rub conditions (combined loading) has been examined by the cases detailed in Tab. 8.
running the BH cases detailed in Tab. 6.
Combined loading BTF change with free bristle height is Effect of Cant Angle
given in Fig. 20 for ∆P=0.3 MPa. As it can be seen from the Case 1 Case 2 Case 3 Case 4
Cant angle, θ [degree] 35 40 45 50
figure, the BTF values increase with the free bristle height. The
Number of bristle rows, NR 11 11 12 13
BTF response of the pressurized seal is the combination of free- PS:Nominal values given in Table 3 have been used for remaining parameters
state BTF and the resistance coming from the inter-bristle and Table 8: Effect of cant angle – Cases for unpressurized nonrotating
bristle-backing plate contacts with friction. The free-state BTF rotor interference conditions
decreases with BH. On the other hand, the pressure stiffening due
to inter-bristle and bristle-backing plate interactions increases as Free-state BTF change with cant angle is given in Fig. 21 for
BH gets larger values, which is mainly due to extended loaded different rotor interference levels. The BTF decreases with
length at backing plate region. Considering the pressure increasing cant angle, and the effect of cant angle on BTF is more
stiffening effect is dominant at combined loading conditions pronounced at small “θ” values. This is consistent with the
under 0.3 MPa pressure load, it is expected to obtain higher BTF analytical methodologies developed in the literature [1], where
values for pressurized seal when higher BH values are used. The tip force is inversely proportional to the sine and cosine of cant
increase in BTF-combined loading conditions is more angle.
pronounced at low interference levels, where the contribution of
free-state tip force to total BTF response is minimal.

Figure 21: Change of BTF (normalized) with cant angle,

Unpressurized-nonrotating rotor interference-loading step
Figure 20: Change of BTF (normalized) with free bristle height,
Pressurized-nonrotating rotor interference-loading step Steady state (ΔP=0.3 MPa)–VM Stress. The effect of cant
angle on bristle stress levels at steady state conditions (pressure
EFFECT OF CANT ANGLE load without rotor interference) has been examined by running
Cant angle, “θ”, is defined as the angle between bristle axis the cases in Tab. 7. FH point stress levels of downstream side
and normal of seal inner circle. As a result of angular contact bristles are examined, which has already been identified as
with rotor, the radial interference bends bristles rather than critical section for a loaded seal due to pressure induced localized
buckles them, which in turn enables seal to survive at long contact at the backing plate corner. The maximum VM stress and
operating periods without bristle loss and performance its change with cant angle is detailed in Fig. 22 under pressure
degradations. However, improper cant angle values may result in load of 0.3 MPa. VM stress levels at FH point of a loaded seal
increased free bristle length and extended frictional contact increases with cant angle due to extended loaded length (while
surfaces under pressure load. In this section, the effect of cant keeping other parameters constant, except the NR change due to

10 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 cant angle). The change in number of bristle rows is also CONCLUSION
effective in determining the VM stress level change with cant Stiffness and stress characterization of brush seals have been
angle (since pressure load per bristle changes with NR). The conducted through correlated CAE models. Characterization
increase in steady state VM stress is more severe at larger cant studies examines the effect of bristle diameter, the free bristle
angle values for the selected nominal seal parameters. Please height and the cant angle on bristle tip force levels and stress
note that the behavior of a seal might be different for different magnitudes during rotor rub as well as under steady state
nominal values for the seal or under different pressure levels, operating conditions. Characterization results give better
which requires more analyses that includes cross-effects between understanding of nonlinear brush seal response to different
the parameters. parameters.
Correlation studies have been conducted by using two
different FE modeling approaches, one of which has C3D8I type
hexahedral elements, and the other one uses B31 type space
beam elements. The contacts and frictional effects have been
included to both models by using the same contact methodology.
Correlation studies are based on comparison of bristle tip force
data of stiffness tests, which have been conducted on a custom
test rig design. The correlation studies show that both FE models
give consistent results with the test measurements under free-
state rotor rub. The FE model with C3D8I elements show good
agreement with the pressurized-rotor rub BTF test data both in
loading and unloading steps. The CAE simulations with B31
elements also show good agreement with the pressurized rotor
rub BTF measurements in the loading step. The VM stress
profiles for bristles are similar for both CAE models, but the
magnitudes are different since the contact localization at the
backing plate corner radius can be better simulated with the FE
model with C3D8I elements.
Figure 22: Change of backing plate side FH point-VM stress levels
The effect of bristle diameter on seal stiffness and bristle
with cant angle, steady state conditions at ∆P=0.3 MPa
stress levels are inspected by using the correlated CAE models.
The free state-BTF magnitudes increases with the bristle
Pressurized rotor interference-BTF. The effect of cant angle
diameter as expected, and the effect of the bristle diameter on
on BTF and bristle stress levels at pressurized-nonrotating rotor
free state-stiffness is more pronounced at higher interference
rub conditions (combined loading) has been examined by
levels. At 0.1mm rotor rub and under pressure load of 0.3 MPa,
running the cant angle cases given in Tab. 7. Combined loading
the bristle tip force levels are not altered with the bristle diameter
BTF change with cant angle is detailed in Fig. 23 for ∆P=0.3
(for the selected seal parameters). The BTF levels under 0.3
MPa. As the cant angle is increased, the BTF for a loaded seal
MPa pressure load have an increasing trend with increasing
decreases, which is a direct result of having smoother contact
bristle diameter for the radial interference levels above 0.1mm.
with rotor at larger cant angle values.
The seal stress levels have also been analyzed for steady state
operating conditions having the pressure load of 0.3 MPa.
Nonlinear FE simulations show that the steady state VM stress
levels do not decrease continuously with increasing bristle
diameter due to localized bristle-back plate contact. For the
selected nominal seal parameters and under pressure load of 0.3
MPa, the stress level decreases as the bristle diameter is
increased up to 0.11mm, and stress stabilizes for larger
diameters.
Characterization studies show that the increase in free bristle
height causes free state-bristle tip force levels to decrease while
it results in higher BTF levels under pressure load of 0.3 MPa.
The increase in pressurized BTF levels is related to increased
pressure stiffening effect for the selected nominal seal
parameters. Bristle stress levels have also been examined under
steady state conditions of 0.3 MPa. For the selected nominal
design parameters and under 0.3 MPa pressure load, the stress
levels decrease with increasing free bristle height, where the
Figure 23: Change of BTF (normalized) with cant angle,
Pressurized-nonrotating rotor interference-loading step
main reason for the decrease is related to the difference between

11 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 aft and forward bending and resulting change in contact AIAA/ASME/SAE/ASEE Joint Propulsion Conference &
localization. Exhibit, AIAA 2009-5168.
CAE analyses for different cant angle values show that bristle [8] Crudgington P F, Bowsher A., Jas W., “Brush seal
tip force levels under rotor rub decrease with increasing cant hysteresis”, AIAA2012-4003.
angle both for pressurized and unloaded seal conditions. The [9] Guardino, C., Chew, J.W, “Numerical Simulation of 3D
smoother contact can be stated as the main reason for the Bristle Bending in Brush Seals”, ASME Turbo Expo,
decrease in the tip force levels under pressure load. At steady GT2004-53176, 2004.
state pressure load conditions (0.3 MPa), the stress magnitudes [10] Lelli, D., Chew, J.W, Cooper, P., “Combined 3D Fluid
tend to increase with the cant angle due to extended loaded Dynamics and Mechanical Modelling of Brush Seals”,
length. ASME Turbo Expo, GT2005-68973, 2005.
Analyses and correlation studies of this paper shows that the [11] Pekris, M. J., Franceschini, G., Gillespie, D.R.H., “Effect of
stress levels and stiffness behavior of brush seals under pressure Geometric Changes in an Idealised Contacting Brush Seal
load are strongly affected by nonlinearities and require nonlinear Bristle Pack on Typical Key Performance Measures”,
CAE analyses to obtain high fidelity. The FE based approach can Proceedings of ASME Turbo Expo, 2011, GT2011-46492.
also be utilized for determining the optimum seal parameters for [12] Anonymous, Abaqus Analysis User’s Manual, Abaqus
harsh turbine conditions. Characterization studies of this paper 6.9.EF1.
focus on the main factors for the selected parameters, and further [13] Bayley, F. J. and Long, C.A., “A Combined Experimental
analyses will be useful for examining the cross effects between and Theoretical Study of Flow and Pressure Distributions in
design parameters and their influence on stiffness and stress a Brush Seal”, ASME Trans. J.Eng. for Gas Turbines and
levels. Power, 115, pp 404-410, 1993.
[14] Turner, M. T., Chew, J.W. and Long, C.A., “Experimental
ACKNOWLEDGEMENT Investigation and Mathematical Modeling of Clearance
This study has been conducted within the content of Tubitak Brush Seals”, ASME Gas Turbine and Aero-engine
supported Brush Seal Characterization project “TEYDEB 1507, Congress Paper ASME 97-GT-282, 1997.
7130891”. The authors would like to thank to Tubitak for their [15] Duran, E. T., “Analysis of Shear Heating and Effect of
support on this project. Temperature Dependent Viscosity on Hydrodynamic Lift of
Oil Brush Seals”, MSc Thesis, Sabanci University, 2006.
REFERENCES [16] Bhate, N., Thermos, A. C., Aksit, M. F., Kizil, H., “Non-
[1] Flower, R., “Brush Seal Development System”, Metallic Brush Seals for Gas Turbine Bearings”,
AIAA/ASME/SAE/ASEE 26th Joint Propulsion Proceedings of ASME Turbo Expo 2004, GT2004-54296,
Conference, AIAA Paper 90-2143. 2004.
[2] Stango, R. J., Zhao, H., Shia, C. Y.,“Analysis of Contact [17] http://www.haynesintl.com/literature.htm
Mechanics for Rotor-Bristle Interference of Brush Seals”, [18] Patent No: US5192084A, “Brush Seal Arrangement”
Journal of Tribology, Vol. 125, 414-421, ASME [19] Kazuhiko, T., “Multi-Stage Brush Seal”, Europen Patent
Transactions, 2003. Application, EP 2 305 956 A2, 2011.
[3] Zhao, H., Stango, R. J., “Effect of Flow-Induced Radial [20] Zhebg, X., “Hybrid Multistage Brush Seal”, WO
Load on Brush Seal/Roto Contact Mechanics”, Journal of 2010/076636 A1, 2008.
Tribology, Vol. 126, 208-215, ASME Transactions, 2004. [21] Kazuhiko, T., “Sealing Apparatus with Multistage Brush
[4] Aksit, M. F., “A Computational Study of Brush Seal Contact Seal”, US 2011/0072831, 2011.
Loads with Friction”, Thesis Project for Doctor of [22] Patent No: US5201530A, “Multi-Layered Brush Seal”
Philosophy, Rensselaer Polytechnic Institute, Troy, New [23] Patent No: CN201671661U, 2010.
York. [24] Dinc, S., Demiroglu, M., Turnquist, N., Mortzheim, J.,
[5] Crudgington, P.F., Browsher, A., “Brush Seal Pack Goetze, G., Maupin, J., Hopkins, J., Wolfe, C., Florin, M.,
Hysteresis”, American Institute of Aeronautics and “Fundemental Design Issues of Brush Seals for Industrial
Astronautics, AIAA-2002-3794, 2002. Applications”, ASME Journal of Turbomachinery, April
[6] Crudgington, P.F., Browsher, A., “Brush Seal Blow Down”, 2002, Vol.124/293.
American Institute of Aeronautics and Astronautics, AIAA-
2003-4697, 2003.
[7] Crudgington, P., Browsher, A., Lloyd, D., Walia, J., “Bristle
Angle Effects on Brush Seal Contact Pressures”, 45 th

12 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use