MEEN 617 - April 2008

An example of system
parameter identification
(Hybrid Brush Seal)
Luis San Andrés (lecturer)
Thanks to Adolfo Delgado, José Baker (RAs) &
support from Siemens Power Generation
 Experimental Facility
Electromagnetic
Structural parameters
Supply pressure
Supporting
springs
Rotor Stinger Shaker Kshaft= 243 lbf/in (42.5 kN/m)
inlet
Ms+d= 9.8 lb (4.45 kg)
Eddy current
sensosrs ζ: 0.01 % (damping ratio)
Pressure
Vessel
Installation:
Flexible
DC Motor 6.550” diameter brush seal
coupling Max. air Pressure: 60 psig
Shaker (20 lb max)
Quill shaft

cm 10
20
30 40
50 60 High pressure air
70
80 90 100
Eddy current
sensor

Roller
Test Rig: Rotordynamic Configuration bearing
assembly Flow
Spring
Disk

Ball bearing

Shaft Flexible
Flow
coupling to
motor

Hybrid brush

GT2008-50532 Detail view of brush seal test rig seal

Brush Seals
Reduce secondary leakage in turbomachinery
Replace labyrinth seals in HP TM (hot side of steam & gas turbines)
Wear and thermal distortions are a reliability problem

Hybrid Brush Seals

Novel improvement over BS. Reduce more leakage and do not
introduce wear or thermal distortion. Allow bi-directional rotation

Bristle Bed
Back Plate

Pad or shoe

High Pressure Side Low Pressure Side

* Close-up courtesy of Advanced Spring Lever
Technologies Group, Inc.
Courtesy of Advanced Turbomachinery Group® Mechanism
Dynamic Load Tests (no shaft rotation)

Fext x
Ks

μs,γs Fext Keq

Meq
L L f Ls Ceq
Lf =244 mm x
Lf =221 mm
L= 248 mm
z
Equivalent Test System

x + Keq x + Ceq x = Fext

Meq 
Parameter Identification (no shaft rotation)
ASME DETC2005-84159

iωt iω t
x = xe F = Fext e Harmonic force
& displacements
F
Z= = ( K eq − ω 2 M eq ) + i ω C eq Impedance Function
x

W = ∫ Fext x dt Work External

2
E dis = πω C eq x
Viscous Dissipation

DRY
2 FRICTION &
Edis = γ eqπK eq x + 4μ F x STRUCTURAL
DAMPING
HBS Dynamic Stiffness vs. Frequency
(no shaft rotation)
Load = 63 N, frequency: 20-100Hz
Pressure ratio (Pr = Ps/Pd) = Discharge/Supply
Model
400
reproduces real
M ωω
Keq K−–M
300 2 2
part of the
eq eq
eq impedance under
[kN/m]

200
(F/X) [kN/m]

100 the given supply

0
pressure
ReRe(F/x)

TESTs Model conditions.

-100
Model Test Data
-200
Pr = 1.7 Pr = 1.7
-300 Pr = 2.4 Pr = 2.4
-400 Pr = 3.0 Pr = 3.0

0 20 40 60 80 100 120
Frequency [Hz]
Frequency [Hz]
 HBS Equivalent Viscous Damping vs. Frequency
(no shaft rotation)
Load = 63 N, frequency 20-110Hz γ eq K eq 4 μ F
Pressure ratio (Pr = Ps/Pd) = Discharge/Supply Ceq = +
ω πω x
10000
Equivalent damping
Test Data
Test Data increases slightly with
Pr = 1.7
[N-s/m]

Pr = 2.4 pressure differential.

Pr = 3.0
Results typical of a
(C eq), [N-s/m]

ΔP +
system with dry-friction
Log(Ceq)

1000
& material damping
energy dissipation
Log

Viscous Model

100
0 20 40 60 80 100 120
Frequency, [Hz]
Frequency [Hz]
Identification of Rotordynamic Force Coefficients
Kyy, Cyy
Eddy current
Electromagnetic sensors Kyx, Cyx
Shaker
Load Cell
Y Kxx, Cxx

Y
X

X
Disk Kxy, Cxy
Ω
Stinger HBS

Excitation force
(frequency ω)

⎡Mxx Mxy⎤⎧x⎫ ⎡Kxx Kxy⎤⎧x⎫ ⎡Cxx Cxy⎤⎧x⎫ ⎛Fx ⎞ ⎛Fix ⎞

⎢M M ⎥⎨⎬+⎢K K ⎥⎨ ⎬+⎢C C ⎥⎨⎬ =⎜⎜ ⎟⎟+⎜⎜F ⎟⎟
⎣ yx yy⎦⎩y⎭ ⎣ yx yy⎦⎩y⎭ ⎣ yx yy⎦⎩y⎭ ⎝ 0 ⎠ ⎝ iy ⎠

Imbalance forces (1X=Ω)

Identification of Rotordynamic Force Coefficients
Kyy, Cyy
For periodic force excitation:
Kyx, Cyx
iω t
F =Fe
iωt
y = ye
Y Kxx, Cxx
X
x x iωt
x = xe .
Kxy, Cxy

EOMS reduce to:

With impedances:
Z xx⋅ x + Z xy ⋅ y = Fx
Z αβ = {K αβ − M αβ ω 2 + iCαβ ω},αβ = x , y
Z yx⋅ x + Z yy ⋅ y = 0

For centered operation

(axi-symmetry)
Zxx = Zyy Zxy = -Zyx
⎡ M xx 0 ⎤ ⎧ x ⎫ ⎡ K xx K xy ⎤ ⎧ x ⎫ ⎡C xx 0 ⎤ ⎧ x ⎫ ⎛ Fx ⎞
⎢ 0 ⎥ ⎨ ⎬+ ⎢ ⎥ ⎨ ⎬+ ⎢ ⎥ ⎨ ⎬=⎜ ⎟+
⎣ y ⎭ ⎣ − K xy
M xx ⎦ ⎩  K xx ⎦ ⎩ y ⎭ ⎣ 0 C xx ⎦ ⎩ y ⎭ ⎝ 0 ⎠
Rotor mode shapes
Measured Mode Shapes (one end fixed)
First Fixed-Free Mode Shape Plot (one end fixed)
Hybrid BrushBrush
Hybrid Seal Rotor/Shaft
Seal Rotor Assembly
Assembly Only first mode excited
1.5 Measured (Nat. Freq. 20 Hz) Prediction (Nat. Freq. ~21 Hz)
Measured (Nat. Freq. 20 Hz) Prediction (Nat. Freq. ~21 Hz)
in rotor speed range (0-
1
1200 rpm)
Normalized Displacement

0.5

0
0.000 0.050 0.100 0.150 0.200 0.250 0.300

-0.5

FIRST MODE
-1
Axial Location [m.] SecondMeasured
Fixed-Free
ModeMode
ShapesShape Plot
(one end (one end fixed)
fixed)
Hybrid Brush Seal Rotor Assembly
Hybrid Brush Seal Rotor/Shaft Assembly
1.5 Measured- (Nat.
Measured - (Nat. Freq.
Freq. 144
144 Hz)Hz) Prediction
Prediction - (Nat.
- (Nat. Freq.Freq.
151.5151.5
Hz) Hz)

1
Normalized Displacement

0.5

0
0.000 0.050 0.100 0.150 0.200 0.250 0.300

-0.5

SECOND MODE
-1
Axial Location [m.]
Effect of rotor speed on rotor-HBS natural frequency

Location of
Gyroscopic effects
0.1 displacement
measurements
negligible for test
Disk
0.075
rotor speeds
[m]
[m]

0.05 Roller bearing

Shaft (600 and 1,200 rpm
Radius, [m]

support
Radius
ShaftRadius,

0.025

0
[20 Hz])
-0.025
Shaft

Location of
Shaft

-0.05 external force

Hybrid Brush
-0.075 Seal location

-0.1
0 0.05 0.1 0.15 0.2 0.25 0.3
Axial Location, [m]
Axial Location, z [m]
Axial Location [m]
Rotor 1st Backward Nat. 1st Forward Nat. 2nd Forward Nat. 3rd
Forward Nat.
Speed [RPM] Frequency, [Hz] Frequency, [Hz] Frequency, [Hz] Frequency, [Hz]
0 30.5 30.5 146 1351
600 29.7 31.4 154 1351
1200 28.8 32.2 163 1351
T=32 Hz
 CROSS-Coupling Effects under rotation
Load=22 N, 600 rpm

1000
[μm]

32 Hz
22 N
X - Displacement [um]

750
For load along X
Due to excitation Synchronous with speed (1X)
direction,
X-displacement

500 rotor principal (X)

250
motions

0
40 40 120 200 280 360 440
>>>
Frequency [Hz]
Frequency [Hz] cross (Y) motions
1000
[μm]
Y - Displacement [um]

750 22 N 3X motions
Synchronous with speed (1X)
Y-displacement

always small
500 32 Hz
Due to excitation

250

0
40 40 120 200 280 360 440
Frequency [Hz]
Frequency [Hz]
Test dynamic stiffness vs. frequency
Load = 22 N, frequency 25-80Hz Fx ⋅ x
Pressure ratio (Pr = Ps/Pd) = Discharge/Supply Z xx= 2
(x + y2 )
400

(K )
Model
300
−ω M
K - Mxxω
xx xx
22
xx reproduces the
[kN/m]

200
TE
measured real
(Zxx) [kN/m]

100
ST
s
part of
ReRe(Zxx1),

100
Model Test Data impedance.
600 rpm, Pr = 1.7 Model
200
1200 rpm, Pr = 1.7 Little effect of
600 rpm, Pr = 2.4
300
1200 rpm, Pr = 2.4 pressurization
400
0 20 40 60 80
Frequency, [Hz]
Frequency [Hz]
Test impedance (imag) vs. frequency
Load = 22 N, frequency 25-80Hz y
Pressure ratio (Pr = Ps/Pd) = Discharge/Supply Z yx= −Z yy
x
150
Test Data
600 rpm, Pr = 1.7
Ima (Zxx) [kN/m]

1200 rpm, Pr = 1.7 Viscous Model

600 rpm, Pr = 2.4
[kN/m]

100
[kN/m]

1200 rpm, Pr = 2.4

Damping is
(Zxx1),
Im(F/X),

TESTs NOT viscous

Im

50

0
0 20 40 60 80
Frequency [Hz]
Frequency, [Hz]
Frequency [Hz]
Equivalent Viscous Damping (Cxx~Ceq) vs. Frequency

600 rpm
Damping decreases
with frequency, with
little effect of supply
pressure. Minimum
value at test system
natural frequency
1200 rpm (~32 Hz)

Pr=1.7 Pr=2.4
HBS predicted & test direct stiffness vs. frequency
Frequency 25-100 Hz
Predicted HBS stiffness
(Ksxx) drops slightly in
140
range from 20- 100 Hz.
[kN/m]

130
600yyrpm Tests show nearly constant
Ksxx
Ksxx = Ksyy, Pr = 1.7
Coefficients [kN/m]

Ksxx = Ksyy, Pr = 2.4

120
Coefficients

KHBS Measured
K P Stiffness,
10 Ks

110

100
Pressure (Pr = Ps/Pd) has
Stiffness

90
1,200 rpm
negligible effect on seal
Stiffness

80 Ksxx = Ksyy, Pr = 1.7

Ksxx = Ksyy, Pr = 2.4
direct stiffness, Ksxx
70 HBS Measured Stiffness, Ks

60
0 20 40 60 80 100
Frequency [Hz]
Frequency [Hz]

ASME GT2004-53614 Code: XLTPGASBEAR

HBS predicted & test cross stiffness vs. frequency
Frequency 25-100 Hz
y
Z yx = − Z xx
4 .10
4
x
Pr = 2.4, Zxy =- Zyx
1200 rpm HBS cross stiffness
2 .10
4
(Ksxy) << direct
[N/m]
Re(Zxy) [N/m]

stiffness (Ksxx)
Stiffness Coefficient

0
Pressure (Pr = Ps/Pd) has
600 rpm
negligible effect on seal
2 .10
4
Test Data cross stiffness, Ksxy
Model

4 .10
4
0 20 40 60 80
Frequency
Frequency [Hz]
[Hz]
Zxy= Zyx (Test Data)
 HBS predicted & test damping vs. frequency
Frequency 25-100 Hz Pressure ratio (Pr = Ps/Pd) = Discharge/Supply

1200 rpm 1
1200 rpm
Pr = 1.7
1 Pr = 1.7 600 rpm Cxx = Cyy, γ = 0.25
1,200 rpm
0.9 Cxx = Cyy, γ = 0.25 0.9
Cxx = Cyy, γ = 0.55

Damping Coefficients [kN-s/m]

Cxx = Cyy, γ = 0.55
Damping Coefficients [kN-s/m]

0.8 Equivalent Viscous Damping, Ceq

0.8 Equivalent Viscous Damping, Ceq
Cxx = Cyy = 0 25
Cxx = Cyy = 0 25
0.7
0.7
Increasing loss factor (γ) 0.6 Increasing loss factor (γ)
0.6
Increasing loss factor γ)
Pr =(2.4
γ ) = 2.4
Increasing loss factor (Pr 0.5 q p g, q
Cxx = Cyy, γ = 0.25
0.5 q p g, q
Cxx = Cyy, γ = 0.25 0.4 Cxx = Cyy, γ = 0.55
0.4 Cxx = Cyy, γ = 0.55
Equivalent Viscous Damping, Ceq
Equivalent Viscous Damping, Ceq 0.3
0.3
0.2
0.2
0.1
0.1
0
0
0 20 40 60 80 100
0 20 40 60 80 100
Frequency [Hz]
Frequency [Hz]

HBS direct damping (Csxx) decreases with excitation frequency. Loss factor
coefficient (γ) models well seal structural (hysteresis) damping
 Conclusions
• A structural loss factor (γ) and a dry friction
coefficient(μ) effectively characterize the energy
dissipation mechanism of a Hybrid Brush Seal (HBS).
• HBS Direct stiffness (Ksxx = Ksyy) decreases minimally
with rotor increasing rotor speed for Pr = 1.7 and 2.4 HBS
Cross-coupled stiffness (Ksxy = -Ksyx) is much smaller than the direct
stiffness coefficients.
• HBS Direct viscous damping coefficients decrease as a
function of increasing excitation frequency.