Fabrication and Testing of a Spar-Actuated Active
Compressor Rotor Blade
by
4
Yiben Lin
B.E., Aerospace Engine, Beijing University of Aeronautics and Astronautics,
Beijing (1998)
MASSA CHUSETTS INSTITUTE
OFTECHNOLOGY
Submitted to the Department of Aeronautics and Astronautics
SE P 1 0 2003
in partial fulfillment of the requirements for the degree of
LIBRARIES
MASTER OF SCIENCE IN AERONAUTICS AND ASTRONAUTICS
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
May 2003
© Massachusetts Institute of Technology 2003. All rights reserved.
..........----------
Author..................................
Department of Aeronautics and Astronautics
ay 23, 2003
Certified by ..............................
James D. Paduano
Principal Research Enjeer of Aeronautics and Astronautics
Thesis Supervisor
Accepted by ............
V
Edward M. Greitzer
H.N. Slater Professor of Aeronautics and Astronautics
Chair, Committee on Graduate Students
AERO
1
Fabrication and Testing of a Spar-Actuated Active
Compressor Rotor Blade
by
Yiben Lin
Submitted to the Department of Aeronautics and Astronautics
In partial fulfillment of the requirements for the degree of
Master of Science in Aeronautics and Astronautics
Abstract
This thesis describes the fabrication and testing of a spar-actuated active
compressor rotor blade. This blade can be actuated using the piezos bonded on the spars.
Hence, the steady shape and dynamic response of the blade can be change. It is an
experimental apparatus for obtaining a database on flutter aerodynamics. The active
blade was constructed by bonding graphite spars to a high strength-to-weight foam blade.
The graphite spars were actuated by piezos bonded on the roots of the spars. The foam
blade has the aerodynamic shape of a GE 90 Fan C rotor blade. Strain gages were
attached to the actuators to measure the root strains. Bending mode vibration was
obtained by actuating the two graphite spars in phase; twisting mode vibration was
obtained by actuating the two graphite spars out of phase. Both modes of vibration can
be excited to investigate flutter aerodynamics.
Various technologies were developed to fabricate the geometrically accurate
active blade. Then the blade was tested for rotation speed up to 5000 RPM with a
vacuum level of 500 micro Torr to 1 milli Torr.
Experiments were conducted before and after the bonding of the foam blade.
Four sets of results are presented. 1) Bench top tests of the graphite spars (Bending,
Twisting); 2) In-situ tests of the graphite spars (Bending, Twisting); 3) Bench top tests of
the assembled blade (Bending, Twisting); 4) In-situ tests of the assembled blade
(Bending, Twisting). The graphite spars and the assembled blade were characterized for
their ability to perform broadband excitation experiments.
For the bench top tests of the graphite spars (Bending and Twisting), the results
are the following: a) For the bending mode, the bandpass bandwidth over which at least
0.5 mm (Peak to Peak) tip bending is maintained 249 Hz, and the bandwidth over which
at least 0.25 mm (Peak to Peak) tip bending is maintained 342 Hz. b) For the twisting
mode, the bandpass bandwidth over which at least 0.5 degree twist is maintained 264 Hz,
and the bandwidth over which at least 0.25 degree twist is maintained 354 Hz.
For bench top tests of the assembled blade (Bending, Twisting), the results are the
following: a) For the bending mode, The bandpass bandwidth over which at least 0.5
mm (Peak to Peak) bending is maintained 38 Hz, and the bandwidth over which at least
0.25 mm (Peak to Peak) bending is maintained 249 Hz. b) For the twisting mode, the
bandpass bandwidth over which at least 0.5 degree twist is maintained 12 Hz, and the
bandpass bandwidth over which at least 0.25 degree twist is maintained 29 Hz.
Thesis Supervisor: Dr. James D. Paduano
Title: Principle Research Engineer, Department of Aeronautics and Aeronautics
3
4
Acknowledgements
I would like to thank my advisor, Dr. James D. Paduano for his encouragement and
guidance during my graduate studies, without his help and advise this thesis would not
exist.
I would like to thank the following staff in the GTL:
Victor Dubrowski, Jimmy Letendre, Jack Costa for coming up with great solutions to all
those problems. Ms Lori Martinez, Ms. Julie Finn, Ms. Mary McDavitt and Ms. Holly
Anderson for their administrative support.
I would like to thank the lab technicians:
John Kane of TELAC and Donald Weiner of Aeronautics and Astronautics student Shop
for their timely help.
I would like to thank my fellow students: Jun Luo, Young Wang, Lixian Liu, Andrew
Luers, Benny Yam, Jhongwoo Peak, Matthew Lackner, Chiang Juay Teo. I would like to
thank my office mates for their friendship, encouragement, and assistance through the
pass two years, Lixian, Neil, Mathieu, Emmanuel, Isabell, caithlin.
Finally and most of all, I wish to thank my wife Tao, my parents Moudao and Aiying, My
three sisters Qiong, Zhang and Fang, and my brother Jijun for their constant support.
5
6
Contents
1
Introduction
17
1.1 Background and Motivation. . . . . . . . . . . . . . . . . . . . . . . . . 17
1.2 The Active Rotor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
1.2.1
Overview of the Active Rotor. . . . . . . . . . . . . . . . . . . . . 19
1.2.2
Spar and Shell Concept. . . . . . . . . . . . . . . . . . . . . .. . . . 20
1.2.3
Bending and Twisting Concept. . . . . . . . . . . . . . . . . . . . 20
1.3 Scope of Thesis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22
2
23
Fabrication of the Active Compressor Rotor Blade
2.1
Obtaining Spars from the Graphite-Epoxy 'Blank'. . . . . . . . . . . . . 24
2.1.1
Problems with Old Cutting Method. . . . . . . . . . . . . . . . . . 25
2.1.2
New Cutting Method . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2 Piezo Package and Wiring Scheme. . . . . . . . . . . . . . . . . . . . . 28
2.3 Bonding Improvement. . . . . . . . . . . . . . . . . .. . . . . . . . . . . 32
2.4 Foam Shaping. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35
2.4.1
Thermoforming. . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.4.2
CNC Machining . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.5 Assembly Procedures . . . . . . .
. . . . . . . . . . . . . . . . . . . . . 38
2.6 Spin Pit Development Work . . . . . . . . . . .. . . . . . . . . . . . . . 41
3
Bench Top and In-Situ Characterization of the Graphite Spar Assembly 45
3.1 Bench Top Characterization. . . . . . . . . . . . . . . . . . . . . . . . . 46
3.1.1
Experimental Setup Summary . . . . . . . . . . . . . . . . . . . . 46
3.1.2
Instrumentation .. . . . . . . . . . . . . . . . . . . . . . . . . . . ..47
3.1.3
Test Procedures and Test Matrix . . . . . . . . . . . . . . . . . . . 55
3.1.4
Results . . . . . . .
. . . . . . . . . . . . . . . . .... . .
7
. . .. 56
3.2 In-Situ Characterization . .
4
. ..
. . . . .. . . . . . . .
. . . . . . . . 67
77
Bench Top and In-Situ Characterization of the Assembled Blade
4.1
Bench Top Tests of the Assembled Blade. . . . . . . . . . . . . . . . . . .77
4.1.1
Experimental Set Up and Test Matrix . . . . . . . . . . . . . . . . 77
4.1.2
Bending Mode Results. . . . . . . . . . . . . . . . . . . . . . . . . 80
4.1.3
Twisting Mode Results . . . . . . . . . . . . . . . . . . . . . . . . 90
4.2 In-Situ Tests of the Assembled Blade . . . . . . . . . . . . . . . . . . . . 97
5
4.2.1
Bending Mode Results. . . . . . . . . . . . . . . . . . . . . .. . .. . 98
4.2.2
Twisting Mode Results. . . . . . . . . . . . . . . . . . . .. . . . . 104
113
Conclusions and Recommendations
5.1
Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . .. .
. . . . . 113
5.1.1
Fabrication of the Spar-Actuated Active Compressor Rotor Blade. . 113
5.1.2
Characterization of the Graphite Spar Assembly and
the Assembled Blade in both Bench Top and In-situ Test . . . . . . 114
5.2 Recommendations for future work. . . . . . . . . . . . . . . . . . . . . 118
. . 121
A
In-Situ Dynamic Balancing Scheme for the Active Rotor . . . . . . .
B
Procedure for Generating Machine Tool Paths by Mastercam. . . . . . . 125
Bibliography. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
8
List of Figures
Figure 1-1: Active Rotor with one composite blade. . . . . . . . . . . . . . . . . . . .19
Figure 1-2: Illustration of the 'spar and shell' concept. . . . . . . . . . . . . . . . . . 20
Figure 1-3: Bending and twisting actuation: (a) Piezo actuators bonded on the
graphite spars (b) bending actuation (c) Twisting actuation. . . . . . . . . 21
Figure 2-1: Graphite-Epoxy 'Blank' . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Figure 2-2: Designed graphite spars in Pro Engineer . . . . . . . . . .
24
. . . . . . . . 25
Figure 2-3: Triangles to be bonded on the Graphite-Epoxy 'Blank'. . . . . . . . . . . 25
Figure 2-4: Obtained accurate projected cutting path. . . . . . . . . . . . . . . . . . . 26
Figure 2-5: Illustration of the attachment of the Graphite-Epoxy 'Blank' to
the mode made by Z Corp. . . . . . . . . . . . . . . . . . . . . . . . . . .27
.
Figure 2-6: Graphite core with cut out spars. . . . . . . . . . . . . . . . ... . . . . .28
Figure 2-7: one set of new lead for making a piezo package. . . . . . . . . . . . . . .30
Figure 2-8(a): Actuators arrangement and wiring on a single spar. . . . . . . . . . . .31
Figure 2-8(b): Actuators arrangement and wiring on graphite
spars for the bending actuation and the twisting Actuation. . . . . . . . 31
Figure 2-9: Aluminum-plastic jig for bonding piezos to lead shapes. . . . . . . . . . .32
Figure 2-10: Illustration of the gap between the piezo and the spar surface. . . . . . . 34
Figure 2-11: Cure cycle used for adding graphite epoxy prepregs to the graphite spars 34
Figure 2-12: Illustration of foam cracking at groove and foam snapping
at maximum curvature due to excess force and limited material
flow during thermoforming. . . . . . . . . . . . . . . . . . . . . . . . . 36
Figure 2-13: Illustration of the piezos to the foam blade. . . . . . . . . . . . . . . . . 38
Figure 2-14: Assembly of the graphite spars to the foam blade. . . . . . . . . . . . ..40
Figure 2-15: Suction side of the assembled blade. . . . . . . . . . . . . . . . . . . . 40
9
Figure 2-16: Pressure side of the assembled blade. . . . . . . . . . . . . . . . . . . . 40
Figure 2-17: Rotor with thermocouple in place. . . . . . . . . . . . . . . . . . . . . .42
Figure 2-18: A bolt with an O-ring. . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Figure 3-1: Experimental setup for determination of transfer functions. . . . . . . . . 47
Figure 3-2: Illustration of the connections and settings for using the audio amplifier. . 48
Figure 3-3: Internal View of Transformer Circuit. . . . . . . . . . . . . . . . . . . . 49
Figure 3-4: Front Panel of Transformer Circuit. . . . . . . . . . . . . . . . . . . . . . 49
Figure 3-5: Diagram of Transformer Circuit. . . . . . . . . . . . . . . . . . . . . . . 50
Figure 3-6: Illustration of the connection of the strain gage amplifier. . . . . . . . . . 51
Figure 3-7: Laser displacement sensor calibration curve. . . . . . . . . . . . . . . . . 53
Figure 3-8: Measured transfer functions from the command signal applied
at piezos to the tip deflection at the leading edge spar (LES) and
the trailing edge spar (TES) . . . . . . . . . . . . . . . . . . . . . . . . . 58
Figure 3-9: Inferred transfer function from the command signal applied
at piezos to the average tip displacement (ATD) . . . . . . . . . . . . . . 58
Figure 3-10: Inferred transfer function from the command signal applied at
piezos to the tip twist . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Figure 3-11: Measured transfer functions from the command signal applied
at piezos to root strains at the leading edge spar (LES) . . . . . . . . . . . 61
Figure 3-12: Measured transfer functions from the command signal applied
at piezos to root strains at the trailing edge spar (TES) . . . . . . . . . . . 61
Figure 3-13: Measured transfer functions from root strains at the leading spar
(LES) to the tip displacement at the leading edge spar (LES). . . . . . . .63
Figure 3-14: Measured transfer function from root strains at the trailing edge
spar (TES) to the tip displacement at the trailing edge spar (TES) . . . . .63
Figure 3-15: Verification the estimated leading edge spar displacement. . . . . . . . .64
Figure 3-16: Verification the estimated trailing edge spar displacement. . . . . . . . .65
Figure 3-17: Inferred transfer functions from root strains at the leading
edge spar (LES) and the trailing edge spar (TES) to the
average tip displacement (ATD) . . . . . . . . . . . . . . . . . . . . . . 66
Figure 3-18: Inferred transfer functions from root strains at the leading edge
10
spar (LES) and the trailing edge spar (TS) to the tip twist . . . . . . . . 66
Figure 3-19: A road map for the result of in-situ tests. . . . . . . . . . . . . . . . . .69
Figure 3-20: Inferred transfer functions from the command signal applied
at piezos to the tip deflection at the leading edge spar by the strain
at the leading edge spar suction side (RPM Tests). . . . . . . . . . . . 71
Figure 3-21: Inferred transfer functions from the command signal applied
at piezos to the tip deflection at the trailing edge spar by the strain
at the trailing edge spar suction side (RPM Tests) . . . . . . . . . . . .72
Figure 3-22: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the leading edge spar by the
strain at the leading edge spar pressure side (RPM Tests) . . . . . . . . 73
Figure 3-23: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the trailing edge spar by the
strain at the trailing edge spar pressure side (RPM Tests) . . . . . . . . 74
Figure 3-24: Inferred transfer functions from the command signal applied at
piezos to the average tip deflection by four strains (RPM Tests) . . . . .75
Figure 3-25: Inferred transfer functions from the command signal applied at
piezos to the tip twist by four strains (RPM Tests) . . . . . . . . . . . .76
Figure 4-1: Experimental set up for bench top tests of the assembled blade. . . . . . 78
Figure 4-2: Two points to measure the deflection by the laser displace sensor. . . . . 80
Figure 4-3: Measured transfer functions from the command signal applied at
piezos to the assembled blade's tip deflection at the leading
edge (LS) and the trailing edge (TS) (Bending Mode) . . . . . . . . . . . 82
Figure 4-4: Inferred transfer function from the command signal applied at
piezos to the average tip displacement (ATD) of the
assembled blade (Bending Mode) . . . . . . . . . . . . . . . . . . . . . . 82
Figure 4-5: Comparison of the average tip displacement of bench top
tests' transfer functions between the graphite spar assembly
and the assembled blade . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Figure 4-6: Measured transfer functions from the command signal applied
at piezos to root strains at the leading edge (LE) of the assembled blade.
11
84
Figure 4-7: Measured transfer functions from the command signal applied
at piezos to root strains at the trailing edge (TE) of the assembled
blade (Bending M ode) . . . . . . . . . . . . . . . . . . . . . . . . . . . .85
Figure 4-8: Measured transfer functions from the root strain at the leading
edge spar (LES) to the tip displacement at the leading edge
(LE) of the assembled blade (Bending Mode) . . . . . . . . . . . . . . . .87
Figure 4-9: Measured transfer functions from the root strain at the trailing
edge spar (TES) to the tip displacement at trailing edge
(TE) of the assembled blade (Bending Mode) . . . . . . . . . . . . . . .. 87
Figure 4-10: Verification the estimated leading edge displacement of
the assembled blade (Bending mode) . . . . . . . . . . . . . . . . . . . .88
Figure 4-11: Verification the estimated trailing edge displacement of
the assembled blade (Bending mode) . . . . . . . . . . . . . . . . . . . .89
Figure 4-12: Inferred transfer function from root strains at the leading
edge spar (LES) and the trailing edge spar (TES) to the
average tip displacement (ATD) of the assembled blade . . . . . . . . . .89
Figure 4-13: Measured transfer functions from the command signal applied
at piezos to the assembled blade's tip deflection at the leading
edge (LS) and trailing edge (TS) (Twisting Mode) . . . . . . . . . . . . 90
Figure 4-14: Inferred transfer function from the command signal applied at piezos
to the Tip Twist (TT) of the assembled blade (Twisting Mode) . . . . . . 90
Figure 4-15: Comparison of the tip twist transfer functions of bench top tests
between the graphite spar assembly and the assembled blade . . . . . . . 91
Figure 4-16: Measured transfer functions from the command signal applied
at piezos to root strains at the leading edge spar(LES) of the assembled
blade (Twisting Mode) . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Figure 4-17: Measured transfer functions from the command signal applied
at piezos to root strain at the trailing edge spar(TES) of the assembled
blade (Twisting Mode) . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Figure 4-18: Measured transfer functions from root strain at the leading edge
spar (LES) to the tip displacement at the leading edge (LE)
12
of the assembled blade (Twisting Mode) . . . . . . . . . . . . . . . . .94
Figure 4-19: Measured transfer functions from root strain at the trailing edge
spar (TS) to the tip displacement at the trailing edge spar(TES) of
the assembled blade (Twisting Mode) . . . . . . . . . . . . . . . . . .94
Figure 4-20: Verification the estimated leading edge displacement of the
assembled blade (Twisting mode).
. . . . . . . . . . . . . . . . . . . .95
Figure 4-21: Verification the estimated trailing edge displacement of the
assembled blade (Twisting mode) . . . . . . . . . . . . . . . . . . . . .96
Figure 4-22: Inferred transfer function from root strains at the leading edge spar
(LES) and trailing edge spar (TS) to tip twist of the assembled blade.
. .
96
Figure 4-23: The assembled blade survived 5000 RPM. . . . . . . . . . . . . . . . . 98
Figure 4-24: Inferred transfer functions from the command signal applied
at piezos to the tip deflection at the leading edge by the strain
at the leading edge spar suction side (RPM Tests of the assembled
blade in Bending M ode) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Figure 4-25: Inferred transfer functions from the command signal applied
at piezos to the tip deflection at the trailing edge by the strain
at the trailing edge spar suction side (RPM Tests of the assembled
blade in Bending Mode) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Figure 4-26: Inferred transfer functions from the command signal applied
at piezos to the tip deflection at the leading by the strain
at the leading edge spar pressure side (RPM Tests of the assembled
blade in Bending M ode) . . . . . . . . . . . . . . . . . . . . . . . . . . . .101
Figure 4-27: Inferred transfer functions from the command signal applied
at piezos to the tip deflection at the trailing edge by the strain
at the trailing edge spar pressure side (RPM Tests of the assembled
blade in Bending M ode) . . . . . . . . . . . . . . . . . . . . . . . . . . . .102
Figure 4-28: Inferred transfer functions from the command signal applied
at piezos to the average tip deflection by four strain gages
(RPM Tests of the assembled blade in Bending Mode) . . . . . . . . . . .103
Figure 4-29: Inferred transfer functions from the command signal applied
13
at piezos to the tip deflection at the leading edge by the strain
at the leading edge spar suction side (RPM Tests of the assembled
blade in Twisting M ode) . . . . . . . . . . . . . . . . . . . . . . . . . . .. 104
Figure 4-30: Inferred transfer functions from the command signal applied
at piezos to the tip deflection at the trailing edge by the strain
at the trailing edge spar suction side (RPM Tests of the assembled
blade in Twisting M ode) . . . . . . . . . . . . . . . . . . . . . . . . . . . .105
Figure 4-31: Inferred transfer functions from the command signal applied
at piezos to the tip deflection at the leading edge by the strain
at the leading edge spar pressure side (RPM Tests of the assembled
blade in Twisting Mode) . . . . . . . . . . . . . . . . . . . . . . . . . . . .106
Figure 4-32: Inferred transfer functions from the command signal applied
at piezos to the tip deflection at the trailing edge by the strain
at the trailing edge spar pressure side (RPM Tests of the assembled
blade in Twisting Mode) . . . . . . . . . . . . . . . . . . . . . . . . . . . .107
Figure 4-33: Inferred transfer functions from the command signal applied
at piezos to tip twist by the four strain gages (RPM Tests of the
assembled blade in Twisting Mode). . . . . . . . . . . . . . . . . . . . . 108
Figure 4-34: A typical burned piezo bonded on the spar. . . . . . . . . . . . . . . . .109
Figure 4-35 Channels of the Slip Ring Wires to conduct signals. . . . . . . . . . . . 110
Figure 4-36: Cables with vacuum passthrough. . . . . . . . . . . . . . . . . . . . . .111
Figure 5-1: Comparison of the average tip displacement of bench top and
in-situ tests transfer functions between the graphite spar assembly
and the assembled blade . . . . . . . . . . . . . . . . . . . . . . . . . . .116
Figure 5-2: Comparison of the tip twist transfer functions of bench top and in-situ
tests between the graphite spar assembly and the assembled blade . . . . .117
Figure A-1: Experimental Set Up for Balancing. . . . . . . . . . . . . . . . . . . . .122
Figure A-2: A diagram used to locate the light spot. . . . . . . . . . . . . . . . . . .123
14
List of Tables
Table 3-1: Actuation power supply specification and set up parameters. . . . . . . . . 48
Table 3-2: Strain gages conditioning amplifier specification and set up parameters
.
. 52
Table 3-3: Nexus conditioning amplifier and accelerometer specification
and set up parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Table 3-4: Laser displacement sensor specification and set up parameters. . . . . . . .53
Table 3-5: Dynamic Signal Analyzer specification and set up parameters. . . . . . . . 54
Table 3-6: Summary of measured transfer functions fro the bench tests
of the graphite spar assembly. . . . . . . . . . . . . . . . . . . . . . . . . 55
Table 3-7: The Inferred transfer functions for the bench top tests of the
graphite spar assembly. . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Table 3-8: The Measured root strain transfer functions for the RPM test
of the graphite spar assembly. . . . . . . . . . . . . . . . . . . . . . . . . .68
Table 3-9: The Inferred transfer functions for the RPM tests of the
graphite spar assembly. . . . . . . . . . . . . . . . . . . . . . . . . . . . .68
Table 4-1: Summary of measured transfer functions of the assembled blade
in bench top tests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .79
Table 4-2: The inferred transfer functions for bench top tests of the
assembled blade. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Table 4-3: The Measured root strain transfer functions for the RPM test of the
assembled blade. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Table 4-4: The Inferred transfer functions for the RPM Tests of the assembled blade. . 97
Table A-1: Typical Rotor Balancing Data. . . . . . . . . . . . . . . . . . . . . . . . 123
15
16
Chapter 1
Introduction
1.1 Background and Motivation
The vibration of rotating blades is one of the main factors limiting the development of
higher performance turbomachinery. Under certain conditions, vibration leads to
pressure perturbations in the fluid, and the fluid provides negative damping to the
structure. Thus, the overall system can become lightly damped or unstable. This
mode of instability is called flutter, which can result in serious damage to
turbomachinery.
In order to address the problem of flutter instabilities, there is a need for reliable
experimental data. Once experimental data are obtained, they would be very helpful
for verifying existing flutter models, and for motivating new models.
A new
experimental concept called the active rotor is being developed to perform such flutter
experiments. The main benefit of the active rotor is its capability of controlling the
motion of individual blades and to change the stiffness and natural frequencies of its
blades. By controlling blades and other operating conditions, it would be possible to
more thoroughly investigate flutter.
The need for the active rotor is motivated
through a description of three potential experiments that would require such a device.
17
The following potential flutter experiments are described in detail in [1].
They are
present here for completeness.
Potential Flutter Experiments
In order to investigate flutter instabilities, three potential experiments that test
different aspects of flutter are described. These experiments could be conducted with
an active rotor.
Experiment I: Investigation of the Effects of Mistuning: Mistuning is the
variation of natural frequencies and mode shapes of individual blades in a rotor,
making it slightly non-axisymmetric structurally.
Mistuning may be introduced in a
rotor as a means of passive flutter control. [1] Mistuning can significantly impact both
the stability and forced response of a compressor rotor.
The experimental
investigation of effects of mistuning parameters is expensive because of the need for
different rotors. However, the active rotor can investigate mistuning in two ways.
First, geometric mistuning may be studied since altering the angle of twist of
individual blades becomes possible. Second, stiffness mistuning can also be done by
applying feedback controls to the piezos to obtain controllable changes in stiffness of
individual blades.
Experiment
II:
Measurement
of
Influence
Coefficients:
Influence
coefficients are defined as the effects of the motion of one blade on another in a rotor.
In general, the motion of a blade will lead to pressure perturbations on other blades.
The characteristics of these perturbations depends on the mode shape of one blade
vibration, the harmonic mode number of the traveling wave of blade vibrations, the
inter blade phase angle, as well as the flow conditions. By using the active rotor, the
motion of some blades may be prescribed and controlled, whereas the motions of
other blades are passively measured.
The generalized forcing function can be
calculated based on measured deflections of passive blades.
Thus, influence
coefficients can be directly measured. Details are given in [1].
Experiment
III:
Measurement
of
the
aerodynamic
damping:
Aerodynamic damping is one of the main parameters used to characterize the
tendency toward flutter.
It is useful to measure and investigate the effect of the
aerodynamic damping on flutter. The following steps can measure the aerodynamic
damping of a blade: First, the blade dynamics can be identified in the presence of flow
18
to measure the total damping, which includes the structural damping and the
aerodynamic damping. Second, the blade may be identified in a vacuum chamber and
a new damping can be calculated. This new damping only includes the structural
damping. The decrement of the calculated damping is the aerodynamic damping.
Currently the experimental investigation of the aerodynamic damping is expensive. A
lot of different blades are needed to investigate the aerodynamic damping over wide
range of frequency because each blade can only measure aerodynamic damping at
certain resonance frequencies.
However, the active blade can measure the
aerodynamic damping over a wide range of frequency by exciting peizos bonded on
the roots of spars, because one of the main features of the active blade is that it can
vibrate in a wide range of frequency. Hence the active blade can effectively measure
the aerodynamic damping over wide range of frequency.
1.2 The Active Rotor
1.2.1
Overview of the Active Rotor
The active rotor is a tool for performing experimental investigations of flutter. Figure
1-1 shows a test apparatus with one active rotor composite blade. The features of the
active rotor concept are that the motion of each blade can be actuated by piezos
bonded on the root of the blade.
Figure 1-1: Active Rotor with one composite blade
19
1.2.2
Spar and Shell Concept
Because the piezos do not develop high enough forces to deform a typical titanium
compressor blade, a 'spar and shell' concept was chosen [3]. Figure 1-2 shows the
'spar and shell' concept. The active blade consists of two parts: A graphite epoxy
core consisting of two spars, and a foam blade (shell). The graphite epoxy core
consisting of two spars can survive the centrifugal force. The high strength-to-weight
ratio foam blade gives the assembled blade its aerodynamic shape.
f.
Graphite core
consisting of two spars
Foam Shell
Assembled blade
Figure 1-2: Illustration of the 'spar and shell' concept
1.2.3
Bending and Twisting Concept
The active rotor uses the 'bending and twisting' concept to actively control the shape
of the blade. Figure 1-3 shows the bending and twisting concept, which enables the
active rotor to investigate both bending mode flutter and twisting mode flutter.
Piezos were bonded on both sides of the leading edge spar and the trailing edge spar.
By correctly applying potentials to piezos with correct polarity, each spar can be
independently deflected.
Actuation of the spars in phase can create an overall
20
bending deformation of the structure, whereas actuation out of phase can create a
twisting deformation.
The deflection goals are 0.5 degree tip twisting for the twisting mode and 0.5 mm tip
bending for the bending mode. The induced pressure perturbation around the blade
by such deflection would be big enough for system identification if these deflection
goals could be met. [12]
Piezo Actuators
Tip Twist
Twisting Actuation
Bending actuation
(b)
(c)
Figure 1-3: Bending and twisting actuation: (a) Piezo actuators bonded on the
graphite spars (b) bending actuation (c) Twisting actuation
21
1.3 Scope of thesis
Chapter 2 describes the fabrication of the active rotor blade.
Chapter 3 describes the bench top and in-situ characterization of the graphite spar
assembly. Actuation testing on the bench was used to verify the proper operation of
the graphite spar assembly, to characterize its non-rotating performance, and to obtain
the transfer functions for estimating tip deflection based on strain gage measurements.
We verified that the estimated tip deflection results are accurate using the laser
displacement sensor. After bench top test, the graphite spar assembly was tested for
its performance in the rotating environment, at various values of RPM.
Chapter 4 describes the bench top and in-situ characterization of the assembled blade.
The results for the assembled blade are presented in this chapter.
Chapter 5 summarizes the work done, and gives an outline for the future work to be
done.
22
Chapter 2
Fabrication of the Active Compressor
Rotor Blade
This chapter describes the fabrication of the active compressor rotor blade. The
fabrication can be divided into the following sections:
*
Obtain graphite spars from the Graphite-Epoxy 'Blanks', which were constructed
previously ([2] and [3])
" Make piezo packages
" Bond piezo packages to the graphite spars
*
Shape a foam blade
e
Assemble the graphite spars to the foam blade
During the process of fabrication, geometric accuracy is critical, since the assembled
blade represents the aerodynamic surface of the blade. The following two improvements
on geometric accuracy were implemented by using appropriate fabrication techniques: 1)
For the fabrication of the graphite spars, a water jet cutter was used instead of manual
machining. 2) For the fabrication of the foam blade, a 3D numerical milling machine
23
was used instead of thermoforming. Both of these fabrication techniques led to good
assembly characteristics and a more accurate geometry of the active rotor blade.
2.1 Obtaining Spars from the Graphite-Epoxy
'Blank'
The Graphite-Epoxy blank, shown in Figure 2-1, was cut to obtain graphite spars. This
blank consists of a complex design and lay up of Hercules' AS4/3501-6 prepreg, molded
and cured to meet the geometric and structural specifications of the active rotor.
Manufacturing details of this blank are given in [2] and [3].
Figure 2-1: Graphite-Epoxy 'Blank'
The geometry of the graphite spars was defined in Pro Engineer as shown in Figure 2-2.
This geometry is 3 dimensional and right now, no three dimension numerically controlled
Graphite-Epoxy cutting tools are available.
24
Figure 2-3: Triangles to be bonded on
the Graphite-Epoxy 'Blank'
Figure 2-2: Designed graphite spars in
Pro Engineer
Furthermore, the blank is composite and standard machining can not be used for cutting
the spars.
2.1.1 Problems with the Old Cutting Method:
The procedure of the old cutting method is summarized as five steps:
1. Design graphite spars geometry in Pro Engineer. (3D)
2. Project the 3D graphite spars geometry to a 2D plane to obtain the cutting path
(2D)
25
3. Discretize the 2D cutting path into small triangle as shown in Figure 2-3. Then
print it on paper (Small triangle patterns are more easily bonded onto the blank).
4. Cut the paper and glue these triangle patterns onto the blank.
5. Use a hand-held cutting tool such as a Dremmel tool to cut out the graphite core
with cut out spars.
There are some problems associated with this method.
Most of them fall into two
categories. The first one is poor accuracy of the cutting path. The other one is hand-held
fitting and cutting methods. Both of these problems lead to poor accuracy geometry of
the graphite spars.
The first three steps of the old procedure are correct. However, in the fourth step, it is not
possible to glue 2D patterns to a 3D blank accurately. Moreover, accurate alignment of
these triangles is not possible. In the fifth step, a hand-held tool is not a good choice to
perform the cutting task, even if an accurate cutting path could be defined.
2.1.2 New Cutting Method
As suggested in [2], a water jet cutter is a potential solution. The procedure of the new
cutting method is summarized as five steps:
e
Design graphite spars geometry (3D) in Pro Engineering as shown in Figure 2-2
e
Project the 3D graphite spars path to a 2D plane to obtain the cutting path (2D) as
shown in Figure 2-4
Figure 2-4: Obtained accurate projected cutting path
26
"
Securely attach the blank to a fast prototyped (3D printed) blank mold made by Z
Corp as shown in Figure 2-5
Figure 2-5: Illustration of the attachment of the Graphite-Epoxy
'Blank' to the mold made by Z Corp
" Use a water-jet cutter to follow the cutting path to obtain the graphite spars.
The following comments should be made:
"
To get accurate cutting results using the water jet cutter, the distance between the
jetting hole to the blank path point should be maintained at less than 5mm.
Otherwise too much graphite material will be cut way and the obtained graphite
spars will be smaller than designed. Hence it is necessary for the operator to stop
the water-jet cutter frequently to lower or enhance the jetting hole positon.
" Double-sided tape and flash tape were used to securely mount the blank to the
mold. Ethanol was used to thoroughly clean the blank and the mold to obtain
better bonding by the double-sided tape.
"
The force exerted by the jet was very large. Securely holding the blank-mold pair
was necessary to guarantee accurate cutting results. Dead weights and clamps
were used to help position.
*
It was found that the new cutting method yielded very precise graphite spars,
based on the assembly of the graphite spars to the corresponding grooves on the
foam blade, which were machined by a 3D numerical milling machine.
27
e
In order to obtain graphite spars from a blank, a corresponding mold is needed to
held and secure the part. Once the blank is cut, the mold is cut also. It cannot be
used again. This mold is made by Z Corporation. The powder used is ZPTM 100
with wax outside to strengthen the mold. The cost is about $300 for each mold.
*
This cutting method is reliable and quick. In all, three blanks were cut. All the
obtained graphite spars are in excellent condition. The machining time for the
cutter is less than 1 hour, including adjusting time.
Figure 2-6 is a graphite core with cut out spars.
Figure 2-6: Graphite core with cut out spars
2.2 Piezo Package and Wiring Scheme
The piezoelectric actuator dimensions were 9mm x 0.254 mm and a length of 36.6 mm
(about 20% of the length of the spar). These dimensions were given in [2]. However,
these dimensions were too large for fitting to the groove on the foam blade. Hence the
piezoelectric actuators were cut using a razor blade to be 7mm x 0.254 mm x 36.6mm.
28
Covering a piezo actuator with an appropriate lead shape made a peizo package. This
peizo package was then bonded to the graphite spars. In order to obtain good actuation
results, the piezo package must be fully bonded to the graphite spars. No slippage or
gaps are allowed, because otherwise the induced strain of the piezo will not be fully
transferred to the graphite spars. From [2], A 3-mil thick Copper-Kapton sheet was
found to be appropriate for making the leads; A commercially available printer was used
to print the designed lead shape. PCB etchant was suitable for etching the lead shape
from the Copper-Kapton sheet. From [3], Epotec 301 was found to be appropriate for
bonding the piezo actuators to the lead shape.
Basically, there are three problems associated with the old lead shape and wiring scheme.
First, the pressure side and suction side actuators were dependent. They were actuated by
the same signal. Hence it was hard to tell which one was the bad one, if one actuator
didn't work. Second, of the four DC voltages applied to the actuators, two were in the
wrong direction. These wrong DC voltages can depole piezos. Hence, the effectiveness
of the actuation can be dramatically reduced.
Third, there is significant geometric
mismatch between the old lead shape and the graphite spars. Hence, piezo actuators
bonded inside of the old lead shape were not fully bonded to the graphite spars. This
problem leads to smaller strain of the spar surface. The smaller strain of the spar surface
leads to the less deflection of the graphite spars or the assembled blade. For the old lead
shape, the two actuators on the leading edge spar use the same lead shape. Hence, it is
difficult to avoid slippage during manufacture. New lead shapes are needed to avoid this
problem.
The new pair of leads for one actuator are shown in Figure 2-7. Every piece of the lead
was made of Copper-Kapton. The copper conducts the command voltage to the actuator
and the Kapton isolates the actuator on the outside. Two pieces of lead and one peizo
strip form a piezo package. Every piezo package connects to a command signal. By this
method every actuator can be excited independently. It is easy to make sure all the
actuators work.
Also the new wiring guarantees that the DC voltages are applied
correctly to prevent depoling.
29
This new lead design can also helps to avoid slippage, because there are no geometric
limitations between the two actuators bonded on the leading edge spar or trailing edge
spar. The four actuators are totally separated. It is easier to bond any of them to the spar
root surfaces.
Figure 2-8(a) illustrates the actuator arrangement and wiring on a single spar. Figure 28(b) illustrates the actuator arrangement and wiring on graphite spars for bending
actuation or twisting actuation. Signal A, B, C and D refer to the AC voltages applied to
piezo actuators.
Piezo
One piezo package
Two Pieces of lead
Figure 2-7: one set of new lead for making a piezo package
30
Signal A
Ground
RED ARROW INDICATES DIRECTION OF POLARITY
PIEZO
SPAR
PIEZO
Ground
Root
Signal B
Figure 2-8(a): Actuators arrangement and wiring on a single spar
Signal C Ground
Spar2
Signal
Ground
Root
Signal B
Figure 2-8(b):
Actuators arrangement and wiring on graphite spars for the
bending actuation and the twisting actuation
31
For the bending actuation, signal A should have the same phase as signal C, and signal B
should have the same phase as signal D. Moreover, both signal A and C should have the
opposite phase with signal B and D. For the torsion actuation, the signal A should have
the same phase as signal D, and signal B should have the same phase as signal C. In
addition, both signas A and D should have opposite phase with signal B and C.
2.3 Bonding Improvement
There are three bonding phases during the fabrication of the active control blade. The
first one is the bonding the piezo to the copper-kapton sheets. Since both piezos and
copper-kapton sheets are flat and small, they can be put inside the aluminum-plastic jig
showed in Figure 2-9. The lower side of the jig is a flat aluminum plate having a port for
a vacuum pump. The upper side of the jig is plastic paper taped on the aluminum plate.
Piezos, lead sheets, and paper towels are placed inside the jig before taping the plastic
paper to the aluminum plate. Paper towels keep a 'passthrough' to the vacuum port to
ensure outgas.
Figure 2-9: Aluminum-plastic jig for bonding piezos to lead shapes
The second phase is bonding the piezo package to the graphite spars. Vacuum was not
used in this process, because the jig with a vacuum port can only be used when the two
32
pieces to be bonded are flat. Since the graphite spars are highly twisted, if we put the
graphite core with cut out spars in the jig, the pressure will untwist graphite spars or even
break them. Therefore, clips, dead weights, and C clamps were used to apply pressure to
help outgas.
A possible way to apply vacuum during bonding is to use a vacuum bag
without an aluminum plate. The third phase is bonding the graphite spar assembly to the
foam blade.
No vacuum was used here either. Details are outlined in section 2.5,
assembly procedures.
Here are some important notes concerning bonding.
The glue layer thickness is
important to the active blade. The glue between the piezo package and the spar surface is
a shear layer. For the same amount of induced strain of the piezo, the strain of the spar
surface is smaller if the glue is thicker. The smaller strain of the spar surface leads to the
less defection of the graphite spars or the blade. Hence low viscosity glue Epotec 301
was chosen as suggested in [3]. Clean surfaces are also important to get good bonding.
Ethanol should be used to degrease the bonding area. Flat mating surfaces are important
to obtain a thin layer of bonding glue.
The previous bonding technique was simply bonding the flat piezo package directly to
the highly curved and twisted spar root surface. It is obvious that this bonding procedure
was not good enough for the actuation. In fact, at some points the gap between the piezo
package and the spar root surface reached more than 1 mm (See Figure 2-10). Hence, a
lot of glue material (Epotec 301) must be used to fully bond them together. Another
problem is that the actuator is fragile and easily broken when external force is applied to
try to get better bonding. The new bonding technique involved trying to obtain a flat
mounting surface from the spar root surface. A milling machine is a good choice for
machining it, however, the root of the spar is too thin to be cut directly. Hence, more
layers of graphite epoxy prepreg were first added to the spar root surface to make it thick
enough to be machined.
33
One piece of piezo
Quarters fit into the
gap between the piezo
and one spar
Gap between the spar
surface and piezo
Figure 2-10: Illustration of the gap between the piezo and the spar surface
Several criteria were used durng the adding and machining to obtain a Ilat surtace. First,
add as few graphite epoxy layers as possible, because the more layers we use, the stiffer
the blade will be. Hence less tip defection will be obtained.
Second, layers are
unidirectional in 0" in order to transfer more strain to the blade, since the root is highly
curved. Totally 15 layers were added on the root of graphite spars.
The cure cycle is 250 0 F two hours followed by 3500 F two hours showed in Figure 2-11.
This curing procedure for the graphite epoxy prepreg is given in [10]. Instead of using a
vacuum bag, C clamps were used to apply pressure to help bonding the additional layers
of prepreg to graphite spars for simplicity.
Temperature (Deg F)
350
250
0.5
4.0
2.0
Time (hours)
Figure 2-11: Cure cycle used for adding graphite epoxy prepregs to the graphite spars.
34
2.4 Foam Shaping
2.4.1 Thermoforming
General procedure
Thermoforming is the process of heating a thermoplastic material, such as the foam used
for the active rotor blade, and then applying pressure by holding it in molds to give a final
shape. When the temperature increases to 3500 F, the cells of the Rohacell WF 2000
become soft. By putting Rohacell in a special set of aluminum molds, and applying
pressure by C clamps and heat, a twisted aerodynamic Fan C shape can, in theory, be
obtained.
Problems with Thermoforming
The major failure of thermoforming is that it is not able to shape the foam to high levels
of curvature and twist. The high strain induced by the thermoforming cycle required for
the fan C blade cracks the foam. Details are outline in [2]. The following is the summary
from this article.
e
Thermoforming did not give the desired airfoil shape with tapered thickness to the
cross section. This is because thermoforming only presses the cells of the foam in
the thickness direction. Thus, tapered thickness could not be done by using the
given blank aluminum molds and available facilities.
"
It was very difficult to achieve the designed twist, even after six thermoforming
cycles. (Each thermoforming cycle needs 4.5 hours.) This is because the fan C
shape is highly twisted.
" After thermoforming is over, the foam tends to spring back. This should be taken
into account while designing the molds for this procedure. However, this was not
done since there was no previous experience with this spring back effect.
"
Thermoforming stiffens the foam during each thermoforming cycle.
35
* Thermoforming did not produce the desired shape at the leading and trailing
edges.
The two cases of cracking (given in [2] ) at the groove and snapping of the foam without
grooves at the maximum curvature due to excess force are illustrated in Figure 2-12.
Foam snapped
here due to
thermoforming
Foam cracked
here due to
thermoforming
Figure 2-12:
Illustration of foam cracking at groove and foam snapping at
maximum curvature due to excess force and limited material flow
during thermoforming
Because of these problems, it was decided that machining the foam would be attempted.
This is outlined in the next section.
2.4.2 CNC Machining
Thermoforming has proved unable to obtain the scaled, twisted aerodynamic Fan C shape.
In order to obtain it, the CNC machine in the Aero Lab at MIT was used to machine a
foam block. This method needs a 3-D milling machine, and an aluminum blade mold.
This mold was used as a base for holding the machined foam blade when one side of
foam blade was finished and the other side of the foam blade was being machined.
e
BLOCK PREPARATION
The appropriate dimensions of the foam block, which was constructed for blade
machining, was 7.65X3.9X3.0 inches. By gluing together three small layers of Rohacell
36
foam, each of dimension 7.65 x 10 x 1.0 inches, a big Rohacell block with dimension
7.65 x 10 x 3.0 inches was obtained. Then the big Rohacell block was cut into three
blocks, two of them with the design dimension 7.65 x 3.9 x 3.0 inches. The last one was
too small, thus not suitable for machining. Epotec 301 was used to glue the materials. A
band saw was used to cut the large Rohacell block.
* MACHINING PROCESS
The pressure side of the blade was machined first. Since this machining stage began with
a rectangular block, it was very easy to fix the block to the tool holder in the milling
machine.
The upper surface was machined in two stages. One was the roughing stage, and the
other was the finishing stage. For both the roughing stage and finishing stage, a 1/8 -inch
ball end mill was used to either remove large quantities of material or to provide an
accurate finished surface.
After machining the upper surface, the block was turned over for machining the lower
surface. In order to hold the block without damaging the finished upper surface, the
block was installed on the corresponding side of the aluminum blade mold by using
double-sided adhesive tape. Then it was secured to fix the aluminum/foam block to the
tool holder in the milling machining by holding the aluminum part. The upper foam
surface could then be machined. Once again, roughing and finishing stages were used for
the lower surface.
* Note on fixing the foam to the aluminum block:
It is important for the foam to be bonded tightly to the aluminum block to give a good
work holding. Loose bonding allows the block to vibrate when it is being machined,
making the surface inaccurate. Therefore, it is necessary to make sure that the foam
block is securely bonded to the aluminum block. In order to achieve this, the foam
surface must be sanded and smoothed, and the foam particles blown away such that the
tape would stick to the main block of material, rather than to a dusty surface.
Furthermore, manual filing is usually needed to ensure that the foam piece fits well into
37
the aluminum block, with no interference points that cause the two parts to pivot rather
than mate squarely.
2.5 Assembly Procedures
The previous concept for assembly leads to two piezos embedded inside the composite
blade. Figure 2-13 shows that, in this case, the two piezos on the suction side of the
graphite spar assembly are exposed after the assembly process.
However, the foam
blade covered the other two piezos (not shown in the graphite spar assembly picture) on
the pressure side after the assembly process.
Two Piezos on the
suction side exposed
after assembly.
Figure 2-13: Illustration of the piezos to the foam blade
The new idea is to expose all piezos outside. This approach is implemented by cutting
foam material out at the base of the blade. The reasons for this approach are as follow:
*
Reduced risk of breakage during the process of bonding the foam blade and the
graphite spar assembly.
" Easier maintenance of the piezos, strain gages, and wiring.
38
*
Better fitting. Since during the process of bonding the piezos to the graphite spars,
more layers of graphite epoxy prepreg were added and then machined to get a flat
bonding surface, the geometry of the piezo on the graphite spars has been changed.
Hence it was not expected to get good matching of the graphite spar assembly
root to the corresponding grooves on the foam blade.
The assembled blade showed that the graphite spars and grooves match very well except
at the root where the foam material was removed.
One drawback of this assembly
technique is poor geometric precision at the blade root.
Some method needs to be
decided to improve the geometric precision, such as fiber cloth and flash tape. The
second drawback is that the structural strength of the blade was reduced, especially near
the root. RPM tests showed that at 3000 RPM, some foam at the leading edge close to
the piezos was lost due to centrifugal and aerodynamic forces.
The final step was fabrication and bonding of the engineering foam (Rohacell) blade to
the graphite spar assembly.
Epotec 301 was applied to the cleaned surfaces of the
graphite spars and foam grooves. The graphite spars fit inside the grooves very well,
which verified the accuracy of graphite spars cutting and foam blade shaping. Paper clips
and C clamps are used to apply load showed in Figure 2-14. This procedure is easy to
implement. However vacuum-bag would probably be a better solution to make the load
more uniform and help outgas the epoxy. The cure is in room temperature and the
minimum cure time is 12 hours. The reason for not using elevate temperature is that the
foam blade will stiffen or become soft at temperatures higher than 3500.
To protect the foam blade from getting smashed, a thin piece of rubber can be inserted
between the aluminum and foam blade.
39
Figure 2-14: Assembly of the graphite spar assembly to the foam blade
Figure 2-15: Suction side of the
assembled blade
Figure 2-16: Pressure side of the
assembled blade
40
This is the first geometrically accurate assembled blade. The graphite spars fit well into
the grooves on the foam blade. All piezos and strain gages are working. This is the blade
used to perform bench top and in-situ tests.
2.6 Spin Pit Development Work
Rig modifications
Various rig modifications to improve the performance of the spin pit have been
implemented.
The piezoelectric accelerometer installed by Farahat [1] for vibration
monitoring was remounted using a nylon stud and a nylon washer, to eliminate ground
loops successfully and provide a higher mounting resonance frequency (15KHz).
Thermocouples were added to monitor heat produced by the rubbing of the lower face
seal against its mating ring (Figure 2-17). The vacuum level was improved from -3500
Micro Torr to -500 Micro Torr to eliminate aerodynamic forces on the rotating blades.
This was accomplished by sealing air leaks from the rotor bolts with O-ring, shown in
Figure 2-18.
41
Thermocouple
line connecting
to thermocouple
meter
Thermocouple
mounted near
seal
Figure 2-17: Rotor with Thermocouple in place.
A red O-ring fits
into the groove
of the bolt.
Figure 2-18: A bolt with an O-ring
42
Rotor In-Situ Balancing
Because various test rotor blades will be put into the spin test facility for testing with
Eddy Current Sensors (ECS), it is necessary to rebalance the rotor when a new rotor
blade is installed. This would normally require sending the rotor to an outside vendor.
Instead, we are implementing an in-situ balancing technique using the vibration sensor
mounted on the main spindle of the spin pit. This will reduce turn-around time for
various rotor measurements.
This procedure of rotor balance requires five runs to complete, since no phase reference
is used for these tests. First, without adding a trial weight, the rotor is run at the desired
balancing speed. The vibration is measured by the accelerometer at the housing of the
bearing.
For each of the next three runs, a trial weight is placed at three different
positions that are all at the same radius, at positions 120-degrees apart on the rotor. The
three corresponding vibration levels are measured, and used to compute the location and
weight of the balance required to eliminate the imbalance from the first test. Finally, the
rotor is run with a balancing weight and the vibration is measured to compare the results
to the first run. Details are outlined in the Appendix A.
43
44
Chapter 3
Bench Top and In-Situ
Characterization of the Graphite Spar
Assembly
Before bonding the graphite spar assembly to the foam blade, bench top and in-situ
characterization of the graphite spar assembly was conducted. Although results vary
significantly with the addition of the foam blade, developing the identification procedure
and estimation techniques provide experience that applies to the assembled blade. The
plan followed to achieve this objective was as follows: Actuation testing on the bench
was used to verify the proper operation of the spar system, to characterize its non-rotating
performance, and to obtain estimation transfer functions. These estimation transfer
functions can be used to estimate the tip deflection based on the strain measured on the
root of the graphite spars. In bench top tests, we verified that the estimation results were
accurate using laser displacement sensor results. After bench top testing, the graphite
spar assembly was tested for its performance at various values of RPM.
45
3.1. Bench Top Characterization
The description of the bench top tests has been divided into three sections - experimental
setup summary, instrumentation and results.
3.1.1 Experimental Setup Summary
The basic instrumentation setup for testing the actuation of the graphite spar assembly is
shown in Figure 3-1.
The spectrum analyzer outputs a sinusoid, which it sweeps in
frequency. This signal is connected to both the power amplifier and channel #1 of the
spectrum analyzer. The sweeping sinusoidal signal, which will be called the command
signal, has an amplitude of 1.0 Vpp. It is amplified via an audio power amplifier, and its
voltage is further raised by transformers.
The signal applied to the piezo actuators is
about AC 90 Vrms with DC 90 V bias. The piezo actuators excite the spars, which in turn
deform. Strain gages measure the root strain and a laser displacement sensor measures
tip deflections. Strain gage amplifiers measure the small changes in resistance associated
with the deformation of the strain gage. The output of the strain gage or the laser sensor
was connected to channel #2 of the spectrum analyer.
The Analyzer measures the
amplitudes of the two input signals, divides them, and plots the result.
Hence, the
analyzer can effectively display the frequency response of strain or tip displacement.
46
Graphite spars with four
actuators
Laser
sensor
Power Amplifier
Figure 3-1: Experimental setup for determination of transfer functions
3.1.2 Instrumentation
Actuation power supply
The specification and set up parameters of the actuation power supply are presented in
Table 3-1. An HP 35665A dynamic signal analyzer is used to generate a 1Vpp sine sweep
signal, which is amplified by 120 using both the audio amplifier and the transformer.
The DC offset instrumentation provides up to 90 V DC offset. Both AC and DC signals
are applied to the piezos. Figure 3-2 shows the connections and settings for using the
audio amplifier. Figure 3-3 shows the internal view of transformer circuit. Figure 3-4
shows the front panel of transformer circuit.
transformer circuit.
47
Figure 3-5 shows the diagram of
Table 3-1:
Dynamic Signal
Analyzer
Audio Amplifier
Transformer
DC Offset
Actuation power supply specification and set up parameters
Source
1Vpp
Model
Yorkville AP4040
Mode
2 channel, stereo
Gain setting
Maximum
High Pass Filter
40Hz
Coil ratio
22 X 115
Gain
90V
Adjust to the maximum gain
for both channel A and B
(a) Front Panel
Input Channel A & B connected
to the command signal by twocircuit stereo %" headphone jack.
Connected to BNC to plug into
signal analyzer
I
Output of the audio
amplifier (connected
internally)
(b) Back Panel
Figure 3-2: Illustration of the connections and settings for using the audio amplifier
48
Input of the Audio Amplifier
Output of the Audio Amplifier
Connected to Transformers
Figure 3-3: Internal View of Transformer Circuit
I High voltage signal
Ground
Output connected to the piezos.
(The connector must be high
voltage BNC.) Shield remains
open at piezo side.
DC offset adjustment
Figure 3-4: Front Panel of Transformer Circuit
49
Figure 3-5: Diagram of Transformer Circuit (By Zhongguo Li)
Ul: Audio amplifier. Manufacturer, Youkvile Sound. Model AP4040
U2: Talema Toroidal Transformer, part number 70085, primary 2 X 115 V, secondary 2
X 22 V.
U3: 12V, 0.7A AC-DC power supply;Manufacturer: Acopian; Model: 12EB70
U4: DC to DC High Voltage Power Supply (HVPS), 0 to 1000 VDC, 1 A
output.Manufacturer: Ultravolt.Model: 1A12-P4-E.
Note: (RJ: 48K Ohm. R2: 50K Ohm (poterntiaometer). R3: 10K R4: 1G. R5: 200K
Ohm. C: 1ONF/JKV DC)
50
Strain gage amplifier
The strain gage conditioning amplifier used is the Measurement Group, Inc 2310 series.
The appropriate strain gages used are EA-06-125AD-120 given in [2]. The specification
and set up parameters of the amplifier are presented in Table 3-2. Figure 3-6 shows the
wiring connections and parameter choices of the strain gage amplifier.
10 kHz low pass
filter
Choose
exitation voltaae
Adjust gain
as 300 ue /v
(a) Front Panel
I
-+
Strain gage input
Strain gage output
(b) Back Panel
Figure 3-6: Illustration of the connection of the strain gage amplifier
51
Table 3-2:
Strain gage conditioning amplifier specification and
set up parameters
Bridge Mode
Quarter Bridge
Active Filter
10 kHz low pass filter
Gain
300 p, /v
Excitation Voltage
10V
Nexus conditioning amplifier and Bruel & Kjaer accelerometer
The type of Nexus conditioning amplifier used is 2692 and the type of accelerometer
used is 4382V. The vibration levels measured by the accelerometer can used to monitor
the condition and health of the spin pit rig. They can also be used to perform in-situ
balancing. The specification and set up parameters are presented in Table 3-3.
The
accelerometer is mounted using a nylon stud and a nylon washer, which eliminates
ground loops. The mounting resonance frequency is as high as 15 kHz. This mounting
resonance frequency is larger than the bandwidth of the accelerometer, which is 8.4 KHz.
Table 3-3:
Nexus conditioning amplifier and accelerometer
specification and set up parameters
High-pass Filter
1 Hz
Low-pass Filter
100 Hz
Output Sensitivity
100 mv/ms 2
Transducer Set-up
3.16 IC/ms 2
Bandwidth
8.4 kHz
Resonance frequency
15 kHz
52
Laser Displacement Sensor
A Keyence LB-72 laser displacement sensor was used to measure spar tip deflection.
The specification and set up parameters are presented in Table 3-4. The calibration data
and best fit line are presented in Figure 3-7.
Table 3-4: Laser displacement sensor specification and set up parameters
Model
Sensor head
LB-12
Controller
LB-72
Reference distance
40 mm
Measuring range
±10 mm
± 4V (0.4 V/mm)
Output voltage
Fast mode, DC to 3K Hz
Response Frequency
50u
pm (at 0.15ms)
Resolution
Calibration Data
10
8
- - -
-- - -- - - -- - - -- - - -- - - -- - --
- - -
6
4
E 2
E)
E
0
ca
-2
0
--I
-4
-6
-8
-10
-4
-3
-2
0
-1
Voltage(V)
1
2
3
Figure 3-7: Laser displacement sensor calibration curve
53
4
HP 35665A Dynamic Signal Analyzer
The analyzer can be used to compute the transfer function of the laser or strain gage
reading to the command signal or the source signal (1Vpp). In swept sine mode, the
analyzer outputs a sinusoid, which it sweeps in frequency. The analyzer then measures
the amplitudes of the two input signals, divides them, and plots the result. The source
(The swept sine output) is connected to input #1 (in addition to the input of the audio
amplifier) and the output of the tip displacement or strain readings under test is connected
to input #2, then the analyzer effectively displays the frequency response of the test. The
specification and set up parameters of the HP 35665A dynamic signal analyzer are
presented in table 3-5.
Table 3-5: Dynamic Signal Analyzer specification and set up parameters
Swept Sine
Mode
Frequency Response
Measured data
50 Hz to 1000 Hz
Frequency range
AC, floating
Input setup
Source Level
1VPP
54
3.1.3 Test Procedures and Test Matrix
Test Procedures
A pure sinusoid signal which sweeps in frequency is used to excite the graphite spars.
The root strains and tip deflections are used to calculate transfer functions.
Table 3-6 shows the summary of the measured transfer functions. Table 3-7 shows the
inferred transfer functions for the bench top tests of the graphite spar assembly.
Notation:
Strain Gage at the Leading Edge Spar (LES), Suction Side
s,
Strain Gage at the Trailing Edge Spar (TES), Suction Side
s2
Strain Gage at the Leading Edge Spar (LES), Pressure Side
S3
Strain Gage at the Trailing Edge Spar (TES), Pressure Side
s4
Tip Deflection at the Leading Edge (LE)
zI
Tip Deflection at the Trailing Edge (TE)
z2
Average Tip Deflection (ATD)
z
Tip Twist (TT)
0
Command Signal
c
Table 3-6: Summary of measured transfer functions for the bench top tests of
the graphite spar assembly
Transfer
s,
s2
S3
S4
z
.ucto
Function
c
c
c
c
c
z2
z_
z
z2
z2
c
si
s3
s2
s4
Note: In all tests allfour piezos are actuated, in phasefor bending mode tests and out of
phasefor twisting mode tests.
55
Table 3-7:
The inferred transfer functions for the bench top tests of the
graphite spar assembly
Transfer Funcitons
z
0
C
C
3.1.4 Results
In order to obtain reliable results in the bench top tests, boundary conditions must be kept
the same. So, the graphite spar assembly was mounted on the rotor, which was clamped
on the bench. The test frequency ranges from 50 to 1000 Hz. The results of frequency
response testing of the graphite spar assembly are presented. Both the leading edge spar
and the trailing edge spar were actuated.
Basically the results can be grouped into four categories.
1. The first sets of results are measured transfer functions from a command signal
applied at the piezos to tip deflection at the leading edge spar (LES) and the
trailing edge spar (TES). The inferred results of this measured data are the tip
bending and the tip twisting transfer functions.
2. The second sets of results are measured transfer functions from the command
signal applied at the piezos to strain gages.
3. The third sets of results are the measured estimation transfer functions, from
strain gages to tip deflection at the leading edge spar (LES) and the trailing edge
spar (TES).
4. The fourth sets of results are the comparisons of the measured and inferred
transfer functions from the command signal to tip deflections at the LES and the
TES, as well as the comparisons of the measured and inferred tip bending and tip
twisting transfer functions.
56
The measured transfer functions from the command signal to tip deflections of the
leading edge spar and the trailing edge spar are shown in Figure 3-8. Based on these
measured data, the mean tip defection was obtained by averaging the two spar deflections.
The inferred average tip deflection is shown in Figure 3-9. The inferred tip twist was
calculated by dividing the sum of the two spar tip deflections by the inter-spar distance at
the tip. The inferred tip twist is shown in Figure 3-10.
The following observations about the measured tip deflections can be made:
a) There are four natural frequencies observed in the frequency range from 50 Hz to
1000 Hz for both the leading edge spar and the trailing edge spar.
b) For the leading edge spar, the first natural frequency is 108.5 Hz with an
amplitude of 16.78 mm/V (Peak to peak). Then second natural frequency is 288
Hz with an amplitude of 3.77 mm/V (Peak to peak). The third natural frequency
is 576 Hz with an amplitude of 1.34 mm/V (Peak to peak). The fourth natural
frequency is 960 Hz with an amplitude of 0.6 mm/V (Peak to peak).
c) For the trailing edge spar, the first natural frequency is 128 Hz with an amplitude
of 8 mm (Peak to peak). Then second natural frequency is 288 Hz with an
amplitude of 2.9 mm/V (Peak to peak). The third natural frequency is 600 Hz
with an amplitude of 1.5 mm/V (Peak to peak). The fourth natural frequency is
900 Hz with an amplitude of 0.2mm/V (Peak to peak).
57
Peak 17 mmN
Peak 8 mmN
Tip Displacement Tranfer Functions
-------------------- a-------Leading Edge Spar
Trailing Edge Spar
-- + ---------- -+--------------------+4-- --5----------------
4.5
3. --
-- - -- - - -------
- ----- -------
------
--- ------
E
E
1.5 ------------------------1
----5
0
I
I
----- ------- T---- - -----
-- --.
500
400
300
200
100
------
-
----- +--------
+----
-------
----
-------
600
- - --
------- -
700
900
800
1000
Frequency [Hz]
Figure 3-8: Measured transfer functions from the command signal applied at piezos to
the tip deflection at the leading edge spar (LES) and the trailing edge spar (TES)
Peak 8.6 mmN
Average Tip Displacement Tranfer Function (Bending mode)
The bandpass bandwidth over
which at least 0.5 mm/V (Peak
to Peak) tip bending is
4 --------------
E
- ------ ----------------
--- ------- -----
3 ------
E
S 2.5 ------
--- + -----
2 ------1.5 -------
00
maintained 249Hz (Red Line)
100
---------
---- -
----- ---
The bandwidth over which at
least 0.25 mm/V (Peak to Peak)
tip bending is maintained
~-----
---------- ------- 342Hz (Green Line)
200
300
400
500
600
700
800
900
1000
Frequency [Hz]
Figure 3-9: Inferred transfer function from the command signal applied at
piezos to the average tip displacement (ATD)
58
The author concludes that the bandpass bandwidth over which at least 0.5 mm/V (Peak to
Peak) tip bending is maintained 249 Hz, which is about 92% larger than the result
presented in [2]. (See Figure 3-9) The bandpass bandwidth over which at least 0.25
mm/V (Peak to Peak) tip bending is maintained 342 Hz, which is about the same
bandwidth as the result presented in [2]
Bandpass bandwidth and Bandwidth:
a) The bandpass bandwidth is defined as the first continousfrequency range over which
the response amplitude remains within 3dB of the target value. Other frequency
ranges may be found within 3dB of the target value. However, these frequency
ranges are not counted into the bandpass bandwidth.
b) The frequency response data only startfrom 50 Hz. No data between the frequency
range 0Hz and 50Hz is available. The following criteria are used to infer whether
this frequency range is a part of the bandpass bandwidth or not. If the frequency
response at 50 Hz is in the bandpass bandwidth, the author infers that the frequency
range between 0 Hz and 50 Hz is in the bandpass bandwidth. Otherwise the
frequency range is not in the bandpass bandwidth.
c) If the bandpass bandwidth includes the frequency range between 0Hz and 50 Hz, the
bandpass bandwidth is called bandwidth. Otherwise it is still called bandpass
bandwidth.
The author concludes that the bandpass bandwidth over which at least 0.5 Degree/V tip
twisting is maintained 264 Hz, which is about 210% larger than the result presented in [2].
(See Figure 3-10) The bandwidth over which at least 0.25 Degree/V tip twisting is
maintained 354 Hz, which is 136% larger than the result presented in [2]. The better
actuation of the graphite spars positively relate to the better actuation of the assembled
blade.
Hence, before the addition of the foam blade to a new graphite spar assembly,
bench top tests should be performed to evaluation the effectiveness of the actuation. By
comparing the data of a new graphite spar assembly with the data of the previous graphite
spar assembly, we can tell whether the new graphite spar assembly has better actuation
results, which will lead to a better actuation result of the assembled blade.
59
Peak 10.2 DegreeN
5
Tip Twist Tranfer Function (Twisting mode)
------- -----4.5 ------- I - ----
------- ------
4 ----------
The bandpass bandwidth over
-
which at least 0.5 Degree/V tip
twisting is maintained 264 Hz
(Red Line)
-
a)
The bandwidth over which
at least 0.25 Degree/V tip
is maintained 354
Hz (Green Line)
2.5
-twisting
-
2-
--
.5----
-
3dB ____
0
0
100
200
300
500
600
400
Frequency [Hz]
700
800
900
1000
Figure 3-10: Inferred transfer function from the command signal applied at
piezos to the tip twist.
Figure 3-11 and Figure 3-12 are plots of both magnitude and phase of the transfer
function of the four strains to the command signal. As mentioned earlier, the root strains
are the signals we used to estimate the tip displacement in the in-situ test.
The following observation about the measured root strains are made:
a) There are four natural frequencies observed in the frequency range from 50 Hz to
1000 Hz for both the leading edge spar and the trailing edge spar.
b) Both strains of the leading edge spar are out of phase. Similarly both strains of
This makes sense because when the spar
the trailing edge are out of phase.
deflects, one side extends, while the other side contracts. This is true for both inphase and out-of phase actuation cases.
c) For the first natural frequency, the strains of the leading edge spar are much larger
than those of the trailing edge spar. Correspondingly, the tip deflection of the
leading edge spar is larger than that of the trailing edge spar.
60
Peak 450 MicrostrainN
Leading Edge Spar Strain Transfer Functions
--- 1 ------------------------ ------,
200 --,
Leading Edge Spar, Suction side
Leading Edge Spar, Pressure Side
-
:E100 --2
100
200
---
---
300
400
L---
-----
500
600
700
---
800
----
900
1000
------- I-------+--------1
----------
-
-- -
--
400
300
--
r--
--
-r----
---200
100
- -
------------
- ----------
- -300 -- +----400-
--
- - -
--
--
-
------------------------- ------- ---------------------- ------ ---------
100 ----
cD-200
CU
---------------------- -------
----
-
-
-
C -100
---
-------------------------
50
0
-------------
-----
150 ----
500
600
Frequency [HzJ
700
800
900
1000
Figure 3-11: Measured transfer functions from the command signal applied
at piezos to root strains at the leading edge spar (LES)
Trailing Edge Spar Strain Transfer Functions
S
2
500
5
0
-o
100
C- -300
-400
--
100
-
------
-------
-------
----------
--
-------
200
300
400
--
-------
500
------ J--------L ------- --- --
600
700
800
900
J
1000
------ ------- ---------------------- ----- ----------------------------
-----
-----------------
100
Trailing Edge Spar, Suction Side
Trailing Edge Spar, Pressure Side
.-
200
300
400
+-----+------
600
500
Frequency [Hz]
700
800
900
1000
Figure 3-12: Measured transfer functions from the command signal applied
at piezos to root strains at the trailing edge spar (TES)
61
Transfer functions from root strain to tip displacements
Estimation techniques have been developed to determine tip deflection based on strain
gages bonded on the roots of the graphite spars.
Until eddy current sensor (ECS)
capabilities are developed, and complementary to the desired ECS measurements, such
estimation is our preferred method to determine tip displacements in the rotating
environment (our laser sensors can only be used in the non-rotating reference frame).
These methods begin with determination of the transfer functions from root strains to tip
displacements.
Bench top tests are performed to obtain these tip estimation transfer
functions. Based on these transfer functions, we can estimate the tip displacements in
both bench top and rotating tests.
In bench top tests, we verify that the results are
accurate using laser displacement sensor results.
results are available at the present time.
In the RPM tests, only strain-based
Ultimately the Eddy Current Sensor
measurements will help to verify the estimation results of the RPM tests. A series of
bench top tests were performed to dynamically characterize the graphite spar assembly.
There are four strain gages bonded on the root of the spars. Each strain gage can be used
to independently estimate the corresponding tip deflection of the spar. Figure 3-13 shows
two estimation transfer functions, which can be used to estimate the leading edge spar tip
displacement by measure the root strain of the leading edge spar suction side or pressure
side.
Similarly, Figure 3-14 showed the other two estimation transfer functions for
estimating the trailing edge spar tip deflection.
For the transfer function from root strain to tip displacements, the input is root strain, and
the output is tip displacement. Hence, The y-axis is deflection over strain, with units of
millimeters per microstrain. Once the strain data is obtained, the tip deflection is very
easy to estimate based on these transfer functions.
62
Leading Edge Spar Tip Displacement Estimation Transfer Functions
Wi
0
E
a)
CU
0.1
Leading Edge Spar, Suction Side
Leading Edge Spar, Pressure Side
---------- -
0.15
--
-- '------- r------ -----
--+-----+
-
------- I-------
-
0.05
0
100
200
300
400
500
600
Frequency [Hz]
700
800
900
1000
600
400
+
---------
------
--- --
-+ ------+ -----
+
--------------
-- -
200
a)
CA,
MU
.I-
0~ --------200
-400
I -- - -
100
200
300
400
--------
------- ---
------ -
- - -
500
600
Frequency [Hz]
700
800
900
1000
Figure 3-13: Measured transfer functions from root strains at the
leading edge spar (LES) to the tip displacement at the leading edge spar
Trialing Edge Spar Tip Displacement Estimation Transfer Functions
I
I
I
E
E
E
-------+ +------
0.1
0.05
II
I
Trialing Edge Spar, Suction Side
Trailing Edge Spar, Pressure Side
C5 0.15
-
-----
----------- -----
-----
------ ------
-I---- -----
--- -- + -
--- ------
-------
---- -------
---
C
100
200
300
400
100
200
300
400
500
600
Frequency [Hz]
700
800
900
1000
600
700
800
900
1000
500
CD
a
0'
-500
CO)
CO
-1000
-1 r:nn
500
Frequency [Hz]
Figure 3-14: Measured transfer function from root strains at the trailing
edge spar (TES) to the tip displacement at the trailing edge spar (TES)
63
Figure 3-15 and Figure 3-16 prove the effectiveness of the estimation transfer functions
for the graphite spars. In Figure 3-15, the red line is the measured leading edge spar
displacement measured by the laser displacement sensor. This red line is plot using the
same raw data as that shown in Figure 3-8. The blue line and black line are the estimated
leading edge spar displacement obtained by the strain gage bonded to the leading edge
spar pressure side (Strain 1) or the strain gage bonded to the leading edge spar suction
side (Strain 3). Similar results are shown in Figure 3-16 for the trailing edge edge.
Peak 17 mmN
Leading Edge Spar Tip Displacement Tranfer Functions
5
4.5 ---- --------------
-
Measured Leading Edge Spar
Estimated Leading Edge Spar, by Strain 1
Estimated Leading Edge Spar, by Strain 3
4
3.5
E
E
-
r-------
rL---------------
3
2.5 ---- -----
----
-
---
-------
-----
------
C
-a
2
- -------
1.5
- -------------
----
----------
------
1
----- -------Ir
- -- -- - ------I-------T-------I--
0.5
100
200
300
400
600
500
Frequency [Hz]
700
800
900
1000
Figure 3-15: Verification the estimated leading edge spar displacement
64
Peak 8 mm'V
P6 Trailing Edge Spar Tip Displacement Tranfer Functions
4.5
-----
-----------
35
E
---- -----------
-------------------------
4--
3
-
+----- ------------------------------------------
.2---
0
Measured Trailing Edge Spar
Estimated Trailing Edge Spar, by Strain 2
Estimated Trailing Edge Spar, by strain 4
100
200
300
400
600
500
Frequency [Hz]
700
800
900
1000
Figure 3-16: Verification the estimated trailing edge spar displacement
Since each curve in Figure 3-14 and 3-15 is the result of a separate experiment, this is a
valid test of the procedure.
The conclusion is that estimation transfer functions are
accurate enough to estimate both leading edge spar and the trailing edge spar tip
deflection.
The ultimate parameters of interest for the active rotor are pitch and plunge. Hence it is
useful to present the data in terms of the tip deflection and tip twist. Based on the data
presented in Figure 3-15 and Figure 3-16, the estimated average tip deflection was
obtained by averaging the two spar deflections. The result is plotted in Figure 3-17. The
inferred average tip deflection showed in Figure 3-9 is also plotted in Figure 3-17. Figure
3-17 shows that the estimated results using any of the four strain measurements match
with the inferred results using the laser displacement sensor very well.
Similar results are shown in Figure 3-18. The estimated tip twist matches well with the
inferred tip twist.
65
Peak 8 mmN
Average Tip Displacement Tranfer Functions
5--------
4.5 -
-
-
--
-
-
-
-
-- - - -- -
----+-----+------
-
----------------------------
--------
-------
- -----
-
-
Measured By Laser
Estimated By Strains
---
- ------------ +-------------
4 ---
3.5
_- -
_
---------
E
E
2 ----
----
------
--------------
------
------- ---
------
------
11
0.5
05
0
--- --- ---- - ------
100
300
200
-----
+----- -, ----- + -----
500
400
600
700
+-L--------------A
800
900
1000
Frequency [Hz]
Figure 3-17: Inferred transfer functions from root strains at the leading edge
spar (LES) and the trailing edge spar (TES) to the average tip displacement
Peak 10.2 Degree N
Tip Twist Tranfer Functions
ae
esrdBBy Laser
I---------------------------Measured
-------------45--- Estimated By Strains
4.5 ---- ----------------------------+I------ ---
+----- -----+ - --4 - - --- - ------ ------- I----
3.5 CD
------
------
3 ---------- ------
------
--------
-
------ J--------L ----------
------- ------- ------- -----
------- ------ ------ -
C
0
100
200
300
400
600
500
Frequency [Hz]
700
800
900
1000
Figure 3-18: Inferred transfer functions from root strains at the leading edge
spar (LES) and the trailing edge spar (TS) to the tip twist
66
Figure 3-17 and Figure 3-18 indicate that the state estimation transfer functions are
accurate enough to estimate both the average tip displacement and tip twist.
3.2. In-Situ Characterization
A series of RPM tests was conducted to obtain strain gage signals from the rotating frame
and to determine the effectiveness of spar actuation in the rotating envirament. The
experimental data are presented.
For the RPM tests of the graphite spar assembly, only one graphite spar assembly was
mounted on the rotor. Hence both static and dynamic balancing were performed. The
graphite spar assembly was hung from two points on the trailing edge spar and the
leading edge spar to locate the spanwise and chordwise center of gravity.
Counter
weights were sequently placed in the rotor clips to statically balance the rotor.
The rotor speed was first set to 1000 rpm. The rotor speed was then increased in 500 rpm
increments, up to 5000 rpm. When the speed reach 2000 rpm, dynamic balancing would
normally be undertaken, to minimize the in plane one-per-rev vibration.
Because of
imperfect static balancing there will be a finite rotating imbalance, resulting primarily in
a one-per-rev vibration.
Table 3-8 shows the RPM experiments conducted. Since the laser displacement sensor
can only be used in the non-rotating reference frame, only four strain gages can be
measured in the spin pit. The average tip displacement and tip twist can only be
estimated by using the estimation transfer functions that were obtained in the bench top
tests. The inferred transfer functions for the RPM tests are summarized in Table 3-9.
67
Table 3-8:
The Measured root strain transfer functions for the RPM tests
of the graphite spar assembly
Table 3-9:
The Inferred transfer functions for the RPM tests of the graphite
spar assembly
Results
In-situ sweep tests of the graphite spar assembly were conducted.
The frequency
responses of the graphite spar assembly at different rotating speeds are presented here.
The frequency sweep tests yielded transfer functions from the command signal to the
strains of the root of graphite spars.
68
Inferred
Measured
Strain at the leading
edge spar suction
side
Strain at the leading
edge spar pressure
side
Strain at the trailing
edge spar suction
side
Strain at the trailing
edge spar pressure
side
Inferred
!F---
Tip deflection at
the leading edge
spar (Fig 3-20 and
Fig 3-22)
Tip deflection at
the trailing spar
(Fig 3-21 and Fig
3-23)
Average Tip
Deflection
(Fig 3-24)
Tip Twist
(Fig 3-25)
Figure 3-19: A road map for the result of in-situ tests
Since the parameters of interest are pitch and plunge, the raw data of the measured strain
gages are not presented. Only the inferred transfer functions are presented.
The excitation frequency ranges from 50 to 1000 Hz. Both the leading edge spar and the
trailing edge spar were actuated simultaneously. The following observation about Figure
3-20 are made:
1. a) For the steady case, one pole dominates the response at a frequency of 108.5Hz,
while other less obvious poles are at approximately 288 Hz, 576 Hz, and 960 Hz.
b) For the operation speed of 1000 rpm, one pole dominates the response at a
frequency of 107Hz, while other less obvious poles are at approximately 287.5 Hz,
591 Hz, and 962 Hz. c) For the operation speed of 1500 rpm, one pole dominates
the response at a frequency of 116.5Hz, while other less obvious poles are at
approximately 297 Hz, 591 Hz, and 962 Hz. d) For the operation speed of 2000
rpm, one pole dominates the response at a frequency of 116.5Hz, while other less
obvious poles are at approximately 297 Hz, 591 Hz, and 962 Hz. e) For the
69
operation speed of 2500 rpm, one pole dominates the response at a frequency of
126Hz, while other less obvious poles are at approximately 297 Hz, 591 Hz, and
952 Hz. f) For the operation speed of 3000 rpm, one pole dominates the response
at a frequency of 126Hz, while other less obvious poles are at approximately 306
Hz, 591 Hz, and 962 Hz.
2. All four curves display some similarity in the plots. One dominant pole and other
three less obvious poles in the frequency range from 50 Hz to 1000 Hz.
Compared with the steady case, the rotating cases have many ripples in the frequency
response curves. Background noise is the main reason for these ripples. When the rotor
is running, the whole system vibrates. Strain gages pick up these vibration signals, which
are not related to the actuation. The output of the strain gages is used to calculate the
frequency response.
Hence these background noise signals appear in the frequency
response curves.
70
20
10
0
20
--------
100
------
200
--
------
-
300
400
500
------
600
700
800
-0 RPM:
-----
900
1000
1000 RPM
------ ------ L------ L ------ ------- -------
------ -----10 ------ - --------
E
+
------
E
20
100
200
-----
10--'---
0
O
20
C
-10100
- --
300
±-----------
400
500
600
700
800
------
--
900
1000
3------I----
I
L------+j------J------I------400
+IIL------500
---------300
200
600------700
800
900------1000
-- 200 RPM
~D
E
0
20
100
200
300
400
500
600
700
10---------------------------
20
100
200
300
400
500
600
800
---
700
900
1000
2500RPM
800
900
1000
3000 RPM
10
0
--100
------200
L-
300
400
500
600
700
---
----
800
900
1000
Frequency [Hz]
Figure 3-20: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the leading edge spar by the strain at the
leading edge spar suction side (RPM Tests)
71
5
-
0
5
100
200
300
400
500
600
700
800
0
5
100
200
300
400
500
700
600
800
200
300
400
500
600
1000
900
1000
1500 RPM
--
a)
900
1000 RPM
-
E
E
0 RPM
700
0a)
CD
0
5
2-
3
100
800
900
1000
-- 2000 RPM
C
0
5
100
200
300
400
500
600
700
800
900
-
2500 RPM
800
900
E
a)
1000
0
C
0
r,
100
200
300
400
500
600
700
-
0
100
200
300
400
500
600
700
800
1000
3000 RPM
900
1000
Frequency [Hz]
Figure 3-21: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the trailing edge spar by the strain at the trailing
edge spar suction side (RPM Tests)
72
10
RPM
-0
5
I
0 F
100
10
200
300
--
-- -+
---+ -+-- -- -+
500
400
--
-+
600
-+
-+
--
--
800
700
-- -
-+
1000
900
-1000RPM
E
E
U
CO
-
------------5 -- - --
100
10
300
200
500
400
------ T-------------- -T--------
600
700
-
0
(U
6
0
10
-
--
100
-
- -
-
300
200
-
-
- -
I-
C
5
C
a)
E
a)
0
10
-
100
L
TO
------
-------------------
300
200
400
5-- 0
10
100
200
300
RPM
--------
500
600
700
800
900
1000
600
700
800
900
1000
1
-- 2000 RPM
- ------- ------
II
- - - - - -
1500
-
2500RPM
----- +------ +-I-----+------ +-I-------
400
1000
- I-I
500
400
900
800
500
600
700
900
800
1000
5-00 RPM
5 - -- -i- 0
100
200
------------ + - -
-- + - ---
300
400
500
600
---
+-----
700
800
+ -
900
--
1000
Frequency [Hz]
Figure 3-22: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the leading edge spar by the strain at the
leading edge spar pressure side (RPM Tests)
73
5
- ORPM|
0
-
100
200
300
II~
400
600
500
700
800
D 0
1000
1000 RPMi
0I--
CD
5
900
1
100
200
300
400
600
500
700
800
900
1000
1500 RPM
CU
100
-o
200
300
400
600
500
700
800
900
1000
E1
-
5
0
- 0
100
200
300
400
500
600
700
8
100
200
800
--
300
400
500
600
700
800
2000 RPM
900
1000
250 RPM
900
1000
Frequency [Hz]
Figure 3-23: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the trailing edge spar by the strain at the trailing
edge spar pressure side (RPM Tests)
74
Average Tip Displacement Transfer Functions
10
5 -0
0
5
10
RPM
290?g 390
19
490
590
690
790
-
890
1000 RPM
100
9uo
0
E
-100RP
M
99
79 -- - 89 20rP
5 -10r ------ ------- 9r-------0Ir-------5I r-------60r-------
10
0
10
1n
100
5 -- ,
10
ni
9W9
goo
800
900
1000
RPM
-2000
250rP
------- ------- ------ ------- ---------------------
390
300
29o
200
490
400
soo
500
690
600
7pn
700
0
0
2500RPM
- ----r------- r------- r------- r------- r----------
2 5 ----
0
10
100
0
200
I
IV
100
200
300
400
300
400
500
600
700
800
900
1000
500
600
Frequency [Hz]
700
800
900
1000
Zoom
In
Average Tip Displacement Transfer Functions
2
RPM
S0
1 ------
1
- --------------
r--------
-----
-
1 00RP
---- -----
-------r----- - ---------------------790
son
so0
4QD
r
3n
7nn
Roo
10
9W0RPML1in
0
100
200
300
400
500
600
700
1KuJ
900
1000
2000 RPM
nVII
IA
--- ------.------.------- ---1 ---0
2
800
100
200
300
400
100
200
300
400
2 0 RP
- -- --- - - -
500
600
700
1000
900
800
300RPM
500
600
700
800
0
900
1000
Frpiinrv lH71
Figure 3-24: Inferred transfer functions from the command signal applied at
piezos to the average tip deflection by four strains (RPM Tests)
75
Tip Twist Transfer Functions
10
0
5
5--
---------
0
1010 1an
2
n
0
10 1on
10
_00
Rn
,Q
n
5+------- ----------------
7on
--------
q90
RQO
1 in
1500RPM
-------
Rrn
rQn
Arn
qr n
---- --- ---------- ------
--
CD
2rn
-
agn
ign
------ -------5---
a)
1 0 RPP
--
------- ------- -
--------- ------ --------
70n
0
RP
1 0
+--- --- -
1-
A~n
RQn
-000RPM
101
n
2on
-- -- 1 --- ----
100
1
0
200
g
Agn
gon
-----
300
n
0
x;90
--
Rqn
-- ------
7n
------- -- -
400
500
600
700
7gn
rQn
rgn
Tip Twist Transfer Functions
Agn
---
RQo
RAg
-3000
In
P
- 1 0
800
Zoom
Annn 10-
900
1000
990
10 -1
PM
0
2
1
1- -------------- ------ -----------------
-20RPM
0
2
1
100
200
300
400
500
600
700
800
900
1000
Frequency [Hz]
Figure 3-25: Inferred transfer functions from the command signal applied at
piezos to the tip twist by four strains (RPM Tests)
76
Chapter 4
Bench Top and In-Situ
Characterization of the Assembled
Blade
After the graphite spar assembly was bonded to the foam blade, bench top and in-situ
tests of the assembled blade were conducted. The results for the assembled blade are
presented in this chapter. The instrumentation for both bench top and in-situ tests are the
same as for the graphite spar assembly given in chapter 3.
4.1 Bench Top Tests of the Assembled Blade
4.1.1 Experimental Set Up and Test Matrix
77
In order to obtain reliable results for the bench top tests, boundary conditions must be
kept the same. So, the assembled blade was mounted on the rotor, which was clamped on
the bench as shown in Figure 4-1.
Figure 4-1: Experimental set up for bench top tests of the assembled blade
The experiment conducted is summarized in Table 4-1.
For the bending mode, ten
transfer functions were measured. Similarly, ten transfer functions were measured for the
twisting mode. The inferred transfer functions for the bench top tests are summarized in
Table 4-2.
78
Table 4-1:
Summary of measured transfer functions of the assembled blade in
bench top tests
Bending Mode
s1
S2
S3
s4
z
z2
z
z1
z2
z2
Transfer Functions
c
c
c
c
c
c
si
s3
S2
S4
Twisting mode
si
s2
S3
s4
zi
Z2
z
i
zi
Z2
Z2
c
c
C
c
c
si
s3
S2
54
Transfer Functions
C
Note: In all tests allfour piezos are actuated,in phasefor bending mode tests and out of
phasefor twisting mode tests.
Table 4-2:
The inferred transfer functions for bench top tests of the
assembled blade
z
0
c
c
Bending Mode
Inferred
None
Twisting Mode
None
Inferred
Mode\ TF
Note about the measuringpointsfor the laser displacement sensor
The displacement of any two points that are sufficiently far apart conveys the information
about twist motion. Hence, the two points used for testing were chosen to be as close as
possible to the tip of the leading edge and the tip of the trailing edge as shown in Figure
4-2. The distance between these two points is 76mm.
79
Point 2 at the
leading edge of the
tip of the
assembled blade
Point 1 at the
leading edge of the
tip of the
assembled blade
Figure 4-2: Two points to measure the deflection by the laser displace sensor
4.1.2 Bending Mode Results
Figure 4-3 to Figure 4-12 are a complete set of results of the bench top tests for the
assembled blade in the bending mode. Figure 4-13 to Figure 4-22 are also a complete set
of results of the bench top tests for the assembled blade in the twisting mode.
For the bending mode, the parameter of interest is the amplitude of the tip bending of the
assembled blade. This is inferred by averaging tip bending at both the leading edge and
the trailing edge. In the bench top tests, the tip deflections of the leading edge and the
trailing edge were measured and the results are shown in Figure 4-3.
80
It is observed that the leading edge and the trailing edge of the assembled blade have
similar frequency response, based on the natural frequencies and amplitudes of the
resonances. This result can be explained by the coupled nature of the assembled blade.
In contrast, Figure 3-5 shows that the leading edge spar and the trailing edge spar of the
graphite spar assembly (with no foam blade) have different natural frequencies, and the
amplitudes of resonances are different.
We conclude that adding the foam blade
significantly couples and changes the characteristics of the spars.
Based on the results shown in Figure 4-3, the average amplitude of the tip deflection of
the assembled blade was inferred and plotted in Figure 4-4. Obviously, this result does
not meet the 0.5 mm tip displacement for the active rotor blade.
It is observed that the assembled blade achieves 0.2mm/V bending between 50 and 300
Hz, and 0.1mm/V bending over a much wider range of frequencies. The first natural
frequency is 74 Hz with an amplitude of 4.75 mm/V. The second natural frequency is
216 Hz with an amplitude of 1mm/V. The third natural frequency is 300 Hz with an
amplitude of 0.345mm/V.
The author concludes that the bandpass bandwidth over which at least 0.5 mm/V (Peak to
Peak) tip bending is maintained 38 Hz (See Figure 4-4) and the bandwidth over which at
least 0.25 mm/V (Peak to Peak) tip bending is maintained 249 Hz.
81
Tip Displacement Transfer Functions (Bending Mode)
--------------- ------- r--------------------------------------------+-Leading Edge
Trailing Edge
4.5 - -------- ------- ------ --------------4 - - - ----- + ------ ----- +------- - -------- -- - + - - 5
3.5
----------
-----------------------------------------------------
E
-
E
1.5 - --------- ------- I------- I------- ------- I ------ ------- I------ ------ I
1
*1
0
400
300
200
100
------ -1
- --+------ +-I
-------I--------- I-----+-
- -
0. -- - -
1000
900
800
700
600
500
Frequency [Hz]
Figure 4-3: Measured transfer functions from the command signal applied at piezos to the
assembled blade's tip deflection at the leading edge (LS) and the trailing edge (TS) (Bending
Mode)
Average Tip Displacement Transfer Function (Bending Mode)
4.5
------
4 -
------ ----------------- -
---
---- -----
-----
3.5 ---- --------------
------
---- -----
----
--------
-----
----
------------
---- - -
-
-------
-
E
E
2.
- ----1.5 ---- ----+------
0.5 01
0
- -----100
200
---300
----- - ----
Ir
-
-----
-
-
600
500
400
Frequency [Hz]
-- - - -
- --
700
800
900
1000
Figure 4-4: Inferred transfer function from the command signal applied at piezos
to the average tip displacement (ATD) of the assembled blade (Bending Mode)
82
Figure 4-5 illustrates the comparison of the average tip displacement in the bench top
tests for the graphite spar assembly vs. the assembled blade.
Average Tip Displacement Transfer Functions (Bench top tests)
1 - - I-- Graphite Spar Assembly
--- Assembled Blade
-
10
-
410
100
'~
300
400
500
600
700
800
900
100[
Zoom In
1
I
0.8
flf
0.6
Graphite Spar Assembly-
-
'A
0.4
a
a
0.2
a
a-
-
-- - -- - - - - -
0
-
~--~ ~-~--
~.
200
-------------
-------
------
------ ------- -------
- -
--
-
2
100
200
300
400
500
a
a
-- -------I
600
700
-
a
-
I
800
900
1000
Figure 4-5: Comparison of the average tip displacement in bench top
tests for the graphite spar assembly vs. the assembled blade
The following observations can be made from Figure 4-5:
1. The amplitude of vibration is reduced after the foam blade is bonded.
The
addition of the foam blade to the graphite spar assembly not only adds mass, but
also adds stiffness to the assembled blade.
2. The natural frequencies are significantly reduced after the foam blade was bonded
to the graphite spar assembly.
3. The maximum amplitude is reduced to about one third not one tenth, which was
predicted by the blade finite element model in [2].
83
Figure 4-6 and Figure 4-7 are plots of both magnitude and phase of the transfer function
of the four strains to the command signal. As mentioned earlier, the root strains are the
signals we used to estimate the tip displacement in the in-situ test.
The following observation about the measured root strains are made:
e
Two poles dominate the root strains transfer functions at frequencies of 74 Hz and
216 Hz.
*
The maximum amplitude of root strains is less than 165 microstrain, which is
37% of the maximum amplitude of root strains before the addition of the foam
blade.
*
Both strains in the suction side have the same phase. Similarly both two strains in
the pressure side have the same phase. This result is consistent with coupling of
the bending mode of the assembled blade.
*
Both strains of the leading edge spar are out of phase. Similarly both strains of
the trailing edge are out of phase.
This makes sense because when the spar
deflects, one side extends, while the other side contracts.
Leading Edge Spar Strain Transfer Functions (Bending Mode)
200
-----_-------------+-
150 - ---------------------
o-0-
-
----_ _-------------__---Leading Edge Spar Suction Side
Leading Edge Spar Pressure Side
---
100 --------- ----- ------ -------- ------- ------ j------- ------- -------100
200
300
400
500
600
700
800
900
1000
-400
0------- - -100
200
300
400
- - - - - - 500
600
700
B00
900
1000
Frequency [Hz]
Figure 4-6: Measured transfer functions from the command signal applied at
piezos to root strains at the leading edge (LE) of the assembled blade
84
200
150
Trailing Edge Spar Strain Transfer Functions (Bending Mode)
------- ------ ------- ------- ------- ------ -------- I
---- - ---------------Trailing Edge Spar Suction Side
Trailing Edge Spar Pressure Side
--------- -------------
0
200
100
500
400
300
700
600
800
900
1000
-200---+-----
100 ------------------------------- ------- ------ ------O0 - ------3
-400
100
200
---
-----
300
400
-----------
500
Frequency
Figure
4-7:
piezos
to
root
strains
at
the
trailing
edge
Mode)
85
900
I
1000
[Hz]
from
functions
transfer
Measured
800
700
600
(TE)
command
the
of
the
assembled
signal
blade
applied
(Bending
at
Transfer functions from root strains to tip displacements
For the tests of the assembled blade, estimation techniques have been developed to
determine tip deflection based on strain gages bonded on the roots of the graphite spars.
These methods begin with determination of the transfer functions from root strain to tip
displacements.
Bench top tests are performed to obtain these tip estimation transfer
functions. Based on these transfer functions, we can estimate the tip displacements in
both bench top and rotating tests. In the bench top tests, we verify that the results are
accurate using laser displacement sensor results.
results are available at the present time.
In the RPM tests, only strain-based
Ultimately the Eddy Current Sensor
measurements will help to verify the estimation results of the RPM tests. With the goal
in mind, a series of bench top tests were performed to dynamically characterize the
assembled blade.
Techniques for measuring transfer functions from root strains to tip displacements used
for the assembled blade are the same as those used for the graphite spars alone. However,
the outputs of estimation transfer functions are different. For the assembled blade, the
outputs are the tip displacements of the leading edge and the trailing edge of the
assembled blade. These two points are shown in Figure 4-2. However, for the test of the
graphite spars, these two points do not exist. That is the reason that different outputs are
used.
For the bending mode and twisting mode, the dynamic responses to actuation are
different. Hence different sets of estimation transfer functions must be measured. The
results for the bending mode are presented in Figure 4-8 and Figure 4-9. The results for
the twisting mode are presented in Figure 4-18 and Figure 4-19. Comparing Figure 4-8
and Figure 4-18. The following observations and comments can be made:
i)
Estimation errors are dependent on excitation frequencies.
ii)
The estimation can only capture the sinusoidal response in steady state, and
can not capture the transient response.
86
Leading Edge Tip Displacement Estimation Transfer Functions (Bending Mode)
Leading Edge Spar Suction Side
Leading Edge Spar Pressure Side
--
U)
0
0.2
------ -------- --------------- ------- ------ ---------------- -----E
a) 0.1
0
I
100
200
300
400
500
600
Frequency [Hz]
700
800
900
1000
500
0
_0
UD
(A
-
----
-500
M~
-1000,
100
200
--
300
400
-------- -------
------
-
500
600
Frequency [Hz]
700
800
900
1000
Figure 4-8: Measured transfer functions from the root strain at the leading edge spar (LES)
to the tip displacement at the leading edge (LE) of the assembled blade (Bending Mode)
Trailing Edge Tip Displacement Estimation Transfer Functions (Bending Mode)
I
I
I
I
Trialing EdgeSpar Suction Side
- Trailing Edge Spar Pressure Side
----------------------- ------------------------
U)
0
0. 2 E
E
- - - - - -- - - --I-- - - - - - -- - -
-- - --1-- - - - --
U)0.
CD
Ca
(U
100
200
-1-
---
I
I
300
400
500
600
Frequency [Hz]
700
800
900
1000
500
U)
U)
0)
0)
0
-o
U)
U)
(U
-c
-500
~I
I
------------------- -----
------I
I
-
- - -
I
I
- - --
-
I
I
- -
- - --
0~
-1000
100
200
300
400
600
500
Frequency [Hz]
700
800
900
1000
Figure 4-9: Measured transfer functions from the root strain at the trailing
edge spar (TES) to the tip displacement at trailing edge (TE) of the assembled
blade (Bending Mode)
87
In the bench top tests of the bending mode, estimated and measured tip deflections are
presented in Figure 4-10, Figure 4-11, and Figure 4-12.
These plots verify that the
estimated tip deflections in the bending mode are accurate.
Leading Edge Tip Displacement Transfer Functions (Bending Mode)
5
4.5
Measured Leading Edge
Estimated Leading Edge by Strain 1
Estimated Leading Edge by Strain 3
--------L------- ------ +1-------+----+-j-----+-L-----+-------
-----------------
4 --------3.5
-
------
---------------- ------ ------- ------- ------- ------- ------- -------
E
E
C
_0
-----------------1------I--------- - - - ------ Ir-------- - - --- - Ir------- - - L
- --- - - - --32 - ------- -L
- - - -- i1.5
--------
0.5
-------
---- L
.I
300
400
.------- +----------------------
S0
100
200
600
500
Frequency [Hzj
700
800
900
1000
Figure 4-10: Verification the estimated leading edge displacement of the
assembled blade (Bending mode)
88
Trailing Edge Tip Displacement Transfer Functions (Bending Mode)
5
Measured Trailing Edge
- Estimated Trailing Edge by Strain 2
Estimated Trailing Edge by Strain 4
--- -+ ----- ----- -----4 -- ----------------------------- -----
4.5
--------------------------
35
----------------
---------------- --------------
------
--------------
E
E
1.5 - -------- ,--------r------ -------- I------- -------1 -------I------- -------
1- -- - -------151 ------ ----0.5
-
+ ------
0
200
100
-----
----- +------- ------ +------
-----
400
300
-----
L----- -
- L ------ ------
700
500
600
Frequency [Hzj
L-- ---
800
-----900
1000
Figure 4-11: Verification the estimated trailing edge displacement of the
assembled blade (Bending mode)
Average Tip Displacement Transfer Functions
Measured By Laser
Estimated By strains
4. ---------
------- -------------------------
45
-------
-----
35
---------
------- ------ ------- -------- ------ ----
----- +-------
---------
+-----
-----
+----
---------------
E
E
2
-8
--
-
.8
--
-
15t
1.5 ------- 0.5
-- - -
100
200
--
- -
- -
-------
-- -
-- --
2 - -- ------- -----
- --
-
-
- --
-
L
-
- --- ----- +------ +-------
-------I-------
-.
400
-
-
----- -- ------- ------- --r------
- - - -L- - --j-L-------
300
-
600
500
Frequency [Hz]
JA- - - L --- --
700
800
900
1000
Figure 4-12: Inferred transfer function from root strains at the leading edge spar
(LES) and the trailing edge spar (TES) to the average tip displacement (ATD) of
the assembled blade
89
4.1.2 Twisting Mode Results
Tip Displacement Transfer Functions (Twisting Mode)
5
- - - -- - - -- - - - - - - - 7 -- - - - - - - - - - - - - - - 0 - -- -
Leading Edge
4.5
-------------------------------------
-------
4 ------3.5
-----------+------
1------ - ----1---------- ++----------
------- --------------- ------- -------------------
------------------
-
Trailing Edge
E
0.5 -
0
1000
900
800
700
600
500
400
300
200
- - -
- - -+
-
---- ------ +-------- ------------ --
-------
100
Y------
- --
- - - - - - ---- - - - - - - - -------------
0.51.
Frequency [Hz]
Figure 4-13: Measured transfer functions from the command signal applied at piezos to the
assembled blade's tip deflection at the leading edge (LE) and the trailing edge (TE) (Twisting Mode)
Tip Twist Tranfer Function (Twisting mode)
---- -- - - --- -- -- -- -- -- - - ---- - ---- ------- - ---- ------- -- -- - ---- ---
45 --
4 ---------
2 .
--
-
-
--
5--
- - -
-
-
- --
-- -- - --
--
--
-
-r
-
-
-
-
-
--
-
--
- -
- -- ---
0
0
-
--
-
0
-
-
30
----
-
--
40
0.5------------------------
-
--
-
-
--
-
---
--
-
--
5
-
-
--
-
-
-
-.-
.-
--
-
---
00
80
70
60
50
-
-
-- -
1.5---------------- -----------------.1.
-r -
-
-
-
-- ---
- -
--- -
-
-----
-----
---
----- ------------
1.2----------------------
----
------- ------------- ------
2.5 -------------------------
.3 ------ -------------3>
-
---
--
-
-
-
----
--------------
-
-
10
.-.
Figure 4-14: Inferred transfer function from the command signal applied at
piezos to the Tip Twist (TT) of the assembled blade (Twisting Mode)
90
From Figure 4-14 and Figure 4-15, it is observed that the assembled blade achieves 0.1
degreeN twist over a wide range of frequencies. The first natural frequency is 74 Hz with
The second natural frequency is 216 Hz with an
an amplitude of 1.87 degreeN.
amplitude of 0.45 degreeN. The third natural frequency is 300 Hz with an amplitude of
0.48 degreeN. The fourth natural frequency is 780 Hz with an amplitude of 0.26
degreeN.
The author concludes that the bandpass bandwidth over which at least 0.5 DegreeN tip
twisting is maintained 12 Hz. (See Figure 4-14) The bandpass bandwidth over which at
least 0.25 DegreeN tip twisting is maintained 29 Hz.
Figure 4-15 illustrates the comparison of the tip twist transfer function in the bench top
tests for the graphite spar assembly vs. the assembled blade.
Tip Twist Transfer Functions (Bench Top Test)
----
S--- ---------
-
------
-----
-----
----------
--
-------------
- ---
-
Graphite Spar Assembly
Assembled Blade
---
----
---
--- -
1a
200
100
500
400
300
800
700
600
900
1000
Zoom In
1.8 CD
-
0
--
--
200
100
': -----------
0.2
~
--0r
t .6J
Graphite Spar Assembly
Assembled Blade
------- -- - -- - - -- - -------ZoomaIn
300
a
a------------
- - ara
S
M
0
100
-
raphite
-
--- ------
200
300
1
400
-
-
a
--
1
500
-
b
As e
-
-
-
-
a-
900
1000
- -I
bade
- a--- - -
-
800
900
1000
800
700
600
500
400
-
-
d
la d
....
1
600
700
Figure 4-15: Comparison of the tip twist transfer functions in bench top
tests for the graphite spar assembly vs. the assembled blade
91
The following observations can be made from Figure 4-15:
e
The twist amplitude is reduced after the foam blade is bonded. The addition of
the foam blade to the graphite spar assembly not only adds mass, but also adds
stiffness to the assembled blade.
" The natural frequencies are significantly reduced after the foam blade was bonded
to the graphite spar assembly.
* The maximum amplitude is reduced to 18%.
From Figure 4-16 and Figure 4-17, the following observation about the measured root
strains for the twisting mode are made:
" The maximum amplitude of root strains is less than 80 microstrain, which is 18%
of the maximum amplitude of root strains before the addition of the foam blade.
" Both strains in the suction side are out of phase. Similarly the two strains on the
pressure side are out of phase. This result is consistent with coupling of the
twisting mode of the assembled blade.
Leading Edge Spar Strain Transfer Functions (Twisting Mode)
150
M0
-400
-G00
Leading Edge Spar Suction Side
Leading Edge Spar Pressure Side
---------- ------- L------ -
100
200
300
400
500
600
-------- ------ - - - --- - --------------
100
200
300
400
600
500
Frequency [Hz]
800
900
1000
--- ------------
-
---
800
900
700
700
1000
Figure 4-16: Measured transfer functions from the command signal applied at piezos to
root strains at the leading edge spar(LES) of the assembled blade (Twisting Mode)
92
Trailing Edge Spar Strain Transfer Functions (Twisting Mode)
------ ------- ------- ----------------------r------- :--------------------Trailing Edge Spar Suction Side
-- Trailing Edge Spar Pressure Side] .
------ ---150 -------
.,200
------------ -- ----- ------- ------- -
100
50
-
100
100
------
------
200
-------------
--------------------
L
300
---7 0------- 0--------------
400
500
-------
700
600
--
800
----
900
1000
------------------ - --------- -- -- - ---------- - -------I------ ---------
4-17:M ---------
- ---- ------
----------n-----t------r
r-----+
a) -200 -- +------+------- +r------I-----+------ +-Ir-----+ ----- +-Ir-----+---------
-L-300 --+ ---------------400
100
200
300
-----400
------500
600
----
------700
800
-- + --900
1000
Frequency [Hz]
Figure 4-17: Measured transfer functions from the command signal applied at piezos to
root strain at the trailing edge spar(TES) of the assembled blade (Twisting Mode)
For the twisting mode, estimation transfer functions are measured and plotted in Figure 418 and Figure 4-19.
93
Leading Edge Tip Displacement Estimation Transfer Functions (Twisting Mode)
I
I
I
I
I
I
I
I
Leading Edge Spar Suction Side
Leading Edge Spar Pressure Side
-
U)
0
-E
S0.2
-------- -- -- - -------------- ------- ------- --------------- -------
E
E
a) 0.1
0I
200
100
300
400
600
500
Frequency [Hz]
700
900
800
1000
500
0)
0)
0)
0)
----
0
0)
(I)
(U
- -- -- -- - --- - -- - - ------ - - ----------- --
-500
-1000
-- - - -- -- -- - - ----- ------- --------------- -------
200
100
300
400
600
500
Frequency [Hz]
700
900
800
1000
Figure 4-18: Measured transfer functions from root strain at the leading edge spar (LES)
to the tip displacement at the leading edge (LE) of the assembled blade (Twisting Mode)
Trailing Edge Tip Displacement Estimation Transfer Functions (Twisting Mode)
I
I
I
I
I
+
0.2
i
I
i
1
Trialing Edge Spar Suction Side
Trailing Edge Spar Pressure Side
--------- ------- ------ ------- ------- ------ ------- ------- #-------
E
E
---------- -------
0.1
-o
ii
C
0
i100
------
-----
------- ------
~1A~
200
300
400
600
500
Frequency [Hz]
700
800
900
1000
200
300
400
600
500
Frequency [Hz]
700
800
900
1000
~-
500
0
ct
-500
-1000
100
Figure 4-19: Measured transfer functions from root strain at the trailing edge spar (TS) to
the tip displacement at the trailing edge (TE) of the assembled blade (Twisting Mode)
94
In the bench top tests of the twisting mode, estimated and measured tip deflections are
These plots verify that the estimated tip
presented in Figure 4-20 and Figure 4-21.
deflections of the leading edge and the trailing edge are accurate. Figure 22 verifies that
the estimated tip twist of the assembled blade is accurate.
Leading Edge Tip Displacement Transfer Functions (Twisting Mode)
4.5
---------------- ------
--
- Measured Leading Edge
Estimated Leading Edge by Strain 1
Estimated Leading Edge by Strain 3
4
-- ------- -- ----- -- ---1-- -- --- r--- -- -- -- -- -- -- -- --- -- -- ---3.5 --- --- -- 4-E
Under-estimate the leading edge
tip deflection off-resonance
2.5
1.5
- ------ -- ----------
+------------
------
------ + -----
0
n
1
3
2
3001
400
500
600
700
800
900
1000
Frequency [Hz
Figure 4-20: Verification the estimated leading edge displacement of the
assembled blade (Twisting mode)
The estimation transfer functions for the assembled blade appear to consistently underestimate the deflection off-resonance (See Figure 4-20, Figure 4-21, and Figure 4-22 for
the twisting mode).
One of the reasons for this problem is the low amplitudes of the
estimation transfer functions off-resonance. A similar problem was also found for the
bending mode estimation (see Figure 4-10, Figure 4-11, and Figure 4-12). This problem
is mitigated somewhat by the over-estimation of the deflection in in-situ tests, in which
the centrifugal stiffening effect causes deflections to be less than those estimated using
the strain gages.
95
Trailing Edge Tip Displacement Transfer Functions (Twisting Mode)
5
4.5
Measured Trailing Edge
Estimated Trailing Edge by Strain 2
Estimated Trailing Edge by Strain 4
------------------
4.
----- +----------------------- ------------------
---------------
- --------- ------- ------- ------- ------- ------ ------- I ------- -------
3
E
E
--- ---4-- - -I- - ----
2.5
2
-
0.5
----
--------
0
20
1
tip deflection off-resonance
---
-+------------
2
Under-estimate the trailing edge
---------
300
400
------------- ------
500
600
700
800
900
1000
L_______jFrequenCy [Hz]
Figure 4-21: Verification the estimated trailing edge displacement of the
assembled blade (Twisting mode)
Tip Twist Transfer Functions
45 -------------- r------ -------
Measured By Laser
----------- Estimated By Strains
-- - - - - - - I
4
4 --
------ -- - +-----
------ ---
+ - -+
------
----
Under-estimate the tip
---
--------
- 2.5
twist off-resonance
CL
15
- -------- ------
0.5 -
-------
---
------ ------------- -------
aI-
------------- ------- I
----- +-------- i-----+-J - ---+L-----+--------
0 1W43001
800%J'~900
Bo
o
l1000e
Frequency [Hz]
Figure 4-22: Inferred transfer function from root strains at the leading edge
spar (LES) and the trailing edge spar (TES) to tip twist of the assembled blade
96
4.2 In-Situ Tests of the Assembled Blade
A series of RPM tests was conducted to demonstrate the feasibility of the active rotor
blade concept.
The RPM experiments conducted are shown in Table 4-3. Compared with Table 4-1, the
measured transfer functions for the RPM tests of the assembled blade are fewer than
those of the bench top tests. The reason is that laser displacement sensor can only be
used in the non-rotating reference frame. Hence only four strain gages can be measured
from the spin pit.
The average tip displacement and tip twist can be estimated using
estimation transfer functions, which were obtained in the bench top tests of the assembled
blade. The inferred transfer functions for the RPM tests are shown in Table 4-4. Figure
4-24 to Figure 4-28 are a complete set of RPM results for the assembled blade in the
bending mode. Figure 4-29 to Figure 4-33 are a complete set of RPM results for the
assembled blade in the twisting mode.
The Measured root strain transfer functions for the RPM test
Table 4-3:
of the assembled blade
Mode\TF
s1
s2
s3
s4
C
C
C
C
Bending Mode
Measured
Measured
Measured
Measured
Twisting Mode
Measured
Measured
Measured
Measured
Table 4-4:
The Inferred transfer functions for the RPM Tests of the
assembled blade
zi
z2
C
C
C
C
Bending Mode
Inferred
Inferred
Inferred
None
Twisting Mode
Inferred
Inferred
None
Inferred
Mode\TF
97
0
When the rotating speed increased up to 3000 RPM, Some foam material at the leading
edge close to the piezos was lost. One reason is due to some small cracks that developed
when the author tried to fit the piezo to the grooves on the foam blade. The rest of the
assembled blade has been tested up to 5000 RPM and is shown in Figure 4-23. Hence
Epotec 301 and the bonding techniques are good enough for rotating speeds up to (and
perhaps beyond) 5000 RPM.
Figure 4-23: The assembled blade survived 5000 RPM
4.2.1 Bending Mode Results
98
Leading Edge Tip Displacement Transfer Funcitons Estimated by strain 1
-
100
200
300
400
500
600
700
800
200
300
400
500
600
800
700
900
1000
1000 RPM
-
100
0RPM
900
1000
2000 RPM
-
E
E
C
05
0
5
100
200
300
400
500
600
700
800
-
0
5
0
5
0
100
100
200
200
300
300
400
400
500
500
600
600
700
700
900
1000
3000 RPM
800
900
-
4000 RPM
800
900
1000
1000
-- 5000 RPM
100
200
300
400
500
600
700
800
900
1000
Frequency [Hz]
Figure 4-24: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the leading edge by the strain at the leading
edge spar suction side (RPM Tests of the assembled blade in Bending
Mode)
99
Trailing Edge Tip Displacement Tranfer Functions Estimated by strain 2
5
0 RPM
0
5
-II
100
200
300
400
500
600
700
800
900
1000
- 1000 RPM
0
5
100
200
300
400
500
600
700
800
900
1000
2000 RPM
E
E
C-
0
5
100
200
300
400
500
600
700
800
1000
3000 RPM
-
-0
900
CO
0
5
100
200
300
400
500
600
700
800
900
1000
4000 RPM
0
5
100
200
300
400
500
600
700
800
-
0
100
200
300
400
500
600
700
800
900
1000
5000 RPM
900
1000
Frequency [Hz]
Figure 4-25: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the trailing edge by the strain at the trailing
edge spar suction side (RPM Tests of the assembled blade in Bending
Mode)
100
Leading Edge Tip Displacement Transfer Functions Estimated by strain 3
0 RPM
100
200
300
400
500
600
700
800
900
1000
-1000RPM
100
200
300
400
500
600
700
800
900
1000
2000 RPM
E
E
aE
0
100
200
300
400
500
600
700
900
800
1000
3000 RPM
2=
0
100
200
300
400
500
600
700
900
800
-4000
0
100
200
300
400
500
600
700
800
100
200
300
400
500
600
700
RPM
900
1000
5000 RPM
-
0
1000
800
900
1000
Frequency [Hz]
Figure 4-26: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the leading edge by the strain at the leading
edge spar pressure side (RPM Tests of the assembled blade in Bending
Mode)
101
Trailing Edge Tip Displacement Transfer Functions Estimated by strain 4
5
-
100
200
300
400
500
600
700
800
100
200
300
400
500
600
700
900
1000
1000 RPM
--
0
0 RPM
800
---
900
1000
2000 RPM
2
-o0
100
200
300
400
500
600
700
800
900
1000
3000 RPM
0
5
100
200
300
400
500
600
700
800
1
900
1000
1
F-4000 RPM]
100
200
300
400
500
600
700
800
-
900
1000
5000 RPM
0
100
200
300
400
600
500
Frequency [Hz]
700
800
900
1000
Figure 4-27: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the trailing edge by the strain at the trailing
edge spar pressure side (RPM Tests of the assembled blade in Bending
Mode)
102
Average Tip Displacement Transfer Functions
5
0RPM
0
100
200
300
400
500
700
600
800
0
5
100
200
300
40
0
5D
10
nn
?n
39Q
490
500
600
700
800
900
1000
1600 RPM
900
1000
2000 RPM
E
3000 RPM
0
55
no
?no
39o
490
100
20
30
40
-50.
600
7Q
SQO
900
10 10
-4000
PM
790
890
9oo
1010
5000 RPM
0
100
200
300
400
500
600
700
800
900
1000
Frequency [Hz]
Zoom In
Average Tip Displacement Transfer Functions
.4
0
0.5
1
--
10
---- - --------- ------- ------- ------- ---
2nn
9n
490
Sp
790
RQO
RP
- 21000 RP
-000RPM
1631o
990
non
0-- - -------- - ----------- ----------------- ------- - --
3 0 RP
0
5000 RPM
>0.5
---------
0.E
1-
------- -------
_
0
0
100
200
300
400
500
600
Frequency [Hz]
700
800
900
1000
Figure 4-28: Inferred transfer functions from the command signal applied at
piezos to the average tip deflection by four strain gages (RPM Tests of the
assembled blade in Bending Mode)
103
4.2.2 Twisting Mode Results
Leading Edge Tip Displacement Transfer Functions Estimated by strain 1
0 RPM
-
100
200
300
400
500
600
700
800
-
0
100
200
300
500
400
600
700
900
1000
1000 RPM
900
800
-
1000
2000 RPM
E
E
-0
0
100
-
200
300
500
400
600
700
800
200
300
400
1000
3000RPM
CU-
100
900
500
600
700
800
900
1000
-- 4000 RPM
0
5
100
200
300
400
500
600
700
800
-5000
0
100
200
300
400
500
600
700
800
900
-
1000
RPM
900
1000
Frequency [Hz]
Figure 4-29: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the leading edge by the strain at the leading
edge spar suction side (RPM Tests of the assembled blade in Twisting
Mode)
104
Trailing Edge Tip Displacement Transfer Functions Estimated by strain 2
5
ORPM
0
5
0
5
100
100
200
200
300
300
400
400
500
500
600
600
700
700
800
900
-
1000 RPM
800
900
1000
1000
2000 RPM
E
E
0
CD
100
200
300
400
500
600
700
800
-
0
5
100
200
300
400
500
600
700
800
--
0
5
100
200
300
400
500
600
700
800
-
0
100
200
300
400
500
600
700
800
900
1000
3000 RPM
900
1000
40 00 RPM
900
1000
5000 RPM
900
1000
Frequency [Hz]
Figure 4-30: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the trailing edge by the strain at the trailing
edge spar suction side (RPM Tests of the assembled blade in the Twisting
Mode)
105
Leading Edge Tip Displacement Tranfer Functions Estimated by strain 3
5
0 RPM
0
100
300
200
400
500
700
600
800
900
1000
1000 RPM
0
5
100
E
E
0
CU
5
CD
-0
100
300
200
400
500
700
600
a
I
a
I
a
I
I
I
I
I
I
I
I
I
I
I
I
I
a
a
*
I
*
I
I
I
I
I
I
I
I
I
I
I
I
I
L
200
300
400
500
600
700
I
I
I
I
I
I
I
a
I
800
900
-
2000 RPM
800
900
I
I
0
5
100
200
j
I
300
400
500
600
700
800
-
100
200
300
1(00
3000 RPM
3
I
1000
400
500
600
700
800
900
1000
4000 RPM
900
1000
5000 RPM
0
100
200
300
400
500
600
700
800
900
1000
Frequency [Hz]
Figure 4-31: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the leading edge by the strain at the leading
edge spar pressure side (RPM Tests of the assembled blade in Twisting
Mode)
106
Trailing Edge Tip Displacement Transfer Functions Estimated by strain 4
-
I
I
I
I
500
600
700
800
l^I
100
200
300
400
200
300
400
500
600
700
-
900
10C0
1000 RPM
-
100
0 RPM
800
900
---
2000 RPM
800
900
1OC 0
E
-o
0
100
200
300
500
400
600
700
M
10C0
3000 RPM
CO
0
0
100
200
300
5
400
500
600
700
800
-
0
1000
4000 RPM
I
600
2
1
700
800
-
0
900
100
200
300
400
500
600
Frequency [Hz]
700
800
900
1000
5000 RPM
900
1000
Figure 4-32: Inferred transfer functions from the command signal applied at
piezos to the tip deflection at the trailing edge by the strain at the trailing
edge spar pressure side (RPM Tests of the assembled blade in Twisting
Mode)
107
Tip Twist Transfer Functions
-- 0 RP
0
1
n
-9--
)0n30n
rQ0
J,0-
790
R90
.6i?0_
790
an
.6f[L
7[
ago
pn
7
Z
990
1000 RPM
-500.
0
n
Agn
A1f
2Q3r
-5r _
0
a -
U
1
0
210
-iQ0
A9g
21p-
3Qn
4p A
spo
~
1 )o
9W0
200RPM
11M
930
-- 3000 RPM
RQn
2I
I
PO
15
0
200
100
-
400
300
I
an Irg
Fp
gn
3I
5000 RPM
1000
900
800
700
600
500
Frequency [HzJ
1no0
9n
4000 RPM
-
0
1010
Zoom In
Tip Twist Transfer Functions
0.5 - --------
------- --------- - - ------
----
------
0 RPM
0
1
0.5
10
100-RP
-
- - ------- - ----------- L------- ------- ------
0
~1I
,
-.
1
% . a
--------
0.5--p----- --
a
n
10
-a ------
-----Rn
rn
Ann
-----
------- -----
------
-------
------
-----
----
20
------
R0n
790
-----
--
P
9An
3000RPM
0
1
0.5
- ------
0
I
---0.5------------
1
0
-------
100
200
300
400
4---
------7
- ______
600
500
Frequency [Hz]
700
-
800
4000 RPM _
900
1000
Figure 4-33: Inferred transfer functions from the command signal applied at
piezos to tip twist by the four strain gages (RPM Tests of the assembled
blade in Twisting Mode)
108
The following observations and comments can be made about the RPM tests of the
assembled blade.
1. The vacuum level should be kept low enough (500micro Torr to 1 milli Torr).
There are two reasons for keeping the vacuum level low. The first reason is that
the foam blade is very thin, especially at the leading edge and the trailing edge. It
is not strong enough to survive the RPM test without an appropriate level of
vacuum. The second reason is to reduce the signal picked by the strain gages due
to aerodynamic forcing.
2. A typical frequency sweep from 50Hz - 1000 Hz lasts 2 minutes (settle time 20
cycles and integral time 20 cyles, 400 data points). The reason that the sweep
starts from 50 Hz is the limitation of the power amplifier. For command signals
less than 50 Hz, the output voltage of the power amplifier is significantly lower
than 90 V AC.
The reason that the sweep stops at 1000 Hz is to prevent
overheating of the piezo. The amount of heat the piezo produces is proportional
to the frequency applied. The experiments conducted have shown that piezos are
very easy to burn at frequencies between 1000 Hz and 1200 Hz. Figure 4-34 is a
typical burned piezo bonded on one of the graphite spars.
Figure 4-34: A typical burned piezo bonded on the spar
109
3. The most critical part in the spin pit is the bearings. Overheating is one of the
causes of bearing failure. The rubbing of the face seals against their mating rings
produces a lot of heat that may lead to the increase of the shaft temperature and
result in the bearings failure.
4. Figure 4-35 shows the slip ring wires that conduct signals to the rotating reference
frame. The channels used to conduct amplified command signals to the piezos
were the Channels 1,9,10,11,12.
Channel 9 was the ground signal for all four
piezos. Channel 12 was connected to the piezo bonded on the leading edge spar
suction side. Channel 11 was connected to the piezo bonded on the trailing edge
spar suction side. Channel 10 was connected to the piezo bonded on the leading
edge spar pressure side. Channel 1 was connected to the piezo bonded on the
trailing edge spar pressure side. The channels used to output the strain gage
signals were channels from 13 to 20
Channel 19 and 20 were connected to the
strain gage bonded on the leading edge spar suction side. Channel 13 and 16 were
connected to the strain gage bonded on the trailing edge spar suction side.
Channel 17 and 18 were connected to the strain gage bonded on the leading edge
spar pressure side. Channel 14 and 15 were connected to strain gage bonded on
the trailing edge spar pressure side.
Figure 4-35 Channels of the Slip Ring Wires to conduct signals
110
5. Lower face seal temperature monitoring was performed in the non-rotating frame.
The wires used with the pass through are shown in Figure 4-36. Vacuum leaking
was found through these wires. Hence, in the RPM tests, it was decided not to
use these wires with pass through to prevent leaking.
Pass through
Figure 4-36: Cables with vacuum passthrough
6. Different assembled blades have different strength due to fabrication. Right now
we do not know at which RPM the centrifugal force will be big enough to break
the component.
7. It is important to stop and check the blade after every RPM test. One must also
balance the rotor every time the RPM is increased.
8. At this stage, the balancing target is 0.1 mils (Peak to Peak) vibration at 5000
RPM, which corresponds to an output of 36 my from the BK signal condition
amplifier with a gain of 100 mv/ms 2 . This corresponds to "excellent" conditions
for the bearings, according to charts supplied by bearing manufactures.
9. A typical balancing scheme will take one and a half hours. The oil free air valve
should be closed during the process of balancing to prevent too much oil in the
bearings.
111
10. The micro connectors on the rotor need to be redesigned. The wiring of the active
blade was connected to the wiring connector by male and female micro
connectors. These connectors are installed with the pine aligned to the radial
direction.
The male connector was secured to the wiring connector and the
female connector only held in place by the pins. The centrifugal force has the
potential to disconnect the female connector. In fact, at 4000 RPM, one female
connector flew off its mating connector. Blue flash tape was used to prevent this
problem. Tests showed that the blue flash tape could tape these connectors until
5000 RPM. Redesign of the micro connectors may be needed if the RPM is
further increased.
112
Chapter 5
Conclusions and Recommendations
5.1 Conclusions
The conclusions have been divided into two sections: 1) fabrication of the sparactuated active compressor rotor blade, 2) characterization of the graphite spar
assembly and the assembled blade.
5.1.1
Fabrication of the Spar-Actuated Active Compressor
Rotor Blade
Obtaining spars from the graphite-epoxy blank
A water-jet cutter was used to follow the cutting path to obtain graphite spars.
The new cutting method yielded very accurate graphite spars. In all, three blanks
were cut. All the obtained graphite spars are in excellent condition. The machine
time for the cutter is less than 1 hour, including adjusting time.
113
Piezo package and wiring scheme
The piezo leads were redesigned to apply voltage to the four piezos separately and
avoid the slippage between the piezo package and the graphite spars during
bonding.
A new actuator arrangement and wiring scheme was applied to
guarantee fully working piezos with correct DC offset to prevent the piezos from
depoling.
Bonding improvement of piezo package to graphite spars
A milling machine and some additional layers of graphite epoxy prepreg were
used to obtain a flat mounting surface from the highly curved root surface of the
graphite spars. The gap between the piezo package and the mounting surface was
thus dramatically reduced. Hence, the thickness of the glue between the piezo
package and the mounting surface was made thinner
Foam Shaping
A 3D numerical milling machine was used to machine out the foam blade. The
finished foam blade has accurate geometry on the pressure side, the suction side,
and the spar grooves. However, more work is still needed to obtain the leading
edge and the trailing edge of the blade. The total machine time is about 7 hours.
Assembly of blade
Epotec 301 was used to bond the graphite spar assembly to the foam grooves.
The graphite spars fits well into the grooves on the foam blade. Epotec 301 and
the bonding techniques are good enough for rotation speeds of at least 5000 RPM.
5.1.2
Characterization of the Graphite spar assembly and the
Assembled Blade in both Bench Top and In-situ Tests
Note: The following results are based on the AC 90 Vrrs with DC 90 V bias signal
applied to the piezos.
114
In bench top tests of the graphite spars (bending mode), the bandpass bandwidth over
which at least 0.5 mm (Peak to Peak) tip bending is maintained 249 Hz, which is
about 92% larger than the result presented in [2]. The bandwidth over which at least
0.25 mm (Peak to Peak) tip bending is maintained 342 Hz, which is about the same
bandwidth as the result presented in [2] (See Figure 5-1).
(Refer to page 59 for
definitions of bandpass bandwidth and bandwidth)
In bench top tests of the assembled blade (bending mode), the bandpass bandwidth
over which at least 0.5 mm (Peak to Peak) tip bending is maintained 38 Hz and the
bandwidth over which at least 0.25 mm (Peak to Peak) tip bending is maintained 249
Hz. (see Figure 5-1)
The following comments can be made from Figure 5-1 for both bench top and in-situ
tests:
1. The amplitude of vibration was reduced after the foam blade was bonded. The
addition of the foam blade to the graphite spar assembly not only adds mass,
but also adds stiffness to the assembled blade.
2.
The natural frequencies reduced after the foam blade was bonded to the
graphite spar assembly
3.
The maximum amplitude is reduced to about one third by the addition of the
foam blade, as determind by bench top tests.
4.
The estimated in-situ tip deflection transfer functions have the same trend as
the corresponding bench top test.
115
Average Tip Displacement Transfer Functions
5
0:J
-- Spar Benchtop
Assembled Blade Benchtop
100
200
300
00
I
400
600
0D
0
100
200
600
50 0
400
300
900
1000
700
800
900
1000
Spar2000 RPM
--
C
0)
800
Spar 1000 RPM
Assembled Blade 1000 RPM
-
E
E
700
Assembled Blade 2000 RPM
200
100
300
600
500
400
200
300
400
800
900
1000
Spar 3000 RPM
Assembled Blade 3000 RPM
-
100
700
00
600
Freq uency [Hz]
700
800
900
1000
(a)
Average Tip Displacement Transfer Functions
1
05
E
0.5
--
100
[,--
300
200
-
-
600
500
400
Zoom In
I
eSpartBenchtop
700
800
900
1OC 10
Spar 1000 RPM
Assembled Blade 1000 RPM
-------
--
o
Assembled Blade Benchtop
- -
------- -------
0.5
V
U
Spa B
1
E
C
0E
MU
1
100
200
---------200
300
800
700
600
A
B
2
Spar 2000 RPM
500
400
--
o.5 - 100
300
600
-
--
0.5
------
0
100
700
800
300
400
900
1000
Spar 3000 RPM
Assembled Blade 3000 RPM
*--------
I
200
1000
- Assembled Blade 2000 RPM
500
400
900
600
500
Frequency [Hz]
700
800
900
1000
(b)
Figure 5-1: Comparison of the average tip displacement of bench top and
in-situ tests transfer functions for the graphite spar assembly vs. the
assembled blade
Tip Twist Transfer Functions
10.
10
100
200
300
5-
Spar Benchtop
Assembled Blade Benchtop
-
----------
S-
400
500
600
700
800
900
1000
Spar 1000 RPM
Assembled Blade 1000 RPM
----
0)0
10
5
0
10
100
200
400
500
----------------
------
600
200
800
300
500
400
200
300
400
900
1000
Assembled Blade 3000 RPM
-
600
700
800
.
,
-- Spar BnhO
RPM
J
900
-
,
1000
Assembled Blade Bh0RPM
----------------------
100
700
Spar 3000 RPM
'-------
100
5- -
0L
300
500
600
700
800
900
1000
Frequency [Hz]
(a)
Tip Twist Transfer Functions
ill?'
I
Spar Benchtop
Assembled Blade Benchtop
--
0.5
------
-
---
100
300
200
400
-
500
600
0
C)
0
100
200
300
1
1
400
800
900
1000
Spar 1000 RPM
Assembled Blade 1000 RPM
-
----------
----
0.5
700
ZoomlIn
500
600
700
800
900
1000
Spar2000 RPM
0.5 Fie52
0i
100
0.5 A100
-
t
200
-
------Cmaio
the-g-r-a-p-h
300
400
Assembled Blade 2000 RPM
t
600
500
fi
-Spar
300
400
800
900
1000
3O0RPM1
Assembled Blade 3000 RPM
-------------
200
ve
700
500
600
Frequency LHzI
700
800
900
1000
(b)
Figure 5-2: Comparison of the tip twist transfer functions of bench top
and in-situ tests for the graphite spar assembly vs. the assembled blade
In bench top tests of the graphite spars (twisting mode), the bandpass bandwidth over
which at least 0.5 degree tip twisting is maintained 264 Hz, which is about 210%
larger than the result presented in [2]. The bandwidth over which at least 0.25 degree
tip twisting is maintained 354 Hz, which is about 136% larger than the result
presented in [2]. (See Figure 5-2)
In bench top tests of the assembled blade (twisting mode), the bandpass bandwidth
over which at least 0.5 degree tip twisting is maintained 12 Hz.
The bandpass
bandwidth over which at least 0.25 degree tip twisting is maintained 29 Hz. (See
Figure 5-2)
The following comments can be made from Figure 5-2 for both bench top and in-situ
tests:
a) The twist amplitude was reduced after the foam blade was bonded. b) The natural
frequencies reduced after the foam blade was bonded to the graphite spars. c) The
maximum amplitude is reduced to 18% by the addition of the foam blade, as
determined by bench top tests. d) The estimated in-situ tip twist transfer functions
have the same trend for the corresponding bench top test.
5.2 Recommendations for Future Work
*
In order to obtain a better bonding of piezo packages to the graphite spars, a
vacuum bag should be used. The vacuum bag not only applies even pressure
to the piezo packages, but also helps outgas the epoxy. Hence a thinner layer
of glue can be obtained.
The expected result is better actuation of the
assembled blade.
*
In order to improve the strength of the foam blade, a layer of graphite epoxy
prepreg can be used to wrap the assembled blade. However the stiffness of the
assembled blade will increase. This is the tradeoff between the strength and
118
stiffness of the assembled blade.
Further experiment and analytical work
should be done to investigate this problem.
*
In the in-situ tests, the estimated tip bending and twisting should be adjusted
for centrifugal stiffening effects.
119
120
Appendix A
In-Situ Dynamic Balancing Scheme
for the Active Rotor
Imbalance, identified by excessive levels of vibration at the same frequency as the rotor
speed, is one of the most common problems causing vibration in rotating machinery. The
extra loading due to imbalance can drastically shorten bearing lifetime. In order to keep
the bearings in healthy condition, an in-situ balancing scheme was performed for the
active rotor tests.
Conventionally, rotor balancing is done by phase measurement. The purpose of phase
measurement is to find the location of the imbalance. However, in some situations, it is
useful to be able to balance the rotor without phase measurement. For example, accurate
phase measurements of the imbalance are often difficult or impossible to implement. For
the spin pit rig, a balancing scheme without phase measurement was used. For each set
of balance, 5 runs were made and vibration amplitudes were measured by the
accelerometer and the B&K charge condition amplifier.
A graphical procedure can
determine the correct phase and proper weight for correcting imbalance.
121
A. 1 Experimental set up
An accelerometer, a B&K charge condition amplifier, an oscilloscope, a dynamic signal
analyzer, and a few trial weights are the equipment used to perform dynamic
Accerometer
Housing of
the Bearing
Figure A-1: Experimental Set Up for Balancing
balancing. The vibration is measured by the accelerometer at the housing of the bearing.
The accelerometer was connected to the B&K charge condition amplifier to amplify the
vibration signal. The oscilloscope and dynamic signal analyzer can be used to read out
the vibration level. Figure A-I shows the experimental set up.
A.2 Procedure
This procedure requires five runs to complete the rotor balance, since no phase reference
is used for these tests. First, without adding a trial weight, the rotor is run at the desired
balancing speed.
The vibration amplitude is measured by the accelerometer at the
housing of the bearing. For each of the next three runs, a trial weight is placed at three
122
different positions that are all at the same radius, at positions 120-degrees apart on the
rotor. The three corresponding vibration levels are measured, and used to compute the
location and weight of the balance required to eliminate the imbalance from the first test.
Finally, the rotor is run with a balancing weight and the vibration is measured to compare
the results to the first run. The data for the four runs might appear as those shown in
Table A-1.
Table A-1: Typical Rotor Balancing Data
RU
N
1
2
3
4
5
Rotating
speed
Position of
Trial Weight
(RPM)
on Rotor
4200
4200
4200
4200
4200
N/A
00
1200
2400
1950
Amount of Trial
weight (grams)
Vibration
Reduction
Vibration
Mils Pk-Pk
and Symbol
0.061
0.130
0.101
0.062
0.023
N/A
13.4
13.4
13.4
10.0
(OA)
(AD)
(BE)
(CF)
(OG)
E
Figure A-2: A diagram used to locate the light spot
123
62%
Figure A-2 is a diagram used to locate the light spot. The procedure is the following:
Using point 0 as a center, draw a circle with a radius of 0.061 to scale and locate points
A, B, C on the circle at the same angular positions as were used on the rotor. Using point
A as a center draw a circle of radius 0.13 to scale. Using point B as a center, draw a
circle of radius 0.101 to scale. Using point C as a center, draw a circle of radius 0.062 to
scale. The circles will all intersect at a common point or nearly so. Take G as the average
point of intersection, and join OG. The proper position for the final correction weight is
parallel to OG. The amount of the final correction weight is OA/OG times the amount of
the trial weight used. In this example, it is 0.061/0.082 times 13.4 grams, which gives 10
grams for the balance weight. A vibration reduction of about 62 percent is obtained. This
is about as much as can be achieved per correction when phase measuring equipment is
used.
A.3 Two important things about balancing
Balancing Speed
Rotating speed has a drastic effect on the measured unbalance vibration level because the
centrifugal force F , due to unbalance, is proportional to speed squared. During the
balancing procedure the rotating speed must be steady at a predetermined speed.
Trial Weights and Balancing Weights
Bolts and nuts were used as weights added on the active rotor. Balancing should start
with very small weights. It is also important to make sure those weights are well fastened
and keep in mind to stay out of the line of flight.
124
Appendix B
Procedure for Generating Machine
Tool Paths by Mastercam
The geometry of the part machined is defined with ProE models (files:
altblmachineu2 and alt blmachine_l-mod). From within ProE, the geometry can be
exported to IGES format files.
The IGES format files are readable from within MasterCAM software available at the
Laboratory for Manufacturing Productivity (LMP) computer cluster. MasterCAM
software is used to generate the machined tool paths and their associate NC code.
" Open MasterCAM software
1. Windows Start menu
2. Program
3. MasterCAM8
4. Mill8.1
" Open IGES File
1. Choose
125
0
Main >File>Converters>IGES>Read file
2. Choose the corresponding files such as altblmachineu2.igs
Part One: Creating a Rough Parallel Toolpath
" Select the surfaces for the toolpath and select the tool
1. Choose
e
Main>Toolpaths>Surface>Rough>Parallel>Boss
2. Choose
e
All>Surfaces>Done
3. Right-click in the tool display area and select the 1/8 "Ball endmill".
0
Click Ok.
" Enter the tool parameters
1. Right click on the tool icon
2. Select the Tool - Spherical End Mill tab
3. Enter the values shown on the following dialog box.
126
Define Tool
Tool -Spherical End MIll Tool TypeI Parmeters I
I
Calc. Speed/FeedI
KCance
4. Select the parameters tab
5. Enter the values shown on the following dialog box.
127
Help
Deine Tool
Tool -SphericaldEr
MI Too Type Paramete
Cale. Speed/Fed
FinishXY step
FinishZ step
RoughXY step (%)
Rough Z step
|.0
O.0
Saes to libray..
10.0
Job setup...
Required pilot dia
Dia offset number
0.0
Length offst number 0
Feed rate
130.0
Plunge rate
15.0
Retract rate
20.0
Spindespeed
13000
Number of flutes
Iaride
Spind rotation
r CW
CCW
cocan
rO1ff
r Food
Z of mtt cutting sped
rmist
Spinde
rMetric Vauaes
%of matL feedper tooh 0.0
Tool Re name
C:\PROGR AM FILES\MCAM
Tool name
1 /8!'Ball endmitl
Manufacue'
toolce
Chuck
I-OK
I
Cncel
6. Select the Job set up tab
7. Enter the values shown on the following dialog box.
128
I
Help
Job Setup
e
Enter the surface parameters
1. Select the Surface parameters tab.
2. Enter the values as shown on the following dialog box
129
Surface Rough Parallel -CTEMP\001.NCI -MPROTMX3
Tool paametes
Suiface paameteft Rough psai parameters
Tip comp
$
C
Absolute
R tratJ
ioOr t
C
Absolute
C
F
rretract
Absolute
SRapid
r-
/20
r
s10
hlave
o drive surfaces
d5
Incremental
incremental
nr..emen a
Top of stoec
ro Absolute
incremental
10.25
Feed plane
[TIP
F~
Us check surfaces
Stock to leave
on check surfaces
Fa Pompt for tool center boundary
incremental
Direcion.j
Regen.I
OK
3. Choose OK.
4. Select the Rough parallel parameters tab.
5. Enter the values shown on the following dialog box.
130
Cancel
Help
Surface Rough Parallel - C:\TEMP\0001.NCI - MPROTMX3x
Tool parametem
Surface parameters Rough paralel parameters
Cut
tolerance
Cutting method
IZigzag
Max stepover...
0.1
Ma
0.0
ng
are
Max stepdown:
Plunge wontrl
( Alow multiple pknges along cut
C Cut from oneside
i Cut from both sides
F Prompt for starting point
r Alow negativeZ motion along surface
W Alow positiveZ motion along surface
Gap setting...I
OK
Hel
_Cance
6. Choose the Cut depths button.
7. Enter the values shown on the following dialog box.
IutDepths
C Absolute
re' Incremental
Inwassmental depths4
Absolute dOppt
Minimum depth
Maximum depth
0.0
Adjustment to other cuts
1-1.0
Select depths.,
00
Adjustment to top cut
Critical depths...
z
Relative to
OK
8. Choose OK twice
131
I
0
Cancel
-.
Help
1
Part Two: Creating a Finish Parallel Toolpath
"
Select the surfaces for the toolpath and select the tool
1. Choose
e
Finish>Parallel
2. Choose
e
AII>Surfaces>Done
3. Right-click in the tool display area and select the 1/8 "Ball endmill".
" Enter the surface parameters
1. Select the Tool Parameter tab.
2. Enter the values as shown on the following dialog box
Surace finish Parallel -
Tool parameters
Surface parameters I Finish paral parametes|I
Left 'cick'on tool to sele;
F
x:
TEMP\0001.NCI - MPROTMX3
FTTaII
Tool#
Tool name
Head #
Feed rate
Dia offset
Plage rate
Len. offset
Retract rate 20.0
*gcliqo.Io
e
Conerradus 1dl0i25
Tool dia
Program #
15.O
wolefne new to
10Spindle
speed
|Off
Coolant
Seq sta
3500
DW
Seqinc.
ChangeNC.
Comment
F-
Home pos,.
.:j, F
Rotary axis...
A
j
r
Ref point...
ru.
T/C plane..
r,
Tool dspla..
r-
anned text.
r To batch
OK
3. Select the Surface Parameter tab.
132
arcel
vaues
Help
4. Enter the values as shown on the following dialog box
Surface Finish Parallel - C:\TEMP\0001.NCI - MPROTMX3
Tool parameters
Surface parameters Finish parael parameters
FClearance..5
r
C
Absolute
C
r* Absolute
r
F
'Or.
I
r
Stocktoleave
0.0
on drive surfaces
Incremental
F Use check surfaces
Feed plane
r- lIncremerita
r Absdkte
:jw
Incremental
2.0
Retract..
r
ITip
Tipcp
Stock to leave
on check surfaces
Rapid retact
To ofstock...
0.0
SAbsoute
Incremental
F Prompt for tool center boundary
D'reton.
WK
Cance
5. Enter the values as shown on the following dialog box
6. Choose OK
7. Select the Finish parallel parameters tab
8. Enter the values as shown on the following dialog box.
133
Help
Surface Finish Parallel - C:\-,TEMP\0001.NCI - MPROTMX3
Tool paametersI Surface paraneters Finishpmaistpare ters
OPm10.W1
Maxseoe.
toceA
Cutting method
r
Izigzag
ahning
Promp for stating point
F Depth limits-
Gp ekgs..]Eg
9. Choose OK
10. Choose Done. Mastercam generates the toolpath, which should look like
the following picture:
Change the gap settings
1. Press [Alt + 0] to open the Operations Manage..
2. Select the Parameters icon for the Surface Finish Parallel toolpath.
3. Select the Finish Parallel Parameters tab.
4. Choose the Gap Settings button.
5. Enter the values as shown on the following dialog box.
134
Gap settings
Reset
Gap size
C Distance
F0.15
ri' %of stepover
Motion <Gap size, keep tool down
Smooth
r
Check gap motion for gouge
- Motion> Gap size, retract
F Check retract motion for gouge
F Optirize cut order
F
Plunge into previously cut area
F Folow tool center boundary at gap
Tangential arc radius:
0.0
Tangential arc angle:
j.0
__iK
.
a
_l Help
6. Choose OK twice
7. Choose Regen Path
8. Choose OK twice
Problems with machining
The Rohacell WF 2000 has cellular structures. The cellular dimensions are of the same
order as the blade tip thickness and groove thickness. Thus, the blade chipped off easily
at these very thin areas.
Another problem of machining blade is that it is very expensive. The cost of machining a
blade is rated at 8 hours* $50/hr =$400. If all the 26 blades of the rotor were to be
machined this way, the total cost of machining blades will be more than $10,000.
135
136
Bibliography
[1] W. A. Farahat. Dynamical Characterization,State Estimation and Testing of
Active CompressorBlades. Master's Thesis, Massachusetts Institute of
Technology, 2000.
[2] D. Sahoo. Manufacture and Testing of an Active compressorbladefor Aeroelastic
studies. Master's Thesis, Massachusetts Institute of Technology, 2000.
[3] G. L. Maahs. Design of an Active CompressorBladefor Aeroelastic Studies.
Master's Thesis, Massachusetts Institute of Technology, 1999.
[4] Crawley, Edward F. and Eric H. Anderson. Detailed models of Piezoceramic
Actuation of Beams. J. ofIntell. Mat. Syst. and Struct., Vol 1, page 4-45, 1990.
[5] Crawley, Edward F. and Javier de Luis. Use of Piezoelectric Actuators as
Elements o Intelligent Structure. AIAA Journal, Vol 25, No. 10, Pages 1373-1385,
October 1987.
[6] A.P.F. Bernhard. Smart HelicopterRotor with Active Blade Tips. Ph.D. Thesis.
Univeristy of Maryland, College Park, 2000
[7] C. E. S Cesnik, S. -J. Shin, and M.L. Wilbur. Dynamic response of active twist
rotor blades. Smart Materialand Structures, 10:62-76,2001
[8] Introduction to PRO-ENGINEER (Release 20). Parametric technology
Corporation Publication (www.ptc.com), 1998
137
[9] James D. Paduano. PersonalCommunications 2001-2003
[10] P.A. Lagace, M. Beaumont, J. C. Breuer, and C.F. Varnerin. TELAC
Manufacturing Class Course Notes. Technical Report 88-4b, Massachusetts
Institute of Technology, September 1992
[11] M.P.Blake, W.S. Mitchell. Vibration and Acoustic Measurement Handbook.
New York Publication, 1972
[12] Willcox, Karen, PersonalCommunications
138