Restriction/Classification Cancelled

Source of Acquisition
CASI Acquired

Restriction/Classification Cancelled
 ^^^^ FFF F ,FC ^FF ^ j ^ ^`9 ^F ^ ^ ^i.P ^^
NACA RM No. SE8A28 x F F F

NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS

RESEARCH MEMORANDUM

for the

Bureau of Aeronautics, Navy Department

VIBRATION SURVEY OF BLADES IN 19XB AXIAL-FLAW COMPRESSOR

III - PRELIMINARY ENGINE INVESTIGATION

By Andre J. Meyer, Jr. and Howard F. Calvert

SUMMARY

Strain gages were used to measure blade vibrations possibly

causing failure in the 10-stage compressor of the 1916 jet-propulsion
engine. The seventh and tenth stages were of great concern as a
result of failures experienced by the manufacturer. Strain -gage
records were obtained from all stages during acceleration, decelera-
tion, and constant speed runs. Curves are presented herein showing
the maximum allowable vibratory stress for a given speed, the change
of the damping coefficient with the mounting of a strain gage at the
base of the blade, the effect of rotor speed , on blade natural
frequency, and the effect of the order of first bending-mode vibra-
tion on stress. It was found that for all stages the lower the
order of vibration the higher the stress but no destructive vibra-
tions were detected.

INTRODUCTION
Blade failures in the 10-stage axial-flow compressor of the
191B ,jet-propulsion engine, presumably caused by vibration, have
been reported by the manufacturer. The failures are believed to
have originated in the seventh and tenth stages at high engine speed
and high pressure ratio. At the request of the Bureau of Aeronautics,
Navy Department, a series of vibration surveys is being conducted at
the NACA Cleveland laboratory to determine the cause of these failures
and to devise a means of eliminating them.
Little published information is available on experimental
measurements of the factors that cause blade vibration; the vibration
characteristics were therefore investigated in a ,jet engine under
actual operating conditions. Prior to the dynamic investigation, the

l `/ 'F 7 3
^ ^ F%^ F FFFl FF ^ FF
.! F ^ ^^ ^
2 NACA RM No. SE8A28

natural frequencies and node shapes of various modes of blade vibra-

tions were determined and with these data, critical-speed diagrams
were plotted for each of the 10 stages of the compressor rotor (ref-
erence 1).

The preliminary dynamic investigation consisted in instrumenting

the rotor and in measuring blade vibrations while the compressor was
driven by an electric drive motor. The results obtained are presented
in reference 2. The data are incomplete for analytical use because
a drive motor with sufficient power to operate the compressor at
rated speed for sea-level conditions was unavailable at the time.
The same compressor was then reinstrumented and assembled within a
complete jet engine in order'to continue the vibration surveys to
maximum rated speed and power. The results are reported herein.

Curves are included to show the-effects of both engine.speed

and the orders of excitations on vibratory stresses and frequencies.
Statically determined damping coefficients and relations between
vibration . amplitude and vibrational stresses or loads are also
reported.

APPARATUS AND PROCEDURE

Before instrumentation, the 19XB rotor used in reference 2 was

returned to the'manufacturer for complete reblading because the
original set of blades had been twisted by the manufacturer in an
effort to improve the compressor efficiency (reference 3).
The compressor rotor was reinstrumented with the types of
strain gage, lead wire, slip-ring assembly, and electric circuit
described and illustrated in reference 2. The only significant
changes in the gage installation were: (1) All gages were so
mounted on the blades in such a position as to detect the first
bending-mode vibration, because no torsional or other bending-mode
vibrations were observed in the first dynamic vibration investigation
and (2) a transparent, polymerizing, coil-impregnating material was
substituted for the ceramic cement previously used to embed the
lead wires in the drilled rotor holes. In the initial dynamic
investigation (reference 2), data from the last five stages were
inconclusive because many lead wires were torn loose by centrifugal
force before completion of the investigation.
After the strain gages were installed, frequency and damping
measurements were made to evaluate the effect of the gage and its
cement on vibration characteristics. The frequencies were determined
by the method reported and illustrated in reference 1 and damping
NACA RA No. SE8A28 3

values were calculated from vibration die-away curves produced by

plucking the blade and recording the vibration with a capacitance
pickup.

The rebladed compressor was then assembled in a late production

19XB engine (fig. 1) and mounted on a sea-level-type test stand. The
inlet guide vanes furnished with this engine transferred part of the
compression load from the rear stages, in which.failures have originated,
to the front stages. A variable-area clam-shell tail cone was sub-
stituted for the original jet assembly in an attempt to make it
possible to vary the back pressure and consequently the pressure
ratio of the compressor.

The rotor speed was varied until blade vibrations were indicated
by the strain-gage signals, at which time the speed was held constant
while photographic records were taken. The engine was accelerated
and decelerated at 400 rpm per second over the entire speed range,
permitting the vibrations to build up to maximums, while continuous
records were being taken. The amount of vibratory stress indicated
was determined by comparing signal heights recorded with the heights
of signals produced by a calibration instrument using similar strain
gages. The vibratory stress signals were corrected for the frequency
response of the recording equipment.

Loads were applied at the tips of the instrumented blades and

corresponding tip deflections and bending stresses were measured.
Thus the observed vibratory stresses were correlated with the tip
amplitudes of the blades.

DISCUSSION AND RESULTS

The critical-speed diagrams plotted in reference 1 indicated

that the blade failures possibly could have resulted from resonant
vibration caused by_a fourth-order excitation (four force impulses
per revolution of the rotor) in the seventh stage or by a sixth-order
excitation in the tenth stage. Exciting forces are generally thought
to be caused either by regularly spaced discontinuities in the air
stream or by load impulses transmitted mechanically from various
engine components. Sources of fourth- and sixth-order excitations
of either type are not readily apparent in the 19XB engine. Four
bearing-support arms are in the front of the engine, but it is
doubtful that their effect would be prominent in the seventh stage
after the air has churned through the six preceding stages. The
first vibration survey conducted on the compressor (reference 2)
definitely determined, however, the presence of fourth- and
sixth-order impulses, as well as many others in all instrumented
stages. The gravity of these fourth- and sixth-order impulses could
4 NACA FM No. SE8A28

not be evaluated in the seventh and tenth stage during the prelim-
inary dynamic investigation because the resonant condition required
speeds higher than those obtainable.

The rotor used in this investigation came from one of the first
experimental 19XB units and apparently a tolerance change or a change
due to the reblading caused the natural frequencies of the new blades
to be higher than those of the original blades. It was found that
the average of the - natural frequencies of the new blades was. 9.8 per-
cent higher than that previously detetmined in reference 1. This
increase in natural frequency and a 4.7-percent increase in frequency
induced by the stiffening effect of the strain-gage installation
made it impossible to reach a fourth- and sixth-order resonant con-
dition in the seventh and tenth stages, respectively, without
dangerously overspeeding the engine. Artificial reduction of the
natural frequencies may permit attainment of these resonances.

A critical-speed diagram (measured vibratory frequency plotted

against rotor speed) showing the orders of vibration encountered is
presented in figure 2. Excitation forces from the third to the
seventeenth order apparently are present within the compressor and
vibrations caused by the orders are exact multiples of rotor speed.
A comparison of calculated and measured effect of _centrifugal force
on blade natural frequency similar to figure 6 of reference 2 is
shown in figure 3. Order of excitation is plotted against measured
vibratory stress in figure 4. For all stages, the lower the order
of vibration the higher the stress and all stages show the same
general slope. All the vibratory stresses plotted are produced by
first bending mode vibrations of the cantilever compressor blades.
Each stage has an exception to the general trend; some orders are
predominant, others are absent. For example, in the tenth stage
(fig. 4(c)).,the seventeenth order results in the highest stress
whereas the eighth, ninth, and twelfth lie below the trend. The
source of the high stress at the seventeenth order as well as the
source of the other exceptions was undetermined.

The records taken during engine acceleration and deceleration

gave a complete account of all vibrations present throughout the
speed range. No obviously serious vibration was present during
either acceleration or deceleration. The vibratory stresses and
consequently the vibration amplitudes for the acceleration and
deceleration runs were very nearly equal to those for conditions in
which the speed was set and held at the peak-resonance point.

The vibrational stresses thus far observed are relatively low

and would not in themselves result in blade failure. The highest
vibratory stress recorded was ±8250 pounds per square inch, which
NACA RS No. SE8A28 5

occurred in the third stage and was excited by a fourth-order source.

In order to determine the allowable vibratory stresses, the first-
and tenth-stage centrifugal stresses were calculated for various
rotor speeds and are plotted in figure 5. The maximum allowable
vibratory stress over the range of rotor speeds is shown
in figure 6. Allowable stresses of the intervening stages lie
between the curves of the two extreme stages plotted. Data for
figure 6 were obtained from calculations of mean working stress
(centrifugal stress plus bending stress.) and the relation given in
reference 4 between mean stress and total allowable stress for the
blade material. Vibratory stresses of approximately 35,000 pounds
per square inch are necessary to cause failure of the seventh- and
tenth-stage blades according to figure 6.

It was suspected that part of the cause for the low measured
vibratory stresses might be the additional damping introduced by
the presence of the strain gage. The damping values were measured
at static conditions for blades with and without gages. In the last
five stages; the vibratory stresses might have been reduced by gage
installation, as indicated by figure 7; the information is inconclu-
sive, however, because the effect of rotational forces on damping
properties is unknown. The inlet guide vanes used also may have
lowered the vibrational stresses as well as the compression load,
consequently reducing the bending stresses in the rear stages of the
compressor. The results of static tests performed to correlate the
vibratory-stress magnitudes with the tip amplitudes and with loads
of .the blades are shown for the seventh and tenth stages in figure 8.
Extrapolation of this information indicates that vibrations with tip
amplitudes of 10.055 and ±0.084 inch must be attained to cause
failure of seventh- and tenth-stage blades, respectively.

When the compressor was driven by an electric drive motor,

throttling the exhaust and thus raising the pressure ratio had a
pronounced effect on the vibration amplitude (reference 2). The
engine used in succeeding investigations was therefore equipped with
a clam-shell arrangement providing a variable-area tail cone. A
resonant condition. of engine speed was set, as in the drive-motor
experiments, and the tail-cone area was varied over the maximum
range permitted by allowable turbine temperatures. No change in
vibration amplitude was noticed, however, probably because the varia-
tion in pressure ratio was small and because the compressor ipherently
operated near the pressure ratio that is the maximum obtainable for
any particular speed of the engine.
6 NACA RM No. SE8A28

SUMMARY OF RESULTS

From an investigation of vibratory stresses in the 19XB com-

pressor conducted by means of strain gages mounted on blades of
each stage, the following results were obtained:
1. First bending-mode vibrations were detected at each exact
multiple of rotor speed from 3 to 17.

2. Fourth and sixth orders of fi3zet bending-mode vibration in

the seventh and tenth stages, respectively, could not be measured
without serious overspeeding of the compressor.

3. No destructive vibrations were observed in the rebladed

compressor throughout its normal operating range as investigated in
the complete engine.

4. For all stages, it was found that the higher the order of
vibration the lower the stress. An exception to the trend was
discovered in each stage.

Flight Propulsion Research Laboratory,

National Advisory Committee for Aeronautics,
Cleveland, Ohio.

REFERENCES

1. Meyer, Andre J., Jr., and Calvert, Howard F.: Vibration Survey
of Blades in Westinghouse 19XB Axial-Flow Compressor. I - Static
Tests. NACA RM No. E6J11, Bur. Aero., 1946.

2. Meyer, Andre J., Jr., and Calvert, Howard F.: 'Vibration Survey
of Blades in 19-XB Axial-Flow Compressor. II - Dynamic Investi-
gation. NACA RA No. E7D09, Bur. Aero., 1947.

3. Downing, Richard M., Finger, Harold B., and Roepcke, Fay A.: Per-
formance of the 19XB 10-Stage Axial-Flow Compressor with Altered
Blade Angles. NACA RA No. E7A21, Bur. Aero., 1947.

4. Noll, G. C., and Lipson, C.: Allowable Working Stresses. Proc. Soc.
Exp. Stress Analysis, vol. III., no. 2, 1946 ) pp. 89-101.

.11
 RESTRICTED

VIBRATION SURVEY OF BLADES IN 19.XB AXIAL-FLAW COMPRESSOR

III - PRELIMINARY ENGINE INVESTIGATION

Andre, J. Meyer, Jr.,

Mechanical Engineer ,_

H and F. Calvert,
Mechanical Engineer.

Approved: 1o^•^^u
ohn H. Collins, J .'
Mechanical Engineer.

jh

RESTRICTFM
NACA RM No. SE8A28 7
ry
o, n
V-
m
m
m
W
NACA RM No. SEBA28 8

2400

2200

2000

1800

1600

1400
a
U

1200
C.
v

d
F+
w 1000

800

600

400

200

0 3

Rotor speed, rpm

Figure 2. - Critical-speed diagram showing measured first bending-mode frequencies

for various stages.
NACA RM No. SE8A28 4

Frequency
p Measured
— -- Calculated
Order of rotor speed
640 5 4 3

560
03
a
U
w

U 480
N

C7'
m
w
400

NACA

320
0 4 8 12 16 x103
Rotor speed, rpm
(a) First stage.
Figure 3. - Critical-speed diagram showing measured and
calculated first bending-mode frequencies.
NACA RM No. SE8A28 10

Frequency
4 Measured
— — --Calculated
Order of rotor speed
6 5 4 3
720

640

to
U
w

560
a^ ®-4-

v
k
480

NACA

v I i 1
400 1 1 f 1 P

0 4 8 12 16x10 3
Rotor speed, rpm
(b) Second stage.
Figure 3. - Continued. Critical-speed diagram showing
measured and calculated first bending-node frequencies.
NACA RM No, SESA28 91

Frequency
O Measured
--- -- Calculated
Order of rotor speed i

800
6 5 4

720

O
Go
P4
a

640
O ^
i

k
i
560 e

NACA

480
0 4 8 12 16 x103
Rotor speed, rpm

(c) Third stage.

Figure 3. - Continued. Critical-speed diagram showing

mea-sured and calculated first bending —mode frequencies.
NACA RM No. SE8A28 12

Frequency
O Measured
— — — Calculated
Order of rotor speed
960 8 7 6 5 4

880

0
0
a, 800
v

v
0
0
o 720
14
w

640

NACA

560
0 4 8 12 16x103
Rotor speed, rpm
(d) Fourth stage.
Figure 3. - Continued. Critical-speed diagram showing
measured and calculated first bending-mode frequencies.
NACA RM No. SE8A28 13

Frequency
O Measured
— — — — Calculated
Order of rotor speed
1280 8, 7 6 5

1200

1120
A
s^

1040

960

NACA

880
0 4 8 12 16 20x103
Rotor speed, rpm
(e) Seventh stage.
Figure 3. - Continued. Critical-speed diagram showing
measured and calculated first bending mode frequencies.
 NACA RM, No. SE8A28 14

Frequency
0 Measured
----- — Calculated
11 Order of rotor speed
19 in Q A 7 A A
1440

1360
a
U
w

m 1280
Q'
m
P
w

1200

NACA

1120
0 4 8 12 16 20x103
Rotor speed, rpm
(f) Eighth stage.

S
Figure 3. - Continued. Critical-speed diagram showing
measured and calculated first bending-mode frequencies.
 0

NACA RM No. SEBA28 15

Frequency
o Measured, blade 78
O Measured, blade 41
— — — -- Calculated

Order of rotor speed

14 10 7
1600

1520

a,
U
N

1440
a^

v{
(D
A4
D:4

1360

NACA

1280
0 4 8 12 16 20x10`
Rotor speed, rpm

(g) Ninth stage.

Figure 3. - Continued. Critical-speed diagram showing
measured and calculated first bending-node frequencies.
NACA RM No. SE8A28 16

Frequency
o Measured
— — — — Calculated
V-
16 14 12 Order of rotor speed
17 15 13 11 10 9 8 7 6
2000

1920

m
v 1840
A
U
m
Q^
1760
w

1680

NACA

1600
0 4 8 12 16 20x103
Rotor speed, rpm
(h) Tenth stage.
Figure 3. - Concluded. Critical-speed diagram showing
measured and calculated first bending-mode frequencies.
 6000
z
D
n
D
Z7

•
5000
z0
U)
M
A
r-I
CO
DN
w CO

N 4000
a^
)1
m

0
3000
A

va^
s^
2000
cd
w

1000

0 4 8 12 16 0 4 8 12 16
Order of excitation
(a) Seventh stage (blade 21). (b) Eighth stage (blade 46).
Figure 4. - Relation of magnitude of observed stresses to order of excitation.
 6000
z
D
n
D
;U
•
Y+ Z
5000 0
o•
m
M
co
D
N
co
CD
m 4000

s,
0

ti 3000
p

b
m
c,

2000
a^

1000

0 4 a 12 16 20 0 4 8 12 16
Order of excitation
(c) Tenth stage (blade 1).. (d) Tenth stage (blade 38).
Figure 4. - Concluded. Relation of magnitude of, observed stresses to order number.

OD
 Z
D
(7
D
40,000
Z
O

a U)
m
CO
^rl DN
C3' 00

30,000
,D
ri

ft
qp

N
4a 20 s 000

10
bo

P
10,000
a^
U

NACA

0 4 8 12 16 20x103
Rotor speed, rpm
Figure 5. Calculated centrifugal stresses at base of
blades of first and tenth stages.

^o
 50,000 Z
nDD
•
to z
0
a' 10
40,000 U)
M
CD
a--i D
N
w CD

4)
F,

w 30,000

> 20,000
a,
,n
c^
30
10,000

k NACA
td

0 4 8 12 16 20 x103

Rotor speed, rpm

Figure 6. - Variation in maximum allowable vibratory stress N
with change in rotor speed. 0
NACA RM No. SESA28 2!

,05

.04

a
9
10

o .03
d
^i
U
v .02
U
on

ti

m •01
a

0
1 2 3 4 5 6 7 8 9 10
Stage of compressor

Figure 7. - Damping of blades from all stages, with

and without strain gages,

t
NACA RM No. SE8A28 22

10,000

8,000

.fl
a
to 6,000
m
+,
Stag
e/
M 7
ao

4,000
^i 10
rQ

2,000

NASA

w n n .^ -^ ^ -3
0 z V s, cs 1V111V

Blade tip amplitude, in.

(a) Tip amplitude.
Figure 8. - Static stress for blade tip amplitude and
loading.
NACA RM No. SE8A28
23

10,000

to
7
8,000
v'
.a

6,000
10
ao
b
4,000
w

2,000

0 2 4 6 8 10 12
Load at tip of blade, lb
(b) Tip loading.
Figure 8. - Concluded. Static stress for blade tip amplitude and
loading,