R. & M. No.

3181
(19,052, 19,901)
A.R.C. Techn;cal Report

,t=,t
QO
t~

r~

MINISTRY

OF A V I A T I O N

A E R O N A U T I C A L RESEARCH C O U N C I L
REPORTS AND M E M O R A N D A

Blade Vibratory Stresses in a Multi-Stage

Axial-Flow Compressor
By D . A. ~KILPATRICK, A. D. S. CARTER a n d L., O ' N E I L L

L O N D O N : HER M A J E S T Y ' S S T A T I O N E R Y O F F I C E
I961
PRICE I75. od. NET

Blade Vibratory Stresses in a Multi-Stage

Axial-Flow Compressor
By D. A. KILPATRICK,
COMMUNICATED BY

A. D.

S. CARTER and

L.

O'NEILL

THE D I R E C T O R - G E N E R A L OF S C I E N T I F I C RESEARCH (AIR)

M I N I S T R Y OF S U P P L Y

Reports and Memoranda No. 3z8 *

January, z958
Summary. A detailed investigation of the stresses in the early stator blade rows of an axial-flow compressor,
which had previously suffered blade failures during its initial performance tests, is described. These vibration
tests show that stall cell excitation is present in the first and second blade rows and causes blade resonances
over a fairly wide part-load condition. Resonances of the fundamental flexural modes of first and second row
stator blades with the fourth and fifth stall cell harmonics give rise to stress peaks in the region of + 10 to + 20
tons/sq in. The inlet guide row also has a stress peak of about _+ 10 tons/sq in. attributable to excitation
associated with the fifth stall cell harmonic.
Away from these resonant conditions stalling flutter (of moderate stress) occurs in the first and second
stator row blades over about 30 per cent to 70 per cent design speed range, the compressor design being such
that these stages are stalled up to about 80 per cent design speed.
Under the forced vibration conditions, the blade phasing, in a row, is consistent with anti-phase vibration
of adjacent blades indicating the effect of mechanical coupling.
Analysis of the initial test conditions indicates agreement between the distribution failures and the number of
cycles of high stress probably experienced by the blades.

1. Introduction.

Blade failures have been experienced in axial-flow compressors under conditions

in which no obvious excitant forces (such as from blade and spider wakes) could be held responsible.
Such failures often occurred in the early compressor stages, more particularly at part-load conditions,
with moderate to high pressure-ratio designs. A previous report (Ref. 1) has described a strain-gauge
investigation on a National Gas Turbine Establishment axial-flow compressor of about 5:1 design
pressure ratio. This compressor had, after comparatively little running, shown blade cracks in the
first three stator rows. T h e majority of these failures were in the second row, and all were consistent
with fatigue failure in the fundamental flexural mode of vibration. It was shown in Ref. 1 that
alternating stresses, of sufficient magnitude to cause blade failures, had been measured and that they
occurred at part-load operation when the stages in question were stalled. The stress peaks and
frequencies were stated to correspond to the lower (rotor speed) forcing orders.
The results of cascade investigations (Refs. 2 and 3) into self-excited forms of blade vibration
had also shown that under stalling conditions in particular, very high alternating stresses could
occur. Application of these results in a semi-theoretical study of some typical, though hypothetical,
compressor designs has previously been made and this confirms that stalling flutter could be
Part I.--N.G.T.E. Report R.201, received 18th February, 1957.
Part II.-.-N.G.T.E. Report R.221, received 19th February, 1958.

experienced in such compressors at part-load conditions. Comparison has also been made in this
study of the limited compressor results of Ref. 1 and these semi-theoretical examples; this indicated
that the vibration measured could have been stalling flutter. Consequently, a more detailed
investigation of the blade vibration characteristics of this compressor was carried out over the operating
range which the previous tests had shown to be most dangerous, and is described in this report.
T h e investigation concentrated on the first and second stator rows, the stresses in several blades
of each, and of one blade in the inlet guide row, being recorded. A detailed survey of the chosen
part-load operating range was made and stress distributions and blade vibration frequencies
obtained. The initial blade failures are analysed in the light of the probable source of the high stress
excitations.

2. Description of Rig. The axial-flow compressor (known by the serial reference number 118)
was driven for these tests by a four-stage experimental turbine (reference number 119), Fig. 1
illustrates the test installation, and it will be seen that the test compressor was fitted with a volute
air intake, open direct to atmosphere. The turbine's combustion system is supplied with a i r b y an
independent (plant) compressor.
2.1. Compressor Design. A summary of the salient features of the N.G.T.E. 118 Compressor is
given below:
Mass flow
Stage temperature rise
N u m b e r of stages
Overall temperature rise
Polytropic efficiency .
Adiabatic efficiency .
Pressure ratio .
Design speed

70 lb/sec
26.3 deg C
8
210 deg C
88 per cent
84 per cent
5.3
9,000 r.p.m.
0-65

Mean gJK~A Ts/u2

All the blades are of C4 profile on a circular-arc (C50) camber-line and are of constant-reaction
design; the stator blades were manufactured in cast H.R. Crown Max and the rotor blades in east
itallium. It might be of interest to note here that the blades were designed assuming a constant
axial velocity and neglecting any radial equilibrium effects. At any radius the blade section of the
rotor and stator of a stage are identical. Some details of the first two stator rows' blading follow:
First stator
Blade height
Camber ~ .
Stagger* .
Pitch/chord ratio*
Root chord
Root thickness/chord ratio
Air outlet angle*
Height/thickness ratio
Design gas bending stress .
At mean diameter.

4.3 in.
26 deg
- 38 deg
0.85
1.21 in.
11 per cent
31.9 deg
32
5.2 tons/sq in.

Second stator
3-7 in.
27.8 deg
- 37- 2 deg
0-79
1- 22 in.
10-1 per cent
30-2 deg
3O
4-65 tons/sq in.

2.2. Compressor History. Four series of tests, prior to those of this report, have been performed
with the compressor. Series I, II and III comprised the initial mechanical and aerodynamic testing
of the compressor, and it was during the ensuing inspection strip that blades in the first three stator
rows were found to be cracked. These failures numbered 1 in the first row, 22 in the second, and 4 in
the third row, the location and form of the cracks being consistent with fatigue in the fundamental
flexural mode. The strain-gauge tests (Series IV) reported in Ref. 1 were done with complete
replacement of the second-row stator blades and 12 new blades in each of the first and third stator
rows. Especial care had been taken to reproduce the original profiles and to ensure sound blade
material.
Upon the conclusion of the Series IV tests, a rotor modification was carried out and the compressor rebuilt with the same blading in both the rotor and stator rows. About 6 or 7 blades in each
of the inlet guide vane, first and second stator rows were strain-gauged, and Fig. 2 shows the
compressor during rebuild.

3. Instrumentation. 3.1. Strain Gauges. All the gauges used were British Thermostat Type
SE/A2, resistance 200 ~2, and were affixed using a nitro-cellulose cement. The lead-out wires,
consisting of 4/0. 006 in. diameter conductors (cotton covered and lacquered), were soldered to the
gauge leads and passed out through holes in the blade roots. Thence they were taken via a shallow
channel between blade root and casing to emerge through small holes in the casing, one hole per
stage per half-casing (Fig. 2). Care was taken to secure and protect the lead-out wires as well as
possible, using small clips and cold-setting adhesive.
3.2. Amplifiers. Two different types of amplifiers were used, the first being based, in a modified
form, on the A.E.R.E. Type 1008 amplifier. Six such amplifier channels were' available, each
comprising a mainamplifier (voltage gain 60,000), power panel, and head amplifier (voltage gain 50).
Two main amplifiers are mounted on one 19 in. panel and share a common power-supply, panel.
The head amplifiers, together with their gauge selection and D.c. polarising supply units were located
near to the compressor, while the remaining amplifying and recording equipment was housed outside
the test house in a mobile trailer caravan.
The second type of amplifier channel, of which two were used, is a commercially available unit
made by Southern Instruments, Ltd., and having an overall maximum voltage gain of 4 x 105.
The complete amplifier, D.c. polarising supply, gauge selection and calibration units were mounted
in two adjacent cases and direct connections made to the strain-gauge lead terminals at the
compressor.
3.3. Recording Equipment. Two cathode-ray tube units, one 6-channel the other a 3-channel unit,
were each fitted with continuous moving-film cameras; a 70 mm camera for the 6-channel unit
and a 35 mm camera for the 3-channel unit. Both cameras were capable of a maximum film speed
of about 80 in./sec, but for these tests were set to run at about 40 in./sec. Five of the modified
'A.E.R.E. type' amplifiers were connected to the Y plates of the cathode-ray tubes in the 6-channel
unit, while the two Southern Instruments channels were connected to the 3-channel cathode-ray
tube unit. The remaining tube in each unit was connected to a common time-marking (1,000 pulses
per second) and compressor revolution marking (1 pulse per 4 revolutions) amplifier unit. Visual
monitoring of all these recording cathode-ray tubes was available between actual exposures, the
necessary time-base (X) sweep being automatically removed during exposure. Initially during the
3
(80624)

A*

tests the 6-channel continuous moving-film camera unit was paralleled by a 6-channel high-speedd r u m camera unit, so that high-speed as well as low-speed film records could be obtained. The d r u m
camera was capable of operating up to about 1,500 in./sec and w a s actually used in these tests at
about 600 in./sec. Due to the mainly low-frequency nature of the recorded vibration waveforms,
such high-speed, limited-length (30 in.) exposures were considered to be unrepresentative, and the
use of the drum camera unit was discontinued.
In addition to the above strain-recording equipment, a single-channel tape recorder was also
used for the constant-speed tests. This was a standard commercial instrument, an E.M.I. Type
T.R.50A, and was connected to the pre-amplifier output of one of the Southern Instruments
strain-gauge channels. Thus the vibration of one blade was recorded both on film and on magnetic
tape.
The various pressure, temperature and speed indicators necessary for the subsequent aerodynamic
analysis of the compressor condition, were grouped in an illuminated instrument panel. This panel
was photographed by a modified Type F24 camera (negative size approximately 5 in. x 5 in.), the
exposure operation being synchronised electrically with the exposure of the other recording cameras
and tape recorder.

4. Description of Tests. As indicated in Section 1, the scope of these tests was purposely restricted
to cover those compressor conditions which the earlier tests had shown to produce significant
stresses, at least in the first two stator rows. The range of test conditions is shown in Fig. 3, along
with the full compressor characteristics obtained in the compressor performance tests, from which
it will be seen that most of the tests consisted of (nominally) constant-speed characteristics with
two additional constant-throttle characteristics. In addition to the characteristics shown in Fig. 3
several other characteristics and isolated points were recorded in the range 6,500 to 7,250 r.p.m.
The full list of test' conditions, including gauge selections, is tabulated below. The blade numbering
convention is based on its location in the top or bottom halves of the stator casing (see, for example,
Fig. 19).
Gauge selection number
Tests

1+5
6 + 11
1 2 + 14
1 5 + 17
18
19, 20
21
22, 23
24
25 -->-28

Condition (r.p.m.) "~

5,000, 5,500, 6,000, 6,400, 6,700

6,500, 6,700, 6,850, 6,850, 6,900, 7,000 .
6,200, 6,700, 6,750t
7,250, 6,400, 6,600-~
5,500+ 7,250+ 6,000 (acceleration then deceleration)
4,000, 7,000
7,000
6,400, '6,600 '.
4,400-)- 6,200 (acceleration, constant throttle)
4,000, 5,000, 6,000, 6,400 (surge points only)

70 mm
camera

35 mm
camera

1
1
1
1
1
1

2
2
2
1
1
1

1
1
1
1

2
1
1
1

Tape
recorder

]Vots$
e Unless otherwise stated the r.p.m, conditions relate to a group of 3/4 points along the constant speed
characteristics.
]- Due to a camera fault, some of these aerodynamic measurements were lost:
4

Gauge selection groups

Selection No. 1

70 mm camera.

35 mm camera.

Tape recorder.

Selection No. 1

Inlet guide row, blade T.5

]st stator row, blade T.5
2nd stator row, blade T.5
2nd stator row, blade T.25
2nd stator rou,,~Tblade B.17

( l s t stator row, blade B.16

L l s t stator row, blade B.27

~Selection No. 2

2nd stator row, blade B.5

2nd stator row, blade B.28

fSelection No. 1
). Selection No. 2

1st stator row, blade B.16

2nd stator row, blade B.5

As indicated in the above tabulation, the procedure for the (nominally) constant-speed runs was
to set the speed and work up the characteristic from the nearly open throttle condition, in three
or four points, almost to surge. The settling time at each point was restricted but was never much
less than 30 sec. Owing to the experimental nature of the turbine running (these tests constituted
practically its first hot running), speed control was at times a little difficuk and a 'course condition'
settling time (i.e., With small speed variation) of several minutes sometimes occurred.

5. Test Results. 5.1. Waveform Analysis.

The basic analysis of complex waveforms, whose

component waves are reasonably steady in amplitude, is well established and provided there are
not too many components it can be performed by inspection methods such as are described in
Ref. 4. In the recording of blade strain and other quantities in a machine such as an axial-flow
compressor or turbine, it may happen that one source of excitation predominates and the resulting
waveform is substantially pure. (For example, Fig. 4, record 1 trace 2 and record 2 trace 2.)
Sometimes though, the waveforms are of quite an unsteady and transient quality, as for example
most of the remaining traces of Fig. 4, in varying degrees. Analysis of these tends therefore to be
somewhat subjective. Application of more mathematical and formal analyses is limited usually b y
inability to discern the repetition cycle in the waveform, or otherwise by the lengthiness of such
operations. Fortunately, however, it does often seem that this unsteadiness of waveform is in fact
more due to a random variation of excitation than to a very complex synthesis of basic waves.
Consequently record interpretation by.simple basic means has given mainly consistent results, and
for the less pure waveforms yields only a small number of component frequencies (in these tests
not more than three), whose amplitudes may vary in random fashion. It is the maximum value of the
amplitude which has here been measured for conversion to the maximum stress figures used in this
report.
There are, of course, mechanical and electrical aids to waveform analysis. Generally speaking,
the mechanical analysers suffer from the disadvantages quoted above in connection with mathematical
methods, and the electrical methods are based on some frequency selective device such as filters or
selective amplifiers. One such electronic instrument available was the Muirhead-Pametrada Wave
Analyser (see Ref. 5), and it was decided to make a magnetic tape record in parallel with one of the

film record channels ~(Section 3.3), for subsequent replay and analysis using this instrument.
Comparison of the frequency analyses of tape and film is made in detail in Section 5.3, but it can be
said here that they correlate exceedingly well. As might be expected the wave analyser measured
more component frequencies than the visual inspection; most of them were of very low amplitude,
however. As far as amplitude correlation is concerned this is more difficult because the wave analyser
gives only mean amplitudes and is normally not fully responsive to transient peak amplitudes.
5.2. Stress Characteristics. The conversion of the recorded signal waveform amplitudes into
alternating stresses was done by the usual straightforward method assuming purely tensile strain
and a constant strain-gauge factor (viz., that quoted by the manufacturer). No correction was made
for gauge position, all the stresses quoted being those measured at the gauge position. The centre
of the gauge was 0.2 in. from the root platform, and a correction factor to give the maximum stress
at the root (fundamental flexure) would be approximately 1.1 in this case.
Simple plots of stress against mass flow for (nominally) constant-speed conditions are not
particularly significant: in addition, some of them are complicated by a speed variation. A much
clearer view of conditions is obtained by superimposing the stress values, using distinctive symbols
for arbitrary ranges of stress level (e.g., + 2--~ + 5 tons/sq in.), onto the familiar pressure-ratio/
mass-flow compressor-characteristics diagrams. Fig. 5 shows such diagrams for the three blades
studied in the first stator row. Less points were available for the lower two diagrams on account of
the gauge selection which was carried out (see Section 4). The stress contour diagrams for the second
stator row blades are shown in Figs. 6 and 7. Only one inlet guide blade was studied and its contours
are included in Fig. 8. It will be noted from these diagrams that the regions of moderate stress for
the first and second stator row blades, say + 5 tons/sq in. upwards, are located roughly between
4,000 and 7,000 r.p.m, and extend to the surge line over most of the flow range between these
speeds. Comparison with the full performance curves of Fig. 3 shows that these significant stress
regions (which are more extensive and continuous for the first than for the second stator rows)
occur just below the 'kink' in the surge line. Inner zones of higher stress, up to + 20 tons/sq in. in
some cases, also occur, there being two such zones for the first stator row blades but only one for
the majority of the blades in the second stator row. In the case of the inlet guide blade the general
stress level is much lower; a localised high-stress zone does, however, occur just below 6,500 r.p.m.
The maximum stresses obtained in the constant throttle tests have also been plotted on a stress
vs. r.p.m, basis. Fig. 8 shows this curve for the inlet guide blade, while the curves for three blades
each of the first and second stator rows are given in Figs. 9 and 10. These diagrams confirm the
previously noted disposition of significant stresses between 4,000 and 7,000 r.p.m. The general stress
levels are seen to be between + 5 and + 10 tons/sq in. for the first stator blades, about + 5 tons/sq
in. for the second stator blades and only about + 3 tons/sq in. for the inlet guide blade. T h e high
stress peaks occur as tabulated below:

(1)
(2)
(3)
(4)

Blade row

Stress maxima

Approximate speed

First stator
First stator
Second stator
Inlet guide

+ 13->- _+ 20 tons/sq in.

_+ 11-> _+ 16 tons/sq in.
+_ 11->- _+ 13 tons/sq in.
_+ 6* tons/sq in.

6,600 r.p.m.
5,300 r.p.m.
7,000 r.p.m.
6,400 r.p.m.

e Peaks of approximately + 12 tons/sq in. were, however, measured at 6,400 r.p.m, during the constant-speed
characteristic runs.
6

The peaks in the groups (1), (2) and (3) above, do not occur at precisely the same r.p.m, for each
blade, but are of the same frequency order. Consideration of blade frequencies and their trends is
made in Section 5.3.
All the stresses described hitherto are fundamental bending-mode stresses but in addition to these
comparatively low frequencies one other higher frequency was in general detectable on each blade.
This represented a much lower stress level, however, as shown in Fig. 11 where the amplitude is
plotted against rotor speed. The only significant stress variation would seem to be a reduction
coincident with high stresses in the fundamental flexural mode.
5.3. Frequency Characteristics. It will be convenient to consider first the case of the highfrequency components, whose (low) stress amplitudes have been plotted in Fig. 11. These measured
frequencies were constant for each blade throughout the test range and they are tabulated below for
comparison with the fundamental torsional (T.1) and the first overtone flexural (F.2) frequencies as
subsequently measured on one blade from each row.
Blades

Inlet guide

First stator

Second stator

Measured frequencies (c.p.s.)

1,110

1,050, 1,050, 1,080

T.1 (c.p.s.) .

1,413

1,335

1,480, 1,490, 1,430,

1,430, 1,480
1,553

F.2 (c.p.s.) .

1,140

1,092

1~493

From this it is clear that the measured higher mode vibration is in the first overtone flexural mode
in all cases.
The majority of the waveform analysis as discussed in Section 5.1 was performed by the conventional visual inspection methods (e.g., Ref. 4). Excluding the higher flexural mode stresses, at the
inost only two component frequencies were occasionally detectable on the inletguide and second
stator row blades, one component greatly predominating. These predominant frequencies are plotted
(against r.p.m.) for first stator row blades in Fig. 12, for second stator blades in Figs. 13 and 14,
and for an inlet guide blade in Fig. 15. The points are plotted regardless of their position on the
pressure-ratio/mass-flow characteristics; they are thus an aggregate of all the available test blade
frequency measurements.
Over most of the speed range it will be observed that the blades are vibrating at their natural
fundamental flexural frequencies (the actual 'scatter' is small and probably within experimental
and analytical limits). For part of the speed range, however, the frequency of vibration increases
approximately linearly with speed, thus indicating a forced vibration. Comparison of Figs. 9 and 12,
10 and 13 respectively shows that the extent of these forced frequency conditions coincides with that
of the high stress peaks, the actual peak occurring when the blade forced frequency equals its
frequeney,
o.p.s.~] is seen to be approximately
natural frequency. The order of the forcing, i.e., /foreed
~ rotor
speed,r.p.s.
l" 94 for the higher stress peaks of the first stage stator blades and 2.45 for their lower stress peaks,
as well as for the main stress peaks of the inlet guide and second stator row blades. It will be noted
that these two orders are very nearly in the ratio of 4:5 (1.94 = 4 x 0.485 and 2.45 = 5 x 0.49).
Thus we can explain every major stress peak recorded in terms of a resonance of the blade
7

fundamental flexural mode with harmonics of a forcing excitation of frequency 48 --> 49 per cent of
rotor speed. The fourth harmonic, on this hypothesis, is responsible for the first stator blade major
stress peaks, and the fifth harmonic for the stress peaks of the second stator and inlet guide row blades
and for the minor stress peaks of the first stators. In addition a minor, in some cases barely
perceptible, peak for the second stator blades in the region of 5,500 r.p.m. (Figs. 6, 7 and 10) is
probably attributable to the sixth harmonic of this forcing excitation.
Further evidence relating to the origin of these stress peaks is to be seen in the minor component
frequencies (with stresses < + 2 tons/sq in.) as detected in the inlet guide and second stat0r row
blade records and plotted in Figs. 15 and 16. From these it is seen that, around the stress peak, there
is present a minor forcing frequency component consistent with excitation by the fourth harmonic
of a 0.48 0.49 rotor speed order, the main forcing frequency component being the fifth harmonic.
Frequency analysis of the limited number of magnetic tape records by the electronic wave analyser
(Section 5.1) is even more revealing. The results are plotted in Fig. 17 and, for the speed range
covering the 6,600 r.p.m, stress peak (Fig. 9), show definite third, fourth and eighth harmonics
and probable fifth and sixth harmonics of 0.48 rotor speed order for the first stator blade. The
'analysis by inspection' described above had produced only the main component, the fourth
harmonic. Sufficient tape-recorded points were not available to cover the lower speed stress peak
(5,300 r.p.m.) for this blade, but the two records taken nearby indicate the presence of the fourth
harmonic forcing component, the main component at that peak being the fifth harmonic. For the
second stator row blade, within the speed range covering its 6,700 r.p.m, stress peak (Fig. 7), the
wave analyser produced definite third and eighth harmonics and probable sixth and seventh
harmonics, in addition to the fourth and fifth (major) harmonics already obtained from the film
records.

6. Discussion of Results. As shown in Section 5 the stresses measured in the first stator row
(3 blades) and in the second stator row (5 blades) are characterised by two main features. These
stresses consist of a generally quite extensive region of moderate stress within which occur higher
stress peaks, two peaks (per blade) for the first row but only one peak for most of the second row
blades (Figs~ 5, 6 and 7). The one inlet guide row blade recorded also shows ~/high stress peak
similarly located, but the general stress level is much lower and seems in the main to increase slowly
with increasing speed (Fig. 8).
The stage characteristics obtained during these tests are shown in Fig. 18 for the first and second
stages only: owing to the much reduced settling times employed there is considerable scatter of
points. It will be noted, however, that over the speed range of the tests both these stages are stalled
and this would thus support the view that the blade vibration (of moderate stress) at the blade
natural frequency is in fact stalling flutter. The inlet guide vane of course is not stalled and thus the
much lower general stresses (maximum + 3 tons/sq in.) are attributed mainly to random excitation
effects such as buffeting.
The nature of the stress peaks on inlet guide, first stator and second stator row blades has been
shown in Section 5.3 to be consistent with forced vibration, and consideration of their frequency
orders as well as of minor frequency components, indicates that the forcing is due to harmonics of a
basic excitation whose frequency is 48 + 49 per cent of the rotor speed. Resonation with the fourth
and fifth harmonics produces the major stress peaks in the first stators, while the fifth harmonic
produces the major peak in the second stator and inlet guide row blades. Such a forcing cannot be

attributed to any normal interference effects (which have integral rotor-speed orders) but is almost
undoubtedly due to stall cell excitation, probably one cell rotating at 48 + 49 per cent rotor speed.
The original aerodynamic test results show that the first and second stages are stalled for a large
part of the speed range, peak pressure rises being reached at surge at about 8,500 r.p.m, for the first
stage and about 7,500 r.p.m, for the second stage. These are only approximate figures on account of
the test-points scatter. Tests on cascades and on a single-stage compressor (Ref. 6) have indicated
that as Va/U is reduced below the peak pressure rise point the onset of stalling flutter stress of significant amplitude occurs at a Va/U of about 15 + 20 per cent below that for peak pressure rise.
Experiments have also indicated that where stall cells are generated they too reach sizeable
proportions at about the same conditions. The tests of this report show that the stresses are beginning
to drop at the upper test speed limit of 7,250 r.p.m, as shown, for example, in Figs. 9 and 10. It
would thus appear probable, from these factors, that no significant 'flutter' stresses will appear much
above 7,500 r.p.m. Stall cells may, however, persist, with much reduced intensity, at higher speeds
and give rise to resonances. The possibility arises in the case of this compressor by virtue of the
neglect of radial equilibrium in the design. Simple radial equilibrium calculations show that the
tip of the rotor would be stalled even at design speed, and there is some aerodynamic evidence to
show that this is actually occurring. It is doubtful though whether any stall cells would have
sufficient intensity to cause appreciable stresses in the first stator row, where the third harmonic
would be in resonance from about 8,600 r.p.m. 9,600 r.p.m, or in the second stator row (fourth
harmonic in resonance from 8,500 r.p.m, to 9,375 r.p.m.). There might perhaps be a further
important peak for the inlet guide row blade at about 8,000 r.p.m. (for a resonance with the fourth
harmonic of the stall cell).
Upon the basis of ~in excitation rotating around the compressor annulus, one might expect that
at or near a resonance condition for the blades in the row, there would be phasing of the blades'
movement consistent with the rotating forcing. Such does not, however, appear to be the case.
Fig. 19 reproduces simultaneous extracts from the two camera records of a condition where one of the
five second stator row blades recorded (B.17) is in resonance with the fifth stall cell harmonic at about
6,700 r.p.m. It will be noted that all the other recorded second row blades are vibrating at the same
frequency, and that those 'farthest' from their natural frequencies have the lowest and least
sustained stress amplitude. The phasing of the stresses is not, however, 'staggered' in a manner
appropriate to the blades' location in the annulus, but the blade movements are in fact very nearly
either in phase or in anti-phase to one another. Consideration of the (recorded) blades' location shows
that just such a condition as recorded would exist if adjacent blades vibrated in anti-phase; other
records (near resonance) confirm this both for the first and second stator rows. This would suggest
that in these tests coupling between the blades, probably mechanical, via the roots, determines the
phasing. The exact mechanism of a blade resonating with a stall cell (or its harmonic) is not at all
deaf, but it is improbable that the cell presents a clearly defined forcing impulse to the blade.
It seems more likely that the 'front' is more diffuse, especially if the time for the cell to pass a
particular blade approaches or exceeds the time for a cycle of blade vibration, in which case any
more 'precise' forcing of the same frequency (such as by root coupling from other vibrating blades)
might be expected to control the phase. It could happen that the blade most close to resonance
would dictate the phasing and would in fact itself be 'properly' phased with the stall cell.
As far as conditions of moderate stress, away from stress peaks, are concerned, in which the blades
vibrate at their natural frequencies, the amplitude distribution was mostly random: it was sometimes

possible to discern a sustained amplitude modulation at the stall cell frequency, but not often.
It is assumed that the blades are fluttering within the stall cell and that the aerodynamic damping
away from the cell, if not actually negative, is very low. The shape of the stress curves of Figs. 9 and
10 supports this, the high off-resonance stresses and the breadth of the resonance peaks indicating
very low damping. By comparison the stress curve for the inlet guide row blade (Fig. 8) shows the
more typical characteristics of a resonant forced vibration with positive damping. Stalling conditions
do not exist in the inlet guide row and this stress peak is a resonance with the forcing pressure
disturbance induced by the stall cell in the adjoining stage.
The co-existence of flutter and stall cell forcing away from resonance conditions will be noted.
A good example is seen in Fig. 17, where the second stator row blade is fluttering at its natural
frequency, at 6,200 r.p.m., and has also components of forced vibration due to the fourth and
eighth harmonics of the stall cell.

7. Assessment of Compressor Failure. The stator blade failures discovered after the initial
performance tests (Section 2.1) were concentrated in the second stator row and were consistent with
fatigue in the fundamental flexural mode of vibration (22 blades were cracked in the second row, 1 in
the first and 4 in the third row). No rotor blade defects were detected (the rotor blade material,
Vitallium, has a much higher fatigue strength than that of the stator blades).
The breakdown of compressor running times, prior to the discovery of these failures, is shown in
Fig. 20c, up to 8,000 r.p.m. An additional 2 hours running was done between 8,000 and 9,000 r.p.m,
No information is available on the actual blade frequencies appropriate to this original build, but
as stated previously especial care was taken to make the replacement blades as nearly as possible
the same as the original ones. After the tests reported here were concluded, tile natural frequencies
of the first and second stator row blades were measured whilst still mounted in the casing. Their
frequency distribution is plotted in Fig. 20a and their resonances with the harmonics of 0.48-order
rotor speed, within the test range, in Fig. 20b. The measured stresses produced by the resonance
of the sixth stall cell harmonic and the second stator row blades have been shown to be only slight
in most cases, but this resonance distribution is included in the Figure for the sake of completeness.
All the other 'harmonic resonances' have produced measured stresses in excess of + 10 tons/sq in.,
especially in the first stator row. Comparison of the distributions of running times and stall cell
resonances shows that 'continuous' excitation of the high stress peak in some of the second stators
occurs at the speed settings of 7,000 r.p.m, and 7,500 r.p.m. (totalling 130 minutes running) while
excitation of the high stress peak in the first stator blades occurs only at the 7,000 r.p.m, condition
(57 minutes running). Thus many more cycles of high stress would have been performed by the
second row than by the first row blades, on this basis. The minimum fatigue strength of cast H.R.
Crown Max is often quoted as _+ 10 tons/sq in. and in any case will not much exceed + 15 tons/sq in.,
both figures being for 107 cycles to failure. Considering the gas bending stress conditions for the
blades (Section 2.1) in addition to the alternating stresses measured it is obvious that considerably
less than 107 cycles would be necessary for failure under the high stress conditions, and that the
actual running times for these conditions were of the necessary order for failure. The preponderance
of second row stator blade failures is, of course, accounted for by the greater number of high stress
cycles which these blades would have undergone.
No vibration measurements were taken on the third stator row blades, but some sort of estimate
can be made of the conditions applying. It is known that this stage is also stalled over a fair range of
10

speed, so that stall cells may well be present, or if they are not, pressure effects from the cells in the
earlier stages would probably be present. Thus blade resonances with associated significant stresses
could arise. An average figure (from a batch of three blades tested) for the fundamental flexural
frequency is 365 c.p.s ~. Such a blade would resonate at 7,600, 6,515 and 5,800 r.p.m, with the sixth,
seventh and eightt~ harmonics respectively of'a 0.48 rotor speed stall cell excitation. Allowing for
scatter in the blade frequencies it is seen from Fig. 19c that the constant-speed condition of 6,000 and
7,500 r.p.m, might be resonance conditions for some of the third stator row blades, thus providing
an explanation of the failures actually experienced.

8. Conclusions. A detailed investigation of the stresses in the early stator stages of the N.G.T.E.
No. 118 Compressor has been made, in order to ascertain the cause of the blade failures suffered
during the initial compressor performance tests. The main results and conclusions can be
summarised as follows:
J
(1) Moderate levels of stalling flutter stress, over a wide portion of the part-load speed range,
are found in both first and second stator row blades.
(2) Superimposed on the above are forced vibration stress peaks which are consistent with
excitation by the harmonics of one stall cell rotating at ,48 ->- 49 per cent rotor speed.
(3) These stress peaks for the first stator blades are in the + 10 + + 20 tons/sq in. maximum
alternating stress range and are due to blade resonances with the fourth and fifth harmonics
of the stall cell frequency, while for the second stator row blades the fifth harmonic produces
stress peaks in the + 1 0 + + 15 tons/sq in. range.
(4) The effects of the rotating stall cell extend to the inlet guide row where a 'sharply tuned'
stress peak of + 12 tons/sq in. (due to the fifth stall cell harmonic) was measured.
(5) For the forced vibration conditions, the blade phasing (in a row) is consistent with anti-phase
vibration of adjacent blades, indicating the effect of mechanical coupling.
(6) Analysis of running times on the initial performance tests indicates agreement with the
distribution of blade failures and the number of cycles of high stress probably experienced
by the blades.
(7) The advantages of simultaneously recording from a number of blades in the same row,
preferably with a wide scatter of natural frequencies, is illustrated. The superiority of
frequency analysis by wave analyser instruments over that by 'visual inspection' is also
demonstrated, especially when handling a fair number of records with complex waveforms.

Acknowledgements. The authors are indebted for the assistance rendered during these tests by
the Component Test Section and in particular by Mr. F. Burnett.

11

REFERENCES

No.

Author

1 J.R. Forshaw

..

Title, etc.
An investigation of the high alternating stress in the blades of an
axial-flow compressor.
A.R.C. 17,132. June, 1954.

D.A. Kilpatrick and J. Ritchie ..

Compressor cascade flutter tests, Part 1.--20 deg camber blades,

medium and high stagger cascades.
C.P. 187. December, 1953.

D.A. Kilpatrick and J. Ritchie . :

Compressor cascade flutter tests, Part II.--40 deg camber blades,

low and medium stagger cascades.
A.R.C. 17,573. October, 1954.

R . G . Manley

. . . . . .

Waveform Analysis.
Chapman and Hall, London, 1945.

5 H . G . Yates

. . . . . . .

Vibration diagnosis in marine geared turbines.

Trans. N.E. Coast Inst. of Engineers and Shipbuilders. Vol. 65. 1949.
A. D. S. Carter, D. A. Kilpatrick,
C. E. Moss and J. Ritchie

An investigation of the vibratory stresses in a single stage axial-flow

compressor.
A.R.C. 17,786. June, 1955.

12

EXHAUST T 0
ATMOSPHERE
~I-IAUST TO
ATMOSPHE'm~

C~PB~-~SF-.D AIR F'ROM

PLANT C O M ~ O f ~
HOUSE

INTAKE fROM
ATMO~I-IERE
\~
1

COMSUSTO~I
SYSTM

PLATE

N~ ,.8 co.p~ssoR
I

PLAN

AH

'
~ / u~'T

~,

"~.oJ,s.u~..~
~ /

~nc vJr.w

FIG. 1.

J"#

PLAN

No. 118 Compressor test installation (Diagrammatic).

View, with top half stator casing removed, during assembly for vibration tests.
Inlet guide vane, 1st and 2nd stator stages only are strain-gauged.

FIG. 2.

N.G.T.E. No. 118

Compressor.

13
(8o624)

"6

"IN$1"AI.I.ATION.- AXIAL AIR INTAKE)

--(NOM.qAU.Y)

k~

VIBIRATION TEST.,
CONSTANT SPEED TESTS

~)-,4tCONSI"ANT THK~0TTI.E ,SETTING TESTS

~SURI[
RATIO

~EDS

(NO.

GAS-

,,s
-TURBNIE

INSTALI.A'rlON~
j VOLUTE AIR
IHTAKE~
7~illl
,

~ 1 ~ 7 4 QI

D5i I
0

I0

30

20

40

50

60

I
70

MASS FLOW IB/SEC.

Fic. 3.

No. 118 Compressor operating condition.

MAXM

STRESS
'roN.~N2

TRACE
I

2-2

_+J9.5

4.7

5.3
4-6.1

5
. . . . . . . . . . . . . . . . . . . . . . . .

,.

. . . . . . . . . . . . . . . . . . . . . .

-"

RECORD I

7-6
12 "2

RECORD 2

4" 3 . 2

8~4

-+

6.0

6-8

4
5

+ 4.4

RECORD 3

FIG. 4.

Blade vibration records.

14

:5.0

BLADE T.~

:~.O

7200

. ~I~

W
01
P-'0
in
1/)
,,i

P-LADE T-5

7?.00

Ig
P-'O
..::i:" " ,

.,.,~:i:ii:!~r-~-~ ~" ':! ~ ii:~ :;' ~ ~'~::~ !II~

~9e%
I'O
0

o_

:5,C

10

20

30
MASS FLOW

40
LB/SEC.

BLADE B.IG

~0

1.0

60

6450.

P-.O

30
40
IV~SS FLOW LI~./SEC

6O

7P_OQ~,
" " "'~:::
~jil[~:~"

6450

~:" ~ I

2-0

50

E,LADE T.P-8

3'0

7200

Id

I0

2"0

~z

hi

-~i?~:~!~" ~ : ~ ' ~ : i : :

"-a
I,O

I'0
0

~-0

I0

5O

~,0
~,0
/5 4.0E~
~MAI~S FLOW LB

BLADE B.27

60

~'O

7~00

I0

~LADE

20

~0

4Q

MAS~ FLOW LB/SF-,C.

B.17

50

60

50

~0

7200
6450. :

".

2.0

I-0

,o

ao

30

MASS FLOW

'
KEY "TO 5TR.55$

~o
L~/SEC.

~o

~o

I0

~0
~0
40
MAS~, FLOW LB/SEC,

o+~--_+s

[--1
~
~

.6. 5 - - ' r i o
~'TON~INe lO--~,5

NOTE ;--NUMBER3 AGAINST CNAR~,CTERI~,TIC5 ARE CORRECTED .SPEEDSr..R.P.NI)

FIO. 5.

I-0

Predominant stress contours in first-stage stators.

KEY TO ,.%T~EISSES

~ "1" 5-'~'"1- IO t TON~IN e

[] +,o_+,sj
N O T E ; - NUMI~,ERS AGAINST. CHARAC.TERI,% CICS ARE CORRECTED :}PEF-.D5(P.-P.M)

F i e . 6.

P r e d o m i n a n t stress contours in second-stage stators.

P01NT5 RECORDED
,ON CON~T 5PEE.D
x l RUN (6-~o~ PNO

I.

B L A D E ~.5

~-0

7200

.....:~:. \

10

u,l
o~

~ 4 5 ~ ! .~.: ~ : . :

fl
tl

cd
:3 P-'O
u~
u)

II
tl

~J

~;..::....:.N

~o..

0
400d

I'C
0

IO

P.O

~0

40

5O

60

5000

(O.)

BLADE

3"0

B.2~

7200

6000

7000

R.P.M

CON,~TANT

THROT~I'LE

5ETTINC~

RLlhl5

NUMBFR5 AGIAINST
CHARACTERLSTIC5 APSE

7P-O0

CORRECTED SPEED5 R~.M

2-0

5500
.-:':.:':-:-'..:-.'-:':.:-:.:-:
,~~'t'~/~.w
...:~'~.::::::..:-.:~:~:::::-;~'.:~::::"

~d

=o.

:':':~':' t?)?...~.?::::::'::::;'~:{:i:-'.?
': :'::.~C"~
I'0
O

I0

~0

50

4O

50

oOO o,

~980~L '

60
.I-O

I
KEY TO STRES,~F.~

O
" "~

tO

P-O

'50
4O
M A 5 5 FLOW L~/$EG.

5O

+2--'5 /
"I'5~:I:Io~TON~IL21N.

> 15

KEY TO STRESSES

]
~

0
A

t "~--"I" 5
-t-5~d:lo

[]

1o~+_15

'TON

NOTE:-NUMBERS AI;IAINSTe'CHARACTERISTICS"ARE CORRECTt~D SPEED5 R,P.M.

(b)

FIG. 7.

Predominant stress contours' in second-stage stators.

S T R E S S CONTOUR51

FIGS. 8a and 8b.

P r e d o m i n a n t stresses i n inlet guide vane.

6,O

~.Bo
BLADE T," 5,
iJ:STATIC~. RISC.P.5
Z

~o

BLADE T. 5
( f STA'r|C "~ zB6)

~s~

+~o
~ O0

ill
II
kID

2(

,~-~-'- ~ o

<C

O
,4.000

5000

~000
R.RM.

BLADE B.16.
tSTATIC~--.210 C.R~

~o

O
4000

7000

z__

cn
ld

as6

5000

6000
R.R M .

7000

5000

&OOO
R.P.M.

7000

ol
Z

BLADE "E. 2S

co
Z

i'~'10

-Ho
a~
hi
o

,,~

<.

<

O
4000

4.000

5OOO

N.

GOOD
R.RM.

7000

BLADE B.27.

BLADE B.17.
(~STATIC ~ 270~

~ STATIC~2.16 .RS
_+ ~o

~10

~219

=,1
I--

~0..~

P
~q

215r~ ~ r ~ . ~

O
4.000

P.70

I
SO0o

E~C,O o

7000,

R.P.M

NOTE:--NUMBER5 ON C U R V ~ SHOW TREND-~ IN VIBRA'rlON

FREQUENCIES, (C. R S.)

Fit. 9. Predominant stresses in first-stage stators

('Constant-throttle setting' runs).

~ooo

~ooo

~ooo

,ooo

R.P.M.

N O ' r = : - N U M B C ~ S ON CURVES SHOW T R ~ N O S

F R = Q U = Nca=s,(c. P.S.)

iN VJBRAT'ON

Fit. 10. Predominant stresses in second-stage stators

('Constant-throttle setting' runs).

230

~z

~C~TATIC:~m 5 C.P.S)

[D

I-I-Z
u~
ld
Iv
pu~

BLADE T.5

(/3
Z
220

+-E

INLET

VIBRATION
FREQUENCY
IIIQ C.P.S.
o

GUIDE

VANE
-+I
__ OO OO O O--__ Oo--o0 - - ,,-o

o-:~

0
z~ ~ 4 Z t

~Or 2KI

8-6 o

go, %-oo

g.

6 " ~ = 1~33 N

ROe

0
4000

SOOO

-<

6000
R.P M,

5000

4000

70 00

6000

7000

R.P.M.-- N
23C

Ist. STAGE 5 T A T O R

"z
Z
o
p-

BLADE

T.5,

+5
/

tn
i/3
M
rr
I-(D

+?_

c/o/

0
0

+-I

O/

--\w/

ou.x~ o

BLADE

VIBRATION
FREQUEklCY
I050 CRS.

DO
~
o
0

c o

Ill

1"

w
u.

\
0
X

O/~
/

0 0

Z~

o\

B.16

~:5TATIC'==.a 13c- R5.)

A A
A

5000

4000

6000

6000

7000

ui

R.P M.

z
(D
Ii
(n
u)
bJ
(M
p.
o3

?nd

BLADE T5.
+2

*-I

- o o :o o o - -

ix

-o~

~"

)~o

o 0
o
o

l" 5TATIC~ 216C...RS

(10~

z~i

B
A

(~

5000

LOCATION

OF"

~000
R.RM.
FUNDAMENTAL

MODES

oO

%'o~.~ /

2OO
4000

Higher mode stresses.

o ~

~50OO

6000

,KEy TO, S'I'RF.~$ES

70.00

8TRESS

7000

R.P.M.--N

&
I~
X

-I--2-,,-_5 "~
TONS/
-~5--~-I-IO !

/IN.~
~10"*-_+15
>-I'. I1~

PEAK~)

FIG. 12.
Fzo. 11.

(~ 210
Ld
b.

O
4000

220

U
Z

V l BRATION
FREGUENE~(
I,BO C.R~.,

STAGE STATOR'

I~LADE. B . 2 7
f,i

7000

/o

R. R M . - - N

250
5000

I-9,4- N

200

Go
0

4000

6,

Predominant frequency vs. rotor speed

(First-stage stators).

300
BLADE T.5
0
I

29t

::'90

-~

ZO ~ ) A Zs

BLADE ~.S

(~STAT,C~ at7 e.~ A

21~0
I
>O
Z
,a

0
270
4000

~000

~000

3'000

R.R M.--N

0
Ld
b.

500

~:'/O

R6C

4000

O
d

5000

~.~M.-N

7000

90~

~0~
BLADE T.28,

//6/~ =2-4~ N

fSTATICN 290C.R$
P-"/O

4000

~000

6000
R.RM.--N

7000

~ STATIC--~="70 C~..~

27C

26C
~000

"

'

,6O00

4.000

~000

6000

o
STRESSE~
m

o
0 ~2"":1:5 ll"ON5~"
~, --I-5~+1o ~,
/ I N ~"
~lO..,,.Z,z,51
- .
xD >+_. is
3

7O00

a--s ]
s-.-_+,o ~..TONS~,~ ,,
IO-~I,S J

7000

R.RM-N

R.RM,--N

KEY TO

~l~e

KEY TO S'I"RSSE5
5000

Z~

~9c

27

~--~o ~m-~,P-~ -~ ~ - -

(~CSTAT,C'~OC.P.S

id

280

u~
o4.

A~ 0

I~LADE ~.17

)..
0
Z

,!
BLADE ~ . 2 ~

o
Z
ul

29C

6000

IN.

FIG. 13. Predominant frequency vs. rotor speed

(Second-stage stators).

F I G . 14.

Predominant

frequency

(Second-stage

vs. rotor speed

stators).

P.9C

/o

P..BO

P..'/O

2BC

~ ~ ;,oo
S;,oO o0

~"/0

Oo

260

(=5x.~,8~0
25O

oi

240

02

g
I

Z30

~'40

Z
bO

~0

ul
Lk

60~= I ' ~ N
(=4X'4~

D-

R3Q

.
~20

21CI

s
"

I
e

R|O

200
19G
4000

~ooo

60o0

7000

R.RM.-N
II~O
5000

4000

KEY TO ~T~E55E~

FIG. 15.

"l,a~5

TO

Predominant and minor frequencies

inlet guide vane.

A
V
Q

BLADE
IbLADE
BLADE
BLADE
BLADE
BLADE

FIG. 16.

R.R M . - - N

T. 5
T.R8 f
I~.t7
T. 5
T.';'8
P-J7

6000

7000

PREDOMINANT FREQUENCY COMPONENT

'~dllgOR" FREQUENCY COMPONFNT
(ALL STRE.5~Eg ~-~_.2. TONS/IN a)

Predominant and minor frequencies

for second-stage stators.

IO
MAXM.

IST STA~[

STRESS
BLADE

~[

rcesl,~

| N O STK-J~
I

II

_+~
B2~93)

I 0 0 0 C. S.

~s(2a~

. . . . . . . .
.. . . .

-,. ,. ,- . ,. , . : v , . . . . v
.

.....

~" ~,

,i

~ v ~ ~ v V ~ -~
~

v'V'~%'VV'~v

% ^ . . . /.
~ ,-'~~h, , t~'~'~,,,~"'v~.'v"~,',.'
. . "... ,~. ^ ~ . ~ . ~ . .;"
' . . ,~' * ".'.^".,t':~.A ~A ,~v~.-w%,"~

COMPRESSOR
2ND

"
o

_+ 6-9

SPEED 6 7 6 0

RPM.,

48-490/~

ROTOR SPEED STALL

"*"

(5TH.

HARMONIC

~O

JI, Jl

I ,

.[I,

,gl,,w

II III

III

5000

6000
R P.M, (N)

io

7OOO

6XO-4~

P.ND
STAGE

N.

OF

II

,, IllttU, IJ

CEll.. )

5XO'46N

.ml

[h

FORCED VIBRATION OF

STAGE STATOR AT 274 CP.S.

--,

J4XO'48~I

5 X O ' 4 8 N =-

+5

....

""" FRQI.m:NC~ C.P.~.

IST

_+e
4-4. 6

ST.STATOR
I.G.V

II

FUNI)AMENTA,L. FREQUENCY (~TATI'~ DISTRIBUTION

id

+_K:).4
. . . . . . . . . .

10

io

............... ... .......................

~,-,(~7,~

bO
1"O

2O0

((1)

I PULSE EVERY4 ROTOR REVS.

/

T.2e~94~

,, I,,h

==================================================================================
6-3

,, IllilJ,ll fhl
7OOO

(280)ETC REPRESENT BLADE NATURAL FP~QUENClES, C.P.S.

(~)

STAU.-C,U. HARMONICS--IM.ADF.. F R I [ q u L N c y R ~ E S .
sso

35 BLADES
T . / ~ . _ " ~. T . ;

. IN E A C H
HALF-

--

- CASING

B.S

B,28

BOOO

R.P.M (~

~-=,oo

~rrc.IND,cA'rE T,~,~ (MIN.~)

/'d'fJ~ m A c r . [ ~ U T m m ANO
/. 0 e c r , ~ T i O m [~[TW[eN
~"CHARACTF.RISTIC-SPF.D.

e.17

T.:,S

5o

T.5

,,,-'51 ---,. ---'61 ~

I~
'b-~Ll'~'"e--~LI ~

Br7

LOCATION OF STRAIN-GAUGED

~VEFORMS

FOR S~MILAR

2NO STAGE STATOR BLADES,

PHASING OF EACH BLAD~

AND STALL CELL
MOVEMENTS.

FIG. 19. Blade vibration records (Stall cell forcing).

SQOO

(,~043

a-Z~'"e"

"*'-Iib'-*,,

7OO0

8O00

R.P.M.

()
N.B.

D~STRISUTiON o r RUNNINq T l k l [ S
DIAGRAMS

(1 AND

FOR Sr-Rtr..~ Z ~- l r T l ~ ' r $ :

BASED ON FREQUENCIES OiSTAINIrD .44FTR

:iERIES 1ZAND'~ VI~ATiON TESI'6, I'L" ON CGNtPI.F.TF.LY

Ni[--~.ADff.ID ND AND PARTIALLY RE-IM.AOED I=ff ~t'AGEIb.

FZGS. 20a to 20c.

Stall cell excitation relationships.

I0

MAXM.
STRESS

BLADE

TONS/IN

z~J
~d
~
OU" 52"'

IST S

N5T~.GD E

4"11

B.5(280)

+6"3

B28/293)

FREQUENCY C.RS.
I000 C.P.S.

(0~

I PULSE EVERY4 ROTOR REVS,

BLADE FUNDAMENTAl- FRE.qUENC (",STATIC") DISTRIBUTION

IO

Id
13

B,,(274
.28 94

4-10"4

S'[STATOR
I.G .V
b~

~'~

WV

35 BLADES
IN EACH

~---~"~T.2&

4-8
4-4.6
_+5

T'!2@

B CASING
HALF5

B.

2NO STAGE STATOR BLADES.

P-ND

(b)

700O

R P.M. (N)
I

6 X O . 4 ~ N __

tI

,, I I ll I,I
5000

STALL-CELL
150

6000

5XO-48N

IJ ll

,, IH,I hi
7000

R.P.M. (N)
HARMONICS--BLADE FREQUENCY

100

B.28

8000

RF-5ONANCES.

ETCINDICATE TIME (MINII,~

E.NTIN ACr..,ELERAT~NSAND
PATI(:~ISBETWEEN
ARACTERISTIC"SPEED.

B.17

Z_

"r.28

5O

'~--I 6"-',-

WAVEFORMS FOR SIMILAR

PHASING OF EACH BLADE
AND STALL CELL
MOVEMENTS.

FIG, 19.

6OO0
I

T.5

B.17
LOCATION OF STRAIN-GAUGED

~o00

IO

B. 5

VLAA~

I =-

COMPRESSOR SPEED 6760 RRM,. FORCED VIBRATION OF

2ND STAGE STATOR AT 274 CRS, (5TH. HARMONIC OF
4 8 - 4 9 % ROTOR SPEED STALL CELL. )
80)ETC REPRESENT BLADE NATURAL FP.EQUENCIES~ C.P.S,

o S

+ 6.9

,~ ,..,v v'X., ,., v,~ v,. v,., L, V V v,,, v v ,/~" v W'V'V%~'v ~/~v

14X O'48 I~

5XO!48N

IST

El'AGE ~

Blade vibration records (Stall cell forcing).

500O

600O

7000

8000

R.P.M.
(C)
N.B.

DISTRIBUTION OF" RUNNING TIME5 FOR SE.RIF-5 "Ir.,~~

TE~TS,"

DIAGRAM5 Q. AND b BASED ON FREQUENCIES Olt.TAINED AFTER

6ERIES"t'e'AND'E VIBRATION TESTS, I.F--ON COMPLETELY
RE--BLADED ~'ND .I~t,NDPARTIALLY RE-BLADED 1ST 5TAGE~.

FIGS. 20a to 20c.

Stall cell excitation relationships.

Blade Vibratory Stresses in a Multi-Stage

Axial-Flow Compressor
By D. A. KILPATRICK*

Summary. A previous build of the compressor had suffered, after a very short time, blade fatigue failures
in the early stator stages. A new blade design was issued involving considerably increased blade chords and
thicknesses, as welI as a change to a material of higher fatigue strength. For this new compressor build it is
shown that there are blade resonances with stall~cell excitation, resulting in stresses of up to 70 per cent of the
fatigue strength.
The major resonances are with the second harmonic of a five-cell rotating stall regime, but other resonances
are attributable to the second harmonic of a six-cell regime and to the eleventh and thirteenth harmonics of
one cell Away from these resonant conditions stalling flutter also is present at much lower stress levels (up to
20 per cent of the fatigue limit). The stalling range of the first and second stages of this compressor is very wide,
extending to about 80 per cent of design speed.
A feature of the results is the smallness of the reduction in the absolute alternating stresses consequent on the
increase of blade stiffnesses. It is thought that differences in the damping capacity of different materials at
different levels of stress is a contributory factor. Although the actual stresses are not much less than with the
original build, the safety factor has been increased from 0" 5 to 1-5 approximately.
1. Introduction. Previous reports (Ref. 1 and the first paper of this R. & M.) have described
investigations into the blade vibratory stresses in the National Gas T u r b i n e Establishment No. 118
Compressor subsequent to the appearance of stator blade cracks during the initial compressor tests.
These reports showed that stresses high enough to cause fatigue failures occurred over a wide range
of part-load conditions, where the early compressor stages were stalled. T h e first paper of t h i s
R. & M. indicated that the blade failures were most probably caused by resonance excitations o f
the blades by harmonics of a flow disturbance, such as a stall cell, rotating at about 48 to 49 per cent
rotor speed.
W h e n the initial blade failures occurred, and before the investigations described in Ref. 1 and the
first paper of this R. & M. were' made, a compressor design modification was put in hand. T h e object
of this was to reduce the blade gas bending stresses and, it was hoped, the superimposed alternating
stresses, by increasing chords and thicknesses. As a f u r t h e r precaution the blade material was to be
changed to one of a higher fatigue limit. This Mk. IL version of the compressor was built after the
tests on the Mk. I build described above had been completed; and its first tests, forming the basis o f
this present report, were devoted to a study of the blade vibratory stress levels in the first two stator
rows. In addition qualitative measurements were made of flow perturbations at inlet to the first stator
stage by means of capacity-type pressure pick-ups. Levels of vibration at the compressor casing were
also recorded over the relevant part-load conditions.

2. Co~npressorDesign. T h e main design scantlings, of the N . G . T . E . No. 118 Compressor are

described in the first paper of this R. & M. It is a~ eight-stage axial-flow compressor giving,, at it~
design speed of 9,000 r . p . m , a mass flow of about 70 tb/sec at a pressure ratio of 5.3/1 r T h e Mk. II
* N.G.T.E. 'Report R.221, received 19th February, 1958.

(80264)

version, as indicated in Section 1, was designed to retain the same ae}odynamic performance, using
blades of greater strength. A comparison of the blading details is tabulated below. In the Mk. I
version the rotor blades were manufactured in cast Vitallium and the stators in cast H.R. Crown Max.
For the Mk. II version bar S.80 material was used for both rotor and stator blades.

Geometrical Blade Analysis

First stator

Blade height (in.)

Camber (deg)*
Stagger (deg)*
Pitch/chord ratio*
Root chord (in.) .
Root thickness/chord ratio (per cent)
Air outlet angle (deg)* .
Height/root thickness ratio
Design gas bending stress (tons/in. a)

Second stator

Mk. I

Mk. II

Mk. I

Mk. II

4.3
26
-38
O. 85
1.21
11
31.9
32
5.2

4.3
26
-38
0.72
1.65
13
31.9
20
1.3

3 "7
27.8
-37.2
0.79
1-22
10.1
30.2
30
4.65

3-7
27-8
-37.3
0.745
1-60
11.84
30.2
19-5
1.15

* At mean diameter.
2.1. Test Installation. As for the test of the first paper of this R. & M., the compressor was
driven by a four-stage turbine and had a volute intake open direct to atmosphere. Fig. 1 illustrates
the installation.

3. Instrumentation.

3.1. Transducers.

Two of the inlet guide blades and nine each of the first
and second stator row blades were gauged wkh British Thermostat Type SE/A2 (200~2) gauges.
The lead-out wires were soldered to the gauge leads and passed out through the compressor casing
to an external terminal board.
Two capacity pick-ups, of the balanced pressure type described in Ref. 2, were fitted at inlet to
the first stator stage, 90 deg apart. They were provided with total-head adaptors to tile nose of the
pick-up, enabling the pressure variation at mean diameter to be indicated (Fig. 2). Measurement of
the vibrations of the compressor casing was carried out by rigidly affixing a seismic pick-up of
R.A.E. design. The electrical output is approximately proportional to the velocity of the vibration
over the frequency range of about 30 c.p.s, to 400 c.p.s.
3.2. Amplifiers. An arrangement of strain-gauge amplifiers and recording equipment slightly
different from that described in the first paper of this R. & M. was used in these tests. Only one
type of amplifier was employed, viz., one of N.G.T.E. construction, based originally on the circuit
of the A.E.R.E. Type 1008 amplifier. The circuit of the amplifier, which is made up of a head
amplifier (situated near to the strain-gauge) and a main amplifier, is detailed in Figs. 3 and 4. It will
be noted that an extra push-pull stage is provided in the main amplifier for use with cathode-ray tubes
of low sensitivity (e.g.; high brightness tubes for high writing speeds). Normally, as in this case, this
stage is unnecessary. The overall gain of the amplifier as connected is about 100,000 and is flat
from 60 c.p.s, to 7 kc.p.s.
24

Seven of these amplifiers were used, being connected to one inlet guide and to tt{ree each of the
first and second stator blade gauges. The screened cables from the compressor-casing terminal
board led 'off to nearby junction, gauge polarising and switching boxes. The polarising circuits
were of the simple potentiometer type, fed by dry cell battery (Kalium type), and provided the signal
inputs to the adjacent head amplifiers. The main amplifiers were located, together with the remaining
amplifying, display and recording equipment in a trailer caravan outside the test house.
The two capacity-type pressure pick-ups were used in an R.F. carrier frequency circuit, the variations in pressure across the pick-up diaphragm producing, in turn, variations in capacity and
frequency modtllation of the 2 mc.p.s, carrier, which is subsequently discriminated and amplified.
Two channels of this equipment, as manufactured by Southern Instruments, Ltd., were installed
in the trailer caravan, the associated oscillator boxes being located near the pick-ups.
The output from the seismic vibration pick-ups is of the order of 1 v/in./sec, and suitable
amplification was obtained in a Mullard oscilloscope, the output to the recording cathode-ray tube
unit being taken in parallel to the oscilloscope's Y deflection plates.
3.3. Recording Equipment. Two recording cathode'-ray tube units A and B, each comprising six
tubes and a 70 mm continuous moving film camera, were used to record the various strains, pressures
and casing vibration. The tube connections were as follows: to unit A, an inlet guide blade, three of
the first stage stator blade gauges and the casing vibration pick-up; to unit B, three of the second
stage stator blade gauges and the two capacity pick-ups. The sixth tube in each unit was connected
to a common marker amplifier giving 500 pulses per second with one pulse every four compressor
rotor revolutions superimposed. The cameras were operated at a film speed of about 40 in./sec
throughout.
In addition to this photographic recording of the dynamic variables, a tape recording of one of the
strain-gauge outputs (first-stage stator blade T.4) was made in parallel with the film recording.
The connection to the single-track tape recorder (E.M.I. Type TR.50A) was made at the main
amplifier input. Connection was also made to the tape recorder of the output from the seismic
vibration pick-up for one additional test also.
The pressures and temperatures relevant to an analysis of the compressor performance were also
recorded by photographing the manometer and instrument panel with a modified aircraft camera,
Type F.24. Exposures in this camera and in the two C.R.T. recording cameras were synchronised,
initiation of the exposures being made by the operation of the record switch on the tape recorder.
A time delay of about one to two seconds was incorporated to avoid the switching transients in the
amplifiers due to the starting of the tape recorder motor.

4. Description of Tests. As with the previous series of tests the scope of these tests was restricted
to cover the part-load conditions associated with the high stresses in the Mk. I version of the
compressor. The test operating conditions are depicted in Fig. 5 and are compared with results
from the subsequent aerodynamic tests.
Short characteristics were taken at 4,000, 5,000, 6,000, 7,000, 7,500 and 8,000 r.p.m. Additional
constant-throttle setting, varying speed tests were made, between approximately 4,000 and 7,000 r.p.m.
No strain-gauge switching was carried out during the tests and no gauge failures were experienced.
The connection of the seven strain-gauges, two capacity pick-ups and one seismic vibration pick-up
is detailed in Section 3.3.
25
(80624)

E*

One additional test, over a speed range of about 4,800 r.p.m, to 7,000 r.p.m., was made for the
purpose of recording the seismic vibration pick-up signals on the tape recorder.
5. Test Results. The major part of the vibration information was recorded on film, being
supplemented to the extent of one channel by magnetic tape recording. Analysis of the film-recorded
waveforms was by 'inspection' methods and the stress amplitudes quoted are the maximum measured
at each condition. Frequency determinations were made in the region of this maximum stress
amplitude and represent an average over about twenty cycles of blade vibration. The large number
of test points (approximately 700) precluded prolonged examination of each waveform, but it is
estimated that the accuracy of the frequency determination is better than 2 per cent.
The tape recordings were analysed using the Muirhead-Pametrada wave analyser, the amplitudes
measured being approximately r.m.s, values, over short intervals. The overall frequency measurement
accuracy is probably better than 1 per cent.
5.1. Blade Stresses. Examination of the stress amplitudes of the predominant vibration component
(fundamental flexural mode), as plotted'in Figs. 6 and 7, shows them to be a maximum betweer/
4,500 r.p.m, and 6,500 r.p.m, approximately. In general two or three stress peaks occur, the maximum
of which reaches nearly + 17 tons/in. 2 for one of the first-stage blades and + 11 tons/in. ~ for one of
the second-stage blades; they are superimposed on a general level of stress of about + 6 tons/in. ~
for first-stage blades and somewhat less for the second stage.
The points from the constant-throttle setting, varying speed tests, used to plot Figs. 6 and 7, are
also plotted together with the constant-speed characteristic points on the overall pressure rise vs. mass
flow diagram in Figs. 8 and 9. Distinctive symbols are used to define the different stress levels.
It will be noted that there is little variation in stresses with respect to position on the constant-speed
characteristic, although certain blades do seem to show consistent increase of stress as the surge
limit is approached. Confirmation of the stress characteristics in first stator blade T.4 is given in
Fig. 10a wherein the tape recorded results are plotted. Comparison with Fig. 6 indicates agreement
in stress peak location, with respect to r.p.m. The stress levels are lower, as might be expected,
being nearer to 'average' values than 'peak', but are similarly disposed.
The stresses measured on one of the inlet guides are shown in Fig. 11, and it willbe seen that the
levels are all very low, only just exceeding + 2 tons/in. 2 at three conditions. The omission of one of
the 'greater than + 2 tons/in. ~' points from the stress contour diagram (Fig. 11b), is due to
instrument errors during one of the short constant-throttle settling runs which precluded calculation
of pressure ratio and air mass flow. The scatter of points in Fig. l l a is high but it does seem possible
to define two stress peaks.
The general picture of stress distribution as indicated in Figs. 6 to 10 is very much the same as
obtained in the tests with the less stiff Mk. I blading (see first paper of this R. & M.). These'earlier
tests gave stress maxima of about + 20 tons/in. 2 and + 12 tons/in. ~ for the first and second stator
rows respectively.
In addition to the predominant fundamental mode stresses described above, one other component
frequency was in general detected in the film record analyses, this being of very much higher frequency.
These higher mode stresses are all very low, lying between + 2 and + 3 tons/in. ~ maximum for the
first-stage stators and less than + 1 ton/in. ~ for the second-stage blades. The variation of these
stresses, with speed, is in general rather random and not particularly significant. Two of the first-stage
stator blades do, however, show higher mode stress peaks similar to the fundamental mode stress peaks.

26

5.2. Blade Frequency Characteristics.--Analysis by Inspection. T h e higher mode detected on the

first-stage blades is associated with frequencies of approximately 1,800 c.p.s. Static laboratory
resonation tests on oneblade gave the second flexural mode (F.2) as 1,792 c.p.s, and the first torsional
mode (T. 1) as 2,117 c.p.s., from which it seems fairly certain that the mode excited in the compressor
is the second flexural mode. For the second-stage blades, however, the frequencies detected are
near to 2,200 c.p.s., and the laboratory tests gave F.2 = 2,020 c.p.s, and T.1 = 2,192 c.p.s. In this
case the mode excited in the compressor would appear to be the first torsional.
T h e frequencies of the predominant components of the blade vibkation are plotted in Figs. 12 and
13. T h e relatively low accuracy of these measured frequencies may or may not be responsible for
the rather confusing scatter of test points, but it does seem that where the stress peaks occur
(cf. Figs. 6 and 7) there is a definite tendency for the blade frequency to vary with speed, indicating
forced excitation. T h e dotted lines in Figs. 12 and 13 indicate the characteristics resulting from
blade forcing by harmonics of a fundamental excitation rotating at 45 per cent rotor speed. It will
be noted that there is a tendency for the measured points to parallel these linesl agreement being
slightly better for the second-stage blades.
T h e following table shows the vibration order at the stress peaks of Figs. 6, 7 and 11a:

Comparison of Vibration Orders at Stress Peaks

At stress peak
Rotor speed
(r.p.m.)

Stress
( + tons/in. 2)

4700
4800
4950
5250
5250
5300
5400
5500
5500
5700
5750
5900
6000
6100
6100
6150

7.5
7.5
8

7
9
10
7-5
2
7
6-5
17
15
11
2
14
3.5

Blade

]3lade frequency
:Rotor angular frequency

(i.e., vibration order)

1S, T.11
1S, T.4
1S, T.31
1S, T.11
1S, T.4
2S, T.19
2S, B.11
I.G.V.
1S, T.31
2S, T.19
1S, T.11
1S, T.4
2S, B.11
I.G.V.
1S, T.31
2S, B.3

5.5
5.49
5.4
4-93
5-02
5-86
5.95
5-02
-4-86
5-45
4-5
4.46
5.35
4.53
4.38
5.3

=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=

12
12
12
11
11
13
13
11
11
12
10
10
12
10
10
12

x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x

0.458
0-457
0-450
0.448
0.456
0.451
0.458
0.456
0.442
0.454
0-450
0.446
0.446
0.453
0.438
0-442

It will be noted that the orders can be equated to an integral times a factor of approximately
0.45, thus strongly supporting the hypothesis of resonance with harmonics of an exciting force of
0.45 x rotor frequency. T h e actual value of the factor varies slightly, showing a general tendency
to reduce with increasing rotor speed.
5.3. Blade Frequency Characteristics.--Wave Analysis of Tape Recordings. T h e first-stage
stator blade T . 4 was permanently connected to the tape-recorder channel (except for the additional
test for tape recording the seismic vibration pick-up output). Subsequent wave analysis of these
27
(80624)

E*

blade stresses revealed a large number of component frequencies, the major of which are shown
plotted against compressor speed in Fig. 14. Tile frequency range shown does not cover the blade
higher mode frequencies which were also found to be present..Of these higher frequency components,
that with the largest amplitude had a frequency of approximately 1,830 c.p.s., and was therefore
identified.with the second flexural mode, F.2 (see Section 5.2). In addition, frequencies equivalent
to modes F.3 (4,350), T.1 (2,150), T.2 (4,600) and T.3 (7,200) were also detected, although of
much smaller amplitudes.
Fig. 14 shows that the largest stress components have frequencies very close to tile blade
fundamental frequency (439 c.p.s.). Curves, showing an increase of component frequency with
increasing compressor speed, can be drawn to pass reasonably through most of the other points.
These curves are not quite straight lines but are found to be represented by equations of the form
60f = pxN,
where f is the blade component frequency, c.p.s.

N'is the compressor speed, r.p.m.

p are integers 3, 4, 5 . . . 12 for each curve
x, instead of being constant as it would be for straight lines through the origin, has values
between 0.43 and 0.46.
It will be noted that the points defining the major fundamental mode stress peak of the blade
(@ Fig. 10, 5,900 r.p.m.) lie on tile curve p = 10, whilst curve p = 11 passes through the one
point indicating the lower fundamental mode stress peak (5,200 r.p.rn.). Thus these stress peaks are
consistent with resonance with excitation from the tenth and eleventh harmonics of a disturbance
rotating at between 43 and 46 per cent rotor speed.
5.4. Alternating Pressures. As described in Section 3.1, capacity-type pressure pick-ups were
installed so as to detect variations in the total-head pressure at mean diameter at inlet to the
first-stage stator blades. Due to the difficulty of effecting a dynamic calibration of these pick-ups
plus total-head adaptors, amplitude measurements are qualitative only.
Generally speaking, the recorded waveforms of the pressure pulsations are somewhat random,
i.e., amplitude and frequency are not very steady at any given condition. Fortunately it is possible
in most cases, particularly between 5,000 and 6,000 r.p.m., where the blade stresses are also high,
to detect a sufficient number of successive cycles for frequency estimation. These frequencies are
plotted in Fig. 15, with indications of the associated amplitude levels. Despite the scatter of points
it does seem possible to draw a famil~r of three curves, near or on which lie the majority of points,
especially those of high amplitude. The curves indicate the same relationship of frequency and
compressor speed (viz., 60f = pxN) as did the characteristics of the blade stress frequency
components (Section 5.3). In Fig. 15 the three curves are for p = 4, 5 and 6 where x lies between
0.44 and 0" 46.
This then is direct evidence of the presence of exciting frequencies in the air flow, their values
being that of the fourth, fifth and sixth harmonics of a disturbance rotating at 44 to 46 per cent
rotor speed. At higher compressor speeds the lower frequencies (/) = 4) seem to predominate,
and as speed is reduced these tend to give way to the higher frequencies (p = 5 then p = 6). The
transitions are by no means clear, considerable overlap occurring.
28

5.5. Stage Characteristics. The variations of the stage pressure rise with flow coefficient are
plotted in Fig. 16, for the first and second stages. Also shown, dotted, are the mean curves obtained
during the aerodynamic performance tests which succeeded the present series of vibration tests.
These performance tests were terminated before the design speed (9,000 r.p.m.) was reached, but
they more than cover the vibration test range and substantiate the general trend of the characteristics
as determined less accurately during the vibration tests. Over the whole of the vibration test range,
the first stage is seen to be stalled, and the second stage is stalled at least to 7500 r.p.m. The rotating
disturbances referred to in previous Sections are therefore most probably stall cells.

6. Discussion of Results. 6.1. Excitation of Blade Stresses. As noted in Section 5.1, the stresses
measured are characterised by the presence of two or three distinct peaks, reaching a maximum
of about + 15 tons/in. 2 for some of the first-stage blades and + 10 tons/in. ~ for some of the secondstage blades (Figs. 6 and 7). Consideration of the frequency characteristics of such stresses, and
especially of those obtained (on a first-stage blade) by instrument wave analysis of tape recordings
(Fig. 14), indicates forced vibration consistent with excitation at harmonic frequencies of a disturbance
rotating at about 45 per cent rotor speed.
Analysis of the alternating pressures at inlet to the first-stage stators also shows that frequencies
equivalent to the fourth, fifth and sixth harmonics of a n approximately 45 per-cent rotor speed
disturbance are present'in the speed range where the stresses are high (Fig. 15). The evidence of the
alternating pressure records is of course direct as regards the character of any rotating disturbances
(i.e., stall cells). Although, since there were only two fixed pick-ups, it is not possible to be certain
about the number of stall cells, it is perhaps reasonable to assume (in the light of evidence of other
compressors) that there were in fact six, five and four cells rotating at about 45 per cent rotor speed,
rather than twelve, ten, and eight at 221- per cent; eighteen; fifteen and twelve at 15 per cent, etc.
Quantitative studies of flow disturbances on other occasions (e.g., Ref. 3) have shown that
successive stall cells often have slightly different amplitudes and shapes and this is certainly confirmed
by the qualitative alternating pressure records of this report. Thus if we assume that there are either
four, five or six ceils rotating in this case, any differences in their amplitudes would introduce harmonics of one cell into the forcing of a blade. Thus the presence of harmonic components three
to twelve of one cell rotating at 45 pet cent rotor speed in the blade stress spectrum does not
invalidate the above conclusions.
It is interesting to note that the stress amplitudes relevant to curves p = 4, 5 and 6 of Fig. 14,
although very small (< + 0.2 tns/in-2), do reach maxima in the same region as the corresponding
pressure pulsations. The stress characteristics for p = 3 and 7 show no definite maxima and are
less than + 0.1 tons/in. .~, whilst those for p = 8, 9, etc., are near enough t o the blade's natural
frequency to be affected by its resonance curve and show stress magnification near to the fundamental"
stress peak (5,900 r.p.m.).
In the first paper of this R. & M. it was noted that at conditions of resonance of one of the blades
in a row, the other instrumented blades appeared to be vibrating at the same frequency and their
phasing such as to suggest that neighbouring blades vibrated in anti-phase. Although, in these
tests, the majority of records taken at or near a blade resonance show some synchronization of
frequency, the phasing does not fit the at~ove hypothesis. The blades are, however, under these
conditions approximately either in phase or anti-phase. It is still therefore probable that coupling
between the blades determines the phasing rather than the stall cell forcing itself.
29

6.2. Rotational Speed of Stall Cells. Figs. 14 and 15 indicated that the relation to compressor
speed of the stall cell excitation frequencies and harmonics is not linear, although nearly so. For
simplicity the equations to both the stress component frequencies and the alternating pressure
frequencies curves was expressed in straight-line form, 60F = pxN, the notation being defined in
Section 5.3. The relative stall cell speed x is seen to be not quite constant, and its variation
is illustrated in Fig. 17, where in Fig. 17a are plotted the values of x obtained from the curves of
Fig. 14, and Fig. 17b shows the values deduced from the alternating pressure m4asurements
of Fig. 15.
The general trend of reduction of stall cell speed is evident and the agreement between the
stress and pressure analyses is good. Average values of cell speed are seen to vary from 45 to 46 per
cent rotor speed at 5,000 r.p.m, dropping to 43 to 44 per cent at 6,500 r.p.m. This trend is also
substantiated by the evidence from the resulting blade stress peaks as presented in the Table
of Section 5.2, where the blade vibration orders are expressed as harmonics bf a fundamental
forcing frequency which varies between 0.46 and 0.44 times rotor frequency as the speed
increases.
6.3. Stress Amplitudes. The peak stresses (at resonance with stall cell excitation) measured on the
Mk. II blades of these tests are, for the three instrumented first-stage stator blades, + 15, + 17 and
+ 13 tons/in, a. In the tests of the first paper of this R. & M. on the same compressor with its' less
stiff Mk. I blading, the stresses reached were + 20, + 14 and + 13 tons/in, a. Similarly, for the
second-stage blades, resonance stresses of + 10, + 3 and + 11 tons/in, a were recorded for the
Mk. II blades, compared with + 12, + 12 and + 11 tons/in, a for the Mk. I blades. The average
reduction in stresses in changing from Mk. I to Mk. II blading is at first sight surprisingly little,
having in mind the considerable increase in stiffness o f the Mk. II blades (the average first-stage
stator natural frequency has been practically doubled, while that for the second stage has increased
by 80 to 90 per cent).
Consideration, however, of the damping capacity of blade materials offers a possible clue as to the
reason for these rather small stress reductions. Determinations 4 of the damping capacity' of
engineering materials at different amplitudes of stress reversals show that, in general, the damping
capacity increases fairly uniformly with increasing stress amplitude up to about 70 per cent of the
fatigue limit, at which point (termed the Cyclic Stress Sensitivity Limit) the damping begins to
increase much more rapidly. In the region below the C.S.S.L. the resonance amplification factor A r
resonance amplitude due to cyclic force 1"~

(defined as ~ c ~ a W ~ - ~ r ~ - ~
/ decreases rather slowly with increasing stress, but above
the C.S.S.L., corresponding to the rapidly increasing damping capacity, the A~ decreases much
more rapidly. The maximum stresses in the Mk. I blading, which was cast in H.R. Crown Max, are
seen to be close to or above the fatigue limit (quoted as between + 10 and + 15 tons/in.a), at which
level the resonance amplification factor A,. would be quite low. The Mk. II blading (S.80 material),
however, has maximum stresses of up to only 70 per cent of the fatigue limit (quoted at + 24 tons/in.a),
and would therefore be subject to a much higher A~.. Thus, on the assumption of the same stall cell
loading on the Mk. I and Mk. II blading, it is evident that the Mk. II blading has a higher resonance
amplification factor, which tends to offset the reduction in stress which would be expected due to
the increased section modulus of the blades only.
It should be emphasized that this hypothesis is tentative only, the following factors being borne
in mind:
30

(1) Complete material damping information of the type indicated above does not appear to be
available for the two blade materials concerned. The information in Ref. 4 includes slightly
different materials and emphasizes the influence on damping properties of small changes
in composition and treatment, etc.
(2) The assumption of similar forcing and neglect of other damping effects (aerodynamic, root
friction, etc.) is rather too sweeping to be fully justifiable.
Nevertheless, it is thought that the effects described above must certainly play a considerable
part in determining the blades' vibrational behaviour. It should be noted that the change to Mk. II
blading, whilst not resulting in as large a reduction of absolute alternating stresses as had originally
been hoped, does increase the factor of safety (Soderberg criterion e) from less than 0-5 to about 1.5.

7. Compressor Casing Vibrations. The output of the seismic vibration pick-up, attached about
halfway along the casing in a direction of major response to vertical vibrations, was recorded on film
throughout the tests. An additional varying speed test was also made with the pick-up output recorded
on tape. The salient results of the subsequent wave analysis of the tape recordings are presented
in Fig. 18.
The most significant result is the large increase in the first shaft order vibration velocity over the
range in which stalling effects are most prominent in the compressor (cf. Figs. 6, 7 and 16). This 1N
vibration peak does not reach values normally to be considered dangerous, but its presence is not
readily accountable although one m!ght expect frequency components representing reaction forces
arising out of the rotation of stall cells. It is also of interest to note the detection of a band of
frequencies consistent with the first-stage stator blade frequencies with peaks conforming to the
resonances of these blades as described earlier.

8. Conclusions. A detailed examination of the stresses in the first two stator blade rows of the
N.G.T.E. No. 118 Compressor Mk. II has been made, together with some qualitative measurements
of unsteady flow conditions. This Mk. II build replaces the original Mk. I build which suffered blade
fatigue failures, and includes blades of considerably stiffer section in a material of higher fatigue limit.
The main conclusions are as follows:
(1) Stall cell excitation is again present, giving rise to moderately high resonance blade stresses.
(2) There appear to be regimes of six, five and four cells as the compressor speed increases
from about 50 per cent to about 70 per cent of design speed. The relative speed of stall cell
rotation varies slightly, decreasing over this compressor speed range from 46 per cent to
43 per cent of rotor speed approximately.
(3) Harmonics of one cell are present in the blade excitation. The major resonances of the
first-stage blades (with stresses up to + 17 tons/in, a) are attributable to the second harmonic
of a five-cell regime. Minor resonances are consistent with the second harmonic of a six-cell
regime and with the eleventh harmonic Of one cell. The second-stage blade resonances are
with the second harmonic of six cells and with the thirteenth harmonic of one cell.

Defined by

g~s bending stress 2v aiternating stress

yield stress
fatigue limit stress

1
factor of safety

31

(4) Away from these resonance conditions stalling flutter is present resulting in stresses of
about + 5 tons/in. ~ for the first-stage and + 3 tons/in. 2 for the second-stage blades.
(5) Despite the considerable increase of blade stiffness in the Mk. I I build, the average
resonance stresses are very little reduced compared with the original blading. It is thought
that the increased resonance amplification factor of the Mk. II blading at stresses lower
than its fatigue limit is mainly responsible. Due predominantly to their increased fatigue
strength, the safety factor of the blading has increased from 0.5 to 1.5.
(6) A significant increase of first, shaft order compressor 'roughness' during the period of stall
effects was noted. An adequate explanation for this seems to be lacking, and the phenomenon
merits further attention, especially as it appears to have been experienced elsewhere.

Acknowledgements. The author wishes to acknowledge the considerable help given to him,
during these tests and their subsequent analyses, by his colleagues.

REFERENCES

No.

Author

1 J.R. Forshaw

Title, etc.

..

An investigation of the high alternating stress in the blades of an

axial-flow compressor.
A.R.C. 17,132. June, 1954.

2 J.R. Forshaw and H. Taylor

The measurement of high frequency alternating pressures in gas

turbines.
A.R.C. 17,134. June, 1954.

3 T.J. t-Iargest and R. Chaplin

4 B.J. Lazan

. . . . . .

..

(Unpublished work at N.G.T.E.)

From Fatigue published by the American Society for Metals,
Cleveland, Ohio. 1954.

32

;~LV

E/,HALI~T T 0
ATMOSPHERE

FJ{HAUST TO

~ALVE

ATMO'~f:t-IERE

CQMP~E6SED A~R "ROM

PLANT COMPRE,3~OR HOUSE

INTAKE FROM
ATMOSPHERE

COMBU,~TII::3N,
SYSTEM

PLATE

I
i
i
I

N~ I1~ COMPRESSOR

N&iI~TU~8~E

PgWER TAKE
- OFF (~EARS
IUMIT

SIDE ViEW

PLAN

~IG. 1.

PLAN

~O, 118 Compressor test installation

(Diagrammatic).

ELECTRICAL COMNECTION
MADE NE~E
BALANCE A ~ '
\
I:~APHF,AGM)'~
CONNECTION

CBAr-.F-.O F

,
1~'~i /
Ir//l'-1~

CAPACITY--TYPE
P~ESSUREPICK UP
{'-~NOWN WITHDP-..A.WN;)

~ //INSULATED ELECTRODE
I T O T A L HEAD
AI3AFI'OR

d
t

I
I
i

'
L . . . . .

i,
i

....

($CHEI~/1ATIG DRAWIN~ONLY")

FIG. 2.

Installation of pressure pick-up.

33

VALVE5 CV35~

/ar

POWER'
INPUT PLUG
H G. 350'V. ~IP_mA
L.q: 3~V, O-2A.)

/&

II

?'l

~
~0UTPUT
[ ~
IPLUG
t
o / I(~o M~IN

c~

l A~).

eUa-CHA65~S

IN~UL~,WF..D FR'OM CAS~

'FIG. 3. Head amplifier.

r--

35OV, ~OmA

'~H?4
tOOO'V

: -/t4
"SALF

20~<
4Azr 0

IIALF G

II---"

2.d IF

5~ I - -

600 :

"il~ -L~
ll/l:.......... l'I
! ..... L

~
OK

JACK OUTPUT
(TO C.~.~)

Fro. 4. Main amplifier.

34

~'TA.G E.
(D~SCONNECT JACK~C,D,
CONIMF__CT JAC K.-PGtH.
CONNECT C-~,"E ~D-,"E)

I : : : : : : AEI~0DYNAM{C PEtE TE~T.5

CONSTANT 5NEED

(.51"~AM TU~F~INE OI~IVE. AXIAL INTAKES)

_~_G_Q_~r_.~NSTANTTHEOTTLEJ" VII~.ATION
TEST..~ (N~-115 E~AS.
TUI~FNNE DelVE, VOLUTE INTAr.E)

-~s--~--~-

8500

P~E~SU~E
IP-.ATIO.

--

NO5. ON r'U~VE5 . ~ E EO~F-.E.TEO

SPEEDS' I~'P'/sl' C ; ~ '~'P'/Vl'~

F~DO0 ~ t .
7500 .%k%. ~

7ooo

4ooo
0

I0

~.0

.~
30

(i /'t

~.

dooo
40

50

60

EOi~ECTED AII~ MA55 FLOW. L ~ . / S E E .

FIG. 5.

Operating conditions for No. 118 Compressor Mk. II.

35

7O

~u

~-AD E "E a-.

":..STATIC

~I-,O.-;-

"=.~-.~.CP...5.
~7

~LADE TI9
.TATE -- 516

.s.

517

10

o/~l

G---"O
0

~-0o0

,5000

~"0-6000

7000

~.RM.

~O00

~000

tO

+-15

-'10

7000

6000

7000

-'15
t~LADE B.3

I~:1..~op_.

~TATIC -~31.C.P.5

6000
I~.P.M.

,q.P_.,=

I~LADE.T.II.

2. 5PI

-+,E

~o

,q.a7/

+1

L~
Z -*,5

"~STATIr" "=" 5 ~ ' 3

-+to

5
z_

4000

<

5000

600o

x<

~EIO0

<
E
-i

1,5
~LADE T.31
STATIC ~.-.~1-46.CP.5

5OOO
~.P,r,1.

z
3

_z

7000

I~.P.M.

r~447
~'~'P- [ I

+-15
t~LADE 15,11
5~5

~'-.STATIE"=" 5 3 5

-*10

-+10
o

~._3

\ sa~l.

+-5

o
o

'q,O00

5000.

r=OOO

7000

~.P-M.
(NOTE :-NUblF~E~5 ON CUt~VE5 .SHOW TI~END~ IN VlP,P--.ATIOIN
F~'O~UEINCIEI~, C.P.-~ .)

FIG, 6. Predominant stresses in first-stage stators

(Constant-throttle setting tests).

4O00

5000

r:~oo

7000

I~.P.M.
('NOTE:-NUMBE~5 ON CUI~VF-..5 5HOW TP--EIND5 IN
VIF~P--.~.TIOINI5TI~E.5.5 FPEQUEINCIES) =.p.s.)

FIG. 7.

Predominant stresses in second-stage stators

(Constant-throttle setting tests).

0131313RPM

613Q0 l~pIv

~'C

. 75L0e M

"t

BLADE

3'0

T.k9
a

60130 I~PMo

"
1

~:

7000 ~PM

7S013 ~PM
Do

6 0 0 0 ~PMoo

l.d

.=~.o

~'0

500 ~

~.~~;.f~!.,.:';:.'!.:'.~

~1313 I~PM
I'0

m
w

5 0 0 0 l~plv
o-

400D I~PM.

o
I'0

lO

~o
MA55

30
FLOW

40

,50

IO

,Su

i~O

LB/~.

"~

513

~13
,=1.0
NA55 FLOW L 6 1 5 ~ i

6U

E50013 I~PI'

I ..oo:.-.1

15000 ~PM
5r
3"0

BLADE

BLADE

B.3.

TII.

I: "

70o0 ~ . ~
@

.oI

~0130 I~PM o ~"

P*C

P ~ O 0 I~PM 0 ~9~

~'O

5 0 0 i~PM

o:

.~

~1313o
(A

W
I'O
I'O

IO

~0

~0
~0
FLOW LI5 /! ~

MA5~

IO

~13

~0

MA55 FLOW

~0

60130 I~PM o
Ul

I1/

~n1313 I~PM o
P-'0

p-n

o:"

13

IO

113

o"*~ ;?J"

~-0
30
4.0
M A 5 5 FLOW L B / 5 ~ 0 .

50

&

FIO. 8.

5 - - ~ l O

-+113---~15

~13
30
~,0
M A 5 5 F L O W L O [J S E C .

513

,Bu

KEY

ro

~TR~55~5.

~ E Y TO

4 0 0 0 !~pNI
1,13

I'D

Oo

tO
m
M

,~001~~ I~PN
o.,~qo

@u

d
W

~PM

7ooO~moOO
: 12

75O0 mPM
~LADE ~31.

3-0

~1300 I~PI
7 5 0 TI

6!-A13E Bill.

,~'0

0 0 0 0 t~PI

6u

50

4o
L 6 /I 5F'C,

IN a

Predominant stresses in first-stage stators.

FIG: 9:

I~]
~

~a~s
~'

-+S ~ + -

-*J0~15

rarqs
113

IN e

Predominant stresses in second-stage stators.

GO

%
Z
0
P
+1
tO
IG
OZ
IL0

+~0

Z
tQ

+-iO

&/
I--

I \

30

o /

*-5

<

<

<Z
Ld

5000

4000

6000

7000

'=t.000

5000

~=000

.P.M. - N

(0.)

5T~ES5

AMPLITUDE

CHAI~ACTEi~ISTIC

A ) C O N S T A N T TH~OTTLF" 5KTTIN~ ~ L I N S ,

I
I

a2

O 4,~0

~00
"

~h

7t/

~o
430

70O0

~.P.M.

450

~d

d-----~-~

6o#'olI,o

45 N ~

3"O

7 5 0 0 I
~.P.M.
700O
!~.R.M.

(It
L 4@0

6OO0
W

5000

6000

7000

i~. P. M , - N

oo

~,'0

5000

I~.P.M.
,o

mo$

8oj'o Io,.o.45 N

/
4OOO

I
~*
: g
o

"

a g
o

'::14300I~.IRM:o o'~
I'O

KEN TO 5 T ; ~ E 5 5 E 5

5.~+10

TONS/IN

tO

~.0

~0

K E Y T O ST~E..~SE5

(6) 5-[eESS

reE~UKNCh"

40

,50

MA5S FL.OW ~ - ~ / ~ E ~ .

13 1 0 ~ +_ 15

~ I-~

T O I N S / I IN a

C H A ~ A , CTKRISTIC
5T~F55

CON5TANT THI~O'TTLE SETTING

C.~NTOLII~5 .

EUNS.')

Fins. 10a and 10b. Predominant component from magnetictape analysis (First-stage stator; Blade T.4)

I~.RM.

FIGS. l l a and llb.

Predominant stresses in inlet

guide vane.

~5"0

5'40

BLADE T 1.9
td

112

Ld

,:l-~:l-O

o!ff ~,~,,, '

bJ

STATIC"=.~-.-~9.C.RS!
'~.~0

,~tO00

O/
0 0,0
, ~ ,~

I\
I\

~,~0

:~

{12
LJ ,se.o

(DO

p=l

/P

~5.=~DOO

,=

~M

p:

,:t~C

7000

(50OO

~,

/
,~
_

U
Z
D
(3
kd

//

4~O

(.o

~ p__

~000

a /I

'~t.0OO

~'000

p=ll

GO00

7000

0 I
o /00
,o . o o

;7t'%

ll;
U
0

/
/o

/
o/1

/o

,,go.
/ c,,,,

BLADE 15.3

%/

/
p
10

o /

-/
?

,/

I p=lP-

i/ p .11

% ,/

,5"10

,~90

7000

,,

5..~0

>u
Z
i

6000

E) ~

/
i/

/ Z, A

/
/ ~
~ ~_
~ll ~
A/ u,..~. ,{ % , 9

5000

i/

==1.5 /

aA

~/
0 (' 0 / ~
~oo{,
, ( ~ / 1 o "e
o
O9/

o,c i

$50

~-DO

x,. /
//~.

~.RM. - N

~LADE T ll
:STATIC --~I.C.P.N.

Oz

'~1

g/

~J:?M.- N .

~O

//

/4
/~l~,.

>.
U

/'% ,,,
%o'\

jISTATIC "=.. SlGc.p.5

a 5000

4000

6000
i I t~.P.M.-IN.

7000

!~.RM. - N .

/
d

~1-60

570

BLADE T.31
FSTATIC"=.r44,B.C.R~ I

112
d

450

/ OO

>U

n'
d

~
(D

/
,,:,l

-~x

/~

o /k ~

~.
,~ / i

oO

KEY TO mT~ES~JES

~
[:3
X

:- D O T T ~

UN~S

FIC. 12.

ae~ ricers 0~

sea c .

.A~,

I
0

O 0 '6 o

11 ~

0 ,,/
A iI

.0

\
p=9

51D

60oo

-+a ~

70oo

~-00O

5000

GOOD

7000

~.P.M. - N .

<-+P

@OT~

zi /

5~0

-/A

p7-11
p-IO
5oooz~ I~.I~M.-N

~00o

550

t~

",.,
'~,'30

BLADE 5.11
~.-%TATIC ~--"535c.p.s

-+5

~-5 ~ r i o
~10~ Z I 5

TONS / IN a

:" ~1,~

KEY TO 5TI~E5

~5 ~-+I0

[]

+-IO--ZI5

TONS

(NOTE :--DOTTED LINES AI~E PLOTS OF" 6 0 3 E = Io x . 0 " 4 5 N ?

r x o.~sN.)

Predominant blade vibration frequencies.

First-stage stators.

FIG. 13.

Predominant blade vibration frequencies.

Second-stage stators.

IN a

lS'r',.-fiTAGE STAT0 ,P-.

~0

i OLAD~ ~ 4~ - - ~,STATIff. ~ 4~9c.p.~

f/

,/ ,I

/
/

,/ /

= =1~.

,Q/

~ / p = l l

) =6

/ . ~_--.p=lo

~00

0;

[.I
[Pl

_~

./

~*""

O
~oc

U
Z

3
~OO

~,

100

"
t-

160
4~

0Q
W
0

~00

1"40
o Q

,,~.~ ,'A'~"

100
IOO

4000

5000

a000

5OOO

6000

5YYI~50L

FI~EGI. COMPONENT
e

A
B

-*5 ~ 1 0
-+10 ~ - ~ 1 5

(~q~-IN EQIN. FOP-.. E:UI~VESp =.,.~ TO p =le)

7000

~.P.M. - - N .

I~.RM. - N
5Y'N ES0/,...,

'~-000

7000

MAX.

O
A

(~INEqN. FOI~CU~VESp=~t-5,6,

ALTERNATING
PI~E55UEE ANJFL, ITUOF_
~- 0"5
0"5 ~ 1 - 0
I'0 ~ 1 - 5

I IN5 H 5 .

PEAKTOP E A K

0-,=1.~, ~ o~ ,~ 0 . 4 6 A P P ~ 0 X .)

O-.=I....N~ co.. < 0.'~,-6 A P P E [ O X , )

FIO. 14. Frequency analysis of magnetic-tape

stress recordings.

FIC. 15. Frequency analysis of alternating

pressure recordings.

0'8
8550 R.RM.

I%"r. 5 T A G

E-

0.7
/
0.6

8UE55EQUE.N-r
~,E.P~OD'YNAM I C
PEP-,FOI~M~,NCE~, /

0"5

q.
4

/
!
!

x
O
Q
+
~,

[~, P. M.
4000
5000
6000
"7000
7500

/
!

0"4
Z
LJ
L;
LI,

1J
O
L2

0.3

O,~

O' 4

O. 5

O.6

~N_o 5 T A G E.
7 5 0 0 ~. P. k.'l.
0,7
8 3 5 0 R.RM

tO
tO
M
n

0"7

0"8

LJ
(0
U3
OZ

0-3

0"S

SU555QUKNT
AERODYNAMIC
PE~FO~MANC~
~TESTS.

0"5

log

R.P.M.

X 4-000

5000
m 6000
+ "7000
h. 7 5 0 0

k V/
/
/
I
I
I
I

0'4
o

0,2
O'~

0"5

E]

O" 40"5
0"6
VLOW C.OE~-FICENT, X/m/u-

FIG. 16. Stage characteristics.

41

0'7

0"47

0,10

p=ll

0"14

Pi~a

a:

s
LJ
i.I

&

0.~3

0.I~

0"I0

Z
O.~,~
~I000

~DO0

A) F~OM BLADE

~OOO

7000

~.RM, " N .

~Tl~'E,...fiSF~EQUENCY ANALY~I~ (FIG.I~)

t)
/0
tJ
> 0,06
Z

_q
0.~14

bP

n/ 0 . 0 4
10

>
0.45

O'Oa

3C

o
4000

5000

6000

7000

Et.R M - N .

1~ 5HA~'-r OEOE~ LN

O"~-3

~,
0"~,~
*~OOO

5000

6000

7OOO

I~.RN. -IN.

15.) F~0N ANALYSIS OF ALTEP-.NATIINQ PeE~SUI~E5 (FIG.'5)

:~

I~OTATIONAL .-~PEED O F Dh..~TUI~ISANE:E

EOTOI~ SPEED

BID[
N.p.

( WHERE 3(: I.~ FI~EQUENCY OF ALE. 5T;=ES~ O~ PI~ESSUi~E r'.R~ ,)

Fins. 17a and 17b.

Stall cell rotational speed.

Is-Ir ~-f,~,GE 5TATOI~ FI~E~UENCIE,~'430->445CP.5.~.

NOTE" ',- THE. HIGFIEle INTEGP-.AL 5 R A F T O ~ D E I ~ S

(UP TO IO N) AP-.E_ P I ~ E . S E N T A T LOWEI~
LEVEL$ THAN 2N, BUT AI~E OMITTED
V l e O M G ~ A P N TO A V O I D C O N F U S I O N .

FIG. 18.

Predominant components of vibration at

compressor casing.

Publications of the
Aeronautical Research Council
ANNUAL TECHNICAL REPORTS OF THE AERONAUTICAL
RESEARCH COUNCIL (BOUND VOLUMES)
I9X Aero and Hydrodynamics, Aerofoils, Airserews, Engines, Flutter, Stability and Control, Structures.
63s. (post zs. 3d.)
I942 Vol. I. Aero and Hydrodynamics, Aerofoils, Airserews, Engines. 75s. (post as. 3d.)
Vol. II. Noise, Parachutes, Stability and Control, Structures, Vibration, Wind Tunnels. 47s. 6d. (post xs. 9d.)
1943 Vol. I. Aerodynamics, Aerofoils, Airscrews. 8os. (post 2s.)
Vol. II. Engines, Flutter, Materials, Parachutes, Performance, Stability and Control, Structures.
9os. (post 2s. 3d.)
I944 Vol. I. Aero and Hydrodynamics, Aerofoils, Aircraft, Airserews, Controls. 84s. (post 2s. 6d.)
Vol. II. Flutter and Vibration, Materials, Miscellaneous, Navigation, Parachutes, Performance, Plates and
Panels, Stability, Structures, Test Equipment, Wind Tunnels. 84s. (post 2s. 6d.)

t945 Vol. I. Aero and Hydrodynamic, Aerofoils. x3os. (post 3s.)

Vol. II. Aircraft, Airserews, Controls. 13os. (post 3s.)
Vol. III. Flutter and Vibration, Instruments, Miseellarteous, Parachutes, Plates and Panels, Propulsion.
r3os. (post as. 9d.)
Vol. IV. Stability, Structures, Wind Tunnels, Wind Tunnel Technique. 13os. (post as. 9d.)
1946 Vol. I. Accidents, Aerodynamics, Aerofoils and Hydrofoils. z68s. (post 3s. 3d.)
Vol. II. Airscrews, Cabin Cooling, Chemical Hazards, Controls, Flames, Flutter, Helicopters, Instruments and
Instrumentation, Interference, Jets, Miseellane0us, Parachutes. i68s. (post 2s. 9d.)
1947 Vol. I. Aerodynamics, Aerofoils, Aircraft. 168s. (post 3s. 3d.)
Vol. II. Airscrews and Rotors, Controls, Flutter, Materials, Miscellaneous, Parachutes, Propulsion, Seaplanes,
Stability, Structures, Take-off and Landing. i68s. (post 3s. 3d.)

Special

Volumes
I. Aero and Hydrodynamics, Aerofoils, Controls, Flutter, Kites, Parachutes, Performance, Propulsion,
Stability. za6s. (post 2s. 6d.)
Vol. II. Aero and Hydrodynamics, Aerofoils, Airserews, Controls, Flutter, Materials, Miscellaneous, Parachutes,
Propulsion, Stability, Structures. 147s. (post 2s. 6d.)
Vol. IIL Aero and Hydrodynamics, Aerofoils, Airscrews, Controls, Flutter, Kites, Miscellaneous, Parachutes.
Propulsion, Seaplanes, Stability, Structures, Test Equipment. I89s. (post 3s. 3d.)
Vol.

Reviews of the Aeronautical Research Council

1939-48

3s. (post 5d.)

1949-54 5s. (post 5d.)

Index to all Reports and Memoranda published in the Annual Technical Reports
19o9-1947

R. & M. 2600 6s. (post 2d.)

Indexes to the Reports and Memoranda of the Aeronautical Research Council

Between lXlos, a351-2449
Between Nos. 2451-2549
Between Nos. 2551-2649
Between Nos. 2651-2749
Between Nos. 2751-2849
Between Nos. a851-2949

R.
R.
R.
R.
R.
R.

& M.
& M.
& M.
& M.
& M.
& M.

No.
No.
No.
No.
No.
No.

245
2550
2650
2750
2850
2950

2s. (post 2d.)

2s. 6d. (post 2d.)
2s. 6d. (post 2d.)
2s. 6d. (post zd.)
2s. 6d. (post 2d.)
3s. (post 2d.)

HER MAJESTY'S STATIONERY OFFICE

from the addresses overleaf

R. & M. No. 3181

Crown copyright I96I

Printed and published by
HER MAJESTY'S STATIONERY OFFICE

To be purchased from
York House, Kingsway, London w.c.2
423 Oxford Street, London w.I
x3A Castle Street, Edinburgh 2
Io9 St. Mary Street, Cardiff
39 King Street, Manchester
5o Fairfax Street, Bristol z
2 Edmund Street, Birmingham 3
80 Chiehester Street, Belfast z
or through any bookseller

Printed in England

R. & M. No. 3181

S.O. Code No. z3-318I