.
4
.
resten
2
?
.
..
:
I OF |
ORNLP
2703
ya
.
.
.
.
.
"
.
?
Bi
EEEFEFEE
MICROCOPY RESOLUTION TEST CHART
NATIONAL BUREAU OF STANDARDS - 1963
..
.
.
ORNU-P-22.02.
Paper to be presented at the Southeastern Symposium in Missiles and Aerospace Vehicles
Sciences, American Astronautical Society, Huntsville, Alahama, December 5-7, 1966
CONF-661204_1
NOV 2 9 1966
Cava
SICES
IH.C. $ 4,00, MN 50

MASTER
THERMAL FATIGUE ANALYSIS OF A CRYOGENICALLY COOTED ROCKET NOZZLE*
A. E. Carden, † D. G. Harman, and E. A. Franco-Ferreira
Metals and Ceramics Division
Oak Ridge Natioria Laboratory
Oak Ridge, Tennessee
m
.

LEGAL NOTICE
RELEASED FOR ANNOUNCEMENT
IN NUCLEAR SCIENCE ASSTRACTS
This report was prepared as an account of Government sponsored work. Neither the Valted
Statos, nor the Commission, nor any persos acting on behalf of the Commission:
A. Makes any warranty or representa uon, expressed or implied, with respect to the accu-
racy, completeness, or usefulness of the informadon contained in this report, or that the use
of any information, apparatua, mothod, or procers disclosed in this roport may not infringe
privately owned rights; or
B. Aanmes day liabilities with respect to the use of, or for damages resultat from the
use of any information, apparatus, method, or proceso disclosed in this report.
As urod in the above, "person acting on behalf of the Commission” includes any om-
ployeo or contractor of the Commission, or ea.ployee of such contractor, to the extent talt
such employee or contractor of the Commission, or employee of such contractor preparos,
disseminates, or provides access to, any information pursuas to his employment or contract
with the Commlusion, or his employment with such contractor.
Research sponsored by the U.S. Atomic Energy Commission under Contract
with the Union Carbide Corporation. Work performed at the Oak Ridge National
Laboratory under Interagency Agreement SNC-17 to AEC-NASA Space Nuclear
Propulsion Office, Cleveland Extension.
TSumner participant from the University of Alabama.
.
THERMAL FATIGUE ANALYSIS OF A CRYOGETICALLY COOLED ROCKET NOZZLE*
A. E. Carden, † D. G. Harman, and E. A. Franco-Ferreira
Metals and Ceramics Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee
INTRODUCTION
Most interplanetary mission profiles require a source of primary thrust
which can be throttled and cycled on and off a large number of times during
a giver journey. One attractive propulsion scheme for a mission of this type
is the nuclear rocket presently under development in the Rover Program.
In essence, a nuclear rocket is a cryogenically cooled exhaust nozzle
coupled to a reactor which acts as a heat sour" for the propellant. The
configuration of the exhaust nozzle is typical of most regeneratively cool.ed
nozzles. That is, the interior surface of the nozzle, which is in contact
with the hot propellant gas, is made up of a multiplicity of suitably shaped,
thin-walied tubes which are bonded to each other. The cryogenic coolant passes
through the insides of these tubes, thereby preventing them from melting. In
many cases it is necessary, in order to provide greater strength to resist high
chamber pressures, to bcnd a relatively heavy reinforcement jacket to those
portions of the tubes which make up the outer surface of the nozzle. A schematic
diagram of a typical cross section of such a configuration is shown in Fig. 1.
It is apparent that, in a nozzle of the type described above, there will
be a very large AT between the tube crowns on the hot gas side and those por-
tions of the tubes bonded to each other and to the jacket. The AT's experienced
in normal modes of operation are capable of producing calculated thermal strains
Research sponsored by the U.S. Atomic Energy Commission under Contract
with the Union Carbide Corporation Work performed at the Oak Ridge National
Laboratory under Interagency Agreement SNC-17 to AEC-NASA Space Nuclear
Propulsion Office, cleveland Extension.
Summer participant from the University of Alabama.
La...
.
..
............. ...
-
-
.
......................
.
...
.
.
.
... ...
.
... ... ... ... ...
...
.
..
.
...
......
......
...
.
..
...
...
.
..:-.
-.1.
.
.
,
.-*
-.
.
.
*
*
*
of as much as 2%. Due to the restraint provided by the jacket, it is likely
that a significant portion of the longitudinal thermal strain will be converted
into iechanical deformation of the tube crowns. Thus, under conditions of
cyclic nozzle operation, it may be possible to induce low-cycle thermal fatigue.
failures in the tubes.
It is not economically feasible to use full-scale nozzle hardware to
analyze experimentally the behavior described above. Consequently, there is
an incentive to devise a scale-model type of test which is capable of evaluating
the effects of all the important parameters which are operative in an actual
nozzle. In the case of thermal cycling stresses, as discussed here, the impor-
tant considerations include (a) the temperature distribution, the physical
and mechanical properties of the nozzle materia. the operating environment,
(a) the geometry of the nozzle elements, and (e) the loading conditions.
This paper describes the thermal cycle testing of specimens which were
designed to simulate and evaluate the conditions exi
geileral configuration shown in Fig. 1. The test results are related to impor-
tant fabrication and operating parameters. An effort is also made to relate
the results to a range of nozzle operating conditions.
SPECIMEN PREPARATION
Each test specimen consisted of a planar array of six thin-walled
Hastelloy X tubes brazed to each other and to a thick Hastelloy X base plate.
The tubes were 1/2-in, diam with a 0.012-in. wall and the base plete was hy in,
wide, 10 in, long, and 7/8 in. thick. A schematic diagram of the specimen is
shown in Fig. 2. Both ends of each tube were plugged. Entry and exit ports
for the coolant were drilled through the base plate and into the tubes near
opposite ends of each specimen.
Each stecimen was instrumented with numerous thermocouples to provide tem-
perature distribution date. These thermocouples were Chromel-F-Alumel with a
wire diameter of 0.020 in. They were resistance welded to the tube crowns and
to the base plate at various locations. For each thermocouple, the individual
wires were welded to the specimen a short distance apart so that a portion of
included in the junction. The thermocouple Locations
and attachment procedure were carefully controlled and reproduced for each test.
The heat source used in these experiments was a quartz lamp radiant furnace.
It was positioned parallel to, and approximately 1/2 in. above, the tube crowns.
The furnace had a lamp area of 8 in. by 2 1/2 in. and an electrical power input
to the lamps of 13.2 kw. This produced a measured heat flux at the specimen sur-
face of approximately 700 w/i ... An overall view of the test setup, with the
furnace tilted up to expose the Lamps, is shown in Fig. 3.

The specimen base plate was kept at or near the coolant temperature by
bolting it to a massive copper heat sink. As evident in Fig. 3, this lieat sink
was contained in a pan which enabled it to remain completely submerged in
coolant during the test.
Coolant flcy was introduced into the insides of the tubes through a copper,
manifold, shown at the left end of the specimen in Fig. 3. This manifold was
positioned beneath the surface of the coolant in the pan to maintain a constant
temperature for the flow of tube coolant.
TESTING PROCEDURE
In all tests, LN2 was used as the coolant. The pan shown in Fig. 3 was
maintained full of LN2 for the duration of each test. A separate source of LN2
was connected to the tube coolant inlet manifold. The LN2 flow which cooled the
tubes was allowed to exhaust directly into the pan at atmospheric pressure.
Provisions were made to maintain an argon atmosphere over the tube surfaces to
protect them from excessive oxidation,
A test run for a given specimen was characterized by thermally cycling the
tube crowns from LN2 temperature to an elevated temperature and back to LN2 tem-
perature. A number of identical cycles were reproduced until tube failure
occurred.
A complete thermal cycle proceeded as follows:
1. The heat sirk and base plate were held at or near LN2 temperature.
The tubes were cooled to LN2 temperature by their internal coolant flow.
3. The quartz lamp furnace was turned on and the tube crowns brought to
the specified peak temperature and held for a predetermined length of
time (hold times of 15 sec and 5 min were studied). Coolant flow
through the tubes was maintained during this step.
The furnace was turned off and the entire specimen was cooled to LN2
temperature.
A typical test thermal ycle, as traced by a strip-chart recorder connected
to the specimen thermocouples, is shown in Fig. 4. This particular cycle was
'for the 5-min hold time;
During test equipment checkout runs using mockup specimens, it was noticed
that small dents in the tubes of the mockup specimens acted as geometric instabil-
ities and led to premature failure. Since the existence of small tube dents as
fabrication defects cannot be ignored, it was decided to include a study of their
effects in this program. Controlled dents were therefore made in some of the
specimen tube crowns. This was done using 9. 1/4-in.-diam cylindrical indenter
: with its longitudinal axis parallel to the plane of the specimen base plate,
but at a 90° angle to the lciigitudinal axes of the tubes. Dents were nade to
depths of 0.005, 0.010, and 0.020 in.
.
A closeup view of the tube crown surfaces of one of the specimens prior to
testing is shown in Fig. 5. Both smooth and predented with two controlled dents
0.020 in. deep) tubes, as well as the attachment points of six monitor thermo-
couples, are shown. The various parameters studied in the entire series of tests
are listed in Table 1.
Table 1. Thermal Cycle Data
Hold Time
at .
Temperature, °F Temperature
Maximum Minimum Maximum
Number of Thermal
Cycles to Failure
First First Crack
Visible Visible Through
Wrinkling Crack Wall
Specimen
Number
Tube
Condition
42
X-I
X-2
X-3
1800
1800
1900
-200
-200
-200
5 min
15 sec
5 min
X-4
1800
-200
5 min
Smooth
Smooth
Smooth
20-mil dent,
20-mil dent'
Smooth
Small sharp
dentc
Medii deep
dentc
Large shallow
dentc
20-mil dent
20-mil denta
20-yil denta
10-mil dent
5-mil dent
· X-5
1600
1600
-200
-200
5 min
5 min
Smooth
web
15
20-mil dentº
Tube crown heating time from -200°F to temperature maximum was approximately
5 sec/cycle.
"Depth of controlled dent made in top of tube crown with a 1/4-in.-diam
indenter.
Description of dents in as-received specimen.
ONO failure.
TEST RESULTS
The results of all the tests are detailed in Table d. For undented tubes,
local creep buckling in the form of ripples occurred in thr tube crowns in approx-
imately 15 to 30 cycles, depending on hold time. The ripples stabilized in
location but increased in amplitude as the cycling progressed. Failure was
always associated with a ripple and nornally occurred in a valley rather than on
a peak. Failure was defined as a crack through the tube wall capable of passing
LN2.
In the case of dented tubes, the dent locations can be considered to be "pre-
rippled" sites. During cycling, the dents deepened while the adjacent smooth
tubing develops ripp.' es. The deepening dents invariably were responsible for
premature failure. The presence of these deleterious dents appears to reduce
tube life, as compared with that for a smooth tube, by a factor of 1/3 to 1/2.
However, there does seem to be a threshold value of dent severity below which
damage does not occur. For example, the 0.005-in.-deep controlled. dent and the
So-called "large shallow" as-received dent had no effect on tube life.
The appearence of the specimen pictured in Fig. 5 at the 22-cycle point in
its test is shown in Fig. 6. Failure has occurred at both of the 0.020-in.-deep
controlled dents and the ripples are well established in the originally smooth
tube surfaces. Figure 7 shows the specimen at the test termination point of 1:.
42 cycles. Both the cracked dents and the ripples have continued to deepen. . The
the top, which was originally smooth, has also failed at a ripple
location which suddenly became unstable and grew considerably deeper than its
neighbors. This general behavior was typical of all the specimens tested.
ANALYSIS OF RESULTS
An erfort was made to relate the data obtained in these tests to existing
fatigue data. A survey of the literature revealed only a small amount of such
datat for Hastelloy X, and it was fo!' isothermal strain fatigue. It was not
possible to obtain a good correlation between this and our thermal fatigue data.
However, appropriate information, obtained by thermal cycle testing of tubular
specimens, was available for a similar alloy, Hastelloy N. This was used for
correlations with our test results.
IM. B. Reynolds, Strain-Cycle Phenomena in Thin-Wall Tubing, GEAP-4462,
(Jan. 30, 1964).
"A. E. Carden, "Thermal Fatigue of a Nickel-Base Alloy," J. Basic Eng.
Trans. ASME, Ser. Ó 87, 237 244 (1965).
The Hastelloy N thermal fatigue data is plotted as AT (thermal cycle test
temperature range) versus Ne (number of cycles to failure) in Fig. 8. The curve
is for a constraint factor - (F) of unity. This constraint factor is defined as
the ratio of actual mechanical deformation in the specimen to the total available
thermal strain (CAT) calculated from the thermal cycle. Figure 9 shows a family
of thermal fatigue curves for Hastelloy N, plotted for three different constraint
factors. The curve for F = 1 is a repetition of the curve shown in Fig. 8, while
the curves for F = 1.2 and F = 1.5 are derived by calculation from the F = 1
curve.
An analysis of the geometry of the specimens used in these tests indicated
that the tube crowns were under very effective constraint. In iact, it appeared
reasonable to consider the constraint factor for a smooth-tubed specimen to be
on the order of unity. Variables such as longer hold times and dented tubes were
expected to increase the constraint factor.
The data generated by these tests have been superimposed on the family of
curves of Fig. 9, This is shown in Fig. 10. The data for the 15-sec hold time
and the F = l curve correlate quite well. The data for the longer hold time of
5 min indicate that the effective constraint factor has been increased to the
order of F = 1.2. It is felt that this is due primarily to stress relaxation of
the tube crowns which can occur under the conditions of time and temperature
studied. The results for the dented tubes show constraint factors in the vicinity
of F = 1.5 due to the strain concentrating effects of the dents.
It appears from this correlation that the techniques used for the determina
tion of the Hastelloy N data are appropriate to the thermal fatigue problems in
regeneratively cooled rocket nozzles. Therefore, work is presently under way to
obtain similar data for Hastelloy X. This data will be associated with the points
plotted in Fig. 20. It should then be possible to extrapolate accurately the data
obtained in the tests reported here to a wide range of nozzle operating conditions.
Additional tests will be performed to investigate the effects of internal pressure
on the fatigue life of the tubes.
CONCLUSIONS
1. Thermal fatigue can be a serious problem for the coolant tubes of
regeneratively cooled rocket nozzles.
2. A laboratory test has been devised which provides thermal conditions.
similar to those seen in service. This test has been capable of producing tube
failure in substantially less than 100 thermal cycles. Thus, the principles of
similitude and modeling appear to be appropriate for the examination or determi.
nation of low-cycle fatigue characteristics of large complex structures.
.
..:-.. ...
.........
3. Increasing high-temperature hold times, up to 5 min, and dents in
the tubes appreciably curtail the fatigue life.
ACKNOWLEDGMENTS
The authors are indebted to Bill Williams, who conducted the tests, and
to C. K. Thomas for assistance in preparing the test facility.
Tubaato Tube Boud.
Tube to Fricket Bond...
X Cryogenic Cooleant
lank Tube

C
ܬܕܝellܘܗܙܙܬܝA-
Reinhor
ng Fucket
CSC
..Sehe.
matic Dinnyeol
Cros Sectio n Typical R
Cooled Nozzle..

ORNI - AC-
གབ
འ
:
8
•
•
Baise Plake.
Inlet Port
ནོ
GE.，： ;
་་
tiབ་ལ་དཔལས༌ :::༤༠ བ་པ་
:
ཙE
་
**་པ;
བབ་
::
1
•
•
ན་ ，
བ་་་
ཟ་*
ei.མ
ORNU-AÉC = OFFICE
,91
oང -4-
Se།རྒྱi༤) དགགsfrཙག །
Cycie Specimen
+
.. ་.…..-ཁ་བ ཁ་ཐཱ་ ད ，，་………
..•
་ ་ - ཁབ ཕ འཆ ...
ཐ་
•

.ORNL - AEC - OFFICIAL

.
..
'
AA.
1:
.
.
.
.
.
1
-.
1
. LAIMAITRE
--".
X FUGL MAIN
*
.
n
.
..
4
-
-
-
..
AV
.
Fig. 3
Overall view of Thermal Cycle
Test Setup
S
'T
Pie
.
A
si
TO
.
.
..
-
Vela
*
**
RNL - AEC - OFFICIAL
**
*

Witavano
WIJIJIO- JIVOTNIO.
*Szi
Jagodi
#
IL OOH Nik ş
WE
LIVIND
LOSUD
UTIOIIT
.
L'
OLIO
D
To!
522201
21
ULEM
KEUZZLE
IL TILLITI TILTOTT
LIVVS
IMMT
LIIIIII
OUVINI
TOWER
IZDEDZETUL
SEIZOST
IN2Z9Si
Il
ORNL - AEC - OFFICIAL
*********
A
.

. Fig. 5 Pretest
locationg and
inade with
Specinen. Nohe thermocouple.
two 0.020" Seero coi trolled sents
Yaoin diancher indeniter.

ew
-
..
comme i mornarices
o
...
N
in the
Kawasisi
XSARA
PIV PRETO
......
Fig. 6. Appearance of specimen after 22 cycles.
Nohe failures at both deat locations and
wrinkling in the remainder of the tobing.
I

ORNL - AC - OFFICIAL
-
:
Fig. 7 Appearance of specimen efter 42 cycles.
8 Nota failure of an originally smooth tube
Ž at a wrinkle locution..

:
ti
-7..
.4.87**--1672 min 5**
CHERAECE OF Ich
*
:
:
.
-..
..-*
......
....
.......
..
ORNI – AEC - OFFICIAL
ORN
::
ORNL-DWG 66-10907
2800
HASTELLOY N, TRANS.
ASME[D], 237-244 (1965)
AT, TEMPERATURE DIFFERENCE (°F)
p.comwe.com motoros
L HASTELMOLA
BEAR 41621, GELEZI
ELECTRI
edhe Deleted
. 100
16, CYCLES TO FAILURE
1000
2000
Fig. 8
..
Thermal Cycle Fatigue Curve for Hastelloy N.
Forca Construint Factor of Unity.
ii-IRTO
1:odiu.

O ?NL-ABC - OFFICIAL
.
.
ORHI - AEC - OFFICIAL
ORNL - AEC - OFFICIAL
ORNL-DWG 66-10905
se

T, TEMPERATURE DIFFERENCE (°F)
Fol.
O
100
FAILUR!
Folia
Falis
1000
NG, CYCLES TO FAILURE
Thermal Cycle Futique corres for
Hastelloy Ni For Constraint Factors (F)
of 1.0, 1.2, 1.5. .
Fig. a
1.
T
ORNL - AEC - OFFICIAL
ORNL - AEC - OFFICIAL
"11;*..
.
:.':.
.:.
ORNL - AEC - OFFICIAL
•
ORNL-DWG 66-40906

O SMOOTH TUBE WITH
15 sec HOLD TIME
O SMOOTH TUBE WITH
5 min HOLD TIME
A DENTED TUBE WITH
5 min HOLD TIME
AT, TEMPERATURE DIFFERENCE (°F)
BASED ON HASTELLOY N
DATA EORUM
CURVEY
17825
"
..
10
100
1000 2000
------- Net, CYCLES TO FAILURE .
Fig.ro A correlation of Hastelloy & thinaia
wall tubing thermal failique data points
with Hastelloys thermal cycle futique ..
curves for various constraint factors...
.
-
.
.
.
DRNI - AC - OFFICIAL
.
.-
..!!,
""
;';.
.
.
1
"
."
.
.
.
3. -
T
.
TI
.
.
.
'
.
+
END
DATE FILMED
12/28/166





.

.
-
-
-
,
..
.
.
.
.
..
.
.
"
i
..
.
A
.N
.
.
2**E
M:
VON
19
.
"
V
... SELIT TILL