Cryogenic Model Materials

16 Pages·2001·1 MB·English

by  Kimmel, W. M.

III
AIAA-2001-0757
Cryogenic Model Materials (Invited)
W. M. Kimmel, N. S. Kuhn and R. F. Berry
NASA Langley Research Center, Hampton, Virginia
J. A. Newman
U.S. Army Research Laboratory
Vehicle Technology Directorate
Langley Research Center, Hampton, Virginia
39th AIAA Aerospace Sciences Meeting & Exhibit
8-11 January 2001
Reno, Nevada
For permission to copy or republish, contact the American Institute of Aeronautics and Astronautics
1801 Alexander Bell Drive, Suite 500, Reston, VA 20191-4344
z
!
i
!
|
!
|
i °
|
AIAA-2001-0757
Cryogenic Model Materials
W. M. Kimmel', J. A. Newman", N. S. Kuhnt, and R. F. Berryt
NASA Langley Research Center
Hampton, Virginia
ABSTRACT require new materials and techniques to provide
An overview and status of current activities increased model performance while eliminating
seeking alternatives to 200 grade 18Ni Steel CVM structural failure at the higher static and dynamic
alloy for cryogenic wind tunnel models is loads. Furthermore, models are tested at
presented. Specific improvements in material cryogenic (-275°F) temperature provide high
selection have been researched including Reynolds number flows. Therefore, model alloys
availability, strength, fracture toughness and must have desirable a number of other desirable
potential for use in transonic wind tunnel testing. characteristics at cryogenic conditions.
Potential benefits from utilizing damage tolerant
life-prediction methods, recently developed fatigue Under high static loads two failure modes exist;
crack growth codes and upgraded NDE methods yielding and fracture. Model alloys need high
are also investigated. Two candidate alloys are tensile strength and fracture toughness to prevent
identified and accepted for cryogenic/transonic yielding and fracture failures, respectively.
wind tunnel models and hardware. Prevention of structural failure under cyclic or
fatigue loading is more difficult. Initially, a
INTRODUCTION structure may be known to contain no cracks large
Early inthe development of the National Transonic enough for failure to occur under anticipated
Facility (NTF) at Langley Research Center loads. However, under fatigue loading small
(LaRC), much research was performed to identify cracks may grow to a critical size where fracture
and characterize materials suitable for building occurs. To prevent fatigue crack failures a
wind tunnel models and support hardware. These damage tolerant life-prediction method has been
activities have not continued or pursued since the developed. Here, fatigue life is assumed
early 1980's. A motivation for these activities has equivalent to fatigue crack growth (FCG) from an
resurfaced in recent years due to the increasing initial defect to a critical crack size. Life prediction
difficulty in procuring 200 grade maraging steel. is a function of final and initial crack sizes,
Due to current trends inthe industry, no depots or anticipated loads, and FCG behavior. Damage
other vendors maintain stock of this material. This tolerance is superior to existing fatigue prevention
has impacted the costs and development cycle schemes, in terms of safety and cost of
times for almost all NTF research hardware replacement parts, and will be evaluated to ensure
projects and forced a mass buy strategy that is safe operation of NTF models. Because reliable
limiting research opportunities. detection of small cracks is important, non-
destructive evaluation (NDE) techniques will also
Additionally, NTF models need to simulate aircraft be studied.
performance under more severe aerodynamic
conditions than in previous decades. Increases in NOMENCLATURE
dynamic pressure (Q) and unsteady loading a,c crack depth, length
AOA Angle of attack
*SeniorMechanicalEngineer,ModelSystemsBranch,NASA-
,, LangleyResearchCenter,Hampton,VA23681 CVN Charpy impact energy
Aerospace Engineer, U.S. Army Research Laboratory,
I_c plane-strain fracture toughness
VehicleTechnologyDirectorate,Hampton,VA23681
1Aerospace Engineering Co-op Student, NASA-Langley N load cycle count
ResearchCenter,UniversityofWisconsin,Madison
Q dynamic pressure
tSenior Inspector, Quality Applications Technology Branch,
NASA-LangleyResearchCenter,Hampton,VA23681 Re Reynolds number
Copyright © 2001theAmerican InstituteofAeronauticsand R stress ratio, _maxk_min
Astronautics,Inc. NocopyrightisassertedintheUnitedStates
Gu ultimate strength
under Title I7, U. S. Code. The U. S. Governmenthas a
royalty-freelicense to exercise allrights underthe copyright
_y yield stress
claimed hereinfor Govemmental Purposes. All other rights
reservedbycopyrightowner.
1
American Institute of Aeronautics and Astronautics
AIAA-2001-0757
OBJECTIVES NTF materials search (Phase I)
The activities presented in this paper seek to The materials search began with an extensive
achieve the following objectives: literature review and discussions with material
manufacturers. A comprehensive search for
• Solve NTF material supply issues by alternate cryogenic model alloys was done by
finding a suitable alternate alloy. Tobler in 1980.2 The primary criteria acceptable
• Find model alloys with higher fracture alternative alloys must satisfy are listed as follows.
toughness, ultimate strength and FCG
characteristics to safely permit testing at 1. Material properties - possess high
higher static and dynamic loads. fracture toughness and ultimate tensile
• Predict and manage fatigue lives of strength at low (-275°F) temperature.
models using damage tolerance methods 2. Metallurgical stability - permit a stable
under anticipated wind tunnel loading. heat treatment for cryogenic-to-room-
• Revise LaRC wind tunnel model design temperature thermal cycling.
criteria, LAPG1710,151, to implement 3. Availability - commercially available in
damage tolerance methods and reduce stock shapes with short delivery time in
model development cycle times. bars up to 18" diameter and plates up to 6"
x 38" x 72".
• Evaluate NDE techniques for detection or
4. NDE - can be evaluated by metallurgical
monitoring small fatigue cracks in potential
and ultrasonic inspection to satisfy LaRC
model alloys.
guidance documents.
PROJECT APPROACH 5. Machinability - desirable machining,
handling and fabricating characteristics.
This project was divided into two phases. Phase I
sought potential alternatives to grade 200
Results of Tobler 2, Wigely 3 and Rush4 indicate
maraging steel from a list of commercially
nickel maraging steels are best suited to meet the
available alloys. A few alloys were selected based
on availability, ease of machining, and material high K4cand high cy requirements of NTF model
behavior (Le. tensile strength, stiffness, and alloys. Figure 1 is a plot of K4cversus cy for several
fracture toughness) at cryogenic and room materials. Most materials shown have properties
temperatures. The list of candidate alloys was within the shaded region. The most desirable
further narrowed to seven based on available properties (high K4c and high (_y) lie along the
funds (i.e. cost). Finally, the list of candidate upper bound of this shaded region. (As seen in
alloys was reduced to two (Vascomax C250 and the lower right corner of the figure, Vascomax
Aermet 100) based on results of further C250 and Aermet 100 are slightly above the
mechanical testing. shaded region.) However, maraging steels a_re
primarily used for aerospace and nuclear
The objectives of the second phase was to select applications and are somewhat difficult to obtain
one of the two candidate alloys based on fatigue due to limited demand. The phase I study
crack growth (FCG) and fracture characteristics. narrowed the list of candidate alloys to the
FCG and fracture behavior are important because following seven:
fatigue life predictions are to be made based on
damage tolerance methods. Phase It consisted of 1, G90cO - (British Steel)
the following activities: 2. 13--8S-u_pertough©- (AIIvac)
3. C250 Maraging - (AIIvac, Carpenter)
1. Determine FCG and fracture behavior of 4. T200©- (Allvac) i
Vascomax C2_50 and Aermetl00 at both 5. Custom465©- (Carpenter) I
. E
cryogenic (-275°F) and room temperature 6. Aermetl000- (Carpenter)
(75°F). 7. PH13-8Mo - (various manufacturers)
2. Use damage tolerance life prediction
methods for maraging steel models. Each of the seven alloys were ordered and
3. Revise LaRC guidance documentation delivered in an acceptable time. Specimens were
LAPG1710.15 to qualify new alloys and machined and sent to a certified lab for Charpy
life prediction methods. 1 impact tests (ASTM E-96), tensile tests (ASTM E-
4. Evaluate the role of NDE in damage 23) and thermal expansion tests (ASTM E-370).
tolerance life management. Charpy impact energy and ultimate tensile
2
American Institute of Aeronautics and Astronautics
AIAA-2001-0757
strength for each material are listed in Table 1. The extraordinarily high Charpy impact energy of
The results of this phase I study were presented at 13-8 Supertough© initially made it an attractive
an NTF workshop in the fall of 1999. As indicated candidate. This alloy is similar to Ph13-8Mo,
by highlighted portions of Table 1, Aermetl00© which was previously found to be acceptable for
and C250 were chosen for additional study in cryogenic testing although care must be taken to
Phase 11. This decision was based on ultimate overcome cryogenic instability related to formation
tensile stress, availability, and fracture toughness. of unstable austenite during heat treatment s. This
material is available in limited production sizes and
350
-275°F has a modest ultimate strength. Thus, 13-8
Supertough© was not selected for further
_" 300 304N consideration.
(/)
_ 25o
NTF materials study (Phase II)
Phase II began in spring 2000 and is scheduled to
_" 200 be complete by February 2001. Inthis phase, FCG
and fracture behavior of both alloys (Vascomax
t-
t- C250 and Aermetl00_) will be determined to
150 3Ni-5Mn
:3 Y,/2X enable fatigue-life prediction and management by
• 100 damage tolerance techniques. Currently, most of
:3 ////////'_"///////_" c250 phase II is completed. Finally, non-destructive
_6
//T i-5AI-2.5 Sn_z"'/J////////_ 2 evaluation (NDE) techniques will be tested on
_ 5o
specimens to determine the smallest detectable
crack size. The ability to detect or monitor
HP-9-4-20 -_
I I I I I propagating fatigue cracks is an important step
5O 100 150 200 250 300 towards applying damage tolerance methods.
yield stress, <_y(ksi) ..Experimental testing and r_,sults
Figure 1 - Plot of K_cversus _yfor several alloys. Existing cracks are known to grow under cyclic, or
fatigue loading. In fact, a significant portion of the
In previous literature, C250 and AF1410 were fatigue lives of most engineering structures is
identified as potential alloys that would be usable spent propagating small cracks to failure.6
for cryogenic models. Microstructural stability tests Damage tolerance life prediction methods are
were favorable and included in the findings of based on FCG behavior and are superior to older
Tobler 2 and Wigely 3. However, these two alloys "safe-life" life prediction methods, in terms of
were not initially accepted for NTF models. safety and part replacement costs. 7Using damage
Instead, Grade 200 maraging steel was selected tolerance to prevent fatigue failure requires (1)
because of higher plane-strain fracture toughness knowledge of how fatigue cracks propagate to
(K_c) at cryogenic temperature. Subsequent to failure and (2) the ability to monitor propagating
phase 1, it was discovered that Aermetl00© is fatigue cracks. FCG behavior of Vascomax C250
essentially a reformulation of the AF1410 alloy and Aermet 100_ will be discussed inthis section.
produced by Carpenter Steel corporation. NDE tests to detect and monitor fatigue cracks are
discussed later.
Table 1- Material properties of Phase Ialloys.
FCG data obtained from laboratory specimens is
Material Cvn (ft-lb) Ou (ksi) related to crack behavior in other components
using the cyclic range of stress intensity factor,
AK, as a crack-tip driving force. Although z_Kis a
13-8 27.8 203.2
function of cyclic stress and crack length, cracks
_-Aerm6t 1001111 "f6..;.9..............3..0..0....3.......... subjected to the same AK have nearly identical
Custom 465 5.2 268.1 fatigue crack growth rates (average increment of
T200 24.3 276.6 crack growth per c_,cle). FCG tests were
performed on both candidate alloys following
ASTM standards (E647). A computer-controlled
13-8 Supe_ough 117.0 215.0 system 8 was used to continuously monitor crack
G90c 21.1 279.7 length throughout testing using compliance data
from a clip gage. 9 This system automatically
3
American Institute of Aeronautics and Astronautics
AIAA-2001-0757
adjusts loads as the crack grows to ensure that associated with artificially high crack closure
programmed stress intensity factors are applied levels. _°-11Constant Kmaxtest data at low AK is
throughout the tests. This allows FCG tests to be free of closure, and is used to characterize the
performed with controlled AK. As outlined in growth of small cracks in model alloys at low _K. _
ASTM E647, FCG tests were performed where z_K
was gradually changed as the crack propagated. Both materials were obtained in forged round bars
In this way, the relationship between fatigue crack of approximately 6" diameter. Each material was
growth rate and ,_K can be determined more commercially available in this shape and delivered
efficiently. Two types of FCG tests were within 2 weeks of order. Disk-shaped compact
performed in this study. Traditionally, FCG tests tension specimens were made from 1" thick slices
are performed where the load ratio (ratio of of the round bars (see ASTM standard E399).
minimum to maximum load, R) remains constant
as ,_K is changed. Constant Kma,tests were also FCG data is presented in Figure 2 for Vascomax
C250 and Aermet 100 as FCG rate versus ,_K.
used for decreasing AK tests. For these tests, z_K
is reduced by increasing the minimum value of (Note that log-log axes are used, which is typical
stress intensity. Constant Kmax tests are desirable of FCG data.) Curves are fit through the data for
because FCG threshold (that is, values of ,_K each material at both cryogenic (-275°F) and room
where fatigue crack growth rates approach zero) temperature (75°F). Each -specimen was
precracked at R = 0.1 and AK = 15 or 20 ksiVin
achieved during constant-R, reducing AK tests are
10-3
10-3
Vascomax C250 (75°F)
104 104 Aermet100(75°F) _
/
_>, 10-5 _>. 10-s
r"
U
::. 10_
10._ =
10-7 Kma, = 20 ksiVin
LL IJ.
10-e n& _ Km._= 50 ksH'in 10 8
• -- daldN = 10"e(AK_,)22 eft)2 2
10-9 10-9 , , , I
2 4 6 810 20 40 6080 1 2 4 6 810 20 40 6080
z_K(ksi_lin) ±K (ksi_/in)
10.3 10 3
Vascoma
104 104 Aermet 100 (-275°F) /
/
G)
-G
._104 10-5
t- ¢-
_) (3
:_¢,-. 10• t- 10"s
(.3 107 (.9 10-7
L) O
LL LL
10-8 J '4: rlI &" K_..--20ksi_'i.n... I 10-8
f daldN = 10-_(AKe,) 29 / I -- da/dN = 10-_(DK.,) 29
10.9 . ...... I 10 9 L i , , = , , , , I
2 4 6 8i0 20 40 6080 2 4 6 810 20 40 60 80
AK (ksiqin) AK (ksiqin)
Figure 2 - Fatigue crack growth data for Vascomax C250 and Aermet 100 at 75°F and -275°F.
4
American Institute of Aeronautics and Astronautics
AIAA-2001-0757
before FCG testing. Data below and above the were expected to be faster than at 75°F.
precracking loads were achieved by reducing and However, the opposite was shown to occur.
increasing AK during testing, respectively. Comparison of FCG data in Figure 2 indicates that
FCG rates were faster at room temperature by
FCG data for both Vascomax C250 and Aermet approximately a factor of 2. This observation is
100 are plotted as FCG rate versus ,_K on the left congruent results on C200 steel by Wagner. 13
and right sides of Figure 2, respectively. (Note the 140 ..&
use of logarithmic axes.) Data generated at 75°F all=•""
and -275°F are shown separately. Both constant
R and constant K_ax test data are shown as
symbols (refer to legend). An "upper bound" line,
_ loo
shown as a solid line, is fit through the data. This
linear relationship (shown in the legends) will later
be used for damage tolerance life predictions.
The line is a good fit to the data at intermediate
values of AK (5 ksiqin < AK < 20 ksi-Vin). In this tco."-)®(n86_0o
range of AK, called Paris regime FCG, the o
relationship between FCG rate and ,_K is nearly 40 c250 c200
linear. At AK < 5 ksiqin, the FCG rate is much less
than an extrapolation of the linear behavior would
_ 20 0 A I standardcompact
predict. Here, FCG rates decrease towards an '4- • • roundcompact _)
apparent threshold, below which no significant
FCG occurs. However, neglecting the fatigue life 0 ' • ' I ' ' ' I ' • ' I ' ' ' | ' ' '
-400 -300 -200 -100 0 100
advantages of FCG threshold will lead to
temperature (°F)
conservative life predictions and greatly simplify
life analyses. As AK increases, FCG rates Figure 3 - Fracture toughness vs temperature.
increase rapidly as Krnaxapproaches the fracture
toughness and fracture is immanent. The fracture Based on known FCG and fracture data, worst-
case FCG conditions occur at 75°F and worst-
criterion of damage tolerance analyses will
account for this deviation in FCG behavior. case fracture performance occurs at -275°F.
Fortunately, the current LaRC practice for crack
It should be noted that both materials have nearly growth life analysis of NTF models uses cryogenic
the same FCG response under similar conditions. fracture toughness properties and room
That is the linear fit for both materials at 75°F and temperature FCG data, which is conservative. To
-275°F are identical. Also these data agree with simplify damage tolerance life prediction and
previous results of C200 steel presented by ensure conservative results, life predictions will be
Wagner _3. Therefore, no significant difference in made using cryogenic fracture data and room
FCG performance is expected between these temperature FCG data. Furthermore, the same
alloys. The only clear difference in these alloys is FCG relation can be used for all model alloys
the cyclic fracture toughness, Le. the _K and Kma× studied here.
at fracture. Although not a standard fracture
toughness test, the data in Figure 2 indicates that This should greatly simplify damage tolerance
Aermet 100 has an edge over C250. Fracture analyses because only a single FCG relation is
toughness tests are planned for both alloys, but needed. Currently, analyses require material
results are not currently available. Existing specific data normally obtained by a series of
fracture toughness data for C200 and C250 steels expensive and time-consuming tests. Typically,
are plotted in Figure 3 as a function of data is not available until late stages of model
temperature. 14 C200 is tougher at room development, even after the design stage.
temperature, but no appreciable advantage is Although material specific fracture data is still
seen at cryogenic temperatures. Based on FCG needed, these tests require much less test time
data in Figure 2, Aermet 100 is expected to than FCG tests. Development and use of a
outperform both C200 and C250. generalized FCG curve should greatly reduce the
complexity of performing damage tolerant life
Because fracture toughness generally decreases analyses on NTF wind tunnel models. It should be
with decreasing temperature, FCG rates at-275°F reasonable to expect that a fatigue crack-growth
5
American Institute of Aeronautics and Astronautics
AIAA-2001-0757
life analysis be performed by the design activity shape (S-shaped) and is normally divided into
early in the model development process. three regions (Stage I-threshold, Stage II - Paris,
and Stage III - crack instability). Details of FCG
DAMAGE TOLERANCE ANALYSES behavior can be found in the literature. 6'16'18'19
An increasing number of structures are designed Several commercially available FCG software
and safely managed using damage tolerance programs have been developed that use FCG
methods. Here, initial defects or cracks are relations to make damage tolerance fatigue crack
assumed to exist even before fatigue loads are growth predictions.
applied. Fatigue life is assumed equivalent to
FCG between initial and critical crack sizes, where Fatigue crack growth codes
fracture occurs. Fatigue lives are predicted using Five FCG codes were used to analyze two crack
FCG and fracture mechanics parameters (such as problems in NTF hardware under two different
AK and K_c),which are related to anticipated loads. load spectra. The FCG codes used included
AFGROW 4.0,20 FASTRAN, 2_ Fracture, 22
Fatigue life predictions are made by integrating
FCG relations (FCG rate, or da/dN, as a function FractureGraphic 2.1,23 and NASGRO 3.0. _7
of AK) to estimate the number of cycles until a Predictions made by these codes for identical
crack configurations and loads are compared.
crack grows to fracture. Several inspection
intervals are normally scheduled within the
AFGROW and NASGRO are available to the
expected fatigue life to ensure an inspection
process has multiple chances to detect growing public and can be obtained over the internet at no
fatigue cracks before failure. Examples of FCG cost. FASTRAN (by Jim Newman) and Fracture
relations are given in Equations 1-3. The linear (by Mike Hudson- and Charles Dunton) are codes
relation associated with Paris15 is shown in developed at LaRC. Fracture has previously been
Equation 1. The Paris relation has the advantage used for life predictions of NTF models.
and disadvantages of simplicity. It greatly FractureGraphic (FG) is a commercially licensed
software code. All five codes calculate FCG using
simplifies fatigue life calculations, but does not
consider non-linear effects of FCG threshold or a variety of crack configurations, each with a
fracture toughness. If the anticipated loads and specific stress intensity factor solution (K solution).
K solutions define the crack-tip stress intensity
crack lengths do not produce a _K near the FCG
threshold, the Paris relation may be adequate. factor in terms of crack configuration and loading.
It is noteworthy that NASGRO, AFGROW, and FG
More complicated relationships have been
all share a common material property database,
developed by others including Forman TM(Equation
orig(nally compiled for the NASA/FLAGRO
2) and Newman, as part of a FCG code
computer program in 1986. 24 This database
NASGRO, 17(Equation 3). The Forman equation
accounts for non-linear effects associated With
fracture and the NASGRO equation accounts for fracture
both threshold and fracture effects.
,-._ Paris /
d---a-=a C(AK)m (1)
dN
(.9
da A(AK)" (2)
0"" _ I1
dN (1- R)K,_- AK
threshold i
(1 AKth) p o
da C AK -- _ (3)
1-
log (AK)
The NASGROW equation (Equation 3) provides a
Figure 4 - Typical fatigue crack growth behavior
good approximation of the entire FCG relation
from threshold to fracture. As schematically
shown in Figure 4, the FCG curve has a sigmoidal
6
American Institute of Aeronautics and Astronautics
AIAA-2001-0757
includes properties for many materials, including tunnel models and used to perform damage
maraging steels used for NTF wind tunnel models. tolerance life predictions. When a cruise model is
FASTRAN incorporates nonlinear or plastic strain tested in the NTF, it typically begins each polar at
material behavior and load interaction effects a low AOA and steps up to higher AOA. These
(acceleration or retardation) using a crack closure step increases in AOA increase the normal force
concept. These features are unique to FASTRAN acting on the model as seen in Figure 5. Figure 5
and allow more accurate life predictions. shows two time plots illustrating the progression of
normal force for a typical test polar. Frequently a
Each of these five FCG codes could be used in test polar ends with violent oscillatory loads
the design process of NTF models. However, associated with an unsteady separated flow
each code uses different means (e.g. different condition, called pitch buffet, normally associated
FCG relations and K solutions) to make fatigue life with large AOA. It is estimated that 30% of the
predictions so each code is expected to predict a 251 test polars ended with pitch buffet
slightly different result. It is the responsibility of conditions. 25
the designer to ensure that the appropriate Trtn$ _ottmodel tlst, M=OS2, ¢Ip2BOOpsi'
5000
material relationships are used for a given
application. For example, NTF models and 400O
hardware are typically subjected to fatigue loads at
low AK and high Kma,. Therefore, it is more 30OO
important to use a FCG relationship with a precise
2OOO
description of threshold than one with a good
description of Stage III crack growth. 1008
0
Load spectra
Traditionally, damage tolerance analyses of NTF -10OE
models investigated two loading scenarios. The
first is a peak load condition that is analogous to -20OC 10o 200 300 408 500 608
Time,s
the maximum load during an AOA sweep in the
wind tunnel at constant dynamic pressure (Q). 50O0 Trmspoqt model tilt, M,O.R, q-|SlO psf
This AOA sweep is referred to as a "polar." The
I
second loading scenario deals with oscillatory 4000 ---
loads (i.e. fatigue loading) during testing. FCG
30O0
data at load ratios of 0 and 0.6 (25% above and
below the mean stress) are used to model the 2000 --
Nf
effects of peak and oscillatory loading, J
10O0....
respectively. Inthis way, effects of an entire load
spectrum is predicted using two constant 0--~--
amplitude (CA) FCG curves. Because load
-100£
interaction effects are neglected, this is an over m
simplification of reality. Better life predictions can
-20OC
50 10O 150 200 25(I 300 350 400 450 500
be made if variable amplitude (VA) loading is
Tlme,S
considered. It is assumed that VA loading implies
Figure 5 - Normal force data vs time. 25
a sequence of varying amplitudes, where the order
of various stress amplitudes is defined. Analyses
A stair stepped VA load spectrum was created to
completed show that CA loading can be
reflect the wind tunnel data for cruise model polar
unconservative, especially for cruise models_ This
testing. For simplicity, the normal force was
is due to large amplitude loads placed upon the
assumed to be the source of loading. It was
model system that are not modeled in the CA
assumed that pitch, yaw, and roll moments, and
approach. VA loading accounts for the loading
axial forces did not contribute to the fatigue
history, and may provide more accurate crack
process. The resulting load sequence of the
growth prediction. This would help achieve the
transport model test then resembles the shape of
project objective of identifying a more accurate
the two plots of Figure 5 with 30% of the polars
and streamlined analysis method.
experiencing buffet loading. The varying
amplitudes of loads can be observed as well as
A load history from a specific NTF transport cruise
the oscillatory nature of the load sequence and
performance model was considered typical of wind
7
American Institute of Aeronautics and Astronautics
AIAA-2001-0757
step increases in load due to increasing AOA. CA life predictions. Results produced by each of the
loading does not account for the pitch buffet loads. FCG codes are plotted as crack length versus
Pitch buffet is important to consider because large cycle count in Figure 7. Results from four of the
amplitude oscillations may overload the model and codes are within 10% of each other. Because
support hardware. Additionally, pitch buffet plane-strain-constraint and crack closure are
conditions produce significantly higher AK loads considered, FASTRAN predicts a longer fatigue
that significantly reduce the model fatigue life. life. Neglecting these effects would produce
Only by considering VA loading can the true load results within the 10% range of the other four
codes.
history, and its effects on fatigue life be described.
The step increases in load were calculated from Simulation and results of NTF model problems
data received from NTF tunnel operations. 25VA After the FCG codes were compared solving a
load sequences are assumed proportional to simple problem, crack growth life predictions were
normal force data taken from two test polars (M conducted on two different parts of a typical NTF
=0.89, q=1910psf and M =0.92, q=2800psf). model configuration. One was the outboard screw
These loads were used to predict stress at a wingtip attachment of a high aspect ratio wing
specific location on the wing. Data was acquired transport model and the second was a swept strut
at a high sampling rate, 320 Hz., creating an sting design with a stress concentration at male
extremely large data set for each force taper joint fillet. The crack configuration of the
component. This assumption is negligible in the wingtip attachment was modeled as a fiat plate
case of the swept sting, where the normal force is with a single corner crack at a hole. The crack
completely dominant. The pitch moment is a minor configuration of the swept sting was modeled as a
load and again, dynamic loading is conservatively surface flaw in a solid rod. Crack configurations
included in the balance data. Buffet onset (large for the wingtip attachment and the swept sting are
amplitude oscillations) was also assumed to occur shown on the left and right sides of Figure 8,
in 33% of the total polars. A plot of pitch buffet respectively. In reality, these structures are made
loads during a high Q test polar is shown in Figure from 200 grade maraging steel with a fracture
6. toughness of approximate!y K_c= 68 ksi_/in. Life
predictions for both cases were made using each
6000 of the five FCG codes under both CA and VA load
spectra previously discussed, and are listed in
5000
Table 2. For consistency, results (cycles) are
4000 reported in numbers of potars. That is, one cycle
is equivalent to one polar (sequence of increasing
Ny 3000
AOA).
2000
IOO0 01E
014 --
0
13000 13250 13500 13750 14000 012 --
Tirrm Isec!
0,1
Figure 6 - Buffet region of high Q test polar. 2s
-- _Fraclure
_ 008
Comparison of FCG codes 006 .. Graphic
L,t O04',
Next each FCG code was used for life predictions
on identical simple FCG scenarios. A surface flaw 002 10000 20000 30000
in the bottom of a trench (notch root) in a flat plate Cycles
that was loaded by CA bending. Component
Figure 7 - Calibration plots of FCG codes.
dimensions and loads were obtained from an
actual test specimen. Later, each FCG code will
be compared with experimental results. Further,
NDE techniques for in-service inspection were
studied with these specimens. Experimental data
(one specimen cycled to failure) provided a
comparison of crack growth rates and specific
crack lengths to evaluate the performance of FCG
8
American Institute of Aeronautics and Astronautics