Kutz M. Handbook of materials selection

Подождите немного. Документ загружается.

3 CREEP AND STRESS RELAXATION TESTING 527

time

strain or elongation

low stress

(or low temp)

high stress

(or high temp)

stage I

stage II

stage III

Fig. 6 Schematic of the different regions of a classic creep curve.

Fig. 7 Loading frame to apply static (or extremely slow cyclic) load conditions.

(Courtesy of the Instron Corporation)

528 TESTING OF METALLIC MATERIALS

extensometry as used in tensile tests can be used on the specimen. However,

higher temperatures are usually involved for most metals, which necessitates

using LVDT, capacitive, or other types of displacement measurement transduc-

ers. The most common transducer is the LVDT coupled with linkages that reach

up through the furnace to attach to the specimen. Special care must subsequently

be taken to ensure that the displacements measured actually correspond to gauge

length deformation and are not biased by linkage or transducer effects.

Stress relaxation testing (ASTM E328) is similar in many ways to creep

testing except that the applied boundary condition is a constant displacement (as

opposed to a constant load). This type of test is more suitable for metallic

materials that have signiﬁcant constraint and hence exhibit little strain increase

as a function of time. One example of this is a high-temperature bolted joint

where a displacement boundary condition is ﬁrst applied and then the stress

relaxes with time. Testing times include both long-term (months or more) at the

service temperature and accelerated tests (days) at more aggressive thermal con-

ditions.

Creep and stress relaxation testing can also be performed on specimens with

a more complex stress state than simple tension. Examples include compression

testing (typically using a carefully machined cylindrical specimen similar to that

described earlier for quasi-static compression testing), bend testing (e.g., with a

cantilever beam specimen), or even torsion testing. Testing has also been per-

formed on tubular specimens that are subject to complex, multiaxial stress states.

4 HARDNESS AND IMPACT TESTING

This section addresses two mechanical test methods that are used to indirectly

determine the tensile and toughness properties of a metallic material. As with

any indirect test method, there are well-known limitations to these techniques,

but they are clearly offset by the reduced cost and increased efﬁciency of the

indirect methods. The test methods, as well as their limitations, will be explained

in more detail in this section.

Hardness refers to a material’s ability to resist deformation when loaded by

an indenter. The indenter can differ in geometry; for instance, it can be conical

(the Rockwell test), spherical (Brinell), or pyramidal (Vickers and Knoop). For

metallic materials, hardness is directly proportional to the yield stress of the

material. Nevertheless, conversions are readily available to relate hardness to

ultimate tensile strength (most commonly for steel although other conversions

for different metallic materials do exist). Hardness testing (ASTM E6, E10, E18,

E92, E1842, and others) is industrially attractive since the testing is rapid (

⬍20

s), low cost, nondestructive, and generally does not require a machined speci-

men. As a consequence of these issues, hardness is often used as an industrial

screening tool to ensure metallic material processing control. An ASTM standard

(E140) exists to convert between the different superﬁcial hardness and conven-

tional hardness scales.

The majority of hardness testers (e.g., those shown in Fig. 8) utilize dead-

weight, springs, or (infrequently) closed-loop computerized systems. Although

many testers provide output on an analog meter, bench-top units with digital

readouts (sometimes with an interface for a PC) are generally more popular

today. When the component being hardness tested is too large to practically

4 HARDNESS AND IMPACT TESTING 529

(a) (b)

Fig. 8 Hardness testers including (a) a benchtop and (b) portable unit.

(Courtesy of the Instron Corporation)

position onto the anvil of the tester, portable test units are available to locally

measure hardness. A typical portable hardness measurer is shown in Fig. 8b.

Regardless of the type of hardness tester used, care must be taken to ensure that

the system is in proper working order by using commercially available calibra-

tion blocks.

Although hardness testing is relatively simple, accurate measurements depend

greatly on careful attention to detail. The anvil should minimize the contact area

with the component. Surface ﬁnish can sometimes be of paramount concern; for

100–150 kgf, a ﬁnish ground surface is sufﬁcient whereas for a 15-kgf load, a

polished or lapped surface usually is required. Surfaces that are ridged due to

machining or that exhibit loose, ﬂaking scale can yield incorrect hardness results.

The test surface should be perpendicular to the indenter since even slight (

⬍5⬚)

angular deviation can cause hardness to shift by several points. Indenter size

should be chosen with the underlying grain structure of the metallic material in

mind. In general, the indenter should be larger than the grain size if a macro-

scopic measurement of hardness is desired. Finally, metallic material homo-

geneity is also important since the indenter effectively samples a region

approximately 10 times deeper than the depth of the indenter penetration.

Instrumented indentation testing is a more recent method developed to derive

more extensive data from effectively a hardness test. During this type of testing,

the instantaneous load–displacement data is recorded continuously. The mea-

sured data can be related to such things as elastic modulus, yield strength, and

strain hardening exponent. Highly empirical punch test methods are also contin-

ually evolving to relate local deformation behavior to properties such as fracture

toughness.

530 TESTING OF METALLIC MATERIALS

Fig. 9 Example of a pendulum impact machine. (Courtesy of the Instron Corporation)

Since some percentage of loading is inevitably dynamic and components often

have to be designed to survive high loading rates typical during an accident,

dynamic fracture testing can ensure structural integrity under this type of load-

ing. During impact testing, a pendulum machine (example shown in Fig. 9) with

a striker is typically used to impart high energy to a notched sample. The notch

toughness of the specimen, typically a ﬂat bar with a carefully machined notch

of various geometries, is measured in terms of the absorbed impact energy that

causes fracture in the specimen. Some other parameters that are commonly ex-

amined include fracture appearance and the deformation noted on the surface of

the specimen. The independent variable controlled during testing is most often

temperature. A typical energy–temperature curve for A36 steel is shown in Fig.

10 contrasting the typical difference between slow bend tests and rapid impact

results.

The Charpy and Izod impact tests both utilize pendulum loading albeit with

specimens of slightly different geometry, especially as related to the specimen

notch dimensions. The primary difference between the two tests though is that

the Charpy V-notch (CVN test as deﬁned in ASTM E23) specimen is three-

point loaded whereas the Izod specimen is cantilever loaded.* Of the two meth-

ods, the CVN test remains the most popular method available, and manufacturing

*Both machines were developed in the beginning of the twentieth century although Charpy was a

Frenchman and Izod was an Englishman.

5 FRACTURE TOUGHNESS TESTING 531

A36 steel

Temperature, F

-400 -200 0 200 400

Absorbed Energy, ft-lb

slow bend

impact (CVN)

upper shelf

lower shelf

Fig. 10 Impact test results illustrating CVN energy versus temperature curves for A36 steel.

lot release requirements using CVN data are actually written into many metallic

material speciﬁcations. The CVN test is rapid and low cost; the most expensive

part is machining specimens although commercial machining templates to ease

fabrication are readily available.

Two other impact tests include the drop-weight (ASTM E208) and the drop-

weight tear test (DWTT as per E436). During drop-weight tests, a series of

experiments are performed where different specimens at a range of temperatures

are impacted with a guided, free-falling weight dropped from a ﬁxed height.

The key ﬁnding is the nil-ductility transition temperature (NDTT), which is the

maximum temperature where fracture was observed. The DWTT test is similar

to the drop-weight test although the specimen used is slightly larger and the test

ﬁxturing is very different. The fracture surfaces are evaluated and the percentage

of shear quantiﬁed. The temperature at which 50% shear occurs is loosely de-

ﬁned as the ductile-to-brittle transition temperature.

CVN testing can also be performed with a fatigue precracked specimen to

eliminate the notch acuity and depth restrictions inherent to the standard CVN

specimen geometry. Considerable effort has also been recently expended devel-

oping instrumented impact testers to record more data. Nevertheless, the primary

shortfall with most of the impact test methods is (a) the relatively small specimen

size (relative to component size) and (b) the high degree of empiricism used to

relate the results to metallic material properties. These disadvantages are offset

by the simplicity and low cost of the test, particularly for production environ-

ments where the test can be used effectively as a quality control tool.

5 FRACTURE TOUGHNESS TESTING

The foregoing discussion introduced the concept of notch toughness, deﬁned

loosely as the energy required to initiate a running fracture from a machined

notch. This is an important design parameter since nearly all structures contain

notchlike stress concentrations. However, in the strictest sense according to the

532 TESTING OF METALLIC MATERIALS

science of fracture mechanics (the study of cracks), fracture is broadly classiﬁed

as either brittle [governed by linear elastic fracture mechanics concepts (LEFM)]

or ductile [elastic plastic fracture mechanics (EPFM)]. Fracture toughness is the

ability of a material to withstand unstable crack propagation (e.g., fracture) as-

suming of course the presence of a crack. Fracture mechanics provides the the-

oretical underpinning that relates the following quantities:

●

Magnitude and range of the remote applied stress

●

Geometry of the component

●

Morphology (size and shape) of the crack

to what is loosely termed the ‘‘crack driving force,’’ or stress–intensity factor

K. Detailed stress analyses of cracks have shown that the magnitude of the local

crack-tip stress ﬁeld is proportional to this K parameter. Although K applies for

LEFM, in the case of EPFM there exists another similitude parameter that is

termed J.

Fracture toughness testing requires closed-loop, computer-controlled servo-

hydraulic test equipment and a suite of accurate displacement extensometers.

Virtually all tests are computer controlled due to the complexity of the test and

the requirements for recorded data. As a minimum, most fracture tests require

measuring the applied load and some displacement measurement across the

crack. Most tests are analyzed by custom software developed speciﬁcally for the

particular test. The most common fracture toughness test is the plane-strain frac-

ture toughness measurement, which yields the K

parameter as described in

detail in ASTM E399. For an ASTM valid K

value, the loading must be such

that there is no signiﬁcant plastic deformation (speciﬁcally, the plastic zone at

the tip of the crack is less than 2% of the thickness and total crack length). The

plane-strain requirement implies that the fracture toughness specimens tend to

be quite thick, most are commonly greater than 0.5 in.

Although E399 allows a variety of specimens, the most commonly used is

the compact-tension specimen, a pin-loaded, edge-notched geometry. The spec-

imen is ﬁrst precracked under fatigue loading to ensure that the crack tip is

sufﬁciently sharp for the fracture test. The specimen is then loaded to failure by

a constant displacement ramp applied between the pins. During the test, the load

and crack mouth opening displacement (CMOD) is measured. Two sample load –

displacement plots are shown in Fig. 11. In Fig. 11a, the behavior is more brittle

with catastrophic failure at peak load. Conversely, in Fig. 11b the trend is more

suggestive of a ductile and somewhat more stable material with slow crack

advance gradually relieving load as the test progresses past the point of maxi-

mum load.

Determining a K

toughness from the load–CMOD data requires ﬁtting the

initial linear portion of the data and plotting a 5% secant offset to that linear

region. The intersection between the measured load–CMOD data and this secant

line deﬁnes the critical load. This graphical construction is evident in Fig. 11a.

However there are stringent limits for qualifying the conditional K

toughness

as an ASTM-valid K

. Most importantly, the crack length and thickness are

required to be sufﬁciently long to satisfy the small-scale yielding criteria de-

5 FRACTURE TOUGHNESS TESTING 533

(a)

Load, kips

(b)

Temp =

1050°F

Material =

beryllium

K =

19 ksi 公in

CMOD displacement, 10

–3

inch

0204060

linear fit

of data

5% slope

offset

Material

beryllium

Temp =

800°F

CMOD displacement, 10

–3

inch

0 5 10 15 20

25 ksi 公in

Fig. 11 Two load–displacement plots from K

fracture toughness tests conducted

on beryllium at elevated temperature.

scribed earlier. Furthermore, nonlinearity is minimized by requiring that the crit-

ical load (corresponding to K

and denoted by P

) must be no less than 90% of

the maximum measured load. The two sets of data indicated in Fig. 11 are clearly

invalid since the conditional P

load is too low relative to the maximum mea-

sured load.

Whereas K

applies for thicker, more constrained sections where catastrophic

fracture occurs rapidly, a different K

fracture test (ASTM E561, also termed a

K–R curve) applies for thinner structure where slow stable crack extension pre-

cedes unstable fracture. The test specimen geometry typically used is a wide,

center cracked panel precracked at low loads to generate a sharp fatigue crack

suitable for fracture. The specimen is monotonically displaced similar to a K

specimen while recording load, CMOD, and often some parameter such as elec-

tric potential drop that can be used to instantaneously provide crack length mea-

surement.

The resistance curve, also called an R curve, is then generated by plotting

crack driving force K against crack extension, termed

⌬a. Crack extension is

deﬁned as the sum of the physical crack length change, usually derived from

either compliance (stiffness) or some other nonvisual crack length indicator, plus

the half plastic zone size. Since the stress state is typically something more like

plane stress, the plastic zones as observed on the surface of the specimen tend

to be quite large. An example R curve is shown in Fig. 12. The K

value is

deﬁned as the K magnitude at the tangent intersection between the R curve and

lines of constant load plotted in K versus crack length (or crack length extension)

space. This load represents the maximum level that can be achieved before the

crack extends (often unstably) and structural integrity is lost. A careful exami-

534 TESTING OF METALLIC MATERIALS

stainless steel

Width normalized effective crack length, 2a

0.35 0.40 0.45 0.50

Crack growth resistance, K

ksi

100

150

200

recorded data

61.2 kip

applied K curves

for different fixed

load values

= 215.2 ksi

55.0 kip

Fig. 12 Example R curve from a K

fracture toughness test performed on stainless steel.

nation of the data in Fig. 12 clearly illustrates that some crack growth was

apparent in this metallic material since catastrophic failure did not occur at the

point of instability.

Fracture toughness testing in the elastic-plastic regime is time consuming, not

trivial, and requires advanced instrumentation to yield excellent results.

After

conducting and analyzing the test, the goal is to generate an R curve, similar to

that described earlier for the K

testing, as shown in Fig. 13. However, instead

of K on the abscissa, the J parameter is included. The J parameter is calculated

by including an elastic portion related to K combined with a nonlinear portion

related to the area under the load–displacement curve (with respect to load-line

displacement). The plane-strain, elastic plastic J

parameter measures the onset

of instability, in this case corresponding to the point where the crack has ad-

vanced a certain amount (after blunting) on the R curve (note this instability

criteria is very different than that described for K

). It is often the case in this

type of test that slow, stable cracking will continue as the specimen is displaced

and the load gradually drops to a small fraction of the peak obtained.

The J

test procedure is usually performed using compact-tension specimens,

although three-point bend specimens are also possible provided care is taken to

ensure that accurate instrumentation is available to measure load-line displace-

6 FATIGUE TESTING 535

Inconel HAZ

W=1.0 inch, B=0.5 inch

Crack length extension, 10

–3

inch

0 20406080100

Fracture Resistance, J in-lb/in

400

800

1200

1600

2000

2400

Fig. 13 Example of an R curve from a J

fracture toughness test performed

on a high-strength weld.

ment. The primary challenge with J

testing is the high accuracy required from

measurements of the load-line displacement. High-resolution measurement is

required so as to infer the extremely small amount of crack tearing necessary

for the J

fracture toughness deﬁnition. This high accuracy is practically

achieved through higher accuracy (minimum of 16 bit), low noise data acqui-

sition boards combined with extreme care and control over the critical test de-

tails, especially as related to loading pin interfaces and clip gauge attachment.

In editions of the ASTM test standards prior to 1998, the test method for the

elastic-plastic J

parameter has always been described in ASTM E813. However,

the E08 committee of ASTM has recently published a combined approach under

the designation E1820. This combined approach outlines a procedure to follow

that can result in any one of the primary toughness parameters, namely K

, K-,

and J-resistance curves. Although these three parameters constitute the most

common fracture test standards, other ASTM methods are also available:

●

Crack-tip opening displacement (CTOD) fracture toughness (E1290)

●

Plane-strain crack arrest fracture toughness (E1221)

●

Fracture testing with surface-crack tension specimens (E740)

●

Chevron notch fracture toughness of metallic materials (E1304)

However, these methods are typically not as wide-ranging nor as commonly

encountered in design as the previously noted fracture toughness test methods.

6 FATIGUE TESTING

The previous section described basic fracture mechanics concepts as well as the

test methods used to quantify fracture behavior given a preexisting crack. In this

536 TESTING OF METALLIC MATERIALS

section the process of crack nucleation (e.g., crack initiation) and stable propa-

gation (e.g., crack growth) under cyclic loading is addressed. Understanding

crack propagation behavior is critically important since some structures, most

notably the current ﬂeet of commercial airliners, adopt a damage tolerance de-

sign philosophy whereby cracks are assumed in the structure, and a safe struc-

tural life requires detecting and managing these cracks. Quantifying crack

nucleation is also important since much of a crack’s cyclic life is spent initiating

or growing when it is small in size.

The test specimens used for fatigue crack initiation testing are similar geo-

metrically to tensile specimens. The two basic test methods for initiation testing

are ASTM E606 and E466 for strain- and stress-controlled testing, respectively.

With both of these methods, the goal is to cycle a specimen until failure occurs.

Strain control is typically used if (a) the loading is with a constant displacement

or (b) the focus is on the short life regime (estimated to be less than 10

to 10

cycles although this is fairly controversial) where applied stress levels often

approach the yield strength of the metallic material. Conversely load control

testing is performed for the mid- to high-cycle life regime. Stress (effectively

load) control testing is relatively straightforward since no displacement moni-

toring or measurement is required. The specimen can be gripped, cycling can

commence, and continue unmonitored (provided the servocontrol loop does not

require any attention) until failure occurs.

One of the most critical details when performing this type of fatigue testing

is controlling bending in the specimen. Poor control of bending typically in-

creases the apparent scatter in the fatigue data generated. Commercial alignment

ﬁxtures are available to mitigate bending. These ﬁxtures utilize a strain gauged

specimen and a ﬁxture that allows minute adjustments of the loading train. How-

ever, these ﬁxtures are not always available, which means less high tech and

pragmatic methods are often required to minimize the effect.

Although guidelines on how to measure and verify specimen alignment are

provided in ASTM E1012, minimizing bending practically requires control of

three issues. First, the specimen must be carefully machined with excellent di-

mensional control, parallelism, and axisymmetry. Second, the ﬁxtures used to

connect the specimen to the loading train must not introduce any asymmetry.

Threaded specimen ends, although commonly employed during fatigue testing,

should be avoided as the tolerances required to engage the thread make accurate

alignment problematic (however, it must be stated that it is not always practical

or possible to avoid threaded ends). If the loading cycle is tensile only (e.g., a

positive R ratio where R is the ratio of minimum to maximum load), a spherical

joint, or more commonly known as a U-joint, in the loading train can minimize

bending. Third, the ﬁxtures in the load train should be designed well enough to

grip repeatably in the same manner so that once a load train has been carefully

aligned with an instrumented specimen it does not vary.

Some typical stress–life (S–N) data are shown in Fig. 14 for two materials,

an aluminum and titanium alloy. S–N data are usually generated at a ﬁxed R

ratio with the results represented as a plot with stress amplitude or maximum

stress on the abscissa and cyclic life on the ordinate. Data are sometimes gen-

erated on specimens that have a groove or hole to simulate stress concentrations

typically found in structure. Furthermore, it is often of practical interest to know