ASM Metals HandBook Vol. 8 - Mechanical Testing and Evaluation

Подождите немного. Документ загружается.

Fig. 8 Comparison of single- and double-converter systems. Calculated displacement, strain, and specific

energy parameters for a highly damped perspex specimen (L = λ/2) are shown. The single-converter

system was excited from the left side. Assumed values for the specimen: E = 48 GPa (6.9 psi × 10

); ρ =

1.2 g/cm

; frequency = 20 kHz. Source: Ref 21

Acoustic horns transmit the resonance developed by the converter to the specimen. One or more acoustic

amplifying horns generally are placed in the wave train to raise the strain amplitude in the specimen to the level

required for fatigue. Design of these horns was developed by Mason (Ref 22) and Neppiras (Ref 15).

Acoustic horns are bars of resonant length with cross-sectional areas that change either continuously or

discontinuously as distance from the input end varies. To maintain the requirement of continuity of particle

velocity along the horn length, the vibrational amplitude must increase in areas of reduced cross-sectional area.

This produces displacement and strain amplification. For the simple stepped-down horn shown in Fig. 9(a),

displacement and, hence, strains are amplified by the ratio of the cross-sectional areas of the horn:

(Eq 5)

where A

is the displacement amplitude on the output end of the horn, and A

is the displacement amplitude on

the input end of the horn. Conversely, an increase in cross-sectional area causes deamplification. A number of

different horn shapes have been designed by detailed mathematical analysis (Ref 23, 24, 25, and 26); typical

examples are shown in Fig. 9 along with the particle velocity and stress distribution along their length.

Fig. 9 Profiles of acoustic horns for amplifying converter output. Variations in particle velocity and

stress along horns are shown below each profile. (a) Stepped. (b) Conical. (c) Exponential. (d) Catenoidal.

(e) Fourier. Source: Ref 23

An obvious problem with the stepped horn is the manifestation of very-high-stress amplitudes at the step. This

eventually causes the horn to fail by fatigue at the step. The Fourier horn (Fig. 9e) is the best overall horn

design for achieving the highest amplification with the greatest strength and stiffness. Ultrasonic converter

suppliers typically carry a variety of acoustic horn designs with amplifications ranging from 1 to 1 to on the

order of 10 to 1. Some manufacturers will design acoustic horns for custom applications, including very high

amplifications, special frequencies, or sealing the entrance of an environmental chamber.

Extension Acoustic Horns. The need for extension horns generally arises when the test specimen must be

isolated in a controlled environmental chamber, furnace, or external load frame. Typically, a 1 to 1 extension

horn is used—that is, a uniform bar of length λ/2, having been modified with a flange for seating an

environmental seal or for attaching the wave train to the external load frame.

It is important to select materials that will not affect the test. Extension horn material should be low damping

whenever possible to avoid losses and excessive heating. For corrosive environments, the horn material should

be such that a galvanic couple is not set up between the specimen and horn. If high temperatures are to be

encountered, the horn material must possess appropriate high-temperature strength. The elastic properties of the

material should be determined for the temperature of the desired test. Some properties of materials used for

acoustic extension horns are presented in Table 2.

Table 2 Typical 20 kHz resonance properties for acoustic extension horn materials

Wavelength of sound at the

resonant frequency (λ)

Young's modulus (E) Material Density (ρ),

g/cm

mm in. MPa psi × 10

Ra-333 8.3 244 9.61 199,300 28.9

Udimet 710 8.1 264 10.40 224,090 32.5

Udimet 720 8.1 264 10.40 226,160 32.8

MP35N 8.5 246 9.68 206,160 22.9

Ti-6Al-4V 4.4 249 9.80 108,940 15.8

AISI 403 7.7 262 10.31 210,960 30.6

17-4PH 7.8 250 9.84 199,960 28.9

Amplitude Detection. The accuracy of determining specimen strain depends to a great extent on the accuracy of

the measurement of the displacement amplitude at the end of the specimen. Much research has been directed

toward the development of sensitive methods for measuring displacement amplitude at ultrasonic frequencies.

The simplest, most reliable method is observation of the trajectory of a feature on the specimen surface with a

dark-field optical microscope having coaxial lighting and a filar eyepiece.

At magnifications of 500× and a reticle scale in hundredths of a millimeter, displacements on the order of 2 to 3

μm can be detected. The trajectory of a point on the specimen surface appears as a bright streak whose length

will be the peak-to-peak displacement amplitude. Visual observation of displacement amplitude with a

microscope does not lend itself to automated recording of the displacement. However, it is the preferred method

for calibrating more automated amplitude detection equipment, because secondary modes of vibration are easily

detected. Secondary modes cause the normally linear trajectory of a point to skew or appear to orbit about

another point.

Other amplitude detection devices more suitable for automated data acquisition or feedback to a closed-loop

fatigue apparatus have also been used. These generally are displacement-measuring devices, which come in

many forms, including capacitance gages (Ref 16, 27), permanent magnet-coil arrangements that use eddy

currents (Ref 17, 28), a microphone pickup (Ref 29), a photodiode arrangement (Ref 30), and closed-circuit

television (Ref 31).

Noncontacting capacitance-type detectors to measure displacement amplitudes at ultrasonic frequency are

commercially available with sensitivities on the order of 3 × 10

μm and linear frequency response up to 50

kHz. Eddy current probes also are commercially available and are frequently used to measure displacement.

Because eddy currents are sensitive to composition and microstructure, eddy current probes require calibration

of the output signal versus displacement for each new material that is tested. These devices generally are not

useful for measuring displacement of nonconducting or low-permeability materials. When the end of the

specimen is inaccessible due to an environment chamber or extension horns, the displacement must be

calibrated from the specimen to some other point on the wave train.

Strain can be measured by directly applying strain gages to the specimen and reading the value from a strain

conditioner that has a frequency response equivalent to the test frequency. Several problems are associated with

strain gages applied directly to the specimen. Strain gages, strain gage leads and the adhesive that holds a strain

gage onto the specimen are subject to fatigue loading and have finite cyclic lifetimes. If they are placed at the

point of maximum strain on the specimen they generally will fail before the specimen does. Attachment of the

gage to the specimen may modify the specimen surface and lower its fatigue properties. The presence of a gage

on the specimen can cause additional heating to occur under the gage. These problems can be avoided by

placing the gage on an adjacent extension horn where the maximum strains are lower than those on the

specimen. The strain in the specimen then can be calibrated to the strain in the extension horn.

It is best to use amplitude detectors that can be located at, or as close as possible to, the specimen.

Consideration should be given as to whether the environment that is used will interfere with operation of the

device. Some devices give outputs that vary when they are positioned away from the specimen and require

calibration for each material; electrodynamic devices operating on eddy currents induced in the vibrating

specimen fall into this category. These devices are more difficult to use than capacitance gages, whose output is

linear over a known range and does not depend on the properties of the test material. Capacitance gages also

need to be calibrated if the working environment changes the dielectric constant.

Crack growth measuring systems for ultrasonic fatigue tests are similar to those used at conventional

frequencies. Optical methods can be used because the fatigue crack grows in the plane of maximum strain,

which is a displacement node. As a result, the crack will not appear to be vibrating. Optical methods using

microscopes on traveling stages can be operated manually or automatically. Automated systems for video

monitoring of the crack length are described in Ref 19. Foil-type crack growth gages can be attached to the

specimen in the path of the growing crack. These gages develop a linearly varying potential as the crack tears

the foil and have been shown to operate linearly under 20 kHz cycling. The usual requirements for monitoring

the symmetry of fatigue crack length in double-edge-cracked or center-cracked specimens at conventional

frequency also apply for ultrasonic frequency testing.

Cooling Systems. As in conventional fatigue testing, the components of fatigue deformation generate heat

within the specimen in an amount equal to the area enclosed by the stress-strain hysteresis loop in each cycle.

At 20 kHz, the specific energy input at high stresses can be as high as several hundred watts per cubic

centimeter for high damping materials. Large temperature excursions can result if the heat is not removed.

These temperature excursions not only change the properties of the test material, but can introduce mean tensile

stresses on the surface of solid specimens because of the differential thermal expansion between the core and

surface of the specimen.

In most cases, the heat generated can be dissipated by forced cooling if the heat transfer path is not too lengthy.

In some cases, the heat generated cannot be eliminated by forced-air cooling, and more efficient external and

internal cooling may be necessary. A hollow specimen can be used to decrease the heat-transfer path through

the metal.

In specimens of materials with high damping and large cross sections, this heat rise can be extreme. Figure

10(a) shows the temperature contour of a 12 mm (0.48 in.) diam resonant steel bar without cooling (Ref 32).

The strain antinode heated to almost 400 °C (750 °F) during ultrasonic-frequency cycling at a stress level of

150 MPa (22 ksi). The highest temperatures are observed at the strain antinode. Efficient cooling of a test

specimen would obviously require the greatest volume of coolant to flow over the strain antinode. Large

temperature excursions can usually be minimized in practice by the choice of a hollow dumbbell specimen

design.

Fig. 10 Infrared thermogram of specimen under 20 kHz excitation without cooling. (a) Resonant

specimen exhibiting nodal heating. (b) Localized heating due to thermocouple epoxied to surface of

nonresonant specimen. Source: Ref 32

The damping characteristic of the material and the efficiency of the coolant determine the choice of a solid or

hollow specimen design for testing. In general, cooling a material with high damping (such as pure copper)

with forced-air heat transfer requires a thin-walled hollow specimen to ensure that the temperature rise is

acceptably low. At the other end of the damping spectrum, titanium alloys generate so little heat at 20 kHz at

modest stress levels that virtually any specimen geometry is compatible with the temperature rise limitations.

Martensitic steel lies somewhere between copper and titanium, and hollow or solid specimens may be selected

depending on whether the coolant is forced air or water. Measurement of the temperature rise at the uncooled

interior of a hollow type 403 stainless steel specimen (1.25 mm, or 0.05 in., wall thickness, with water cooling

at the external surface) revealed a temperature rise of less than 10 °C (18 °F) even during fatigue at stresses

high enough to cause failure in 10

cycles.

Monitoring the temperature rise is essential to obtaining reliable data, particularly when testing in air. Attaching

a thermocouple to the specimen at the maximum strain point presents two basic problems. First, the specimen

could be damaged and premature fatigue failure could occur. Second, infrared data show that attachment of the

thermocouple to a nonresonant bar causes localized heating in the vicinity of the thermocouple (Fig. 10b).

The localized heating implies that thermocouples may not be reliable for measuring the specimen temperature

during ultrasonic vibration. It has been proposed that the localized heating could be caused by vibration of the

leads of the thermocouple due to active displacement at all positions of a nonresonant specimen. Hence, placing

the thermocouple at the displacement node of a resonant specimen might minimize this local heating. This

viewpoint has not yet been verified.

Two alternate temperature-sensing methods that work well in air environments are infrared imaging of the

specimen and application of temperature-sensitive paint to the specimen surface. These methods provide only

surface temperature information. The subsurface temperature is hotter. In liquid environments, the alternate

temperature measurements will not work, and thermocouple data may be the only possible method. In liquid

environments, particularly water, temperature control is somewhat easier because the liquid acts as a coolant

and moderates the temperature excursion. Generally, the liquid is controlled to some temperature below the

desired temperature so that the excess heat generated from testing will be dissipated in the liquid. For elevated-

temperature testing, the furnace temperature must be controlled so that the heat generated in the specimen helps

achieve the desired temperature.

A forced-air cooling system with an infrared temperature monitoring system is shown in Fig. 11. Air jets are

aimed at the strain antinode. With an air line pressure of 0.5 MPa (72.5 psi) and a venturi-type air cooler in the

line, an air jet exit temperature of 0 °C (32 °F) can be obtained. A liquid cooling system requires that the liquid

flow over the surface of the specimen.

Fig. 11 Test facility (20 kHz) showing positioning of forced-air cooling, infrared temperature monitor,

and external load frame for mean load. 1, converter; 2, booster horn; 3, connecting horn; 4, specimen; 5,

capacitance gage; 6, cooling ring; 7, four air inlets; 8, venturi air cooler; 9, air supply; 10, upper and

lower support plates; 11, hydraulic pistons; 12, window; 13, infrared camera

The major difference in ultrasonic fatigue testing compared to conventional fatigue testing is that the specimen

should not be totally immersed in a bath of coolant. Immersing the specimen could prevent the system from

obtaining constant-amplitude resonance. If resonance can be obtained while immersed, cavitation erosion

damage may occur at fillets and ends of the specimen. This can be prevented by the application of a thin film of

rubber compound, such as carboline neoprene adhesive, to these areas of the specimen. A typical liquid cooling

system is shown in Fig. 12. In this system, the liquid flows onto the specimen and drains off to be recirculated.

This system can also be used for environmental testing.

Fig. 12 Environment supply system for liquid cooling or corrosion-fatigue testing

In a liquid cooling system, an inert coolant is necessary to achieve baseline fatigue property data approaching

those of air tests. The coolant must not be corrosive to the material and must have a heat capacity large enough

to remove the heat from the specimen. Deionized, low-oxygen-content water is a very high-heat-capacity,

minimally corrosive coolant for most materials. It is a little more difficult to use low-oxygen deionized water as

a coolant because an environmental system is necessary to control the gaseous traces in the water. Acid-free

transformer oil is another coolant frequently used for ultrasonic fatigue tests. Liquid cryogen coolants, such as

liquid nitrogen, generally are less effective due to their tendency to vaporize on contact with the specimen. The

vapor forms a boundary layer at the specimen surface, which reduces effective cooling instead of increasing it.

Fatigue property data obtained with a nonaggressive coolant generally are slightly lower than data obtained

with air cooling. Baseline data obtained with an inert coolant are usually necessary for the interpretation of

corrosion fatigue data.

External Load Frame. The wave train arrangement can be used without further attachment for completely

reversed tension-compression testing. The wave train can also be placed in an external load frame, such as a

tensile test machine, to provide static mean loading or superposition of large-amplitude low-frequency cycling

on top of the high-frequency cycling (Ref 18, 32). The external load frame is attached to the wave train at the

displacement nodes on acoustic horns on either side of the specimen, as shown in Fig. 11.

Design of specimens to be subject to superimposed external loads must take two additional factors into account.

First, the elongation of the specimen due to external straining must be considered to stay within the bounds of

the resonance conditions. Second, there have been observations of softening of metals during simultaneous

tensile or compressive mechanical deformation and high-frequency straining, which is known as the Blaha

effect (Ref 33). These mechanisms are discussed in more detail in Ref 32. Test engineers should be aware of

this effect because it can result in additional plastic deformation of the material during testing.

Environmental Fatigue. Ultrasonic fatigue testing can be performed under most environmental conditions. One

possible exception is vacuum, in which testing is narrowed to a few very-low-damping materials at low stress

levels. Environmental testing requires the normal ultrasonic fatigue testing apparatus with the addition of an

environmental chamber around the test specimen and an environmental supply system.

Elevated Temperatures. For high-temperature testing, the furnace serves as the environmental chamber and

supply system. A high-temperature test stand with mean load capability can be constructed by placing a furnace

around the specimen in the test system shown in Fig. 11. Some tuning generally is required to design extension

horns that will be resonant in the temperature gradient from the furnace midpoint to the ambient temperature

outside the furnace, because the resonant frequency is temperature dependent. For high-temperature fatigue, the

specimen displacement amplitude usually will have to be calibrated to a displacement antinode outside the

furnace (Ref 34).

Aggressive Liquid Environments. Testing in corrosive liquid environments requires both an environmental

chamber and an environmental recirculation system. This is essentially the same equipment needed for inert

liquid cooling, as shown in Fig. 12. Additional features are incorporated into the environmental recirculation

system to provide control of the solution composition, purity, and temperature. Ports are incorporated so that

environment composition samples can be taken for documentation and solution pH can be adjusted. An inert

gas overpressure is usually maintained throughout the system to control the dissolved oxygen content of the test

solution. Appropriate plumbing and seals are incorporated so that the specimen chamber can be purged of air

prior to circulation of the environment.

A controlled-corrosion fatigue chamber is shown in Fig. 13. This chamber exhibits features needed for

electrochemical corrosion fatigue study (Ref 20). The chamber is composed of an outer chamber and an inner

chamber constructed of Teflon. The inner chamber contains a finite volume of liquid around the gage section of

the specimen. A platinum electrode is fitted into the inner chamber for anodic or cathodic polarization of the

specimen. A window is placed at the side of the inner chamber to enable viewing of the amount of liquid in the

inner chamber. A port is placed in the front of the inner chamber so that a standard reference electrode can be

inserted to measure the electrochemical potential of the specimen. The electrodes can be removed for normal

corrosion fatigue testing.

Fig. 13 Environmental fatigue test chamber for electrochemical 20 kHz testing

Proper selection of horn material is important in corrosion fatigue testing, because a mismatched specimen and

horn material combination may set up a galvanic couple when the joint is wetted with a conductive solution.

Depending on the galvanic couple, the horn material may cause the gage of the specimen to be

electrochemically more active or passive than normal. This could have a pronounced effect on the corrosion

fatigue properties that are being determined. It is advisable to make the specimen and the extension horn out of

the same material. If the horn and specimen must be made out of dissimilar metals, care must be taken to ensure

that the joint is not exposed to the conductive test solution.

In fatigue crack growth testing in liquid environments, the effect of the liquid inside the growing fatigue crack

also must be considered. The effect of the liquid fatigue crack growth rate is currently being investigated.

Depending on the fluid properties, the fatigue crack may be wedged open, causing errors in the da/dN and

threshold stress intensity range, ΔK

question is whether the environment ever extends to the crack tip. It has

been suggested that at high crack growth rates, the environment has little influence on rapid crack growth, and

the crack tip behaves as if it were in vacuum (Ref 35).

References cited in this section

15. E.A. Neppiras, Proc. ASTM, Vol 59, 1959, p 691

16. J.K. Tien, S. Purushothoman, R.M. Arons, J.P. Wallace, O. Buck, H.L. Marcus, R.V. Inman, and G.J.

Crandall, Rev. Sci. Instrum., Vol 46, 1975, p 840

17. W. Kromp, K. Kromp, H. Bitt, H. Langer, and B. Weiss, Ultrasonics International 1973 Conference

Proceedings, IPC Science and Technology Publications, Guildford, Surrey, U.K., 1973, p 238

18. I. Hansson and A. Tholen, Ultrasonics, March 1978, p 57

19. S. Stanzl and E. Tschegg, Met. Sci., April 1980, p 137

20. R.A. Yeske and L.D. Roth, in Ultrasonic Fatigue, J.M. Wells, O. Buck, L.D. Roth, and J.K. Tien, Ed.,

TMS-AIME, Warrendale, PA, 1982, p 365

21. P. Trimmel and W. Kromp, in Ultrasonic Fatigue, J.M. Wells, O. Buck, L.D. Roth, and J.K. Tien, Ed.,

TMS-AIME, Warrendale, PA, 1982, p 37

22. W.P. Mason, J. Acoust. Soc. Am., Vol 28, 1956, p 1207

23. J.R. Frederick, Ultrasonic Engineering, John Wiley & Sons, New York, 1970

24. G. Amza and D. Drimer, Ultrasonics, Vol 14, 1976, p 223

25. E. Eisner, J. Acoust. Soc. Am., Vol 35, 1963, p 1367

26. L. Balamuth, Trans. IRE (Ultrasonic Eng.), Vol 2, 1954, p 23

27. B.S. Hockenhull, C.N. Owston, and R.G. Hacking, Ultrasonics, Vol 9, 1971, p 26

28. A. Thiruvengadam, J. Eng. Ind. (Trans. ASME), Vol 11, 1966, p 332

29. H. Konagai, S. Tanaka, and T. Sakurai, J. Soc. Mater. Sci., Jpn., Vol 24, 1975, p 753

30. G.C. George, Corrosion Fatigue: Chemistry, Mechanics and Microstructure, O. Devereux et al., Ed.,

National Association of Corrosion Engineers, Houston, 1972, p 459

31. V.A. Kuz'menko, G.G. Pisarenko, and A.K. Gerikhanov, Probl. Prochn., Vol 4, 1977, p 120

32. R.B. Mignogna and R.E. Green, Jr., in Ultrasonic Fatigue, J.M. Wells, O. Buck, L.D. Roth, and J.K.

Tien, Ed., TMS-AIME, Warrendale, PA, 1982, p 63

33. F. Blaha and B. Langenecker, Die Naturwiss., Vol 42, 1955, p 556

34. G. Whitlow, L.E. Willertz, and J.K. Tien, in Ultrasonic Fatigue, J.M. Wells, O. Buck, L.D. Roth, and

J.K. Tien, Ed., TMS-AIME, Warrendale, PA, 1982, p 321

35. J.K. Tien and R.P. Gamble, Met. Trans., Vol 2, 1971, p 1933

Ultrasonic Fatigue Testing

Test Specimens

Ultrasonic fatigue test specimens must be designed to resonate at the desired test frequency. The first step in

designing an ultrasonic fatigue test specimen, acoustic horn, or resonant bar is to obtain the appropriate

properties and constants of the materials. The material density, dynamic modulus of elasticity, and the half

wavelength of sound in the material at the desired testing frequency must be determined.

Frequency, Wavelength, and Speed of Sound. The longitudinal resonance frequency of a bar test material is

measured experimentally, as shown in Fig. 14. A uniform bar of test material is excited by a small converter

coupled to a variable-frequency oscillator. The converter can be an electrodynamic vibrator or any other

vibrator capable of ultrasonic frequencies. A piezoelectric pickup is placed against the opposite end of the bar

to monitor the amplitude of vibration. For most pure metals and alloys, the bar should be about 100 to 150 mm

(4 to 6 in.) long and 4 to 10 mm (0.16 to 0.4 in.) in diameter. This diameter is comparable to the diameters of

most test specimen gages.

Fig. 14 Experimental measurement of the longitudinal resonance frequency of specimens

The bar should be similar in size to the eventual specimen gage diameter, because measured resonant

frequencies vary with large-diameter bars. This directly affects calculation of the dynamic elastic modulus, and

ultimately affects calculation of the fatigue stress amplitude. Frequencies on the order of tens of kilohertz

should be measured for a bar length in this given range. Sweeping the oscillator through the frequency spectrum

produces a large increase in output for some frequency; this is the resonance frequency. The length of the bar is

λ/2 for the experimentally determined resonance frequency. Resonant wavelengths for several pure metals at

test frequencies of 20 kHz and 2 MHz are given in Table 3. The resonant wavelength is inversely proportional

to frequency. The need for a macroscopic test specimen frequently precludes high-frequency testing in the 2

MHz range. The speed of sound through the material can be calculated by relating the speed of sound, C, to

frequency, f, and wavelength, λ:

C = λ

= λ

(Eq 6)