Capitulo 19

Documento extractado del libro Manufacturing Systems

Tema

Fracture testing

1

Chapter 19

FRACTURE TESTING

*19.1 WHYFRACTURE TESTING? There are several good reasons why we should be interested in fracture behavior and fracture testing of metals. Fracture of any material (be that a recently acquired child's toy or a nuclear pressure vessel) is gene rally an undesirable happening, resulting in economic loss, an interruption in the availability of a desired service, and possibly damage to human beings. Besides, one has good technical reasons to do fracture testing. These include:[l) to compare and to select from candidate material s the toughest (and most economic) one forgiven service conditions; to compare a given material's fracture characteristics against a specified standard; to be able to predict the effects of service conditions (e.g., corrosion, fatigue, stress corrosion, etc., on the material toughness); and to study the effects of metallurgical changes on material toughness. One or more of these reasons for fracture testing may be involved during the design, material selection, construction, and/or operation of metallic structures. There are two broad categories of fracture tests, qualitative and quantitative. The Charpy impact test exemplifies the former, and the plane-strain fracture toughness (KIc) test illustrates the latter. We describe briefiy important tests in both these categories. *19.2 IMPACT TESTING We saw in Chapter 3 that stress concentrations, like cracks and notches, are sites where failure of a material starts. It has been long appreciated that the failure of a given material in the presence of a notch is controlled by its fracture toughness. A number of tests have been developed and standardized to measure this "notch toughness" of a material. Almost all of these are qualitativ~ and comparative in nature. As pointed out in Chapter 3, triaxial stress state, high strain rate, and low temperature all contribute to a brittle failure of the material. Thus, in order to simulate the most service conditions, almost all of these tests involve a notched sample to be broken by impact over a range of temperatures *19.3 CHARPY IMPACT TEST The Charpy V -notch impact test is an ASTM standardJ2] The notch is located in the center of test sample. The test sample, supported horizontally at two points, receives an impact from a peIldulum of a specific weight on the side opposite that of the notch (Fig. 19.1). The specimen fails in fiexure under impact (E - 1()3 S-1).

2

In the region around the notch in the test piece, there exists a triaxial stress state due to plastic yielding constraint there. This triaxial stress state and the high strain rates used propitiate the tendency for a brittle failure. Generally, we present the results of a Charpy test as the energy absorbed in fracturing the test piece. An indication of the tenacity of the material can be obtained by an examination of the fracture surface. Ductile materials show a fibrous aspect, whereas brittle material s show a fiat fracture. A Charpy test at only one temperature is not sufficient, however, because the energy absorbed in fracture drops with decreasing test temperature. Figure 19.2 shows this variation of energy absorbed as a function of temperature for a steel in the annealed and in the quenched and tempered state [3] The temperature

3

Figure 19.2 Energy absorbed versus temperature for a sted in annealed and in quenched and .tempered states. (Adapted with permission from [3], p. 825-829.) at which there occurs a change from a high-energy fracture to a low-energy one is called the ductile-brittle transition temperature (DBTI). However, as in practice there does not occur a sharp change in energy but instead there occurs a transition zone, it becomes difficult to obtain this DBTI with precision. Figure 19.3 shows how the morphology of the fracture surface changes in the transition region. The greater the fraction of fibrous fracture, the greater the energy that is absorbed by the specimen. The brittle fracture has a typical c1eavage appearance and does not require as much energy as the fibrous fracture. BCC and HCP metals or alloys show a ductile-brittle transition, whereas FCC structures do noto Thus, generally, a series of tests at different temperatures is conducted which permits us to determine a transition temperature. This ductilebrittle transition temperature, however arbitrary, is an important parameter in material se1ection from the point of view of tenacity or the tendency of occurrence of brittle fracture. As this transition temperature is, generally, not very well defined, there exist a number of empirical ways of determining it, based on a certain absorbed energy (e.g., 15 J), change in the fracture aspect (e.g., the temperature

4

corresponding to 50% fibrous fracture), or lateral contraction (e.g., 1 %) that occurs at the notch root. The transition temperature depends on the chemical, composition, heat treatment, processing, and microstructure of the material. Among these variables, grain refinement is the only method that results in an increase in strength of the material in accordance with the Hall-Petch re1ation and at the same time reduction in transition temperature. Heslop and Petch [4] showed that the transition temperature Te depended on the grain size, D, in the following way:

Figure 19.3 Effect of temperature on the morphology of fracture surface of Charpy steel specimen. Test temperatures Te < Tb < Te < Td. (a) Fully brittle fracture; (b, c) mixed-mode fractures; (d) fully ductile (fibrous) fracture. where B is a constant. Thus, a graph of Te against D will be a straight line with slope -11/3.

5

19.4 DROP-WEIGHT TEST This test is used to determine a reproducible and well-defined ductile-brittle transition in steels. The specimen consists of the steel plate containing a brittle weld on one surface. A saw cut is made in the weld to localize the fracture (Fig. 19.4). The specimen is treated as "simple edge-supported beam" with a stop placed below the center to limit the deformation to a small amount (3%)

Figure 19.5 Charpy V -notch curve compared with drop-weight NOT.

6

and prevent general yielding in different steels. The load is applied by means of a freely falling weight striking the specimen side opposite to the crack starter. Tests are conducted at 5-K intervals and a break/no break temperature, called the nil ductility transition (NDT) temperature, is determined. NDT temperature is thus the temperature below which a fast unstable fracture (i.e., brittle fracture) is highly probable. Above this temperature, the toughness increases rapidly with temperature. This transition temperature is more precise than one of the Charpy-based transition temperatures. The drop-weight test uses a sharp crack that movesrapidly from a notch in a brittle weld material, and thus the NDT temperature correlates better with the information from a K1c test, described in Section 19.6. This test pro vides a usefullink between the qualitative "transition temperature" approach and the quantitative "K1c" approach to fracture.fl). The drop-weight test provides a simple means of quality control through the NDT temperature. It (the NDT temperature) can be used to group and classify various steels. For some steels, identification of the NDT temperature can be used to indicate safe minimum operating temperatures for a given stress. That this drop-weight NDT test is more reliable than a Charpy V -notch value of transition temperature is illustrated by Fig. 19.5 for a pressure-vessel steeUl) The vessel fractured in an almost britde manner near its NDT temperature, although according to the Charpy curve it was still very tough. The drop-weight test is applicable primarily to steels in the thickness range 18 to 50 mm. NDT temperature is unaffected by section sizes above about 12 mm; beca use of the small notch and the limited deformation due to brittle weld bead material, sufficient notch-tip restraint is ensured. 19.5 INSTRUMENTED CHARPY IMPACT TEST The Charpy impact test described above is one of the most common tests for characterizing the mechanical behavior of materials. The principal advantages of the Charpy test are the ease of specimen preparation and the executiun of the test proper, speed, and low costo However, one must recognize that the common Charpy test basicaliy furnishes information of only a comparative char

7

Figure 19.6 (a) Typical oscilloscope record of an instrumented Charpy impact test. (b)

Schematic representation of an in-strumented Charpy impact test.

acter. The transition temperature, for example, depends on the specimen thickness (hence, the need to use standard samples); that is, this transition temperature can be used to compare, say, two steels, but it is not an absolute material property. Besides, the common Charpy test measures the total energy absorbed (ET), which is the sum of energies spent in initiation (E¡) and in propagation (Ep) of crack (i.e., ET = E¡ + Ep). In view of this problem, a test has been developed called the instrumented Charpy impact test.[5] This instrumented impact test furnishes, besides the absorbed

8

energy, the variation of applied load with time. The instrumentation in vol ves the recording of the signal from a load cell on the pendulum by means of an oscilloscope in the form of a load time curve of the test sample. Figure 19.6(a) shows a typical oscilloscope record and Fig. 19.6(b) shows a schematic representation. This type of curve can provide information about the load at general yield, maximum load, load at fracture, and so on. The energy spent in impact canal so be obtained by integration of the load-time curve. From the load-time curve, one can obtain the energy of fracture if the pendulum velocity is known. Assuming this velocity to be constant during the test, we can write the energy of fracture as

where E' is the total fracture energy based on the constant pendulum velocity, Vo the initial pendulum velocity, P the instantaneous load, and t the time. In fact, the assumption of a constant pendulum velocity is not a valid one. According to Augland,[7]

where Et is the total fracture energy, E' = Vo fJ P dt, a = E'/4Eo, and Eo is the initial pendulum energy. The values of total energy absorbed in fracture computed this way from the load-time curves show a one-to-one correspondence with the energy-absorbed values determined in a conventional Charpy test[8] Based on this, we can use Eq. 19.2 for cotnputing, at a given temperature, the initiation and propagation energies. This information, together with the load at yielding, maximum load, and load at fracture, gives the capacity to identify the various stages of the fracture process.

It is well known (see Section 3.4) that the plane-strain fracture toughness (K¡c) test gives a much better and precise idea of the material tenacity and is a material property. However, this test, as will be seen below, possesses certain disadvantages: equipment and specimen preparation is rather expensive, the test is relatively slow and not simple to execute, and so on. Consequently, there have been attempts at developing empirical correlations between the energy absorbed in a conventional Charpy test (Cv) and the plane-strain fracture tough-ness (K¡c ))9] The reader is warned that such correlations are completely empirical and are valid only for the specific metals tested)9] The instrumented Charpy test, with samples precracked and containing side grooves in order' to assure a plane-strain condition, can be used to determine the dynamic fracture toughness KID. For ultra-high-strength metals (CTy very large), Km = K¡c. Thus, we may use instrumented Charpy test, with samples precracked and containing side grooves in order' to assure a plane-strain condition, can be

9

used to determine the dynamic fracture toughness KID. For ultra-high-strength metals (CTy very large), Km = K¡c. Thus, we may use instrumented Charpy test to determine K¡c or KID for very high strength steels. But there is still the need to verify this better and we must check the results obtained from the instrumented Charpy impact test with those obtained from standard ASTM K¡c test, as described in the next section. 19.6 PLANE-STRAIN FRACTURE TOUGHNESS TEST The fracture toughness K¡c may be determined according to the following standards: ASTM E399/79 or BS 5447/77)10] The essential steps in the fracture toughness tests involve measurement of crack extension and load at the sudden failure of sample. As it is difficult to measure crack extension directly, one measures the relative displacement of two points on the opposite sides of the crack plane. This displacement can be calibrated and related to real crack front extension. The typical test samples of tension and bending that are used in the fracture toughness tests according to the ASTM standard are shown in Fig. 19.7. Figure i9.7(c) shows how the size of the specimens can be varied. Tensile and Charpy specimens are alsoshown for comparison. The relation between the applied load and the crack opening displacement depends on the size of the crack and

10

Figure 19.7 Typical ASTM standard plane-strain fracture toughness test specimens: (a) compact tension; (b) bending; (e) photograph of specimens of various sizes showing Charpy and tensile specimens for comparison purposes. (Courtesy of M P A.

11

Stuttgart)

Figure 19.8 Typical load-displacement curves in a Ktc test (schematic) thickness of the sample in relation to the extent of plastic zones. When the crack length and the sample thickness are very large in relation to the quantity (K¡cfuy)2, the' load-displacement curve is of the type shown in Fig. 19.8(a). The load at the brittle fracture that corresponds to K¡c is then well defined. When the specimen is of reduced thickness, a step called "pop-in" occurs in the curve, indicating an increase in the crack opening displacement without an increase in the load Fig. 19.8(b). This phenomenon is attributed to the fact that the crack front advances only in the center of the plate thickness, where the material is constrained under plane-strain condition. However, near the free surface, plastic deformation is much more pronounced than that at the center, and it approaches the conditions of plane stress. Consequently, the planestrain fracture advances much more in the central portion of the plate thickness, and in regions of material near the specimen surfaces the failure eventually is by shear. Figure 19.9 shows the plastic zone at the crack front in a plate of finite thickness. At the edges of the plate (X3 ~ + B/2) the stress state approaches that of plane stress. At the center of a sufficiently thick plate, the stress state approaches that of plane strain. This is due to the fact that the €33 component of strain is equal to zero at the center as the material there in this direction is constrained, whereas near the edges the material can yield in the X3 direction and €33 is different from zero. When the test piece becomes even thinner, the plan e-stress condition prevails and the load-

12

displacement curve becomes as shown in Fig. 19.8(c). To make valid fracture toughness measurements in plane strain, the influence of the free surface, which relaxes the constraint, must be maintained small. This enables the plastic zone to be constrained completely by elastic material. The crack length must also be maintained greater than a certain lower limit. Up to this point, the conditions of sample size and the crack length have been discussed in a qualitative way. The lower limits on width, thickness, and crack length all depend on the extem of plastic deformation through the (K1c/ o-y)2 factor. In view of the lack of current knowledge about the exact plastic zone size for the crack in mode I (crack opening mode), it is not yet possible to determine the lower limits of dimension of the test piece theoretically. These lower limits above which K1c remains constant are determined by means of trial tests. Samples of dimensions smaller than these limits tend to overestimate the K1C limit.

In the fracture toughness tests, the crack is preferably introduced by fatigue from a starter notch in the sample. The fatigue crack length should be long enough to avoid interference in the crack-tip stress field by the shape of notch.

13

Figure 19.11 Procedure used for measuring the conditional value KQ.

14

Under an app1ied load, the crack opening disp1acement can be measured between two points on the notch surfaces by various types of transducers. Figure 19.10 shows an assembly for measuring disp1acement in a notched sample. E1ectrical resistance measurements ha ve al so been used to detect crack propagation. Calibration curves are used for converting disp1acement measurements and resistance measurements into crack extension. The load-displacement curves generally show a gradual deviation fram linearity and the "pop-in" step is very small (Fig. 19.11). The procedure used in the analysis of load~isplacement records of this type can be explained by using the Fig. 19.11. Let us designate the linear slope part as OA. A secant line, OPs, is then drawn at a slope 5% less than that of line OA. The point of intersection of the secant with the load~isplacement record is called Ps. We define the load PQ, for computing a conditional value of K1c, called KQ, as follows: If the load on every point of curve before Ps is less than Ps, then Ps = PQ (case 1, Fig. 19.11). If there is a load more than Ps and before Ps, this load is considered to be PQ (cases 11 and 111 in Fig. 19.11). In these cases if P max/ PQ > 1.1, the test is not a valid one; KQ does not represent the K1c value and a new test needs to be done. After determining the point PQ, KQ is calculated according to the known equation for the geometry of the test piece used. A checklist ofpoints is given in Table 19.1, and Fig. 19.12 shows schematically the variation of Kc with flaw size, specimen thickness, and specimen width.

TABLE 19.1 Checklist for the K1c Test 1. Dimensions of test piece

a. Thickness, B 2: 2.5(K1C/CTIj)2 b. Crack lcngth, a 2: 2.5(K1clCTIj)2

2. Fatigue pre-cracking a. Kmu./Klc :::; 0.6 b. Crack front curvature :::; 5% of crack length c. Inclination :::; 10° d. Length between 0.45 W and 0.55 W, where W is the width of the test sample

3. Characteristics of load displacement curve. This is effectively to limit the plasticity during the test and determines if the gradual curvature in the load-displacement curve is due to plastic deformation or crack growth. a. Pmu./PQ :::; 1.1

15

19.7 CRACK OPENING DISPLACEMENT TESTING For crack opening displacement (COD) testing, there exists a British Standards lnstitution BS 5762: 1979 method for crack opening displacement (COD) testing." Basically, the test sample for determining Oc is a slow bend test specimen similar to theone used for K¡c testing. The proposed BSI method is very similar to the ASTM E399 method for K¡c. A clip gage is used to obtain the crack opening displacement. During the test, one obtains a continuous record of load, p, versus opening displacement, D.. (Fig. 19.13). In the case of a smooth P-D.. curve, the critical value, D..c, is the total value (elastic + plastic) corresponding to the load maximum [Fig.19.13(a)]. In case the P-D.. curve shows a region of increase in displacement at a constant or decreasing load, followed by an increase in load before fracture, one needs to make auxiliary measurements to determine that this is associated with crack propagation. Should this be so, D..c will correspond to the first instability in the curve. If the P-D.. curve shows a maximum and D.. increases with a reduction in P, then either a stable crack propagation is occurring or a "plastic hinge" is being formed. The "D..c" in this case [Fig. 19. 13(b)], according to the British Standards

16

Institution, is the value corresponding to the point at which a certain specified crack growth has started. If it is not possible to determine this onset of crack propagation, one cannot measure the COD at the start of crack propagation. However, we can measure, for comparative purposes, an opening displacement om, computed from the clip gage output D..m, corresponding to the first load maximum. The results in this case will depend on the specimen geometry.

19.7.1 Computation 01 5c Experimentally, we obtain 6.c, the critical displacement of the clip gage. We need to obtain oc, the critical CTOD (crack-tip opening displacement). The British (COO) Standard contains vaIÍous methods, all based on the hypothesis that the deformation occurs by a "hinge" mechanism around a center of rotation at a depth of r( w - a) below the crack tip (Fig. 19.14). Experimental calibrations,

17

using specimens of up to 50 mm thickness, have shown that for COD in the range 0.0625 to 0.625 mm, to a very good approximation oe can be obtained from the relation

This relation is derived on the assumption that the deformation occurs by a hinge mechanism about a center of rotation at a depth of (w - a )/3 below the crack tip (i.e., r = i). However, r can be smaller for smaller values of l::..e. r =: O in the elastic case (very limited plastic deformation at the crack tip) and r =: i for a totally plastic ligament. 19.8 J-INTEGRAL TESTING The J-integra1 test was standardized by ASTM in 1981. The "Standard Test for he. A Measure of Fracture of Toughness" bears the designation E813-81 and covers the determination of he, the critica1 va1ue of J. As pointed out in Section 3.8, the physica1 interpretation of the J-integral is related te the area under the curve load versus the load-point displacement for a cracked samp1e. Both compact tension.

18

Figure 19.15 Method for determining J/e: (a) load identical specimens to dilferent displacements; (b) measun: average crack extension by heat tinting; (e) calculate J far each specimen; (d) plat J versus t:.a to find J/e. and bend specimens can be used. The ASTM recommended procedure requires at least fom specimens to be tested. Each specimen is loaded to different amounts 01' crack extensions (Fig. 19.15). One calculates J for each specimen 1'rom the expression where A is the area under load versus the load-point displacement curve, B the specimen thickness, and b the uncracked ligamento The value of J so derived is plotted against Óa, the crack extension of each spccimen. One way of obtaining Óa is to heat-tint the specimen after test and then break it open. A "best line" through the J points and a "blunting line" from the origin are then drawn. This blunting line (indicating onset of crack blunting due to plastic deformation) is obtained as follows:

Where

Is the yield stress and OUTS is the ultimate tensile stress

The intercept of de J line and the blunting line fives Jic. J is related to K:

Jic defines the onset of crack propagation in a material where large-scale plastic yielding makes direct measurement almost impossible. Thus, one can use l-integral testing to find K/c value of a very ductile material from a specimen of dimensions too small to satisfy the requirements of a proper K/c test.

19

EXERCISES 19.1. The unique feature of grain refining is that it is the only strengthening mechanism that increases

strength as well as toughness (as measured by the Charpy impact test). However, the grain size, generally, does not show any effect on uniform deformation. Why?

,19.2. How does the Charpy impact transition temperarure of a steel vary with: (a) Pearlite content? (b) Free-nitrogen content?

(e) Substitutional solutes? 19.3. Why are impact tests commonly used for steels but not so commonly for nonferrous

metal s and alloys? 19.4. Describe a few methods of monitoring crack length, pointing out the advantages

and disadvantages associated with each one. 19.5. The fracture behavior of metals is affected by temperature, strain rate, and thickness. Show

schematically the effect of these three variables on the Kc behavior of a metal. 19.6. Plot schematically the Charpy curves, energy absorbed versus temperature, for

three different thicknesses tI> t 2 > t 3, 19.7. A "stiff" machine is one that imposes a given external displacement irrespective of the load

level, while a "soft" machine is one that imposes a given loading rateo What do you think is the effect of machine stiffness on the determination of KIC?

19.8. "Acoustic emission" is a term applied to low-amplitude stress waves emitted by a solid when it is deformed. Extremely sensitive transducers can be used to study the acoustic emission from structures containing cracklike flaws. Write a short description of the application of acoustic emission to fracture mechanics. (Ref.: Acoustic Emission, ASTM STP 505, ASTM, Philadelphia, 1972.)

19.9. Describe the use of photoelasticity for determining stress-intensity factors in some fracture mechanics prob1ems. 19.10. Comment on the use of K1SCC as an engineering design parameter. REFERENCES [1] W.J. Langford, Can. Met. Quart., 19 (1980) 13. [2] Standard Methods and Definitions for Mechanical Testing of Steel Products, ASTM Standard Method A370, ASTM Annual Standards, Part 10, ASTM, Philadelphia. [3] J.c. Miguez Suarez and K.K. Chawla, Metalurgia-ABM, 34 (1978) 825. [4] J. Heslop and NJ. Petch, Phil. Mag., 3 (1958) 1128. [5] Insrrumented Impact Testing, ASTM STP 563, ASTM, Philadelphia, 1974. [6] K.K. Chawla and M.R. Krishnadev, unpublished research. [7] B. Augland, Brit. Weld. J., 9 (1962) 434. [8] G.D. Fearnehough and c.J. Hoy, J. Iron Steel Inst., 202 (1964) 912. [9] S.T. Rolfe and J.M. Barsom, Fracture and Fatigue Control in Structures, Prentice Hall, Englewood Cliffs, N.J., 1977, p. 140.

20

[10] Standard Test Method for Plane Strain Fracture Toughness of Metallic Materials, ASTM Standard Method E399-78, ASTM Annual Standards, Part 10, ASTM, Phi1adelphia. SUGGESTED READING ASTM STP 466, "ASTM, Phi1adelphia, 1970. ASTM STP 514, ASTM, Philadelphia, 1972. ASTM STP 536, ASTM, Philadelphia, 1973. ASTM STP 563, ASTM, Philadelphia, 1974. ROLFE, ST., and J.M. BARSOM, Fracture and Fatigue Control in Structures, Prentice

Hall, Eng1ewood Cliffs, N.J., 1977.