CHAPTER 1

INTRODUCTION
Reliability of materials and structures in the form of thick-walled cylinders is of critical importance to many industries including power, nuclear, chemical, armament, and food processing industries. Catastrophic failure of these cylinders can put the human life and the surroundings at very high risk. For this reason, the integrity of the cylinder should be guaranteed. The purpose of this chapter is to present a general overview of the title. The chapter comprises the following topics: The basic concepts of the fatigue process Thick-walled cylinder The problem description, research strategy and the objective of the present research Overview of the thesis

1.1 Fatigue: Progressive fracture

Fatigue is defined as: The process of progressive localized permanent structural change occurring in a material subjected to conditions that produce fluctuating stresses and strains at some point or points and that may culminate in cracks or complete fracture after a sufficient number of fluctuations [1]. Under the cyclic application of stress below the ultimate tensile strength of a material, localized hardening or softening occurs due to plastic deformation. The localized plastic deformation may occur at points of stress concentrations or even in absence of a stress raiser; for example in persistent slip bands extrusion formation can result in

crack initiation. This localized structural change may develop cracks in the structural member after a certain number of cycles. The loading is called fatigue loading and the subsequent fracture is called fatigue failure. The corresponding number of load cycles or the time during which the member is subjected to these loads before fracture occurs is referred to as the fatigue life of the member. It is believed that most common failures are due to fatigue; many experts of the field suggested that 50 to 90 percent of all mechanical failures are fatigue failures [2] and usually these failures are unexpected. Fatigue failures occur in every field of engineering and also in interdisciplinary engineering fields. They include thermal, mechanical fatigue failure in electrical circuit boards involving electrical engineers, bridges involving civil engineers, automobiles involving mechanical engineers, farm tractors involving agricultural engineers, aircraft involving aeronautical engineers, heart valve implants involving biomedical engineers, pressure vessels involving chemical engineers, and nuclear piping involving nuclear engineers. Thus, all fields of engineering are involved with fatigue design of metals. Fatigue failures encompass problems in simple items like door springs, tooth brushes, tennis racquets, electric light bulbs, and repeated bending of paper clips and include more complex components and structures like pressure vessels, ground vehicles, ships, aircraft, and human body implants. The latter include automobile steering linkage, engine connecting rods, ship propeller shafts, pressurized airplane fuselage, landing gears, and hip replacement prostheses. The integrity of nearly all engineering structures is threatened by the presence of cracks. Structural failure occurs if a crack larger than a critical size exists. Although most well designed structures initially contain no critical cracks, subcritical cracks can grow to failure under fatigue loading, called fatigue crack growth (FCG). Because it is

impossible or impractical to prevent subcritical crack growth in most applications, a damage tolerant design philosophy was developed for crack sensitive structures. Design engineers have taken advantage of the FCG threshold concept to design for long fatigue lives. FCG threshold ( Kth) is a value of K (crack-tip loading), below which no significant FCG occurs. Cracks are tolerated if K is less than Keffth.

However, FCG threshold is not constant. Many variables influence microstructure, environment, and load ratio. 1.1.1 Stages of fatigue failure

Kth including

The fatigue life of a metallic material is divided into several stages: crack nucleation, micro-crack growth, macro-crack growth, and failure. Crack nucleation is associated with cyclic slip and is controlled by the local stress and strain concentrations. Although the slip-band mechanism of crack formation may be necessary in pure metals, the presence of inclusions or voids in engineering metals will greatly affect the crack-nucleation process. Micro-crack growth, a term now referred to as the small-crack growth regime, is the growth of cracks from inclusions, voids, or slip bands, in the range of 1 to 20 m in length. Schijve [3] has shown that for polished surfaces of pure metals and commercial alloys, the formation of a small crack, about 100 m in size, can consume 60 to 80 % of the fatigue life. The AGARD [4, 5] and NASA/CAE [6] studies on small-crack behavior in a variety of materials showed that about 90% of the fatigue life is consumed in crack growth from about 10 m to

failure. This is the reason that there is so much interest in the growth behavior of small cracks. Macro-crack growth and failure regions are the regions where fracture mechanics parameters have been successful in correlating and predicting fatiguecrack growth and fracture.

Generally, a fatigue fracture has three distinct stages; crack nucleation, its growth and ultimate failure. Equally present in the first two stages is some kind of repetitive load. Typically, this load is mechanical; in the case of the crank of a combustion engine, it can be a repeating pressure; in the interior of an aircraft or fixation of wings in the fuselage of such aircrafts, it can also be of thermal origin; e.g. in boilers, heat exchangers, etc. Fig. 1.1 shows the sketch of the three stages of fatigue as seen on the fracture surface.

II
Striations One line per cycle

III I

Fig. 1.1 Sketch showing three stages of fatigue as seen on the fracture surface

The fatigue life of a member, i.e. the total number of cycles to failure, is the sum of cycles at the first and the second stages: Nf = Ni + Np where Nf : Number of cycles to failure Ni : Number of cycles for crack initiation Np : Number of cycles for crack propagation In high cycle fatigue stresses are predominantly elastic and studies show that in this case Ni is relatively high [3]. In low cycle fatigue, bulk plasticity is involved and stress levels are usually above the yield strength of the material. With increasing stress level, Ni decreases and Np dominates. (1.1)

1.1.2 Fatigue crack growth curve In general, the fatigue crack growth process is characterized by three distinct regions [7] as shown in Fig.1.2. The sigmoidal shape can be divided into three major regions. Region I is the near threshold region and exhibits a threshold value, Kth, below

which there is no observable crack growth. Cracks which form below the fatigue strength are called non-propagating cracks. This region is associated with the growth of cracks at low stress intensity factor ranges ( K) and is commonly believed to account for a significant proportion of the fatigue life of a component. Region II is the stable crack growth region and has been extensively studied for its technological importance [8 - 19]. This region, normally known as the Paris region, shows essentially a linear relationship between log da/dN and log K. This region has received the greatest attention as it is in this region the Paris crack growth law [20] can be applied, viz: da/dN = C K
m

(1.2)

Here m is the slope of the line obtained from the above equation and the coefficient C is found by extending the straight line to K=1 MPa m. Both C and m are

experimentally obtained constants. The second stage prevails for an appreciable time until finally the material fails. In region III the fatigue crack growth rates are very high as they approach instability, and little fatigue crack growth life is involved. This region is controlled primarily by fracture toughness KIC of the material.

I Log da/dN

II

III

Paris region da/dN = C Km

Kth

Log K

KIC

Figure 1.2 Typical fatigue crack growth curve

For a given material and environment, the fatigue crack growth behavior is essentially the same for different specimens or components because the stress intensity factor range is the principal controlling factor in fatigue crack growth [2]. Thus the fatigue crack growth rate (da/dN) versus K data obtained on simple specimen

configurations, under constant amplitude conditions, can be used for engineering design. Knowing the stress intensity factor expression, K, for a given component and loading conditions, the fatigue crack growth life of the component can be assessed by integrating the sigmoidal curve between the limits of initial crack size and final crack size. Schematic in Fig. 1.3 shows a constant amplitude load cycle with zero mean stress.

+
u

max

Stress

Time

min

Fig. 1.3 Alternating stress cycle with zero mean stress The related terminology is as under: Stress range Stress amplitude Mean stress Stress ratio
u

= =( =( =

ma max

mi min mi min)/2 mi min)/2

(1.3) (1.4) (1.5) (1.6)

ma max -

ma max

mi min

ma max

In a structure with crack size a, the stress intensity factor is defined as: Stress intensity factor and Stress intensity factor range K = K =

a . F(a)

(1.7)

a . F(a)

(1.8)

where F(a) is the shape factor related to the geometry of the component. Fig. 1.4 shows the crack tip stress field and the schematic of the three crack modes. deformation modes

Mode I Opening

Mode II Sliding

Mode III Tearing

Fig. 1.4 Crack tip stress field and crack deformation modes

Many fatigue crack growth data have been obtained under constant load amplitude test conditions using sharp cracked specimens. Mode I fatigue crack growth has received the greatest attention because this is the predominant mode of macroscopic fatigue crack growth. KII and KIII usually have only second order effects on both crack initiation and crack growth rates. 1.1.3 Fatigue crack propagation The second stage of fatigue fracture is crack growth or propagation. A crack forms on the slip plane of a persistent slip band and initially propagates as a stage I crack. Beyond a grain or two it becomes a stage II crack, propagating on a plane perpendicular to the principal tensile stress until the member breaks in a rapid tensile fracture mode. During stage II propagation, striations or ripples are formed on

portions of the fatigue crack surface perpendicular to the tensile direction. The growth of the crack from intrusion to the stage II propagation is a rapidly accelerated process. Hence, the fatigue crack growth process is strongly controlled by the initiation of the intrusion. Fig. 1.5 is a schematic showing the nucleation and propagation stages during fatigue. In this second stage, initially the crack will grow along lines of maximal shear and later on, when the crack has grown, along the lines of maximal tensile stress. At this stage, the micro-crack becomes a macro-crack; the growth depends solely on the properties of the bulk material and not on the microscopic or the local properties. In the crack growth process, the crack tip experiences a succession of tensile and compressive stresses. As compression changes to tension the crack opens and a system of shear stresses begins to operate at the crack tip. As the tensile stress increases plastic deformation also increases and ductile blunting of the crack tip takes place leading to crack growth. When the cycle enters the compressive phase the shear stresses are reversed and the state of maximum compression closes the crack almost completely; in the process the plastic flow reverses. The process induces a striation on the fracture surface. Subsequent tensile stress reopens the crack. The process repeats itself thereby generating striations on the fracture surface as shown schematically in Fig. 1.6. Crack closure also effect the fatigue crack growth and various sources of closure have been identified, residual such stress, as plasticity, fluid oxide or debris, phase

roughness/microstructure,

viscous

penetration,

transformation, etc. Plasticity-induced crack closure has been one of the most widely studied research topics in the area of fatigue crack growth. During loading, large tensile plastic strains are developed near the crack tip, which are not fully reversed upon unloading. This leads to the formation of a plastic wake behind the crack tip and

subsequently reduces the driving force for fatigue crack growth. However, crack closure may not be as important in stage II as in near-threshold range. The second stage, commonly known as the Paris regime prevails for an appreciable time until finally the material fails in stage III.

Crack Loading direction Free surface Stage I Nucleation Stage II Propagation

Fig. 1.5 Schematic showing the nucleation (I) and propagation (II) stages during fatigue in a polycrystalline material

Striations
(c)

(a)

S
(b)

(b)

(d)

(c)

(a)

(e)

(d)

t
(e)

Fig. 1.6 Illustration of Stage II crack growth showing the formation of striations [21]

10

The fatigue life of a member is affected by many factors e.g.; 1. 2. 3. 4. The type of load (uniaxial, bending, torsion) The nature of the load-displacement curve (linear, nonlinear) The frequency of load repetitions or cycling The load history (cyclic load with constant or variable amplitude, random load, etc) 5. 6. 7. 8. 9. The size of the member The material flaws The grain size and microstructure The manufacturing method (surface roughness, notches). The localized surface irregularities (e.g. stamping) can be the point of high stress concentration 10. The operating temperatures (creep due to high temperature, brittleness due to low temperature) 11. The environmental operating conditions (corrosion, vacuum)

1.1.4 Study of the fractured surface (fractography) f f The information stored in specimen during fracture becomes the subject of study [22] even after the specimen undergoes fatigue failure. Sometimes, it is possible to determine the exact location of crack initiation and one can locate the source of crack initiation. A fatigue fracture surface is distinctive in appearance and consists of three regions; these are (i) a smooth portion, often possessing the nucleation site (ii) beach markings showing the growth of the fatigue crack up to the moment of final failure and (iii) the cleavage or shear which is the final fracture region.

11

The initiation and propagation phase normally generates a cracked surface that is fairly flat. In an electron microscope, it is possible to distinguish tiny ripples on the surface; each ripple actually corresponds a loading cycle very similar to the growth rings of a tree. A scanning electron micrograph depicting striations on the fractured surface of a fighter aircraft turbine blade that failed due to fatigue is shown in Fig.
-6 1.7. A crack growth rate of ~ 0.5 x 10-6 meter per loading cycle can be obtained from

the micrograph at high magnification. Furthermore, it is possible to determine seasonal differences, which reflect, for instance, change in the amplitude of loading. The final fracture zone can be smooth and shiny in case of brittle materials or may have a lustrous structure indicating ductile failure.

20 m

2 m

(a)

(b)

Fig. 1.7. Scanning electron micrographs depicting striations on the fractured turbine blade a) low magnification b) high magnification

Thick-walled 1.2 Thick-walled cylinders Thick

Thick-walled cylinders (TWC) in the form of boilers, gun barrels, nuclear reactors, pipelines and high-pressure containers are essential structural members for many industries. These include chemical, petroleum, nuclear, armament and food industries. The general function of these cylinders is to retain the processes, gas, fluids or

12

machinery inside them and isolate it from the surroundings. In many applications the cylinders are prone to cyclic stresses during their normal operation. The origin of these stresses may be the fluctuating pressures inside the pressure vessel, the pipe containing fluid, or instead, the vibrations of the rotating machinery inside them. Large internal pressures produce high tension hoop stresses along the inner surface of the cylinder; the latter may result in the nucleation of the internal surface cracks due to cyclic action of high-pressure pulses. Cracks nucleate at the surface of the bore, grow into arrays of longitudinal and/or radial cracks and reduce the strength of the structure resulting in premature failure at pressures which are even lower than the design capacity. Catastrophic service failures result in loss to human life and have significant impact on the economy. For this reason, proper material and precision in design is necessary to ensure integrity of the cylinder during the service life. Defects in the thick-walled cylinders may be generated during manufacturing or during the service; these defects are the potential site of crack initiation. In addition, during the normal operation, thick-walled cylinders may suffer from a number of degradation mechanisms including stress corrosion cracking, creep etc. Initiation of cracks can never be ruled out during the normal operation and must be considered while analyzing against fatigue failure. If the primary crack growth mechanism is slow, the cracks will be detected during routine maintenance by non-destructive testing (NDT) so that corrective measures can be taken before crack growth moves into a high risk regime. It is thus quite necessary to analyze, in detail, the crack propagation behavior in industrial applications.

13

1.3 Problem description and research strategy

1.3.1 Extruded cylinders In present work, the component being studied is a thick-walled extruded cylinder under fatigue loading. In industries of a wide variety, the thick-walled cylinders are the main critical components. The cylinders are manufactured through extrusion; the most popular extrusion procedure is based on splitting the ingot to be extruded in three or four segments, which subsequently get welded together upon passage through a specially designed die. The cylinder produced in the process has seams in the wall parallel to the extrusion direction. After the final heat treatment processes, the seams become an integral part of the cylinder. However, the probability of certain manufacturing defects may not be entirely ruled out; for instance lack of precise control of the processing parameters may result in incomplete welding of the segments creating a crack usually on the inner surface. The length of the crack so generated may be small or could extend the entire length of the cylinder as shown in Fig. 1.8. To avoid seams the double action extrusion process is used; see the schematic of the process in Annexure I. 1.3.2 Fatigue process in thick-walled cylinders A fatigue crack passes through three major stages and these are crack initiation, stable crack propagation and final rapid fracture. Normally, the first stage takes quite a long time but in a cylinder which already contains internal surface cracks, the first stage is assumed to be already completed and fatigue consideration in the design is of crucial importance. Under fatigue loading conditions the cracks present on the inner surface of the cylinder may grow outwards in the radial direction and result in the complete failure of the cylinder. Therefore a comprehensive study of these cylinders under the

14

conditions described above is indispensable. In this work the crack growth design problem associated with a thick-walled cylinder is being studied. 1.3.3 Definition of the problem being studied The component which is the subject of present study is an extruded cylinder of an aluminum alloy AA 6061 in T6 heat treatment condition. Repeated pressure applied to the cylinder from the inner surface produces cyclic stresses in the cylinder body. The objective of the work is to predict the life of the cylinder in the presence of fatigue loading with internal surface cracks along the cylinder axis. The problem is analyzed with the help of finite element method (FEM) using the commercial structural analysis ANSYS programme. The results may be used to avoid failure of cylinders under the conditions studied as well as to derive life extension strategies.

Axial crack

Pressure

Fig. 1.8 Section of the thick-walled cylinder with internal surface crack along the cylinder axis 1.3.4 Research strategy The first, rather difficult but important, step of a fatigue analysis is the accurate determination of the materials resistance to fatigue loading. Normally, a fatigue analyst first looks for the accurate fatigue data and more than often decides to create data relevant to his own research task [23]. In the present work, the research strategy

15

was based on the development of the fatigue data of the cylinder under investigation. Since the full-scale fatigue crack growth test of the thick-walled cylinder subjected to internal pressure involves a significant amount of time and cost, middle tension M(T) samples taken from the cylinder were alternatively used for the fatigue crack growth simulation of the actual cylinder [24]. Samples from two different orientations, representing the axial and the hoop stresses in the cylinder, were subjected to various stress levels and crack growth rates and fatigue life of the samples thus accomplished. The experimental work was replicated through modeling and simulation of the same geometry with the help of commercially available structural analysis software ANSYS. Research undertaken proceeded as follows: 1. Materials characterization including metallurgical aspects and mechanical testing under uniaxial tension 2. Experimental fatigue crack growth testing of the TWC and determination of the material fatigue properties using M(T) specimens f f 3. Numerical simulation of fatigue crack propagation using finite element analysis of the experimental geometries under fatigue conditions 4. Finite element analysis of the thick-walled cylinder under static loading and fatigue conditions using the experimental fatigue data and specified boundary conditions 5. Interpretation of the results 1.3.5 Research objectives Material defects due to manufacturing processes in the presence of aggressive environment under cyclic loadings can trigger the failure of the cylinder; this may subsequently result in safety concerns and expensive outages. This thesis deals with

16

the aspect of fatigue crack propagation in a TWC. The objective of this research work is to carry out theoretical and experimental investigations of an aluminum alloy thickwalled cylinder under fatigue loadings. The study is performed to comprehend the fatigue process, from the macroscopic stress-strain relations to microscopic fatigue crack growth mechanisms, predict the fatigue life and suggest measures to improve the fatigue properties. The research objective was achieved through material properties. characterization, cyclic testing, and numeric modeling.

1.4 Overview of the thesis

Chapter 1 Introduction covers the basic concepts of the fatigue process, introduction to thick-walled cylinders, the problem description and the objective of the present research. Chapter 2 provides the historical review of fatigue and a brief literature review relevant to this dissertation. Chapter 3 gives the details of the experimental work conducted during this study. This includes the equipments, standard methods, material characterization technique and the details of the tests to determine the monotonic tensile properties and fatigue f crack growth rate data of the alloy. Finally the detail of the fractographic study in scanning electron microscope (SEM) is provided. Chapter 4 Thick-Walled Cylinder; Material and Structural Characterization provides metallurgical and mechanical characterization of the TWC. The chapter presents the findings from the procedures employed (EDX analysis, light microscopy and scanning electron microscopy) to better understand the material and the understand manufacturing characteristics of the extruded TWC as well as their monotonic

17

properties. Post fracture analysis provides the micro-mechanism of the fracture process under monotonic loading. Chapter 5 Experimental Fatigue Crack Growth Study presents the results of the crack growth experiments of the TWC in two orientations. The fatigue life of the specimens and the crack growth rate data of the material at different stress levels are given. The fractographs showing the topographical variations with parametric change are also presented. Chapter 6 Numerical Simulation of Fatigue Crack Propagation presents the results of modeling and simulation of the experimental geometries under fatigue conditions using finite element analysis. Chapter 7 provides a detailed finite element analysis of the TWC under static and fatigue loading. Chapter 8 Summary and Conclusions presents the conclusions of the research study and based on the results recommendations for further research are suggested.

18

CHAPTER 2

Literature Review
Fatigue of materials involves a very complex interaction of different metallurgical, mechanical and technological factors and is still only partly understood [2]. These include the material intrinsic properties, thermal and mechanical processing, surface conditions, the geometry of the component and finally the cyclic loading conditions. Each of these factors includes a list of variables which significantly influences the mechanical properties of the material including the fatigue behavior. In the following sections, a brief literature review of the fatigue and finite element method is presented. The goal of the literature survey is to develop a practical methodology to analyze the fatigue crack propagation in thick-walled cylinders under fatigue loading. Different aspects related to fatigue can be best understood by the historical overview enlightening the evolution of the process with time.

2.1 Historical overview of fatigue

2.1.1 1800 Era The word "fatigue" was introduced in the 1840s to describe failures occurring from repeated stresses which were below yield stress. The first major impact of failures due to repeated stresses involved the railway industry in the 1840s. It was found that railroad axles failed regularly at shoulders [25] and immediately elimination of sharp corners was recommended. During 1850s and 1860s, August Wohler introduced the concept of the fatigue limit. He performed the first systematic laboratory investigation of fatigue tests under repeated

19

stresses. Wohler has been named the "father" of systematic fatigue testing. Using stress versus life (S-N) data, he showed how fatigue life decreased with higher stress amplitudes and that below a certain stress amplitude, the test specimens did not fracture. He pointed out that for fatigue, the range of stress is more important than the maximum stress [26]. During the 1870s and 1890s, different researchers substantiated and expanded Wohler's classical work. Gerber, among others, investigated the influence of mean stress, and Goodman proposed a simplified theory concerning mean stresses. Their names are still associated with diagrams involving alternating and mean stresses. Bauschinger [27] in 1886 showed that the yield strength in tension or compression was reduced after applying a load of the opposite sign that caused inelastic deformation. This was the first indication that a single reversal of inelastic strain could change the stress-strain behavior of metals. This finding was the forerunner of understanding cyclic softening and hardening of metals. 2.1.2 Early 1900 era In the early 1900s, Ewing and Humfrey [28] used an optical microscope to pursue the study of fatigue mechanisms. Localized slip lines and slip bands leading to the formation of micro cracks were observed. Basquin [29] in 1910 showed that alternating stress versus number of cycles to failure (S-N) in the finite life region could be represented as a log-log linear relationship. His equation, further modified by others, is currently being used to represent finite life fatigue behavior. In the 1920s, Gough and associates contributed greatly to the understanding of fatigue mechanisms. They discussed the combined effects of bending and torsion (multi-axial

20

fatigue). Gough published a comprehensive book on fatigue of metals in 1924 [30]. Moore and Kommers [31] published the first comprehensive American book on fatigue of metals in 1927. In 1920, Griffith [32] published the results of his theoretical calculations and experiments on brittle fracture using glass. He found that the strength of glass depended on the size of microscopic cracks. If S is the nominal stress at fracture and a is the crack size at fracture, the relation is Sa = constant. With this classical pioneering work on the importance of cracks, Griffith developed the basis for fracture mechanics. He thus became the "early father" of fracture mechanics. In 1924 Palmgren [33] developed a linear cumulative damage model for variable amplitude loading and established the use of the B10 fatigue life based upon statistical scatter for ball bearing design. McAdam [34], in the 1920s, performed extensive corrosion fatigue studies in which he showed significant degradation of fatigue resistance in various water solutions. This degradation was more pronounced in higher-strength steels. In 1929-30, Haigh [35] presented a rational explanation of the difference in the response of high tensile strength steel and of mild steel to fatigue when notches were present. He used the concepts of notch strain analysis and residual stresses, which were more fully developed later by others. During the 1930s, an important practical advance was achieved by the introduction of shot-peening in the automobile industry. Fatigue failures of springs and axles, which had been a very common problem, thereafter became rare. Almen [36] correctly explained the spectacular improvements by compressive residual stresses produced in the surface layers of peened parts, and promoted the use of peening and other processes that produce

21

beneficial residual stresses. Horger [37] showed that surface rolling could prevent the growth of cracks. In 1937 Neuber [38] introduced stress gradient effects at notches and the elementary block concept, which states that the average stress over a small volume at the root of the notch is more important than the peak stress at the notch. In 1939 Gassner [39] emphasized the importance of variable amplitude testing and promoted the use of an eight-step block loading spectrum for simulated testing. Block testing was prominent until closed-loop electrohydraulic test systems became available in the late 1950s and early 1960s. 2.1.3 Mid 1900 era During World War II, the deliberate use of compressive residual stresses became common in the design of aircraft engines and armored vehicles. Many brittle fractures in welded tankers and Liberty ships motivated substantial efforts and thinking concerning preexisting discontinuities or defects in the form of cracks and the influence of stress concentrations. Many of these brittle fractures started at square hatch corners or square cutouts and welds. Solutions included rounding and strengthening corners, adding riveted crack arresters, and placing greater emphasis on material properties. In 1945 Miner [40] formulated a linear cumulative fatigue damage criterion suggested by Palmgren [33] in 1924. This criterion is now recognized as the Palmgren-Miner linear damage rule. It has been used extensively in fatigue design and, despite its many shortcomings, remains an important tool in fatigue life predictions. The formation of the American Society for Testing and Materials (ASTM) Committee E09 on Fatigue in 1946, with Peterson as its first chairman, provided a forum for fatigue

22

testing standards and research. Peterson emphasized that the fatigue notch factor, K was a function of the theoretical stress concentration factor, Kt, the notch and component geometry, and the ultimate tensile strength [41]. In 1953, he published a comprehensive book on stress concentration factors [42] and an expanded version [43] in 1974. The Comet, the first jet-propelled passenger airplane, started service in May 1952 after more than 300 hours of flight tests. Four days after an inspection in January 1953, one of the Comets crashed into the Mediterranean Sea. After much of the wreckage had been recovered and exhaustive investigation and tests on components of the Comet had been made, it was concluded that the accident was caused by fatigue failure of the pressurized cabin. The small fatigue cracks originated from a corner of an opening in the fuselage. Subsequently, one more Comet aircraft failed catastrophically. By September 1953, a test section of the cabin had been pressurized 18000 times to 57 kPa in addition to 30 prior cycles between 70 and 110 kPa. The design stress for 57 kPa was 40 percent of the tensile strength of the aluminum alloy. Probably the first 30 high load levels induced sufficient residual stresses in the test section so as to falsely enhance the fatigue life of the test component and provide overconfidence. All Comet aircraft of this type were taken out of service, and additional attention was focused on airframe fatigue design. Shortly after this, the first emphasis on fail-safe rather than safe-life design for aircraft gathered momentum in the United States. This placed much more attention on maintenance and inspection. Major contributions to the subject of fatigue in the 1950s included the introduction of closed-loop servo hydraulic test systems, which allowed better simulation of load histories on specimens, components, and total mechanical systems. Electron microscopy

23

opened new horizons to a better understanding of basic fatigue mechanisms. Irwin' [44] introduced the stress intensity factor KI which has been accepted as the basis of linear elastic fracture mechanics (LEFM) and of fatigue crack growth life predictions. Irwin coined the term "fracture mechanics," and because of his many important contributions to the subject at this time, he is considered as the modern father of fracture mechanics." The Weibull distribution [45] provided both a two- and a three-parameter statistical distribution for probabilistic fatigue life testing and analysis. In the early 1960s, low-cycle strain-controlled fatigue behavior became prominent with the Manson-Coffin [46, 47] relationship between plastic strain amplitude and fatigue life. These ideas were promoted by Topper and Morrow [48, 49] and, along with the development of Neuber's rule [50] and rain flow counting by Matsuishi and Endo [51] in 1968, are the basis for current notch strain fatigue analysis. The formation of the ASTM's Special Committee on Fracture Testing of High-Strength Steels in the early 1960s was the starting point for the formation of ASTM Committee E-24 on Fracture Testing in 1964. This committee has contributed significantly to the field of fracture mechanics and fatigue crack growth and was combined with ASTM Committee E-09 in 1993 to form Committee E-08 on Fatigue and Fracture. Paris [52] in the early 1960s showed that the fatigue crack growth rate (da/dN) could best be described using the stress intensity factor range KI. In the late 1960s, the catastrophic crashes of F-111 aircraft were attributed to brittle fracture of members containing preexisting flaws. These failures, along with fatigue problems in other U.S. Air Force planes, laid the groundwork for the requirement to use fracture mechanics concepts in the B-1 bomber development program of the 1970s. This program included fatigue crack growth life considerations based on a pre established

24

detectable initial crack size. Schijves [53] in the early 1960s emphasized variable amplitude fatigue crack growth testing in aircraft, along with the importance of tensile overloads in the presence of cracks that can cause significant fatigue crack growth retardation. In 1967, the Point Pleasant Bridge at Point Pleasant, West Virginia, collapsed without warning. An extensive investigation [54] of the collapse showed that a cleavage fracture in an eye bar caused by the growth of a flaw to a critical size was responsible. The initial flaw was due to fatigue, stress corrosion cracking, and/ or corrosion fatigue. This failure has had a profound influence on subsequent design requirements established by the American Association of State and Highway and Transportation Officials (AASHTO). In 1970, Elber [55] demonstrated the importance of crack closure on fatigue crack growth. He developed a quantitative model showing that fatigue crack growth was controlled by an effective stress intensity factor range rather than an applied stress intensity factor range. The crack closure model is commonly used in current fatigue crack growth calculations. In 1970, Paris [56, 57] demonstrated that a threshold stress intensity factor could be obtained for which fatigue crack growth would not occur. During the 1970s, an international independent and cooperative effort formulated several standard load spectra for aircraft, offshore structures, and ground vehicle usage [58, 59]. These standard spectra have been used by many engineers in a variety of applications. In July 1974, the U.S. Air Force issued Mil A-83444, which defined damage tolerance requirements for the design of new military aircraft. The use of fracture mechanics as a tool for fatigue was thus thoroughly established through practice and regulations. This practice also emphasized the increased need for an improved quantitative, nondestructive

25

inspection capability as an integral part of the damage tolerance requirements. 2.1.4 End 1900 era During the 1980s and 1990s, many researchers were investigating the complex problem of in-phase and out-of-phase multi-axial fatigue. The critical plane method suggested by Brown and Miller [60] motivated a new philosophy concerning this problem, and many additional critical plane models were developed. The small crack problem was noted during this time, and many workers attempted to understand the behavior. The small crack problem was complex and important, since these cracks grew faster than longer cracks based upon the same driving force. Definitions became very confusing. Interest in the fatigue of electronic materials increased, along with significant research on thermomechanical fatigue. Composite materials based on polymer, metal, and ceramic matrices were being developed for many different industries. The largest accomplishments and usage involved polymer and metal matrix composites. These developments were strongly motivated by the aerospace industry but also involved other industries. During this time, many complex and expensive aircraft components designed using safe-life design concepts were routinely being retired with potential additional safe usage. This created the need to determine a retirement for cause policy. From a fatigue standpoint, this meant significant investigation and application of nondestructive inspection and fracture mechanics. In 1988 the nearly fatal accident of the Aloha Boeing 737, after more than 90000 flights, created tremendous concern over multi site damage (MSD) and improved maintenance and nondestructive inspection. Corrosion, corrosion fatigue, and inadequate inspection were heavy contributors to the MSD problem that existed in many different airplane types. Comprehensive investigations were undertaken to understand the problem

26

better and to determine how best to cope with it and resolve it. Also during the 1980 and 1990s, significant changes in many aspects of fatigue design were attributed to advances in computer technology. These included software for different fatigue life (durability) models and advances in the ability to simulate real loadings under variable amplitude conditions with specimens, components, or full-scale structures. This brought significantly more field testing into the laboratory. Integrated CAE involving dynamic simulation, finite element analysis, and life prediction/estimation models created the idea of restricting testing to component durability rather than using it for development. Increased digital prototyping with less testing has become a goal of twenty-first-century fatigue design. Additional readings on the history of fatigue can be found in [61, 62].

2.2 Fatigue crack propagation; effect of microstructure

The mechanical behavior of materials depends strongly on their microstructure. It is known that an aluminum alloy exhibits very different properties depending on whether it is cold rolled or heat treated under different temper conditions. Metal fatigue is also significantly influenced by microstructure [2]. This includes chemistry, heat treatment, cold working, grain size, anisotropy, inclusions, voids/porosity, delaminations and other discontinuities or imperfections. If the actual fatigue data are available, microstructural effects are inherently accounted for and, therefore, do not have to be accounted for again [2]. Inclusions act as stress concentration sites and are common locations for microcracks to nucleate under cyclic loading. Anisotropy caused by cold working gives increased fatigue resistance when loaded in the direction of the working than when loaded in the transverse direction. This is due to the elongated grain structure in the direction of the original cold working.

27

Investigations of the aluminum alloys under fatigue loading have been extensively done [63 - 83] and the microstructural features including the constituent particles are found to play an important role in the nucleation and fatigue crack growth process of these alloys. In a study of aluminum alloy AA 2026 extrusion bars, it was observed that reducing the density of constituent particles significantly enhanced the fatigue resistance of the alloy [84]. Zhu et. al. [85] have studied the effect of microstructure and temperature on fatigue behavior of a cast aluminum alloy and reported that the influence of microstructure on the fatigue resistance is greater than the influence of the testing temperature. In a study on aluminum alloy, Merati [86] pointed out that grain size and orientation could play crucial roles while determining the fatigue life. Generally, it is accepted that the microstucture is important in crack initiation, relatively unimportant in propagation where Young's modulus is the most important factor. Suresh et. al. [64] have highlighted the influence of the grain structure and slip characteristics in aluminum alloys. They pointed out that microstructural effects have a strong influence on fatigue crack growth rates near threshold. Fonte et al. [73] have studied the effect of microstructure and environment on fatigue crack growth resistance of aluminum alloys and concluded that these significantly affect the process of fatigue crack growth. An autofrettage process is commonly used in pressure vessels to enhance the fatigue life of the component. This process produces compressive tangential residual stresses near the bore of the pressure vessel. These residual stresses counteract the large tensile tangential stress at the inside surface of the pressure vessel subjected to a cyclic internal pressure.

28

The compressive tangential residual stresses near the inside surface due to the autofrettage process retard crack formation and growth. References [7-19, 73-89] cover some recent research work on fatigue crack growth process in metals. These research works include the effects of loading parameters, environment, surface treatment and microstructure on fatigue crack growth of different alloys.

2.3 Finite element method

Finite element analysis (FEA) is a useful and powerful technique for determining stresses and strains in structures or components too complex to analyze by strictly analytical methods. With this technique, the structure or component is broken down into many small pieces (finite number of elements) of various types, sizes and shapes. The elements are assumed to have a simplified pattern of deformation (linear or quadratic etc.) and are connected at "nodes" normally located at corners or edges of the elements. The elements are then assembled mathematically using basic rules of structural mechanics, i.e. equilibrium of forces and continuity of displacements, resulting in a large system of simultaneous equations. By solving these large simultaneous equations system with the help of a computer, the deformed shape of the structure or component under load may be obtained. Based on that, stresses and strains may be calculated [90]. The finite element method (FEM) is probably the most versatile way of calculating stress intensity factors. This method primarily involves the evaluation of displacements at nodal points of the body which has been idealized into a system of elements connected at the nodal points. The FEM has become a powerful tool for the numerical solution of a wide range of engineering problems. The FEM has been extensively used to solve problems

29

involving irregular regions and complicated cracks [91]. The application of FEM in the solid mechanics problems was first adjusted by Turner et al [92]. In 1961-1962 [93, 94], the extension of the FEM to three-dimensional problems occurred with the development of a tetrahedral stiffness matrix. In 1969-1970 [95, 96], the uses of finite element methods in fracture mechanics were initiated. In LEFM problems, the conventional constant elements and the higher order elements did not satisfy the singularity at the crack tip. In 1971, Tracy [97] made the earliest attempt on the development of special crack tip elements that directly model the 1/ r near the tip elastic strain field singularity in order to solve the LEFM problems successfully. From the early 1950s to the present, enormous advances have been made in the application of the FEM to solve complicated engineering problems. 2.3.1 Modeling through FEM The finite element method is a method of approximation, in which a discrete model is generated by a set of small continuous functions defined over a number of finite regions called elements. The FEM approximates any continuous quantity, such as displacement, temperature or pressure. When the FEM is used in fracture mechanics, the cracked body of interest is broken up into a collection of pre-selected finite elements which are connected at nodal points. According to the minimum potential energy principle, the global nodal forces [F] and the global nodal displacements [d] are related through use of the global stiffness matrix [K] as follows [98]: [F] = [K] [d] where [F] and [K] are assembled from the corresponding element matrix. Once the displacements [d] at the nodes are obtained, the field quantity of any point (2.1)

30