A history of fatigue

~ Pergamon Engineering Fracture Mechanics Vol. 54, No. 2, pp. 263-300, 1996 Copyright © 1996 Elsevier Science Ltd. 0013-7944(95)00178-6 Printed in Great Britain. All rights reserved 0013-7944/96 $15.00 + 0.00 A HISTORY OF FATIGUEt WALTER SCHLITZ IABG, D-85521 Ottobrunn, Germany Abstract--The history of fatigue from 1838 to the present is described in detail, with special emphasis on the German contribution in the time period of 1920-1945. A number of distinguished scientists and engineers, and their contributions to the further development of fatigue knowledge are specifically mentioned. Copyright © 1996 Elsevier Science Ltd. 1. INTRODUCTION MANY BOOKS and papers about fatigue start with a more or less detailed account of the historical development of this branch of technology; they are, however, mostly limited to the description of results. With the present paper, the author strives for an evaluation of the importance of scientists and engineers and their work for the further development of fatigue technology and knowledge. For this evalution, two criteria were established: • Were the results of the work useful for the following generations or not? A positive example would be, for example, the Palmgren-Miner rule, still being employed the world over, 50 respectively 71 yrs after its publication. A negative example would be the "Damage Line" of French, which has only caused confusion, or the "over-" and "understressing" works of Kommers, which uselessly haunted people's minds for decades. • Does the work in question only contain results or did the researcher also draw conclusions? A positive example would be W6hler's allowable stresses for railway axles in the finite life region. Since the author obviously knows the important German fatigue efforts better than those of foreign engineers, this paper possibly has an entirely undesired nationalistic German touch. On the other hand, practically all Anglo-American historical descriptions of fatigue give a biased account as well, because they do not mention the decisive German contribution in the period of 1925-1945. In that period, however, the foundations were laid for what we know today in fatigue and fracture mechanics. This will be discussed in detail in Section 6. Out of the large number of engineers and scientists who have worked on fatigue problems, because of limited space only three are described in detail, namely W6hler, Thum and Gassner, who all three fulfill the criteria mentioned above. Of course, many other names are mentioned, albeit briefly. Some border areas of fatigue, for example non-destructive inspection, are not discussed. Others, such as metallurgy or the development of fatigue testing machines, will be mentioned only briefly and where absolutely necessary. Some earlier papers on the history of fatigue [1-3] have made the author's work much easier. When studying the old works some interesting points attract attention: • Our predecessors were, in some respects, very modern. W6hler, for example, as early as 1860 suggested design for finite fatigue life [4]. In other respects, however, their opinions were astonishingly primitive and erroneous. • Knowledge about certain methods was highly developed in one location, while a few kilometers away it was nonexistent, for example on shot peening. • Decades after final clarification of certain problems, they continue being discussed over and over again in the literature and, what is more, they still haunt people's minds, for example the influence of testing frequency on fatigue life. "['The original German paper, published in Materialwissenschaft und Werkstofftechnik 24, 203-232 (1993), was dedicated to Prof. J. Schijve on the occasion of his 65th birthday. The present English version is slightly modified and modernized, and was translated by Dr J. Lincoln, WPAFB. 263 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 附注 去找参考文献1-3 Administrator 线条 Administrator 线条 Administrator 线条 264 WALTER SCHOTZ • It takes a long time, sometimes decades, until parameters known for ages are treated scientifically. Further decades pass before this scientific knowledge is applied to engineering practice. One example is the scatter of fatigue life, implicitly taken into account in 1860 by W6hler[4], first specifically mentioned in 1927 as "scatter", scientifically treated from 1945 onwards, but still not considered even today in some branches of industry. • Many well-known old papers treat additional important problems besides the subjects for which they are still remembered today. For example, W6hler measured the service stresses on railway car axles before his well-known fatigue experiments, and Palmgren in his paper of 1924 stated not only the linear damage accumulation but also presented a four-parameter equation for the entire SN curve and already specified a B10-1ife. • Some well-known fatigue scientists and engineers wrote only a few papers on the subject, others were extremely productive for decades; Miner only authored five papers, Palmgren only three, but Thum and coworkers no less than 574! • Fatigue failures of machines, plants and vehicles in service always led to significant efforts and advancements of the state of the art. This was as true for the work of Albert in the 1830s - - in the mines of the Oberharz the conveyor chains failed - - as for the work of W6hler and his predecessors in England - - when the axles of railway cars and locomotives broke. This is also valid for the German efforts of the 1930s - - here the cause was the fatal fatigue failure of a Dornier "Merkur" wing strut. The work of Thum was prompted by many fatigue failures of machine and vehicle components in service. After WW II, a number of aircraft crashes, notably of two "Comets", resulted in more aircraft fatigue problems and in 1969 the fatigue fracture of an F-111 wing led to a complete change of the structural specifications of the US Air Force, combined with an immense fracture mechanics programme that is continuing to this day. In more recent times (1988) the near-fatal accident to an Aloha Airlines Boeing 737 was the cause of renewed efforts and investigation activities into the structural integrity of old and poorly maintained aircraft. • In every time period there are one or more unrealistic ideas and solutions which at the time are followed enthusiastically by some distinguished scientists and engineers; with the benefit of hindsight, however, their delusions appear incredible! 2. 1837-1858. THE TIME BEFORE WOHLER The history of fatigue begins with Albert [5], who was a Royal Hannoverian "Oberbergrat" (civil servant for mines). In 1837 he published in Clausthal the first fatigue-test results known. For this purpose he constructed a test machine for the conveyor chains which had failed in service in the Clausthal mines. As early as that, he therefore tested actual components, not just the material! Since chains at the time could only be replaced by hemp rope which had to be imported at great cost, Albert invented the wire rope - - surely more important than those first fatigue tests. In 1842, Rankine [6], better known from thermodynamics by the "Rankine process", discussed the fatigue strength of railway axles. He suggested that these axles be forged with a hub of enlarged diameter and large radii, so that the grain flow would not be cut more than necessary by machining. York [7] conducted experiments with railway axles. In 1853 the Frenchman Morin in his book Resistance des Matbriaux [8] discussed reports of two engineers responsible for horse-drawn mail coaches. The replacement of the axles of the coaches was prescribed after 60 000 km, an early example of the "safe life" design approach. The axles of other mail coaches were to be inspected thoroughly after 70 000 km, cracks to be repaired by "fire-welding". It was noted that those cracks mainly occurred at section changes. The term "fatigue" was mentioned for the first time by the Englishman Braithwaite in 1854 [9], contrary to a widespread belief which ascribes it to Morin. Braithwaite, however, says [9] that a Mr Field coined the term. In his paper Braithwaite describes many service fatigue failures of brewery equipment, water pumps, propeller shafts, crankshafts, railway axles, levers, cranes, etc. Allowable stresses for fatigue-loaded components are also discussed. In this period many disastrous railroad accidents due to fatigue occurred; for example, on 5th October 1842, a locomotive axle broke at Versailles, claiming the lives of 60 people [10], about the same number as were lost in the two "Comet" crashes of 1954. In the history of the "Institution of Mechanical Engineers" in London of 1854 it is mentioned that a member had seen a collection Administrator 线条 A history of fatigue 265 of hundreds, if not thousands of failed railway axles. As late as 1887 English newspapers reported the "most serious railway accident of the week", and in many cases these were due to fatigue failures of axles, couplings and rails, and claimed many lives. More references on this period can be found in refs [11-16]. 3. 1858-1870: W¢)HLER W6hler, Royal "Obermaschinenmeister" of the "Niederschlesisch-Mfihrische" Railways in Frankfurt an der Oder, measured the service loads of railway axles with self-developed deflection gages as early as 1858 [17] and 1860 [4]. To the author's knowledge, this has not been noted before by the many authors describing W6hler's work. Specifically, this was accomplished for a number of four-wheeled and six-wheeled freight and passenger cars on trips between Breslau and Berlin as well as Frankfurt an der Oder and Berlin. The measurements were carried out for 22 000 km. The deflection of the axle was scratched on a zinc plate by a scriber by means of a compound lever system. Only the largest deflection per trip was measured. According to W6hler: "In order to know the force necessary for a certain deflection, the axle was bent by a dynamometer, which was fastened to the rims of the wheels". This means in our words that W6hler even then calibrated the forces acting on the axles. In the corresponding figures of his paper those forces are given. W6hler then discusses the largest axle deflection per trip and the corresponding service load, and calculates the bending and torsional stresses of the axle. He then compares the measured bending forces with those caused by the static axle load and arrives at a factor of 1.33; that is, in our present-day terminology, he determined an impact factor of 1.33. W6hler then draws the following conclusions from his measurements: "The number of such cycles per trip is considerably smaller than the number of miles the axle travels during its life. Therefore, the safety requirements are met if the axles can withstand the maximum stresses measured as many times as its expected life in miles. If we estimate the durability of the axles to be 200 000 miles with respect to wear of the journal bearings, it is therefore only necessary that it withstands 200 000 bending cycles of the magnitude measured without failure". Thus W6hler implicitly suggested design for finite fatigue life, taking into consideration even the scatter of fatigue life, or in other words, the probability of failure. Since no fatigue-test data were available to him at that early date, he estimates them and arrives at an allowable axle load for 200 000 cycles of 136 "Zentner" (ca 6800 kg). Beginning in 1860 [4, 18-21], W6hler published the results of fatigue tests with railway axles. Since the rotating-bending test machine he designed and built ran at a very low frequency, he designed new machines for carrying out axial-bending and torsion tests on different notched and unnotched specimens. In 1870 [20, 21, 21a] he presented a final report containing the following conclusions, often called "W6hler's laws": "Material can be induced to fail by many repetitions of stresses, all of which are lower than the static strength. The stress amplitudes are decisive for the destruction of the cohesion of the material. The maximum stress is of influence only in so far as the higher it is, the lower are the stress amplitudes which lead to failure". W6hler therefore stated the stress amplitudes to be the most important parameter for fatigue life, but a tensile mean stress also to have a detrimental influence. From his quantitative results he draws the following conclusions about this mean stress influence: "Components loaded in tension and compression like connecting rods, wheels, balances, etc. must be stronger by a factor of 9:5 than components loaded only in tension, like bridge members or roof beams. The springs of railway cars are loaded by small amplitudes, but high maximum stresses. Therefore the allowable maximum stresses can be much higher than usual. And indeed this is the case, with the spring steel of a loaded car reaching a stress of 180 hundredweight per square inch. Quite often this value even reaches 900 or 1000 hundredweight per square inch and with good steel that is safe. This is important for the smooth ride of railway cars". After a discussion of why a safety factor is necessary, W6hler comes back once more to finite life design: "It must be taken into consideration whether unlimited or limited life is required for the component. It follows that different components need different safety factors. In any case two such factors are necessary, one for the relation between the maximum stress in service and static strength, and the other for the allowable stress amplitude." Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 Administrator 线条 266 WALTER SCH~)TZ W6hler then suggests a safety factor of two for static strength and an additional one of two for fatigue strength. In his opinion this is adequate for all circumstances. These factors, however, are only valid for unnotched sections, because "the strength of joints in the form of riveted joints, keyed joints and such kind, and different shapes require special tests. The results of the tests with sharply notched specimens have proved the necessity of such special tests". Thus W6hler correctly does not present the additional safety factors for these joints, but requires special tests. The safety factors given above are only valid for infinite design life, because W6hler continues: "For components with finite fatigue life other considerations apply: if, for example, it is known that the maximum bending stresses on a railway car axle occur when traveling over switches, and if the number of such switches during the life of the axles is known, it is in accord with the requirements of safety that the allowable stresses in the axle are those which lead to failure after many millions of cycles, that is for iron 160 hundredweight per square inch and for cast steel up to 220 hundredweight per square inch". To Dr Fischer, Thyssen-Kassel Research Centre, the author is obliged for the relevant information that even today the allowable stresses for railway car axles are roughly fixed by the number of switches traversed. In another paper of 1870 [21] which to the author's knowledge has never been cited before, W6hler describes the dimensioning, design and material selection for railway ~r axles. According to W6hler "the axles were sized so that they should never fail in service according to experience. It was therefore all the more embarrassing when they failed in large numbers. The cause of these failures was not immediately apparent nor were means of preventing them available. Crystallization due to vibrations, by earth magnetism and other dark notions were resorted to before finally it was decided to believe that the axles had failed because they were too weak. When this was finally realized, the loads due to which the axles broke were soon found". W6hler then describes the forces acting on the axle in service, for example the static load, lateral loads due to cornering, wind pressure, etc. He calculates the service stresses via the measured loads and the axle diameter. By comparing these stresses with the result of his fatigue tests he concludes that axles are completely safe. Furthermore he describes the allowable axle loads according to the "Technical Regulations of the German Railways", which depend on the material, diameter, etc., and which also contain rules about the size of the radii between the axle and journal diameter. The "metallurgical size effect" was already taken into account at that time, i.e. the allowable stresses for thinner axles were higher than those for thicker axles, "because it was assumed that smaller dimensions allow the material to be worked better and therefore would result in higher fatigue strength". Even crack propagation is already mentioned: "Those seams which are not entirely superficial, especially those which are radial, propagate in service. This property of crack propagation in service is more pronounced the more carbon the material contains. Most striking in this respect is cast steel. The author has observed several times that fine, hardly visible longitudinal cracks in cast steel axles which only appeared to be forged-in seams, after several years in service, propagated into the axle for 20 mm and more, and made its replacement necessary". In summary one can only admire the work of W6hler in its entirety, encompassing the measurement of service loads, the calculation of the corresponding service stresses, the design for finite life including scatter (probability of survival) up to the observation of crack propagation and the quantitative suggestions for the decrease of the notch effect. The next known mention of design for finite life occurred about 75 yrs later (Almen and Boegehold [22] for differential gears, and Thum and Bautz [23] in a more general sense). The next service measurements known to the author date from the 1920s and 1930s, the first measured load spectra by Batson and Bradley [24], and Kloth and Stroppel [25, 26]. W6hler incidentally represented his test results in the form of tables, Only his successor Spangenberg [27-29] as director of the "Mechanisch-Technische-Versuchsanstalt" in Berlin plotted them as curves, although in the unusual form of linear abscissa and ordinate. The SN curves were called "W6hler curves" since 1936 [26]. Not until 1910 did the American Basquin [30] represent the finite life region of the "W6hler curve" in the form "log a, on the ordinate, log N on the abscissa" and describe it by the simple formula Administrator 线条 Administrator 线条 A history of fatigue 267 a, = CR", which is still used even today. In a large table Basquin gives some numerical values for C and n, based for the most part on W6hler's tests 50-60 yrs before! We can deduce from this that W6hler was well known to the research scientists of the following generations and that not very many new data had been obtained in the meantime. In 1867 the English technical journal Engineering commented on W6hler's exhibits at the Paris World Trade Fair: "Long after the many exhibits which have won medals at this fair will have been forgotten, the fundamental work of W6hler will be remembered" [31]. The Englishman Gough honoured W6hler in 1924 in his book, the first on fatigue [32], in the following way: "W6hler's work will survive the time as a monument to his genius as an engineer an