The Nature of Fatigue

Abstract

Abstract While the fundamental nature of fatigue is not well understood, many of the phenomenological aspects of the problem have been sorted out. A general outline of the problem, methods of presenting data, types of fatigue and the factors involved are discussed. The fundamentals of the problem and what is known about fatigue today are outlined, and the implications of this knowledge in terms of possible treatments, inspection, etc., particularly as related to the drilling industry, are presented. Introduction Fatigue is a strange and exceedingly dangerous problem. Fatigue failure occurs with little or no warning under repeated applications of a load which the metal would support indefinitely if the load were applied statically. The problem was first recognized in 1850, but it has become progressively more prevalent as technology has developed machines and equipment subjected to repeated loading and vibration. Today fatigue accounts for about 90 per cent of all service failures due to mechanical causes. Causes of Fatigue Failure Much has been written about fatigue and a great many people have studied and continue to study it, using the most modern techniques and ingenious experiments. While a great deal is known about the phenomenon, we do not understand the basic mechanism or nature of the problem. Three basic factors are, however, necessary to cause fatigue failure. These are: a maximum tensile stress of a sufficiently high value; a large enough fluctuation in the value of the applied stress; and a sufficiently large number of cycles of the fluctuating applied stress. While these are sufficient to cause fatigue, there are a host of other variables which alter the conditions for fatigue, such as temperature, crystal system of the metal, grain size, environment (corrosive or otherwise), metallurgical structure, stress system, etc. Fluctuating StressThe general types of fluctuating stresses which can cause fatigue are illustrated in Fig. 1. The first represents an idealized situation wherein the net resultant stress is zero, fluctuating in a sinusoidal fashion from tensile to compressive. This is the most common form used to study fatigue in a laboratory, but is also approached in service by a rotating shaft operating at a constant speed without overload. The second shows the sinusoidal stress form but here the resultant (or mean) stress is not zero. The third shows an irregular or random stress cycle of the type most frequently encountered in actual service, and of course is of the type found in drilling operations. S-N Curves The most common method of presenting fatigue data is by means of the stress-number (S-N) curve, which relates the number of cyclic stresses imposed on a specimen to failure with the maximum stress applied. These data are generally obtained by imposing a sinusoidal stress pattern with a net resultant (or mean) stress of zero as shown in Fig. 1.Fig. 2 shows a typical S-N curve for a ferrous and nonferrous material. The fatigue life of the nonferrous material is very short for high stresses, but this life becomes increasingly longer as the magnitude of the maximum stress is reduced. For the case of the ferrous metals, a plateau is reached when the stress is reduced below a certain level. This means that for a stress equal to or less than the plateau value, the specimen will support an infinite number of cycles without failure. This plateau value is the maximum stress for infinite life, and is called the fatigue or endurance limit. Most nonferrous metals such as aluminum, magnesium and copper alloys do not have a true fatigue limit because the S-N curve never becomes horizontal and they will eventually fail at any value of an applied cyclic load. JPT P. 869ˆ