Design for High-Temperature Applications: Part Two

Fracture at Elevated Temperatures

The constant load creep rupture test is the basis for design data for both creep strength (minimum creep rate or time to a specific creep strain) and failure (time to rupture). The various ways in which such data are presented, correlated, and extrapolated are addressed in subsequent sections.

At this point, it is appropriate to consider the processes leading to fracture. Plastic instability in ductile materials has already been reviewed. This process may lead directly to fracture in pure metals and contribute significantly to fracture in engineering materials at moderately high stresses. However, of much greater concern are the processes leading to intergranular fracture with reduced ductility at low stresses and high temperatures. Here again, many of the basic studies have been conducted on pure metals and solid-solution alloys.

Crack Nucleation and Morphology. Two types of cracking have been identified: wedge-shaped cracks emanating from grain-boundary triple points and the formation of cavities or voids on grain-boundary facets often oriented perpendicular to the applied tensile stress. Although much work continues to model the nucleation and growth of these cracks and cavities, there are uncertainties in the mechanism of nucleation and in the identification of a failure criterion.

Another major problem is the effect of temperature and stress on the extent of cracking at failure. Most theories assume that failure occurs at some critical cavity distribution or crack size. However, it has been shown that the extent of cavitation at failure or at any given fraction of the failure life is very sensitive to the test conditions. Thus cavitation damage at failure at a high stress may be comparable to damage in the very early stage of a test at low stress. For stress-change experiments, there is therefore a loading sequence effect on rupture life, which is discussed later in this article, for engineering alloys.

Embrittlement Phenomena. As pointed out previously, rupture life is primarily a measure of creep strength; fracture resistance would be identified better with a separate measure that reflects the concern with embrittlement phenomena that may lead to component failure. Most engineering alloys lose ductility during high-temperature service. This has been shown to be a function of temperature and strain rate so that there is a critical regime for maximum embrittlement. At a fixed strain rate, for example, ductility first decreases with increasing temperature. This is believed to be caused by grain boundaries playing an increasing role in the deformation process leading to the nucleation of intergranular cracks. At still higher temperatures, processes of recovery and relaxation at local stress concentrations lead to an improvement in ductility.

Environmental Effects

It has long been known that test environment may affect creep-rupture behavior. Until quite recently, however, the work has been largely empirical with creep tests being conducted in various atmospheres and differences noted in creep rates and rupture lives.

The effect on rupture life, in particular, was often less than a factor of ten in environments such as oxygen, hydrogen, nitrogen, carbon dioxide, and impure helium compared with vacuum. In many cases, it was not clear how inert the vacuum was, and little account was taken of specimen thickness. Often, effects on ductility were not reported, and there were very few studies of crack propagation.

Embrittling Effects of Oxygen. At about the same time that the ideas on environmental attack at an intergranular crack tip were being developed, it was also shown that short-term prior exposure in air at high temperature (greater than about 900°C, or 1650°F) could lead to profound ernbrittlement at intermediate temperatures (700 to 800°C). This was shown to be caused by intergranular diffusion of oxygen that penetrated on the order of millimeters in a few hours at 1000°C (1830°F).

Combined Effects of Oxygen and Carbon. Of special interest relative to the previous discussion of creep cavitation is the reaction between diffusing oxygen and carbon. In nickel, it was found that if this reaction were prevented, creep cavitation could not develop during creep tests. Prevention was achieved either by removing the carbon (decarburizing) or by applying an environmental protective diffusion or overlay coating.

Effect of Other Gaseous Elements. Hydrogen, chlorine, and sulfur may also cause embrittlement as a result of penetration. Sulfur is particularly aggressive in that it diffuses more rapidly and embrittles more severely than does oxygen. It is also frequently found in coal gasification and oil-refining processes as well as industrial gas turbines operating on impure fuel.

Creep Rupture Data Presentation

Laboratory creep tests are typically run between 100 and 10,000 h, although a few are run for shorter times (for example, for acceptance tests), and occasionally some testing is conducted for longer times. Since most high-temperature components are expected to last ten years or more, service stresses are obviously lower than those used in the longest creep tests to generate data for most of the alloys used.

Therefore, to provide data for creep rates and rupture lives that are appropriate for the setting of design stresses, it became necessary to develop methods for extrapolation. Over the years, a tremendous amount of effort has gone into optimizing methods of data extrapolation.

One of the major considerations in such procedures must be statistical issues, such as the best estimate of the stress associated with a given median life or creep rate, the use of stress or time as the dependent variable in the data fitting, the treatment of variability among heats of the same alloy, and the analysis of data with run-outs. All of these issues have been treated with considerable rigor and shown to be important relative not only to the proper interpretation of data, but to the proper design of experiments. In addition, there are different practices among testing laboratories that may have appreciable effects on results. These include specimen geometry, loading procedure, specimen alignment, furnace type, and temperature control.

Despite all these concerns regarding proper statistical treatment of data, a methodology has been developed based on time-temperature parameters that are now in widespread use. The approach may be used to achieve the following major design objectives:

Damage Accumulation and Life Prediction

Engineering procedures for life management of operating components assume that the material is progressively degraded or damaged as creep strain increases and operating time accumulates.

Damage may be in the form of precipitate changes that may result in softening (overaging) and reduced creep strength, or embrittlement and reduced resistance to fracture. The embrittlement may be due to segregation of harmful species, either from the interior or from the external environment, to interfaces, especially grain boundaries. Damage may also occur as a result of progressive intergranular cavitation and cracking, as previously described. Some of this damage may be reversed by suitable heat treatment or by hot isostatic pressing and may allow the possibility of component rejuvenation.

There are two basic approaches to using the concept of damage accumulation for life assessment:

Conclusions

Time-dependent deformation and fracture of structural materials are among the most challenging engineering problems faced by materials engineers. The critical role played by high-temperature energy-conversion machines in modern society attests to the remarkable success of the design methodology developed during last decades.

However, modern design needs, including accelerated evaluation and development of advanced materials, and improved remaining life assessment methods for operating equipment have identified some ways in which the methodology might be improved. It is desirable to decouple the creep strength and fracture resistance criteria. This could lead to new accelerated short-time testing in which the objective is not to attempt to incorporate microstructural evolution and damage-development in the test, as in the traditional long-time creep-to-rupture test. Rather, the accelerated test may be used to measure separately the consequences of these changes on creep strength and fracture resistance.

The generally neglected anelastic or time-dependent recoverable component of creep may be the dominant strain component in many service situations at low stresses and needs to be incorporated in design analysis. This is true for ceramics and metals as well as polymers. It may also provide, in some cases, a critical link between deformation and fracture.