In the case of metals, both mechanical and thermal fatigue are types of mechanical fatigue. However, they normally take quite distinct forms.

Normal high cycle fatigue describes the failure by applied cyclic stress well within the elastic limit of the material (i.e. well below the yield or proof strength). The number of cycles to failure is often in the range of 104 to 108 cycles.

However, in the case of thermal fatigue, temperature cycling of one part of an engineering structure can quickly lead to strains which are up to ten times over the elastic limit, causing huge plastic straining on each cycle. This often occurs because the surrounding engineering structure is rigid, but the cyclically heated and cooled portion is localised within it or by it, and is mechanically constrained, prevented from expanding and contracting. The number of cycles to failure can be as little as 1 to 100, or occasionally more depending on the material, its strength, and the effective applied strain. A typical instance of thermal fatigue is experienced in an automotive cylinder head by the small bridge of metal between the exhaust ports of a 4 valve per cylinder. The thermally induced strain occurring by starting and stopping the engine induces strains of over 1 per cent (compared with about 0.1 per cent strain at the yield point). The development of a crack across this bridge allows water into the cylinder and the engine fails. Such failure in aluminium alloy heads is avoidable by ensuring that no oxide bifilms are present in the bridge to initiate the fracture. Thus up to 50% of racing car cylinder heads which were poured by gravity, entraining a large population of bifilm cracks because of the surface turbulence, failed on the test bed, compared to zero failures when filled by a counter-gravity process as in the now well-known history of the development of the Cosworth Casting Process (Qualtiy Castings; A personal history of the Cosworth Casting Process. John Campbell 2014. Aspect Design Publications; currently in publication.)

Practically both are different and can not be correlated. If you need further detailed discussions please be specific ion your question.

While there is clearly a difference between the thermal component which influences fatigue beaviour of materials and the influence of the mechanical component, it has to be said that the thermal component, especially at temperatures approaching that required to activate atomic diffusion/vacancy formation (typically 0.075Tm(K))in solid state systems, does have a direct influence on the contribution of the mechanical influence to fatigue behaviour. Atomic diffusion begins with the creation of different kinds of gradients (thermal, chemical, compostional) in materials and thus itdoes alter the mechanical behaviour of most engineering materials, including hot forming tools. For a start see https://dl.dropboxusercontent.com/u/6469685/Thermo-Mechanical-Fatigue_Made_Easy%5B1%5D.pdf, to help you better appreciate the definition of both influences.

Thank you very much. Those two explanations kinda resume what I was thinking, but it was hard to put together. You helped a lot.

Please also note that in polymers there is a clear distinction between thermal and mechanical fatigue: In thermal fatigue, cyclic loading causes a significant temperature increase of the specimen that reduces strength/stiffness and may lead to ultimate failure.

In the case of metals, both mechanical and thermal fatigue are types of mechanical fatigue. However, they normally take quite distinct forms.

Normal high cycle fatigue describes the failure by applied cyclic stress well within the elastic limit of the material (i.e. well below the yield or proof strength). The number of cycles to failure is often in the range of 104 to 108 cycles.

However, in the case of thermal fatigue, temperature cycling of one part of an engineering structure can quickly lead to strains which are up to ten times over the elastic limit, causing huge plastic straining on each cycle. This often occurs because the surrounding engineering structure is rigid, but the cyclically heated and cooled portion is localised within it or by it, and is mechanically constrained, prevented from expanding and contracting. The number of cycles to failure can be as little as 1 to 100, or occasionally more depending on the material, its strength, and the effective applied strain. A typical instance of thermal fatigue is experienced in an automotive cylinder head by the small bridge of metal between the exhaust ports of a 4 valve per cylinder. The thermally induced strain occurring by starting and stopping the engine induces strains of over 1 per cent (compared with about 0.1 per cent strain at the yield point). The development of a crack across this bridge allows water into the cylinder and the engine fails. Such failure in aluminium alloy heads is avoidable by ensuring that no oxide bifilms are present in the bridge to initiate the fracture. Thus up to 50% of racing car cylinder heads which were poured by gravity, entraining a large population of bifilm cracks because of the surface turbulence, failed on the test bed, compared to zero failures when filled by a counter-gravity process as in the now well-known history of the development of the Cosworth Casting Process (Qualtiy Castings; A personal history of the Cosworth Casting Process. John Campbell 2014. Aspect Design Publications; currently in publication.)

Thermal fatigue on its own needs considerable cycling to obtain results. You can accelerate time to cracking by applying mechanical loads. You can check for more information in http://www.sciencedirect.com/science/article/pii/S1877705811003031,

http://www.doaj.org/doaj?func=fulltext&aId=1062435 and http://library.witpress.com/pages/PaperInfo.asp?PaperID=17968.

It is clear that the term 'thermal fatigue' is surrounded by confusion, almost certainly as a result of the term being at best unhelpful, and at worst, redundant. It would be best to avoid using the term altogether.

The link quoted by Brano Okolo in the first exchange of this series, "Thermomechanical Fatigue made Easy" by Darrell and Benjamin Socie is probably the most clear and complete statement defining the various forms of fatigue. However, I think most readers will find their 'Easy' explanation difficult.

In plainer language, it is worth spelling out some red-herrings: of course any metal suffering repeated strain as a result of repeated stress will heat up somewhat. However, in practice this is usually negligible as a result of the good conductivity of metals which ensures that the resulting temperature differences are small, and hardly affect the applied cycling strain.

Another unnecessary confusion arises from mixing up the concepts of fatigue at high temperature, which is necessarily complicated by the simultaneous effects of creep, together with dislocation and vacancy effects.

As I mentioned earlier, both so-called thermal and conventional fatigue are names to describe mechanical cyclic straining. Both can occur at low or high temperatures.

'Conventional' fatigue results from the cyclically applied mechanical strains, usually as part of a test procedure to evaluate the life of a component when subjected to design stresses below, usually well below, the elastic limit of the material. In most fairly clean metals, failure usually requires thousands or millions of cycles, earning it the name of 'low stress high cycle fatigue'.

'Thermal' fatigue is a mechanically cyclic straining condition which arises in a location which experiences cyclic temperature changes but whose thermal expansion is constrained mechanically. In the example of the bridge between exhaust ports of the 4 valve cylinder head, no additional applied strain is necessary to promote eventual fracture (in any case the mechanically applied strains are negligible compared to the thermally induced strains which commonly extend well into the plastic region) and failure has been commonly observed to occur in the first cycle, or sometimes a few hundred or a few thousand cycles. This condition is commonly known as 'high strain low cycle fatigue'.

Despite the high strain, in fact no failures are observed if the bridge is free from the oxide cracks which are often found in metals poured turbulently during casting. The improved casting conditions of modern 4 valve per cylinder engines is the reason why although thermal fatigue conditions in modern engines, especially diesels, are even more severe, failure no longer occurs as a result of the freedom from casting defects (bifilms in particular).

Mechanical fatigue data should allow one to rank the sucsceptability os a series of alloys, but it seems to me that it is impossible to exactly match the stress-strain cycle of a thermal transient with a mechanical one.

Thermal conductivity would limit the frequency range you can study with a mechanical test.

The strains one can develop with differential thermal expansion is quite limited.

P.G. Shewmon

Sorry for entering this discussion so late, but I was on summer vacation. I suggested Ana Paola to contact ResearchGate with this question, hoping to generate this kind of discussion. I agree with John Campbell that one should not mix thermal fatigue with high temperature fatigue (creep fatigue) and that, at least for low temperatures, where diffusion is negligible, the main driving forces for thermal fatigue should be equivalent, but my observations of parts which suffered thermal fatigue (typically large steel ingot molds made of cast iron and rolling cylinders made of high chromium white cast iron, in the skin) show two distinct aspects of thermal fatigue: 1- there is a visible difference in height in the different thermal fatigue cells which form the crack network, which points to me that, somehow, mode II loading play a part in thermal fatigue and 2- in most cases these are non propagating cracks, they stop before getting critical. This second aspect I believe can be justified by the decreased stiffness of the surface layers due to the formation of the crack network, but the first one is a puzzle to me, does anyone have an idea?

Based on my understanding, fatigue is classified as:  

1. Mechanical fatigue includes high cycle fatigue and low cycle fatigue which is distinguished by amplitude of structural load and whether plastic strain occurs.

2. Thermal fatigue or thermal-mechanical fatigue (TMF) includes thermal stress induced  fatigue, creep and oxidation. Thermal stress induced fatigue is theoretically similar as low cycle fatigue and the load is caused by difference of thermal expansion.

Ade's straightforward definition seems to me to be excellent.

However, Claudio poses questions which do not seem so easily answered. What is certain is that both his ingot moulds and cast rolls will both have been cast rather turbulently, and so will contain a large population of bifilm cracks as a result of the folding-in of the surface oxide film during the pouring of these castings. Many of these cracks (which are present from the casting process, and have not originated as a result of any stress) will be near the stressed surface. Bifilm cracks are characterised by being extremely thin, and so easily overlooked, but their area varies widely as a result of their randomly turbulent formation, perhaps from only a few micrometres to several centimetres. Only relatively few will be favourably oriented in the applied stress regime so as to develop into fatigue cracks which will eventually lead to macroscopic failure. The majority of the cracks would remain unchanged, and so appear to be non-propagating.

For the future, it is not necessary to pour castings turbulently, so these huge populations of casting defects need not be made. Good designs of filling systems for liquid metals (usually costing little or nothing) can avoid the entrainment of defects, creating essentially defect-free components. The fatigue lives (whether thermal or other varieties of fatigue) then become very large (possibly infinite?) because the crack initiators have been removed.

Evidence for these conclusions is summarised in my "Complete Casting Handbook 2015", which also gives details of effective casting techniques to achieve these results.

Prof. Campbell, I appreciate very much your effort to contribute to this discussion, which is until today a concern, cause I'm writing my PhD thesis. I'd like to ask what you think about the changes in mechanical and thermal properties of the material along the different regions, because of the thermal gradient caused by insufficient heat transfer. How does it affect the stresses? Should it be considered in component design?

Dear Ana Paola

I appreciate your concern. All I can contribute to the discussion is my own experience, which I like to think is in line with common sense.

In the case of high cycle fatigue, the stress is always well below the yield stress, and the lifetimes are correspondingly long.  Temperature is of course, an additional independent variable; the fatigue process can be carried out at low or high temperatures.  If the temperature becomes especially high, then a certain amount of creep strain starts to contribute towards the eventual failure process. However, all this is clear, and is well known.

Turning now to thermal fatigue, it is common for thermal effects (actually temperature differences) in components to introduce strains of well over the yield strain, resulting in stresses well over the yield stress. Strains of the order of 1% are common, which is approximately 10 times over the yield strain.  The thermal effects could be distant in the component, so that the temperature of the thermally fatigued portion of the component need not be significantly changed.  Thus the thermal fatigue process could happen at any temperature, the 'thermal' action being distant and completely unrelated.  For this reason 'thermal fatigue' is rather unhelpfully named. It should really be called "low cycle fatigue" since, in principle, the process can occur independent of the action of temperature.  I have given above an extreme example to make a point, but the examples I have seen in industry are often rather similar to this.

In those cases where so-called thermal fatigue is carried out at high temperature, as in the case of high cycle fatigue, "creep processes start to creep in".

Failure mode is separate feature which tends to distinguish thermal fatigue. Because of the high strains, well into the plastic flow regime, ductile shearing, leading to 'necking' between major cracks or pores, leads to rapid failure; lifetime cycles of between 10 and 100 are common. The familiar 'beach marks' denoting the stepwise progress of the crack are usually absent.

In all cases of fatigue I have experienced, both high and low cycle fatigue, the initiators of failure have been, in every case, bifilms.  Quite often, the majority of the fracture surface is seen to be composed of a single large bifilm - it is not surprising that such material was not able to withstand the application of strains above the yield point.  The 50% failure rates of racing car cylinder heads, mentioned in one of the earlier missives, dropped to zero overnight when making the castings by an improved filling process which eliminated bifilms. The thermal fatigue continued to operated on the overheated bridge between the exhaust values, but failure never occurred.

I hope you will find these notes helpful.

With best wishes for your research.

John Campbell

 Dear Ana Paola

Professor John Campbell has given you the full am\nsver to your question. I have more anything to add.

Good lick to in researches

 Yours Faithfully Tamaz Eterashvili

First of all, I wish you a happy new year for everybody!

In addition to your clever answers, I just want to say that for composites (with polymer matrix) and polymer materials, thermal fatigue and mechanical fatigue are very different. Polymer materials are very sensitive to temperature and strain rate as you know. Their thermal conductivity is also very bad. That induces specific phenomena which do not appear in the case of metallic materials.

These specific behaviors were studied in the past, not so often in my point of view, and should be improved. The understanding of these phenomena induced mainly useful phenomenological laws, but are quite poor.

Thank you Prof. Campbell for net a clear information

Komplex answer for Your question is in my book "Preseeure equipment in European Union" (Alfa Košice 2014) or in my paper"Thermal stresses ..."

Conference Tlak 2017 - Hradec Králove - CzR .

This question was discussed 2 years ago. Still there are ambiguities?

Dear Ana exactly its not equivalent

Not equivalent.

Thermal fatigue is influenced by temperature.

Mechanical fatigue is influenced by the number of repeated cycles during loading.