Perhaps one should first define what one means by plastic instabilities. As long as we stick to tension, there is no confusion, because the manifestation of instability (necking) is both easily observed and well-defined mathematically.

On the other hand in a test like torsion, it is difficult to establish any geometric instability since the geometry is in principle invariant. Failure simply occurs when a certain condition is reached (consider for instance the max. shear stress criterion); do we consider this point as a plastic instability?

Going further to tests such as compression, where the geometry does change, but in a beneficial manner, making failure less likely. 'Instabilities' in this case would arise from the experimental setup, non-conservation factors like friction, and so on. These are very indirectly related to plasticity.

So at a macroscopic level, the definition is difficult.

At a microscopic level, we may observe certain manifestations; such as flow localisation. People have put forward instability criteria for such cases. These include empirical criteria (eg. Semiatin) as well as criteria developed on energy input/output considerations (eg. rate of energy storage to rate of energy dissipation).But the question again is, at what point do we call it an instability? Stage-I fatigue crack growth is governed by plasticity considerations and Stage-II is governed by fracture mechanics -which of these two is better described as a plastic instability? On one hand it should be Stage-I, because that is 'plastic', on the other hand Stage-II is a more evident 'instability'.

I agree with Ashranth answer

Also these links may help you

http://www.hrpub.org/download/20170130/EEE2-14508434.pdf

http://www.kstreetstudio.com/science/natedwkshp/files/Meier-NEW-2001-1.pdf

Plastic instability is the result of crack initiation in RCF cycle. It depend on the materials and it is for sure will affect the stress cycle. As plastic zone increases the mechanical properties will change during this fatigue cycle.

Thank you for three good answers.

For Aashranth I should have been specific that when I referred to torsion I was thinking of a thin walled tube, which would buckle eventually. Even so in torsion of a solid bar once a shear crack forms plastic instability might affect the progress of the crack.

I also presume we should differentiate between plastic instability on a macroscopic scale (across many grains in a metallic material) and plastic instability on a microscopic scale. This might help differentiate between necking/buckling and the plastic instabilities which might occur during fatigue cracking. I wonder whether to understand plastic instability at a microscopic scale whether we would need to consider crystal plasticity modelling?

Or is it more simple than that? Could it be the case that plasticity at a microscopic scale, i.e. when some grains are deforming plastically and others in a structure are still elastic then this is not an instability. However, once all(or a critical number of) grains are deforming plastically across a section then instability in the structure occurs?

Is there anything we can learn from superplasticity?

Sorry for thinking out loud.

Hello Michael! Hello everyone!

Very interesting topic for discussion.

In my opinion all the standard material characteristics, which are being obtained from mechanical testing are macrocharacteristics, therefore they are affected by plastic or any other instabilities on microlevel. We can not avoid this in any other way, but by using characteristics, that describe material on microlevel. For instance local strain is very good one in my experience. By “local” I mean strain, measured using small enough measurement base (or spacing) in range from 5 to 50 micrometers. In this paper [Article Determination of the Local Strains Near Stress Concentrators...

Article STRAIN ENERGY DENSITY AS THE LINK BETWEEN GLOBAL AND LOCAL A...

What I trying to say is that standard material characteristic is really outdated. Lots of errors in engineering calculation using FEM are made by using standard characteristics, especially in nonlinear solutions. The new characteristics, invariant to specimen geometry should be developed and standardized.

I think if you use some coating such as TiN or WC it will great to avoid superplasticicy in.RCF application or highly compressive pressure application. We found too using FEM simulator that we can reduce Von Mises stresses and decrease the rate of wear.

I think there are two important things to consider:

- Is my test used to make relative comparisons or as indirect input (e.g. by afterwards using/applying semi-empirical data/approaches, for instance when designing for fatigue), or is my test used to determine absolute physical properties? Example is for instance a fracture toughness test: localisation will definitely influence the outcome, but we only look at the macroscopic behaviour in very-well defined (as in constrained) test conditions, so we always will have similar results.

- If my test is used to determine absolute physical properties, is their value based on a macroscopic or microscopic approach? The former will typically assume some sort of homogeneous state, and then localisation will lead to calculated values with large deviations from the real values. If on the other hand, the properties are measured on the same or smaller scale compared to the localisation phenomenon, then you should find the real physical value.

The latter issue can be nicely illustrated with FEM. If I simulate a tensile test using a single cell mesh, I would have to put in the macroscopic failure strain in order to get failure at the same point as during the experiment. If on the other hand I am simulating the material with an extremely fine mesh, I will require to put in the maximum local fracture strain (as obtained by doing an analysis of the necking region) to find the same result.

Thank you Frederik

That is a very informative answer.

Putting plastic instabilities to one side for a few seconds:

If I measure Elastic Modulus for a metallic material, on a polycrystalline, isotropic bar with dimensions of 10s of mm long and a few mm diameter. I can using dynamic methods measure a “Physical property” of that metallic material (and avoid any problems with plastic instabilities).

However, this does not mean I would know the Elastic moduli in all crystallographic orientations for each of the individual grains, which I would need to understand the physical properties of a single crystal.

So we understand, from these examples, that the “Mechanical Properties” and even some “Physical Properties” which are to be used in a FEM analysis need to be appropriate to the length scale of the component being analysed.

This makes predictions of real world plant components a little more complex than simply making sure that the boundary conditions of our “Mechanical Properties” and the boundary conditions of component being analysed by FEM are similar.

FEM WILL SHOW GREATLY THE INFLUENCE OF PLATICITY ON STRESSES AND SPECIALLY IN RCF APPLICATIONS

Perhaps we need to distinguish between:

For example, in a tensile test on a metallic material it is the effects of plasticity which give rise to the yield properties, the proof stresses and the work hardening rate. Indeed, I would suggest the main reason for doing a tensile test is to measure the effects of plasticity. However, it is plastic instability which controls the maximum load (which is then used to estimate the tensile strength sometimes called ultimate tensile strength). Plastic instability also gives rise to necking and which means that the elongation at failure is not independent of specimen geometry. In addition, we know from the use of high stiffness testing machines that plastic instability is sensitive to the test machine! Which means that the elongations and reductions of area measured on different test machines can be different! Again the effects of plastic instability.

I would suggest that tests such as rolling contact fatigue would be in the bracket of tests, which are affected by plasticity. However, I don’t think these tests are affected by “plastic instability”. I am happy to be corrected.

I will give another example a uniaxial fatigue test may well be affected by plasticity: In a high cycle fatigue test very localised microscopic plasticity such as at a pore, inclusion or surface flaw. Or in a low cycle fatigue test (preferably strain controlled) the macroscopic plasticity is the dominant factor in the test result. Up until the crack has reached a certain size, I would suggest that “plastic instability” does not play any role on HCF or LCF crack nucleation, although “plasticity” is a controlling factor in crack nucleation. However, once the crack has reached a certain size “plastic instability” could then become a significant factor in the remainder of the test. For example in a fatigue test which is taken into compression (such as an LCF test with R

crack initiation and crack propagation is a major issue here in addition to wear

Plastic instability should effect three point bending test also. Further plastic instablity must be changing with ending speed also

Stress raiders happen when you go from elastic to plastic and you have crack initiation in toughmaterials under node I and II

In the case of a tensile test, it is the microscopic behaviour that leads to strain hardening. It occurs regardless of the state of stress (eg. compression, tension, torsion etc.). Instability in tension is due to the specific geometry of the test: a reduction in load carrying cross section of the specimen during deformation. In torsion tests instability can arise due to adiabatic heating of the specimen. Under compression the formation of adiabatic shear bands lead to instability. Bucking can also be considered to be an instability. In all these cases mechanical instability is a manifestation of bifurcation - two possible solutions to the problem for a given set of boundary conditions and material properties. In other words instability is tied to how we do the experiments. If we are doing an experiment to measure a particular property then it may be a good idea to make sure that both the stimulus and the measured response are homogeneous.

Instability also depend on the materials. If the material is brittle micro cracks will affect the stress greatly when compared to other materials or other properties

Those interested in the effect of plastic instabilities on tensile tests should read the paper by F.A. Nichols, "Plastic instabilities and uniaxial tensile ductilities", Acta Metallurgica, 28 (1980), 663-673. Nichols makes reference to the classical Considere (1885) and Hart (1967) papers on the theory of the tensile test. Nichols' paper will answer your questions regarding the several measures of tensile ductility; total elongation, uniform strain, necking strain and reduction in area. Nichols shows that each of these is dependent on strain hardening and strain rate sensitivity of the tensile flow stress and the method used to conduct the test (constant loading rate, constant strain rate). W.J. Duffin developed comparable instability analyses for groove and patch like defects in multiaxial stress fields. Unfortunately I can't find a reference for this paper.

Plastic instability variation depends greatly on the materials and coating used during the process. For example if we coat M50 steel with TiN we reduce plastic instability greatly due to the safety envelope developed.

Dear Meryl

Thank you for adding your answer to this question, your answer is exactely the response I was looking for, indeed when I saw it elsewhere I was inspired to ask this question. Are there any similar analyses for other tests? For example can the same analysis be applied to a constant load creep test?

Dear Gabi

You mention TiN coatings. Nevertheless, does the same apply to any hard surface layer, for example a nitrided steel?

From my experience it should apply. Also try WC FOR THE SAME PURPOSE

Michael - See Y. Estrin and H. Mecking, "A unified phenomenological description of work hardening and creep based on a one-parameter model" Acta. Metall. (1984) 55-70. See also the first 4 references provided in this paper and search forward on Estrin's papers. Is not creep just a slow growing plastic instability?

In the latest experiments and FEM model we found that induced microcracks will create huge plastic instability. Coating such as TiN and WC will inhibit such instability by totally and uniformally distributing the stress. Thus creating safety envelope especially in M50 steel

Gabi and Meryl

Yes Meryl

I agree "creep is just a slow growing plastic instability" this is exactly where I am heading with this question. It is no surprising then that our design methods for creep can be a long way off the mark. Nevertheless, if we work with the problem rather than against it then I think we can develop reliable design methods.

The issue being that laboratory mechanical property tests can be highly influenced by plastic instability. Whereas real plant components might respond very differently to plastic instability. We need to understand this in order to use laboratory test data to predict real plant components.

For example, engineers often use the term "constraint" to hand wave away difficult problems in complex components and those involving multiaxial states of stress. However, perhaps all that so called "constraint" means is that that plastic instability is being inhibited and brittle failure mechanisms are being helped. For example creep cavitation and intergranular cracking.

I have mentioned creep nevertheless, the same issues are present in fracture mechanics also. There are geometries and loadings in which plastic instability actually dominates fracture and others where LEFM dominate fracture.

In fatigue also both plasticity and plastic instability affect fatigue such as Gabi has been mentioning RCF. Fatigue has the further complexity that we need to think of both crack nucleation and crack growth. Gabi has also contributed regarding the effect of surfaces; clearly, very important for both plasticity and plastic instability. Not least because plastic instabilities commonly start at surfaces.

Good discussion, this is really helping get some complex problems clearer in my mind.

Is there anything more that can be said about fracture toughness tests and plastic instability?

If you have concentrated Hertzian pressure then fracture toughness will play an important role

The papers (2) published by W.A. Duffin use a perturbation analysis to model the plastic instability and growth of two-dimensional "groves" and "patches". If someone can find these papers I would be grateful. The Duffin and Nichols papers provide a basis for discussing plastic instabilities and the relationships to measures of fracture strain.

See also the paper by F.A. Nichols on "How brittle is brittle fracture", Eng. Fract. Mech. 12 (1979) 307-316. In this paper Fred, a materials engineer/scientist, is trying to organize his understanding of "brittle" and "ductile" fracture during a time when "fracture mechanics" was undergoing an explosion of development and being forced on material and mechanical engineers involved in materials testing, selection and component design.

Relative to fracture toughness tests and plastic instability one needs to read papers on the theroy of the J-R "ductile tearing test". See J. R. Rice, W. J. Drugan and T-L. Sham "Elastic-Plastic Analysis of Growing Cracks" in: Fracture Mechanics: Twelfth Conference, ASTM STP 700, American Society for Testing and Materials, 1980, pp. 189-221. Also see Graham G. Chell and Ian Milne, "Ductile Tearing Instability Analysis: A Comparison of Available Techniques", Elastic-Plastic Fracture: Second Symposium, Volume II--Fracture Resistance Curves and Engineering Applications, ASTM STP 803, C. F. Shih and J. P. Gudas, Eds., American Society for Testing and Materials, 1983, pp. II- 179-II-205.

Note again that "stiffness" of the load train has a significant effect on plastic instabilities of both the smooth surface tensile test and cracked body fracture toughness (J-R curve) test. Most test labs use stiff test machines in displacement control to minimize machine stiffnes effects on fracture toughness test results. This is in contast to the conventional dead weight creep test where the load train is fully compliant. Note neither of these test strategies may represent the stiffness of an engineered component.

Also potentially having a major effect are the loading mode and rate (constant loading rate, constant displacement rate and constant strain rate). Stress and strain states can have a major effect on plastic instabilities and measures of ductility; for example, strain to fracture in a plane strain tensile test is usually much lower than that in the conventional uniaxial tensile test. Plane stress toughness can be significantly greater than plane strain toughness in a conventional cracked body fracture toughness test.

In a previous comment I stated that engineered components are limited by design to stresses below yield. I should have said "primary" stresses are limited to below yield. A primary stress is due to "active" "load controlled" mechanical loads and satisfy force and moment equilibrium. Primary stress that exceeds the yield stress by some margin will result in failure ("plastic colapse)". "Secondary" stresses are those arising from geometric discontinuities (stress concentrations), local thermal "hot spots" and residual stresses due to component manufacture and installation. Secondary stresses are termed self-limiting by the ASME code: that is, once the yield point has been passed locally around the stress concentration, the direct relationship between load and stress is broken, due to the reduced post-yield stiffness of the material. Good design limits these stresses to a fraction of the cross section and requires that they "shake down" to "elastic action". Of course this does not necessarily prevent initiation and propagation of fatigue, corrosion fatigue and stress corrosion cracks.

When you have plastic instability concentrated stress coming from Hertzian pressure create a state of normal and shear that can not be predicted by FEM model sometimes specially if we have point contact. Thus we should rely on some experimentations and microscopy to verify these.

Dear Meryl

Your answer is 10/10. You are a true expert in this field and you have made this question by your answers. I only wish we could have worked together.

I trust that my platitudes are not too clumsy my hope is that by asking these questions in RG and getting high quality answers that we capture knowledge and understanding plus those old references, which would otherwise become forgotten.

In some years to come a keen researcher will search RG for some answers to their problems and they will find your answers, which will greatly help their work.

Best regards

Mike

Thank you much and I appreciate the discussion

Gabi

You mentioned "Hertzian pressure create a state of normal and shear that can not be predicted by FEM", why is this? With a realistic inelastic materials models and correct mesh refinement I would have hoped that FEM could predict most things. Do you mean "can not be predicted by elastic FEM"?

I wish I could contribute more on RCF and Hertzian stress fields, however, I have never worked on these. I have therefore, been leaving you to make your own points. We do nevertheless, have some interesting cases of surface hardened layers on components (carburisation). These layers are brittle, but strong. In HCF they seem to increase life by preventing the formation of a microstructurally small crack. Perhapse this is because surface plasticity is inhibited by the strong surface layer. However, in LCF the life is decreased. Perhapse because the brittle layer nucleates a crack very easily once any plasticity occurs at the surface. But I think this is all due to whether or not plasticity occurs and not specifically to uncontrolled "plastic instability".

Nevertheless, during fatigue crack growth "plastic instability" seems to play a role.

Thank you for your contributions.

Sometime the depth to the radius ratio is too small specially at the point contact or if you have solid precipitate debris that act like point loads so you have several dynamic variations and we found that FEM will not take these into considerations.

So much more complex than a radius pressing onto a surface.

I meant the depth with respect to the radius at the contact

Mainly I wish to say "Hi!" to John Brear. I think this conversation demonstrates the need to qualify the context of a question such as this. The classical question to necking of a tensile bar has been, as they say in court, "Asked and answered" long ago. Given this is now 2018, I have deduced that the original question therefore had a much more specific goal in mind - but I am unable to find what is is. I ask because I have questions of my own about necking, or more precisely the failure to neck, in the creep range, which might have relevance

I kind of agree

Dear Douglas

I hope you get a reply from John Brear it is always great to chat with him and I was hoping he might contribute something to this question regarding the difference between different shear modes of failure.

As for my question; I usually have multiple goals in mind when I post a question on RG:

1. Education; mainly mine but also to have a searchable record of the answer for others to find. In particular the answer by Meryl M. Hall, Jr. is very useful as a record for the future. As an aside following up on the references reminded me of just how influential the laboratory of GE at Schenectady has been. Not only in the USA but also in the UK which has benefited from people working at GE and then returning to the UK and taking up education and industry posts on their return. I almost think GE Schenectady deserves a RG page all of it’s own.

2. The classical answer “as they say in court, "Asked and answered" long ago.“ Is a bit of a problem as I suspect the answer has been largely forgotten and is not being taught at University. This is a problem because a lot of applied research in mechanical properties is still being carried out on new materials and yet the test results for tensile strength, rupture strength and ductility are being used as if they have real physical meaning. Without the understanding learnt in the past. I think that there is always a need to re-invent the wheel every few decades. Otherwise even good published research can become buried under decades of “routine and practice”. For my own part I can’t remember being taught about the effects of plastic instability at University, although I can remember being warned about the so called “tensile strength” but the theory was not explained.

3. As for my current technical goal, my other question

https://www.researchgate.net/post/Can_the_point_of_plastic_instability_be_identified_from_the_results_of_a_creep_test_on_a_metallic_material

is more to the point.

Finally, you mentioned that you have questions of your own regarding “the failure to neck, in the creep range”. Can I ask; do you mean, very low ductility failures where creep failure occurs without significant plasticity? Or do you mean; very large uniform elongations where the material behaves like chewing gum?

Michael, In reference to your response, I would say a little bit of both. My experience is that "chewing gum" failures only happen in labs where investigator are trying to do the job in a hurry and use too short a timescale. The worrisome range is 1 to 10%, high enough for the material to still be usable in a practical situation, but not 100s of %. Instead of drawing out into a neck, the material deforms surprisingly uniformly, way past the point I would imagine a neck to form. My thought is that it is a form of superplasticity. A faithful model of a tensile test in FE does this but my analytical skills are not up to developing a simple model

People need to get a copy of F.A. Nichols, "Plastic instabilities and uniaxial tensile ductilities", Acta Metallurgica, 28 (1980), 663-673. I have located the papers by Elliot Duncombe, Int. J. mech. Sci. Pergamon Press. 1972. Vol. 14, pp. 325-337, and Int. J. Solids Srructures, 1974, Vol. 10, pp. 1445-1458. If possible I will provide a link to these papers on my RG page. Read these papers and you will be your own expert on the subject.

Plastic instability is caused when surfaces can not absorb high hertzian pressure and you will have small crack initiation which in turn will propagate faster when a point load will hit it fast and thus the boundary will shatter and plastic deformation will be in full force

Here are the plastic instability papers. Plastic instability leading to initiation and growth of a "neck" in a tensile test is governed by the material's strain hardening, n, and strain rate sensitivity , m, of the flow stress. Necking is initated when the "uniform strain" reaches a critical value. In a fixed strain rate test, the critical strain is e = n. In a fixed extension rate test, the critical strain is e = n/(1+m). The total elongation to fracture increases with increasing strain rate sensitivity. Literature data show that as m approaches 1, total elongation can exceed 2000 % (superplastic behavior).

Secondary creep occurs when the strain hardening is zero, that is, when mechanical strain hardening and thermally activated strain softening are equal n = n(+) - n(-) = 0. This says that once a tensile creep specimen has exhausted the primary (hardening) creep stage and enters the secondary (non-hardening) creep stage, n = 0 so that no additional strain is necessary to initiate necking! The specimen may not be recognized as necking due to the large width of the neck, which may be the entire gage section. Note that a material with m = 1 is "superplastic". Superplasticity is often assigned to materials having 0.3 < m < 0.8.

Plastic instability will render mechanical property tests inefficient if the crack or plastic deformation s are large enough

Dear Meryl M. Hall, Jr.

Thank you very much for the papers. As I don't have access to a well stocked library it can take me some time to get papers for myself.

Your note on secondary creep is consistent with the work of Edward Hart

Chapter A Critical Examination of Steady State Creep

and I am very pleased to see you linking in with superplasticity.

With regards to a uniaxial creep test the interpretation suggests that the so called tertiary creep is the result of localised stress increase due to non-uniform deformation and possibly some thermal softening. Unfortunately, I also find that primary creep is highly influenced by how the specimen was loaded. Therefore, with regards to a creep test both the primary and tertiary phases are unreliable! Which also means that the time to rupture and ductility are unreliable also. This rather limits the application of creep data for Design to the most simplistic and conservative treatments.

I wonder if this will be good enough for the 21st century? Don't get me wrong the creep test data was very useful in the 20th century. I am just wondering whether the demands of higher thermal efficiency for any fossil fuelled application (land, sea and air) and interest in very high temperature fission not to mention fusion reactors; means that it is time for us to rethink the good old creep test?

Thank you Gabi

If I split up mechanical property tests into to two groups:

Then we can then see that:

1. Tests on uncracked specimens, will be affected by both (i) the presence of a crack and (ii) plastic instability. Therefore, we have to be really careful then when analysing the results of such tests. Unfortunately, when we analyse the data from these tests we usually make two assumptions (i) the specimen is un-cracked right up to the instant of failure and (ii) the strain is homogeneous. It is my position that since, these two assumptions are untrue we actually need to be really carful when we draw conclusions from our data. Or perhaps we need to improve the analyse methods so that we don't actually have to make these assumptions at all! This could be particularly important since one of the main reasons why we would perform a test on an uncracked specimen is to measure deformation properties. Therefore, plastic instability could well make a real mess of our data.

As for 2. Tests on cracked specimens: We have already made a start by implicitly considering the crack. However, plastic instability might still be a cause of problems with our data analysis. You see in most tests on cracked specimens we try to limit plasticity by choosing highly constrained specimen geometries (min-size CT specimens and side-grooves). However, this does not mean that plasticity is eliminated completely, hence the need for elastic plastic fracture mechanics. Also it does not necessarily mean that plastic instability is eliminated, or does it? To be honest I don't know whether plastic instability can occur in a CT specimen. If it does it would be very localised to the crack tip. Nevertheless, it could still mess up the data analysis.

It is Ok if you do that but you should use optimization techniques to get the best results

Gabi

I absolutely and completely agree. Which is why I am asking these questions, because I need to understand the mathematical models to describe plastic instability in order to be able to write down the equation set that needs to be optimised, for each type of test. The papers by Hart and Nichols describe these models and I am working them into deformation and damage models at the moment. The next problem I will have to contend with will be that non-uniform deformation creates a multiaxial state of stress. Then I will have another decision to make; (i) shall I ignore stress state effects (ii) or shall I try to take account of the effects of state of stress?

Michael

Stress state definitely does affect plastic instability. See Elliot Duncombe's papers which deal with the this subject in great detail. He deals with plastic instability of 2 geometries; bi-axially stressed grooves and locally thinned patches in cylinders and plates. The math makes for slow going but when you complete the task you should be better equiped to develop the models that you need.

The adequacy of any mechanical test depends on the application. If your engineered structure involves long slender rods loaded by dead weights in a high temperature furnace, then the "good old creep test" may be what you need. I know of no engineered component that fits this description. However, my experince is that more often than not what is needed is results of a stress relaxation test, which does not involve plastic instabilities. For example, the heads of reactor pressure vessels are fastened using bolts or studs that are pre-tensioned/torqued to a stress that will not allow the head to "lift off" during normal and off-normal (accidental) pressure operation of the vessel. This requires a bolt/stud material that is not subject to rapid stress relaxation at normal and off-normal temperatures of operation. What is needed for this application is a constant-displacement stress relaxation test and a constutive equation to describe the test results.

Another application from my experience is the creep rupture of a pressurized tube, (e.g., reactor fuel rods) either closed end or open end (2 different stress states). Of course what is needed are pressurized creep rupture tests, which measure creep rates and rupture strains, and a creep rupture model (Duncombe's model, which predicts the strain at which plastic instability may initiate, was developed for this case). Of course, although creep could not be avoided and needed to be studied, the goal was to avoid excessive distortion and interference of components, which may occur long before creep rupture could occur. I never had an engineering need to know or teritiaray creep rates. Creep strain to initiate rupture (not time to rupture) is what is needed for engineering models.

All other potential contributors ignored (development of porosity, cracks, etc), tertiary creep in the dead weight creep test is due to the fact that stress is not constant in this test. As the test specimen cross section decreases, either uniformly or locally, the stress increases while the strain rate also increases. This continues until dlnS/dlne' > m, that is, until the ratio of the rate of increase in stress/to the rate of increase in strain rate exceeds the materials strain rate sensitivity so that there is positive feed back and stress increases rapidly until the fracture stress is exceeded.

It also depends on the materials and if it is coated or not or if there was certain treatment that prevents plast deformation