Failure Analysis - Science topic

There are many good guidelines, books, training, and even software for effective Root Cause Analysis available. Look them up. Take a course. Learn how to do it correctly the first time.

A starting point is - was it a brittle fracture or a ductile failure? Usually can tell from visual inspection. Also can you run a sample through a Charpy V notch or other testing?

What alloy was it?

What is past history?

Was a heat treatment involved? (probably not with an ingot)

Any extraordinary loads? My father ran into a case where fins were breaking off a casting as a worker was hitting the casting with a sledgehammer to "remove the sand".

@sunil Bhat, with UTS, what do you mean?

Many thanks Joshua I really appreciate your answer

Rhys:

This is a question that has been raised for half a century or more - I can remember Keith Miller, when I was a graduate student, opining that delta-K could not be used to describe fatigue-crack growth - nevertheless fracture mechanics is still successfully used today. As long as one is properly aware of the basis and limitations of governing parameters such as K and J, I still believe that they have a powerful application. However, this is not always the case when they are used and accordingly fracture mechanics has become one of the most abused form of mechanics.

If you think that this approach should be dismissed though because of its "limited validity and questionable interpretation", could I respectively ask what you would recommend as a replacement?

cheers Rob

The inclusion of fracture toughness based on the K-factor on an Ashby diagram can be considered an ingenious solution. The K-factor is a measure of the stress intensity factor at the tip of a crack and is a critical parameter in determining the fracture toughness of materials. By including this parameter on the Ashby diagram, it allows for a quick and easy comparison of the fracture toughness of different materials, regardless of their class or composition.

However, it should be noted that the K-factor only provides one aspect of the material's fracture behavior and other parameters such as crack growth resistance and microstructure should also be considered when selecting a material for a specific application. Additionally, the values on an Ashby diagram are typically plotted for a specific set of testing conditions and may not accurately represent the fracture toughness under different conditions. Therefore, while the inclusion of K-factor on an Ashby diagram is a useful tool, it should not be the only parameter considered when selecting a material for a specific application.

Foluso Awolola These links might be useful:

1. https://github.com/topics/water-distribution-networks

2. https://github.com/yingqianzhang/water-pipes-failure-prediction

I do not think there is a weakening factor. The values you have are depending on the material properties.

No thanks, however just for the sake of curiosity, I invite you to read a litle about 'auxetic materials', if you are familiar with forget my advice. My Regards

Strain is any kind of deformation, including elongation. Elongation is the word used if we're talking specifically about tensile strain which the sample deforms by stretching and becoming longer. Usually the percent elongation is calculated, which is just the length of the polymer sample after it is stretched (L), divided by the original length of the sample (L0), and then multiplied by 100.

Dear sir,

during my master degree my research work was on laser additive manufacturing.

i would like to pursue my doctor degree on additive manufacturing because of my previous work.

if there is any vacancy on additive manufacturing please let me know.

thank you

I suggest to use FIDES methodology since it has models for Li-ion batteries.

If you want to calibrate a model with a compression test (epoxy shows different response in tension and compression), just use j2 plasticity, available in ABAQUS. All you need is the evolution of the plastic strain and the yield stress.

The torque – tension relationship is highly sensitive to normal variations in the coefficients of friction between threads and between the turning head and the surface of the joint. Refer to the following paper:

Effect of Tightening Speed on the Torque-Tension and Wear Pattern in Bolted Connections

Nature of load, shear load, ductility etc.

higher stiffness

Jesus verifica el FMEA del producto y de ahí ismo determinas si es Weibull, lognormal, normal, etc. Saludos.

Si es una sola probabilidad de falla o sea alta es Weibull

Si son varias grandes es lognormal.

Si son tres o cuatro muy parecidas es normal.

Eso lo sacar del RPM del FMEA .

Also, you need to findout the Hashin Damage model parameters for the axle material from the previously published experimental studies.

Dear Sanchit Saran Agarwal ,

Please have a look at these articles which talks about prior probability of warm spare gate in a dynamic Bayesian network

Yes, that helps, but, is there a laser inspection method for the same? Hossein Keshtkar

try this link it may help you

https://www.adamantvalves.com/common-failures-of-safety-valve.html

Hello Oussama!

I advise you to use dynamic Bayesian networks. I think it is a very suitable method for dynamic systems.

Good luck

Hi Saif Aldin ALI Madhi,

You may find these articles useful.

Akash:

It depends on the size and geometry of your test specimens.

Elongated dimples which form in shear are very common in uniaxial tensile test fracture surfaces. The classical example is the "cup-and-cone" failure mode. In the center of the samples - the "cup" - where the stress-state is dominated by the applied tensile stresses, the process of ductile dimple formation occurs by microvoid coalescence, i.e., voids nucleate at particles (either by the particle cracking or separation at the particle/matrix interfaces); these voids then grow, primarily driven by the triaxial stresses at the center of the sample, until they coalesce or more likely the separating ligaments between the voids "neck down" due to plastic instability. Nominally the same process occurs near the surface of the sample - the "cone" - but now the process is dominated by the near-surface shear stresses. The result is the coalescence of voids formed around particles in shear, which naturally results in elongated dimples.

As you describe your aluminum sample as "sub-size", it is highly likely that the process of microvoid coalescence is dominated by the near-surface shear stresses, in which case elongation dimples would be the result.

ROR

P.S. It is interesting to note here that the formation of dimples both in tension and in shear can even occur at the center of a uniaxial tensile sample in the "cup" region. In many low-alloy steels, the initial voids responsible for their microvoid coalescence ductile fracture are formed at inclusions, e.g., MnS inclusions which readily debond from the matrix. These voids then grow under the triaxial stresses, as described above, but the necking down between these larger voids by a plastic instability can occur by shear-induced microvoid coalescence of cracks in the much smaller carbide particles. The latter is know as a void sheet instability.

Please refer to: https://www.weibull.com/hotwire/issue116/relbasics116.htm

If you are still interested on generating randomly distributed cylindrical particles please have a look at this journal paper. I described the algorithm step by step in this paper which enables manipulating the orientation of the particles as well. Basically, with this method you can create a randomly dispersed and oriented cylindrical particles.

Thanks,

Thank you for your response. I will try to built the model step by step as you advised and share the results.

Dear Marcello Fera

I know maybe it is difullcut :) but generally its possible.

Thanks for your answer anyway.

Best.

hi.

Read papers that are similar to your work and fill those parameters.

It would be good to break the problems into bits:

1. When the rod is pushed, as Prof. Hariharan rightly pointed, the sleeve would act as a mechanism to extrude the rod into thinner diameter. However, where I differ from Prof. Hariharan is that extrusion cannot happen when the sleeve is of constant diameter. The sleeve should be somewhat like a funnel to guide the rod into the die. Also, extrusion involves severe plasticity, which doesn't seem to be your case.

2. I would rather think of the boundary condition as a simple friction clutch (discounting heat effects - assuming quasi static) where the sleeve just throttles the rod.

3. In such a case, when an axial compression is applied and assuming there is no lateral instability (such as linear or non-linear buckling), then the rod should travel past the sleeve - of course, leading to tremendous friction depending on the axial load. Oh! My God!!! There could be local failures such as surface spalling, fretting, heat affected zone, etc.

4. If the sleeve is sufficiently strong and the friction is high, the rod would start to bulge / barrel, if not allowed to buckle.

5.... and so on!!!!

So, you see, you have only told us a small fragment of the problem, leading to multiple assumptions from my side! :-)))

Best would be to draw a free-body diagram that clearly outlines the forces and moments, and perhaps you could use John Harvey's book and use the equations of a cylinder within a cylinder problem... with some tinkering. Sounds easy enough!

To sum it all up, first try approximations and get a hang of the problem before using ABAQUS or ANSYS.

Bests!

knn, kss, ktt should be taken as large as possible to avoid collaps of contact point. the usual convention is E/t where E is the modulus and t is the thickness of adjacent layer. similarly G12 and G23 for kss and ktt. for damage initiation both shear can be assumed to be inter-laminar shear strength i.e S12 and normal as strength of matrix. these assumptions are quite appropriate. this is extended for damage energies too.

Maybe Aidlab (229$)? Here's the article how can you connect Aidlab's ECG readings with Kubios: https://medium.com/@Aidlab/exporting-your-ecg-readings-with-aidlab-4dd5c049078a

it is for over values of the range!

By definition, a slope fails with a FoS going below 1.0. As long as the FoS is above 1.0, a slope is considered stable. Thus, I do not see how your figure could be drawn.

Of course, the FoS is a simplification and can not be measured directly. However, the definition is rather clear and if you follow it a FoS>1 relates to zero probability of failure.

I am not getting your question. What you want to know? What you you want to investigate, is it for developing new composition, new heat treatment procedure or to improve mechanical property? Which aspect you want to know or want to explore in Grade-E casting?

Please refer atteched reference book by Donald Wulpi on failure analysis of industrial components.

SHIVRAJ SINGH KASANA

Is there any reference stating that "it is not recommended to average internal variables?" Actually plastic strain is also based on an internal variable (plastic multiplier or whatever you call it). If the plastic strain tensor is averaged over some elements, then the difference obviously is that in one case a tensor is averaged as opposed to a scalar.

I dont know what specific user subroutine is necessary to gather information from neighboring elements, but surely there must be one, if it is not UMAT.

As mentined before, to know your problem it is necessary to define the options that you are using in oyur model.

Dear Prof. Naderpour.

Thank you very much. I appreciate your reply.

Best wishes,

Hamzeh

Foundation failure analysis depends on the shear strength parameters of the soil, foundation shape and depth and water table level.

Dear Manoranjan Jena

It was a very nice and informative conversation.

I wish you the best things.

Best Regards

@ Aleem:

Although, element erosion should not be used to delete elements because if the erosion deletion criterion is set too high, element distortion will occur, and if it is set too low, premature deletion will occur. If it is necessary for you to use element erosion, especially for large deformation in the post-failure region, you could make the failure principal strain a higher magnitude of the peak strain. This also depends on the material model you are using in your simulation.

There are more than 35 wear mechanisms available for failure of metal under sliding wear. This wear on this picture appears to be adhesive wear.

The terminologies brittle and ductile failure is commonly applied to materials failure under tensile load.  

Hi Deniz,

Can you share your graphs?

 It seems a contradictory result.

I am really keen on to dig its root cause.

Sincerely,

Amin

Plastic and surface

Meanwhile we found some literature and scientific names for the structures. The structures resembling natural valeys/rivers (my first 3 pictures) can be addressed as: fingering, viscoelastic fingering or air fingers. To my understanding, they are no fractals, as the self-similarity stops at an early scale. The honeycomb-like structures (last picture - red) are adressed as cavitation in the literature.

According to literature and some own experiments, fingers and caviations depend on the following aspects: In the case of air (or an other low viscosity material) beeing present, factors that facilitate the flow of the adhesive lead to fingers. Conversely, factors that inhibit the flow of the adhesive lead to small fingers and promote the appearance of honeycomb structures (cavitation). Fingers can only occour if air is present. Fingers are promoted by a low internal friction (viscosity), a slow speed of separation of the parts and a high thickness of the adhesive layer. If no air can penetrate into the adhesive, only cavitation are formed. Cavitations originate from unavoidable preexisting bubble nuclei in the range of a few µm that enlarge when pulling forces are applied

Amar MB, Bonn D, 2005. Fingering instabilities in adhesive failure. Physica D: Nonlinear Phenomena 209, 1–16. DOI: 10.1016/j.physd.2005.07.002.

Nase J, Derks D, Lindner A, 2011. Dynamic evolution of fingering patterns in a lifted Hele–Shaw cell. Physics of Fluids 23, 123101. DOI: 10.1063/1.3659140.

Poivet S, Nallet F, Gay C, Fabre P, 2003. Cavitation-induced force transition in confined viscous liquids under traction. EPL (Europhysics Letters) 62, 244.

Poivet S, Nallet F, Gay C, Teisseire J, Fabre P, 2004. Force response of a viscous liquid in a probe-tack geometry: fingering versus cavitation. The European physical journal. E, Soft matter 15, 97–116. DOI: 10.1140/epje/i2004-10040-2.

Saffman PG, Taylor G, 1958. The Penetration of a Fluid into a Porous Medium or Hele-Shaw Cell Containing a More Viscous Liquid. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 245, 312–329. DOI: 10.1098/rspa.1958.0085.

Tanguy F, Nicoli M, Lindner A, Creton C, 2014. Quantitative analysis of the debonding structure of soft adhesives. The European physical journal. E, Soft matter 37, 3. DOI: 10.1140/epje/i2014-14003-8.

Hi Sir,

Its very interesting topic. The following article links may help you,

https://www.researchgate.net/project/Mode-I-Fracture-Parameters-Using-Steel-Fibers-in-High-Strength-Concrete.

i think, it needed to take an exponential of predicted T if transformed logarithomic values of T were used into the model

ASM Handbook on Fractography, but rather old (1987):

First  you have to model that zone then you have to add the material properties in a different material model. Then during meshing you have to chose the mesh tool > global set (click)> new window come there you have to put that material no. which have CZ properties?> then mesh it.

Hello,

I recommend:

Solutions predictive diagnostics of vehicles and infrastructure.

Implementation by railway industry Big Data technology is predictive analytics.

Best Regards

Arkadiusz Kampczyk

AFT models usually focus on the time of failure, not explicite to the event. Problematically the fact is more complicated, if you analyzing recurrent events.

There can be many causes of the problem, but the most likely is the partial crystallization of the amorphous SiO₂ to cristobalite. This crystallization means a large volume change, which may cause cracks.

An X-ray diffraction analysis can prove the formation of cristobalite.

Against cristobalite formation may help lower sintering temperature, the faster heating rate and shorter holding.

I have some experiences with cohesive cracks and XFEM. Did you resolve the step that lead to crack initiation? And if you use small steps the problem you mentioned becomes negligible.

Please let me know.

Best regards,

Phu

Dear Meko,

you might want to have a look on Advanced Monte Carlo methods such as Line Sampling and Subset simulation

They work in quite well in high dimensional space

or if you are looking for some efficient methods for computing the gradient in high dimensional space

Regards

Edoardo

One easy approach is to penalize the objective function. Specifically, in a optimization project of minimization, results that against the predefined constraints will be assigned to a great value (penalty). In this case, the proposed solution that fail to satisfy the constraint(s) will never be the future candidates in the following generation. 

Dear A. Morasch,

Thank you ,my confusion is clear now.

Thanks Baeker Sir and Ramsay Sir.

Thank you Mr. Hamad for your excellent suggestion. Its working perfectly and I am very happy.

I have tested my model for a lower load like 0.5mm (in terms of displacement) where there is no failure and the model works perfectly. The loading and unloading both  curves are linear and in same way as shown in attached figure.

For higher load also now it's working well using the method suggested by Mr. Hamad. The new Reaction force vs. Displacement curve is attached for displacement upto 1mm. There is no jump in reaction force during 1st unloading time and also residual reaction force amount has also decreased from 11.4N to 3.5N which is very less compared to the maximum reaction force.

Thank you once again all of you.

Hi. it is depend on your strategy. you can use the sudden degradation models  based on the stress analysis and some relevant criteria such as Hashin-Rotem or use the gradual degradation model (Crack density based model) using the energy criterion. 

Hi,

First of all, stress from FEA will not be calculated at the nodes, they are calculated at integration points (also known as material points). For full integration, there are 8 integration points and you will have stresses at all these integration points for single element. If you choose reduced integration there will be only 1 integration point and you will get single value of stress for an element.

Though I do not know about this failure criterion, my suggestions will be independent of the failure criterion used. You can do following things.

Apply failure criterion for each integration point and check the status. If all material points of an element shows failure, that means the element has also failed.

or

You can check, stress values at all integration points. If they are not very different from each other, then probably you can take value of stress at the Centroid of the element (that will be simply the average of stress from all 8 points). Considering stress at Centroid to be representative stress of that element, you can check the failure criterion, if satisfied you may consider element as a failed element.

Hope this helps.

May I know? the Mn-Steels based either ferritic or austenitic composition range.

As long as the different models are similar in functions, it is more efficient to maintain a single FMEA to cover them. Within the single FMEA, it is possible to denote some failure modes as specific only to Model X.  Or you could add an extra column to input which are the applicable equipment for each line item.

Thank you very much Dr Zekry

Dear friend, very interesting question. I am interested to your question and will explain in detail soon (similar to my case). A bit hurry for this time, please remind me.

However, the easiest way is searching the related document by typing the keywords into google scholar. You will find some related articles.

If yet to find the articles, do not hesitate to let me know. InsyaALLAH I will help you in detail.

Good luck. Dr Zol Bahri - Universiti Malaysia Perlis

Hi,

I don't have direct access to any white papers on flop SEU and FIT rates but I can say that in typical designs, the FIT rate of sequential logic, at least in terms of retention cross-section (without accounting for logic or timing derating - if you want to include this, and assuming a flop has the same fail rate as SRAM, derating will reduced the actual flop error rate to 5-10% of the measured retention value in most logic designs), is about the same as that of the SRAM - in other words if you put and array of flops in a neutron beam/alpha source and an SRAM array with the same number of elements as the flop array, the failure rate will be similar. It is true that flops with higher drive transistors (wider gates) can have a reduction in failure rate but we usually see all flops from a technology library from the weakest to the strongest are about 2x higher than SRAM (in the same technology) to about 15x lower failure rate, In a FIINFET technology I would expect the same trend to be followed since the sensitive element in the SRAM and the flop is the same or similar FinFET transistor.

Sorry if this is to general for you needs, but if you are trying to design experiments, the rule of thumb above has worked for many generations of bulk and SOI devices and I assume will work for FinFET as well.  

One note, I am assuming you are referring to standard flops and not DICE or some other redundant flops which, of course, would have a much, much lower failure rate.  

Usually, using F-ion to remove Si-ion, and C-ion, H-ion to reactionwith O-ion to reduce the concentration of F-ion, other bombardment was reaction also for expose ionic bond or removed some hard removed chemical such as Ti. Any quesiton, you could reach me by mail. tao.wang@diasemi.com

Dear Murkute: an additional aspect is the level of adhesion observed after sliding. Pay attention on it and this point can be useful for your definition. Regards.

No ready-made data is available, at least to me. But you can proceed in the following steps:

1) Collect failed parts data from warranty and/or testing

2) Use the data to estimate reliability

If you do not have and/or can't arrange for any data, then look for MIL-STD (freely downloadable). You may find gross estimate of failure rate.  

I think fatigue life test in general is a destructive test. If you are interested in obtaining experimental values , I think you should proceed till obtaining the required no. of cycles then stop, If the blade persists I think this means that the test is successful.

ANSYS does cover RC. You can use solid65 element with different nonlinear plasticity models with cracking and crushing capabilities and with rebar information in any direction you want, with pretty easily adjustable concrete material model in 3D (including principal energies for fractures). Even though the solid65 includes rebar model, it is wiser to turned that off and model reinforcement using 1D link elements. The Prestressing can be applied on "other"  link elements that traverse the solid45 concrete elements as prestressing tendons. Simply look for solid65, and for Concrete material option in the TB command.

Hi Sandip,

Similar work has been done in Abaqus focusing on steel reinforcements.

You can get ideas from this report:

Regards,

George

Jitendra, 

I have read your papers. It is a huge part of my PhD thesis, the concept of NLU and your work is the main inspiration. One of the first papers I read.