User login


You are here

Fatigue damage initiation - computational methods

pragtic's picture

Dear iMechanicians,

It seems that nobody here is active in fatigue life prediction focused on the initiation period. I believe it should be here and do not understand why the fracture mechanics has its place here but the initiation of cracks has been forgotten. Although cracks are very interesting there is a lot industrial sectors, where they are not wanted. Is not the reason of the current situation that the available computational tools that are free are mostly related to crack growth?


Jan Papuga


pragtic's picture

Well, I should admit that I started this forum also for another reason. Except for the research in fatigue prediction methods themselves I am the developer of fatigue freeware PragTic.

PragTic is intended as a tool usable to work either on FEA data or in so called isolated points - i.e. points with no relation to any FE-model. Its main focus is currently set to uniaxial and multiaxial methods. If you are interested, you can read more about the project on and, naturally, you can download it there.

I am the only developer now, so the further development is not so quick as I would like. Nevertheless, I think PragTic is perfect at least for a research and educational use. Recently, I prepared an automatic reporting of bugs available for registered users. The more users will use the program, the more response I will get and PragTic becomes better (well, and I will get more satisfaction Smile).

Of course, if you know about any computational fatigue related research in your company, where direct post-processing of FEA-results and the ability to implement new computational methods could be of use, let me know.


Jan Papuga

pragtic's picture

The last entry today:

I am organizing the Workshop on Computational Fatigue Analysis 2008 & The 2nd PragTic Users' Meeting in the end of October 2008. If you are interested, see Everybody is welcomed.




pragtic's picture

I would not like to do this forum one man show, but nobody has any comment it seems.

Well, I would like to post the question above, which is also related to the work on PragTic project. I am administering the webpage, where I try to document links to other fatigue related software (or any other fatigue related webpages), that can be used under any circumstances (being purchased, developed or lent). It is quite possible that there are more of them, so if you know about anything else, please let me know.

I am also interested in experience of users of common commercial software as MSC.Fatigue, FemFat, Fe-Safe, etc. My effort in PragTic development is driven by feeling that the documentation provided with these packages is not sufficient. I am afraid that many of the users have only vague notion, what they are computing and how precisely above all. If you look at the link, you can find there a proposal for PragTicA project prepared for a call on FP7 funding. One of the main target is a collective validation of commercial fatigue solvers. 

It is very well possible, that the project will be rejected. The other way then is building a similar project from the bottom, by gathering the interested institution and companies. But I need more information for something like that. So, it is your turn now.



my gmail id

Hi Jan,

I agree with you that the Fracture theory is very connected to Fatigue theory even the Fracture theory's success probably mostly applied to Fatigue. I found the obscure thing here that you should be more pricise and specific on what you are doing !




pragtic's picture

Hi Hamanh,

I am sorry, but my English is not so strong and I do not understand the meaning of your second sentence. Could you explain it in a bit more words?

I'm currently on leave in Croatia and do not have a good access to the Internet. Any my reply thus can take some time.




 I am mechanical engineering, and I woring with ANSYS CLASSIC 11. I dont now how could be setting Ansys to resolve some fatigue problems.

Can help me some body with some indication, or maybe steps with the manner in which I can setup the problem?

Thank you for understanding and best regards




pragtic's picture

Hi ingcartun,

Well, I myself have no experience with the fatigue module in Ansys. But what I read about its features some two years ago, it is a very basic module. In fact the automated work concerns reading the stress tensors in individual nodes and relating the local stresses to S-N curves. This necessitates definition of various modifying coefficients, which has to be either input or selected by you. The problem can arise by highly complicated (i.e. more complexly shaped) components, where search for this value is not easy. Anyway, even if you tune the computational results to some experimental tests that you could have, it relates only to the specific notch you examine now. If your goal is to prepare various design proposals, you do not know, if the critical place is not shifted to another locality, which can be quite near.

This is the reason why I keep my fingers crossed to FemFat, with its stress gradient concept, that could at least a bit simplify the work. But still I am not sure how well it works...

I would be quite interested how far the fatigue module in Ansys has got through the time. Have anybody else any experience? Can it assemble the local loading state from non-correlated loads in various load channels?

Back to you, ingcartun - I am not sure, how far away you live, but I would recommend you to visit the PragTic Users' Meeting (see somewhere in the start of this forum). No kidding, this is no promotion. The things you have to do before any such analysis are quite similar and similar should be also the processing of your input. So, it could help you to orient in any program of this kind.





pragtic's picture

Hi All,

I have already mentioned the name PragTicA before. It is a project that was prepared as a response to a call of the European Community under the FP7 scheme. As the time passes by and nothing happened I should prepare myself for its rejection. I assume, it could provide highly important data and thus I would like to present it also to other iMechanica users in order to explain its expected assets.

We have got to a situation, where there are already some commercial fatigue computation oriented packages (e.g. FemFat, Fe-Safe and other) available. You import there the FEM-model and results, make some clicks and selections and gets the coloured map of fatigue damage over the structure. The problem is, that such solution is very tempting as regards its simplicity, but a general intrinsic property of fatigue methods is that they are of empiric nature and thus any use outside of verified space is dangerous.

Producers of commercial systems do not provide any warranty on the use of the software itself or of the methods implemented in it and do not inform the user about potential deviations under specific conditions. The quality of the help depends on the particular software package, but there is no comparison of various usable methods anywhere. What is also of importance, the methods implemented e.g. for multiaxial loading analysis are quite dated and do not give so good results as they could.

I am programming PragTic already for years. Its goal was at first local - I wanted to compare various multiaxial solutions. But as it grew in my hands, I decided to provide it as a freeware ( and try to complete the blank space with a more thorough analysis of various computational methods, so that the users could anticipate the quality of predicted results.

PragTicA consortium was thus established with several goals: 1) analysis of the state of the art in specific fatigue areas; 2) a further development of PragTic so that it could be used for the most of common fatigue problems; 3) improvement of existing methods or development of new solutions with direct application and testing with PragTic; 4) experimental work to analyze potential experiments critical to the existing fatigue computation methods; 5) mutual comparison of different fatigue solvers including PragTic on retrieved experimental data; 6) open Internet dissemination of all obtained results.

PragTicA project focuses on the fatigue areas concerning: a) notch effect analysis, b) fatigue calculation under multiaxial load conditions, c) seam and spot welded structures, d) riveted structures, e) composite components.

PragTic's source code is offered for joint further development to all participants in the consortium. As it now looks like the project is close to rejection I am considering if not to give out the source code to companies or universities that would be interested in the next development and could help me to sustain the idea of PragTicA project living.

So, this is the sketch. As I mentioned in some other comment a more comprehensive reading on the project can be found on The answer to the application should be published till the end of October and I has to wait for it. But if you find it interesting, please contact me in any way, we can think of what to do together in both variants - if the project succeeds or fails.



Mike Ciavarella's picture

Dear Jan

   since I see your forum is a little "single-authored" by yourself, I try to enlarge the discussion.  I have myself been interested in fatigue crack initiation and propagation -- is the distinction that clear? I see your efforts are mostly in "initiation", but given today we deal with large number of cycles easily (either extended life or high frequency forces us to do so), what permits us to be sure about these classical methods?  Anyway, I have recently tried to be critical about various generally "accepted" theories and concepts.  I wonder if you would like to comment --- so as to make the forum more active!

For example:-

1) Paris' law -- is it really a "law":  some recent investigation, with many data. So what is the precision possible with such law?  When are we interpolating, and when are we taking risks in extrapolation?

no one, and one hundred thousand crack propagation laws: A generalized
Barenblatt and Botvina dimensional analysis approach to fatigue crack growth

Journal of the Mechanics and Physics of SolidsIn Press, Accepted ManuscriptAvailable online 16 September 2008
Michele Ciavarella, Marco Paggi, Alberto Carpinteri
Open Preview   Purchase PDF (949 K)
| Related Articles

2)  how difficult it is to "unify" initiation and propagation.  See:

On the possible generalizations of the Kitagawa–Takahashi diagram and of the El Haddad equation to finite life
International Journal of FatigueVolume 28, Issue 12December 2006, Pages 1826-1837
M. Ciavarella, F. Monno
Open Preview   Purchase PDF (1053 K)
| Related Articles

 A generalized Paris’ law for fatigue crack growth
Journal of the Mechanics and Physics of SolidsVolume 54, Issue 7July 2006, Pages 1333-1349
N. Pugno, M. Ciavarella, P. Cornetti, A. Carpinteri
Open Preview   Purchase PDF (280 K)
| Related Articles

3) how modern and classical notch formulations are not that distant from each other:

On fatigue limit in the presence of notches: classical vs. recent unified formulations
International Journal of FatigueVolume 26, Issue 3March 2004, Pages 289-298
Michele Ciavarella, Giovanni Meneghetti
Open Preview   Purchase PDF (216 K)
| Related Articles

I have also recently proposed a generalization of the El Haddad equation along the lines of some of the above papers. It is just submitted, so I attach only the abstract.



A simplified "damage tolerance" design approach for "crack-like" and blunt notches
Politecnico di Bari, 70125 Bari, Italy, presently at LMS-Ecole Polytechnique, Paris;

A very simple extension of the heuristic El Haddad formula is proposed for finite life. In
the limit of static as well as infinite life design, the new equation reduces to a form which
has received independent verification. The key starting point is to assume, in analogy to
the Basquin power-law SN curve for the fatigue life of the uncracked (plain) specimen, a
second power law (KN) defining a "finite life" fracture threshold: from the two laws, a
"finite life" intrinsic El Haddad crack size is naturally defined, also a power law of number of
cycles to failure. Good agreement is found using recent fatigue data by Susmel and Taylor, which they used to validate their method. Similar results are expected in fact, since we show analytically their method closely corresponds, especially when using two calibrating
SN curves (plain and a specimen with a sharp large crack) for their best-fitting strategy, to
our "K-N" finite life "threshold" power law. In view of this analogy, we also generalize some
equations considering blunt notches, including a recent Ciavarella-Meneghetti equation. The present approach permits therefore with simple equations to define generalized Kitagawa-Takahashi diagrams for notches at finite life which could be a useful alternative to the classical "damage tolerance" approach, requiring more sophisticated calculation, as well as the Susmel-Taylor recent proposal.

Keywords: Fatigue design, crack propagation, critical distance approach

Hello Jan,

   let me introduce myself- Rajesh working at Georgia Inst of Technology, Atlanta. I recently came across your view points on crack initiation and propagation. I agree with you on the efforts to understand fatigue crack initiation. I am doing a lot of work on the same. To make it clear I strongly feel that detailed understanding on fatigue crack initiation is necessary especially in high cycle fatigue to very high cycle fatigue regime. This is true because studies have shown that crack initiation and small crack growth could consume around 60-80% of the service life of the components intended for HCF and VHCF applications. Secondly, we need to understand the crack nucleation process at the grain scale due to the heterogeniety involved at that scale. Also, understanding of other intrinsic and extrinsic microstructure features that affect the service life is critical. I do see a lot of efforts on such studies and I am very happy to be involved. I do understand that its much more cost effective to design materials based on propagation such as Paris Law but I have some reservations using such rule for all fatigue applications. My feeling is that we  try to be too conservative with such design methods and thereby lose significant useful life of the component. Future is more ' application based materials design' where researchers and industries will focus on modifying the microstructure of the material to achieve desired fatigue resistance. I also see tremendous innovations on process route design in future. With the recent thrust towards alternative fuels for transportation, especially after the recent burst in crude oil price, I envision challenging and lucrative fatigue problems in the automotive and aerospace sectors. 

Regarding the fatigue software Pragtic that you are working on, I have not gone into the details but will spend some time on that and get back to you with more details.


Mike- You have posted some interesting a view points specifically coupling initiation and propagation. I will have to read those in detail. Also, I would like to actively participate in this and related forums.

Once again, I appreciate your efforts to open a discussion on fatigue initiation problems.



pragtic's picture

Dear Rajesh and Mike

Thanks for your posts. It is nice to see that there are more people interested in this topic. I do not have that much time (the end of the year is approaching, which means a lot of work to fulfil all the proclaimed goals), so I have read only the oldest paper recommended by Mike until now. You, Rajesh, mention some reservation to the use of fracture mechanics in the HCF domain. Well I had seen it similarly until I have read this Mike's paper.

In fact, all the methods trying to take into account the complex stress state around an examined point, work with some of these variables:

  • stress gradient at the point
  • structural stress in some defined distance from the point
  • stress along some path starting from the point
  • stress field at some area or volume around the point

I have accepted these approaches as simple engineering methods that are trying to do their best to include the effect of stress distribution. The paper that I am referring to says that the fatigue limit is there because of some small crack could not be created and thus the load state in some distance from the critical point is the decisive variable. If the load had been higher, the crack would be there and we could forget the fatigue limit. So, this is the reason why it is called unified theory - it unifies the crack initiation and the approaches used in fracture mechanics.

Well, Mike, it is quite possible that I did not write it as you would say it, so if you have any commentary, you have a lot of space here.

I skimmed at the paper by Norberg and Olsson (S. Norberg and M. Olsson, "The effect of loaded volume and stress gradient on the fatigue limit", International Journal of Fatigue, 2007, Vol. 29 (No.12), pp. 2259-2272). Their results should be quite interesting for you, Mike (anyway, I think you have read it already), because of they've got to the best results with the volume method. Well, it depends on many factors - I've been playing with the stress gradient only and you immediately see, that  it is very much depending on the definition of equivalent stress used for the stress gradient. The study has to be very broad and complicated if we would like to check all the potential degrees of freedom. What is very important from my point of view is that they used computational methods, to which I have reservation as regards their results - see FatLim for better understanding why.

The notch effect evaluation is also a part of the PragTicA project. I think, if we get the money, a lot of work can be done, because the evaluation by PragTic can be very efficiently automated. But, as I also wrote, it looks that we are heading to an unlucky end with PragTicA.

Well, I have to get back to my work. I'll write again, when I will read more.




Hello Jan,
Nice to see you here! I have several email discussion with Mike on the topic. I have been working on a method to unify the fatigue crack initiation and propagation into a single framework. One benefit on this approach is that the definition of fatigue damage initiation is not required. The fatigue damage initiation is really unclear and depends on the different applications. The other benefit is that the stress gradient effect, e.g., notch effect, can be included. I recently published two articles discussing on this topic.

Probabilistic fatigue life prediction using an equivalent initial flaw size distribution
International Journal of Fatigue, In Press, Corrected Proof, Available online 4 July 2008
Yongming Liu, Sankaran Mahadevan

Fatigue limit prediction of notched components using short crack growth theory and an asymptotic interpolation method
Engineering Fracture Mechanics, In Press, Corrected Proof, Available online 27 June 2008
Yongming Liu, Sankaran Mahadevan

The methodology is not completed yet and only limited validations have been performed. Several other papers are under review and I will forward them to you once they get published. It is a different way to look at the fatigue damage initiation.




pragtic's picture

Dear Yongming,

It is nice to hear from you once again. I hope you are satisfied at the Clarkson University.

Well, it seems that the topic of unified theories joining the stress field distribution and small crack growth (or "non-growth") is very popular today. Some time before we focused on multiaxial fatigue, we left our trace there and continue to next horizons... No kidding, I do not want to be ironic, but I am fascinated by the fact, how the research topics are stylish. Is anybody comparing all those new methods? And the older ones?

Since that my conference is approaching, I do not have that much time, so I only looked at your papers. Well, I think, this is not the right discussion - I started with talking about PragTic and FatLim, Mike referred on his papers and now you about those yours. It is more like a series of advertisements.

This is a forum intended for engineers. What about to try to describe your ideas, Yongming and Mike, to common engineers and to highlight the substantial differences to other approaches (I mentioned some of them before)? I only touched this topic in one of my previous comment. We are just talking, so this is the right time for philosophical interpretation of our findings (and minimum of equations!).




pragtic's picture

Dear iMechanicians,

Anybody who has looked on the previous "discussion" can see that the computational fatigue problems still do not seem to attract an adequate attention. Some change is necessary.


So, it seems it is the right time for a contest on the best "PragTic in Use" paper. If interested, look on BTW, the current exchange rates are 1USD~22.17CZK or 1EUR~27.96CZK.

Best regards,


My recent contribution, this is very simple yet effective way to deal with unifying initiation and propagation.The approach is more engineering type


Vikram Shenoy PhD

hi jan and others,

let me intoduce myself first, i'm a post graduate student pursuing my in engineering analysis and design. as you have mentioned in the earlier posts, a comparison between the newer fatigue life prediction methods and the older methods is truely necessery. the fatigue life prediction method taking into consideration the crack intiation and the crack propogation dates back to the Manson's double linear theory (where total no. of cycles for failure = no. of cycles for crack initiation + no. of cycles for crack propagation). the prediction of fatigue failure was moderately efficient. the theories of fracture mechanics also have a close relation with the crack initiation and hence the prediction of fatigue failure and fatigue limits, and therefor a need for comparision become mandatory.regarding the fatigue software which you are working on "pragtic", its really greek and latin to me. havent used such a software.well i must must say that you have been providing resourceful dicussions.



prasad shet

Subscribe to Comments for "Fatigue damage initiation - computational methods"

Recent comments

More comments


Subscribe to Syndicate