Hello Salvatore Nasisi

Thanks for a very detailed reply.

I do not want a step by step procedure. I have designed for fatigue earlier also but never for fatigue due to thermal loading. In our organization we follow practices very close to ASME-BPVC and design curves are taken from ASME.

As you have mentioned no singularities should be there, but I am getting a stress field where four stress contour of von Mises plot (in ANSYS with default contour option) are passing through an element of 1 mm size, (component size is much larger), so situation is much similar to singular case. My feeling is I will get a similar stress field in real situation due to thermal shock during the beginning of transient, so how should I interpret the results. Should I use the maximum value of von Mises or average value across element or weighted average or there is some other method. As I am in design stage so either method must be as per some code or there must be some good logic for doing so. I have a variable amplitude loading but I can take care of that. Stresses are mutiaxial…

My doubt is only about stress number. Which number I should take for further calculation?

Your valuable suggestions will be very helpful.

Regards

Naveen

Dear Salvatore....thanks for a quick reply....

From your explanation i am getting a feel that my stress field may not be accurate, although i am very sure about it but still i will recheck (solved with axisymmetric)......Just two things i would like to know.....

1- are the practices of fatigue design/analysis (including result interpretation) are valid when it is due to thermal load?

2- When i saw the results first I also felt it like singularity......but i got convinced with the logic that when a very high temperature steam (around 600 deg C) will impinge on a surface of 30 deg C outer fiber will stretch and immediate inner fiber will be at room temperature. If in geometry, there is a discontinuity there may be a very high stress gradient because of such loading. So I felt stress field is correct.....was i right or should i improve my mesh/simulation. In the outer fiber (my 1mm element) stress is varying by 30%) so how should i proceed in terms of numbers...

during the course of literature search on it, i found some people are using the temperature difference on some portion of radius and calculating manually....

We solved the problem again with more refined mesh.....element of 1 mm size was divided in 5 so now every element is around 200 micron.....results are good now....peak value has increased but there is no abrupt change in stress value across the thickness of element.....

previously when we solved......gradients were very high........reason was......although mesh of 1 mm was quite fine for a large body (even for fatigue calculations also) but it was not sufficiently fine in comparison to the discontinuity at that location.......especially when we are interested in peak stress for fatigue calculation

My feeling that mesh is quite refined in previous case caused a stress field that will not be there on real component......

Thanks to all of you for participating in the discussion and sharing your valuable knowledge....

Ciao Dear Salvatore Nasisi…thanks for all the previous discussions and sharing your views.

Hello everyone…….

I would like to discuss the topic once again but especially in light of ASME code practices, as we follow DBA of section VIII div 2 for design.

1. During transient thermo-structural analysis maximum thermal stresses comes to component at around 250 deg C maximum temperature (at one location in the component) and finally the component go up to 560 deg C, at this maximum temperature thermal stresses are very less as thermal gradients are very low. As per ASME code, I should use the fatigue curve of material at maximum operating temperature i.e. 560 deg C. Why so…why not at 250 deg C. I am finding a very tough time for designing against a very harsh startup.... Are there some other practices in some other codes/standards/books….

2. Can anyone please suggest some paper on THERMOMECHANICAL fatigue analysis in ANSYS or fatigue due to thermal stresses…..there are so many on google but are not really very helpful….please suggest some really good papers, that are discussing design/result interpretation of FEA results…..

Dear Salvatore Nasisi…First of all thanks for a detail reply…

I agree that I am not having any strong reason from departing from ASME….It is just some questions arises during work…….I am analyzing the problem with my best efforts but I m getting a very strong feeling that I need to sharpen myself 

I would like to reply why higher thermal stresses at 250 deg C not at 560 deg C…..when my component is at RT steam of 250 deg C will directly impinge on it and after that there is a soaking at this temperature till the body reaches to isothermal state….after that temperature of steam ramps slowly up to 560 deg C. So highest thermal gradient comes at beginning….this behavior is quite similar to behavior of any component of fossil power plant that comes in steam path…where maximum stress comes at the beginning of steam impinging. Due to thermal shock very sharp thermal gradient comes at component…

I would like to keep this question live for discussion of fatigue problem….….any type of views about fatigue design from anybody are welcome…..we can use it as a forum…have a nice time everyone….

mateial is a Ni based superalloy and strains are elastically calculated plastic strain..

Could someone suggest me a free pdf download which gives step by step fatigue calculation as per ASME from the results of a FEA simulation (ANSYS specifically).

you can use any prediction method either stress-life method or strain-life method.