S-N data correlation 2

[zerow3]

Hi, I am trying to do some fatigue calculations on a component, but find that there are not enough material data (S-N data). So I am trying to generate my own data (mainly with higher stress ratios) to use in conjunction with those available. However, the data available are at a different heat treat condition. So my question is what sort of procedures is needed to correlate the data? Is it recommended to do such a correlation? Also, how many stress ratios should I replicate? Should I just replicate 1 stress ratio and assume all other stress ratio are related in the same manner? Are there any standards available for correlating S-N data?? Please give me some advices.
Thanks in advance.

 

[metengr]

zerow3;
Here are my thoughts;

You should familiarize yourself with fatigue testing, presentation of data, and statistical analysis of fatigue curves if you are going to perform testing to obtain new data.

I would suggest that you subcontract the actual mechanical fatigue testing to a reputable material test lab that routinely conducts fatigue testing of metallic materials. Most reputable material test labs will be able to guide you on suggested number of fatigue tests per stress ratio.

You will need to generate unique S-N curves for a particular heat treatment condition because there are no shortcuts or correlations to manipulate fatigue data for metals that can be thermally treated to various strength levels.

Try to review the following ASTM standards for background information;

E466 Conducting Force Controlled Constant Amplitude Axial Fatigue Tests of Metallic Materials
(Load controlled fatigue; no measurement of strain)
E 468 Presentation of Constant Amplitude Fatigue Test Results for Metallic Materials
(Reporting of load controlled fatigue data)
E 739 Statistical Analysis of Linear or Linearized Stress-Life (S-N) and Strain-Life (?-N) Fatigue Data
(Statistical analysis of fatigue life curves)
E 1049 Cycle Counting in Fatigue Analysis Cycle
(counting procedures for spectrum loads)
E 1823 Terminology Relating to Fatigue and Fracture Testing (Definitions used in fatigue testing)

 

[zerow3]

Many thanks.

Any suggestions/reference on how to select the number of specimens needed for each curve?

 

[metengr]

http://scitation.aip.org/getabs/ser...00124000004000421000001&idtype=cvips&gifs=yes

 

[TVP]

The reference provided by metengr is excellent, with lots of data. As an example of what's required in order to develop statistically meaningful fatigue curves, here are some numbers for samples needed when determining the fatigue limit by the staircase method:

5-9 Rough estimate of the mean value
13 Mean value with confidence interval
17 Useable value for standard deviation
25 Good estimation for the standard deviation

As you can see, sample size canchange dramatically, depending on how much confidence you need in the data.

 

[zerow3]

Thanks