Topics relate to Granta Design and more

Non-constant amplitude fatigue issues

    • Jared McFadden


      I am trying to do a fatigue analysis on two models but am getting results that seem incorrect. Both models use the same part with same S-N data but one has a larger load applied, about 2x more than the other. I created an ASCII file for both that put the load history as a ratio of the maximum load. The issue is 1.) the results of the model with the smaller load are saying that the fatigue life is less, which I know is incorrect and 2.) the fatigue life result of the lower load model is showing as one value, even though I am getting a varying safety factor. I made sure that my S-N data encapsulated all stresses experienced by the part. With that being the case, I do not know what to put for the infinite life value. I noticed that changing this value has a drastic change on the result. When infinite life is changed, the fatigue life of the lower load model just changes linearly and the fatigue life of the higher load model does not change linearly. I have attached my S-N data, the load history plot and the result for the lower load. 

    • Daniel Shaw
      Ansys Employee

      I assume that you are using the Mechanical Fatigue Tool with History data loading.  Some thoughts:

      1. Your SN curve does not seem realistic.  SN only applies to high cycle fatigue (HCF) where the calculated stresses are below yield (Although it is sometimes used for streses slightly above yield).  For ductile materials that typicaly means cycles above 10e4.  You have some very low cycle points on your SN curve.
      2. The SN curve has a big jump in cycles around 100 MPa.
      3. I have never seen SN curves with more than 1e9 cycles.  How did get a SN numer at e28 cycles? 2 MPa is a tiny alternating stress.
      4. I suscpect that the alternating stresses ave very low, so the FT is essentially predicting no fatigue damage.  Sometime FT results are difficult to interpert when the calculated alternating stress is less than the lowest stress from the SN curve.  What are the stress levels in the model and how do they comapre to the SN curve?  You should be able to do a simple hand calculations using the FE and predict what the maximum Salt.  How does that number compare to SN curve.
Viewing 1 reply thread
  • You must be logged in to reply to this topic.