Thank you so much for you explanation. 

So, to check the model :

I switched the load to Fy=1N and I load the Fy history data file : 

I get a life of 47 585 Cycles. But my damages are all over 1, they should be less than 1.

How can I use these damage results to determine after both analysis (with fy and fz) to determine the life in cycles using Miner's rule ? 

In Apdl script I had a table with damage less than 1 and I add whole node damage to get the life.