Learning Forum

[Federico Parisini]

hi, I really need help understanding how to implement a fatigue simulation of threaded connections. i have correctly defined the material of the screws (S-N curves). 
The point is that performing a fatigue analysis on a two-step analysis, in which in the first step I have the bolt pretension and in the second the application of the loads, makes me doubt whether the software considers that the preload is also applied cyclically. 
So I eliminated the bolt pretensions and only considered the loads, but I'm not sure it's right not to consider the preload of the screws. How can I consider the static load of the screws and the cyclic loads at the same time? thank you.

 

 

[danielshaw]

The Mechanical Fatigue Tool (FT) is an entry level fatigue product.  It cannot evaluate the fatigue damage caused by a constant stress (preload) and an alternating stress (external loading).  You need to use Ansys nCode DesignLife for that type of calculation.  From a fatigue damage perspective, the constant stress caused by the preload is a mean stress that does not contribute to the alternating stress.  You could neglect the constant stress in the FE and include its effect in the fatigue damage by manually adjusting the S-N curve (e.g., using Goodman).  This approach requires extra effort and would only be applicable to one mean stress, but it should work.  You could make this approach more general (and conservative) by adjusting the S-N curve using the yield stress.  This approach assumes that the entire model experiences stress alternating about a constant yield stress.

[Federico Parisini]

Thank you! So what should I do? Remove the bolt pretension? After I've removed them, I don't understand how to consider their effect using Goodman, though.

[danielshaw]

You have 3 options, but options 2 and 3 are not “good” options.  They are require substantial user effort, because you are stretching the capabilities of the entry level FT.

1.       Using Ansys nCode DesignLife is the best option if you have access to an Ansys nCode DesignLife license.  It has advanced fatigue capabilities which make this type of simulation straightforward.

2.       Adjusting the S-N curve is a conservative option that is difficult to implement.  Goodman is a mean stress correction theory (Gerber is another one) that calculates an adjusted alternating stress (Salt) based on the mean stress and the ultimate tensile stress (UTS) of the material.  If you are not familiar with Goodman (or Gerber), you can find descriptions of them on-line.  If you are using the S-N method, then mean stress from the preload should be less than the yield stress (YS) of the material.  So, a conservative approach is to just assume that the mean stress caused by the preload is equal to the YS throughout the entire model and then use Goodman along with the mean stress (assume to be the YS), the UTS, and the S-N curve to develop an “adjusted” S-N curve that includes the mean stress effect.  Then use that adjusted S-N curve along with just the alternating stress from the FE to estimate the fatigue life.  This approach requires substantial effort and may be difficult to implement if you need the preload to accurately calculate the alternating stress in the FE.

3.       You can use the FT with a constant amplitude, non-proportional loading to evaluate your type of loading, but you need to adjust the loading in the FE model.  To use this approach, you  need to create two Mechanical environments and use a Solution Combination to combine their effects. 

 

With 2 Environments and constant amplitude, non-proportional loading , the FT calculates the alternating stress to be one-half of the difference between the two load cases (Salt = 0.5*(E1 - E2)), and the mean stress to be the average of the  two load cases (Smean = 0.5*(E1 + E2)).  So, you need to set up the environments to ensure that the combination produces a varying load that alternates about the static load.  This type of loading should work.

Environment 1 (E1) = varying load/2 + static load = external load/2 + preload

Environment 2 (E2) = -varying load/2 + static load = -external load/2 + preload

When those load cases are combined within the FT, they will produce the following mean and alternating stresses.

Smean = 0.5 * (E1 + E2) = 0.5 * [ (vl/2 + sl) + (-vl/2 + sl) ] = static load = preload

Salt = 0.5 * (E1 - E2) = 0.5 * [ (vl/2 + sl) - (-vl/2 + sl) ] = 0.5*varying load = 0.5*external load

To use this approach, you need to:

·         apply these two loadings in the FE to create the two result sets,

·         combine the result sets in a Solution Combination (just add them together),

·         insert a FT under the Solution Combination, and

·         specify constant amplitude, non-proportional loading in the FT.

This approach requires extra effort to create the 2 new structural enviroments to be used by the FT, but it should work.

[Federico Parisini]

Thank you for the time you're dedicating, but I have a question for each of the options you proposed.

1. I have Ansys student 2023 R2, and I have Ncode Mechanical embedded design life, which I'm not sure if it's the correct add-on. Is there a manual for its use?

2. I'm not clear on what I need to do for this option. I have the material data for the screws defined with S-N curves related to specific mean stresses, taken from a paper. I thought that subsequently, the program would automatically make its considerations regarding these curves. What you're suggesting is for me to manually calculate the S-N curve related to the yield strenght and use that in the simulation? How do I set it if I've defined the material with multiple curves to determine which curve it should refer to?

3. Have I to perform the analyses in the two environments in two steps each (1: bolt pre-tension, 2: external force)? 
Then if I have an external force of 2000 N, do I need to input a force of 1000 N in the first environment and a force of -1000 N in the second one? Even if the load is zero-based and not fully reversed?

Thank you again.

[danielshaw]

If you have access to the Ansys nCode DesignLife embedded Add-On, I strongly recommend using it.  It is the simplest and most correct method.  You just need to run the preload and preload+external load cases in Mechanical and specify Time Step loading in DesignLife with both result sets selected.  Note:  You can also include additional loadings if appropriate.  DesignLife will internally determine the mean and alternating stresses from all selected result sets.  You can then specify any applicable mean stress correction method (Goodman, Gerber, Interpolate, etc.) and DesignLife will apply it.

For the 2^nd option, yes you need to manually create new S-N curves that include the effect of mean stress.  I am not recommending this method, but it is theoretically doable.

For the 3^rd option, you need to apply whatever combination of two loadings that produce the desired Salt and Smean using the FT equations.  I suspect that is +/- ½ the applied load, but I did not thoroughly check the math.

[Federico Parisini]

Hello, I appreciate your help earlier. I have another query. Is it problematic to have frictional contact? I've come across guides online indicating that frictional contact results in a non-proportional load for fatigue tool, yet I haven't found any practical examples illustrating how to address this scenario. I find it quite perplexing that after days of searching, there's a lack of information on simulating fatigue in bolts. Additionally, the Ncode tool in my Ansys appears to be malfunctioning; while I can set up the analysis, I'm unable to execute it. Any insights or suggestions would be greatly appreciated.

Is it possible to pay for an hour or two of consultation from someone experienced in the field? I need assistance in making the correct assumptions, as this project is crucial for me.

[danielshaw]

Frictional contact is ok, if you are using Constant Amplitude non-Proportional loading or Time Step loading in Ansys nCode DesignLife because the final stress states are used.  It is questionable when used with other loading types because the stress states are scaled.

I do not know why the embedded Ansys nCode DesignLife Add-On is not working for you.  I recommend opening a SalesForce Case using the Customer Portal.  A support engineer will assist you.

You need to discuss consulting with your Ansys account manager.

[Federico Parisini]

I am a student ????