JohnnyEnglish
Mechanical
Hi,
I'm currently using Ansys and attempting to predict the fatigue life of a sheet metal structure containing many rivets subjected to a random vibration spectrum. I am using 4 noded 3D elastic shell elements for the sheeting.
My first issue is how to model the rivets? I have used coupled nodes before and found that the model produces reasonable modal results but something is amiss with the reported rivet forces. I backed this up by doing a few simple static analyses on lap joints. There are numerous references in books and on the internet which point toward using a beam element and a modified stiffness (Michael Niu - Airframe Structural Design and others) - however, none of the stiffness models for the beam hold true especially in more complex joint geometries. Any ideas?
My second problem is obtaining a usable stress to calculate the fatigue life. Ideally I'd like to resolve the in plane nodal loads acting on the thinnest sheet and calculate a bearing stress (area = rivet dia x sheet thickness). This would not account for any friction effects which should yield a conservative result. The SCF might be taken from Niu again using the severity factor concept, but if you've tried to figure it out it's not that straight forward. Maybe I should just use 3 to be on the safe side. Any more ideas?
Thirdly - The result of the random vibration analysis gives me a statistical result (one sigma) which tells me that the sheet will be subjected to this bearing stress (or less) for 68.XX% of the time. I can then obtain the two and three sigma stresses and, using Miner's rule and the appropriate S-N material fatigue curve, combine them to provide some sort of indication of fatigue survival from the random vibration. The result is conservative since it assumes the stress is at the maximum level for the whole period. (The input spectrum is actually that of an accelerated vibration test to represent "real life" fatigue conditions, so I'm really just trying to ensure the structure passes the test - the test table is capped at three sigma accelerations so this assumption for the Miner's calculation is valid.)
I have attempted to find answers on the web, but to no avail. My colleagues have similar issues, so no answers there; and the all other attempts (including the local Ansys guys) so far have failed to address the problem.
Any shared experiences or help related to rivets and/or vibration fatigue would be much appreciated. Hopefully I've included enough detail.
John.
I'm currently using Ansys and attempting to predict the fatigue life of a sheet metal structure containing many rivets subjected to a random vibration spectrum. I am using 4 noded 3D elastic shell elements for the sheeting.
My first issue is how to model the rivets? I have used coupled nodes before and found that the model produces reasonable modal results but something is amiss with the reported rivet forces. I backed this up by doing a few simple static analyses on lap joints. There are numerous references in books and on the internet which point toward using a beam element and a modified stiffness (Michael Niu - Airframe Structural Design and others) - however, none of the stiffness models for the beam hold true especially in more complex joint geometries. Any ideas?
My second problem is obtaining a usable stress to calculate the fatigue life. Ideally I'd like to resolve the in plane nodal loads acting on the thinnest sheet and calculate a bearing stress (area = rivet dia x sheet thickness). This would not account for any friction effects which should yield a conservative result. The SCF might be taken from Niu again using the severity factor concept, but if you've tried to figure it out it's not that straight forward. Maybe I should just use 3 to be on the safe side. Any more ideas?
Thirdly - The result of the random vibration analysis gives me a statistical result (one sigma) which tells me that the sheet will be subjected to this bearing stress (or less) for 68.XX% of the time. I can then obtain the two and three sigma stresses and, using Miner's rule and the appropriate S-N material fatigue curve, combine them to provide some sort of indication of fatigue survival from the random vibration. The result is conservative since it assumes the stress is at the maximum level for the whole period. (The input spectrum is actually that of an accelerated vibration test to represent "real life" fatigue conditions, so I'm really just trying to ensure the structure passes the test - the test table is capped at three sigma accelerations so this assumption for the Miner's calculation is valid.)
I have attempted to find answers on the web, but to no avail. My colleagues have similar issues, so no answers there; and the all other attempts (including the local Ansys guys) so far have failed to address the problem.
Any shared experiences or help related to rivets and/or vibration fatigue would be much appreciated. Hopefully I've included enough detail.
John.
