AnsweredAssumed Answered

How accurate are SW 2019 FEA results for Hertzian cylinder contacts?

Question asked by Rod Newstrom on Jan 23, 2020
Latest reply on Feb 4, 2020 by Bill McEachern

I see a number of responses in these forums indicating SW FEA doesn't handle Hertzian cylinder contacts properly, but they're all older. I'm wondering how accurate they are in SW 2019.


I'm designing a cam driven engine. I do all the calculations in Excel then export the results as equations to Solidworks where I create 3D models and run analysis to confirm my Excel calculations. The Excel calculations related to the most heavily stressed cam (R1 = 0.290", R2 = 2.468", 2,768 lbf force) indicate peak contact pressure of 248 kpsi over a contact length of 0.993" and width (2b) of 0.014" yielding Von Mises stress of 231 kpsi. The material is Maraging 350 steel having 319 kpsi yield, 0.32 Poisson Ratio, and 29,733 kpsi modulus of elasticity. I have confirmed my 2D calculations with several on-line calculators, so I'm confident they are correct. The 231 kpsi Von Mises result is, however, uncomfortably near the 319 kpsi yield.




When I run the same conditions in Solidworks FEA, I get a peak Von Mises stress of 213 kpsi. At first glance, that seems to correlate well with my 2D Excel result of 231 kpsi. Unfortunately, the Solidworks result reflects a hot element, and the rest of the surface shows a much lower Von Mises stress, typically below 138 kpsi. I extracted the stress and contact force data from the contact patch and confirmed it is a 3D Mohrs Circle calculation, so it's possible I'm merely seeing an inherent difference between 2D and 3D analysis of these type contacts.

Should I trust the Solidworks 3D results or my own 2D calculations when evaluating the design for margin? I already know the "safe" answer, "use the worse 2D numbers," but that doesn't help much in this case.


Follow Up: Per the selected answer, 3D FEA simply isn't a good way to approach Hertzian contacts. When I ran the 2D simulation, I got different answers than produced by my Excel code. Solidworks did, however, yield results very close to this free software: Hertz Contact Stress Calculations and Graphs with HertzWin. Since my cams are designed in Excel using splines with XYZ points exported to Solidworks curves, I really need my Excel calculations of stress to be accurate, so I strove to fix them. After some close examination, I found I had swapped X and Y in some calculations. I corrected the error and now obtain precisely the same results as the free software linked above. I have attached the Excel spreadsheet which includes summaries of and links to the sources in case anyone else finds it useful.