Hello everyone,
Part of my project is to investigate the effect of surface roughness on stainless steel 17-4 PH in a 3-point bend simulation on Abaqus. I was to use 2D planar mode and manually add a very small amount of roughness at the centre of the tensile surface (as this is where max stress would be). I made the roughness Ra15 as shown in the second image. The rough width is 100 microns and I added a very fine 1 micron mesh (100 micron x 20 micron rectangular partition) in the rough region as required from my supervisor.
I am a novice in FEA and Abaqus. Would someone be able to recommend how I analyse these results of the simulation? For example, would it be useful to extract load vs displacement data for this model and compare it to the load vs displacement data for a smooth model which I have also done? However, since the rough region is very small, I don't think this will give a significant difference in load vs displacement data. Perhaps going down the route of stress concentrations is better for my analysis?
My understanding is that a rougher surface provides more nucleation sites for crack initiation, which explains their lesser mechanical performance.
Hello Azam,
interesting question, I am also working on simulating a 3 point bending test. With regard to FEM you will only get the information for a continuous, “perfect“ material. Failure modeling and fracture mechanics are therefore, in my eyes, necessary to describe the behavior you want to test and model. If you really should model the roughnesses geometrically, as your image suggests, or if they should be rather given parametrically is indeed a good question and could potentially require quite large computing power in the first case. However, when modeling geometrical you obtain the tiniest of elements for the roughness and, in the case of your picture, large differences in element size to the rest of specimen which could potentially influence your results. As side remark I suggest to use symmetry planes in your model.
Now from the experimental side, I totally agree that comparing force over displacement is the examination to aim for. You can then directly compare to your FEM results, which is a good baseline. Even with a continuous, perfect, FEM you can check out if the model is running good, then take care about failure/roughness implementation. For identifying failure due to localization, a coupled optical measurement of strain during the experiment could be rewarding to correlate strain peaks to your load curves, in a quantitative and qualitative way.
what does literature say about taking roughness In FEM into account?
So long these are my ideas, best regards,
Peter
Dear Azam,
Actually the effect of surface roughness will be neglected in the abaqus analysis. If you are interested in the effect of the same, you use fracture mechanics approach,
Measure the average peaks and valley height, consider it as a crater crack, simulate the model you will get the result. Also conduct the experiments on the same samples and deduce a mathematical model. For this you have to conduct more number of experiments.
Thank you K V Arun and Peter Frohn-Sörensen for your very helpful advice I really appreciate it, apologies for the late response!
- Badges
- Science topic