I can speak to this I think (work for a DMLS machine OEM). The biggest drawback from 3D printing in regards to mechanical properties is from the fatigue strength of the material. You’re right @DEVLINCC, Ti64 is more sensitive to surface cracks because it has a lower elongation than 316L, and in a sense, a 3D printed surface has little teeny cracks all over it. This makes the probability of one of those “cracks” being big enough to cause a failure go up over machined/forged.
The challenge I see here (new to the forum and framebuilding) is understanding this mechanism and how to design for it when the only data available (usually) is tensile. Tensile is less sensitive to surface roughness than fatigue so being 95% tensile strength doesn’t equate to 95% as tough necessarily… curious to hear the community’s thoughts here! How are people ensuring robust part design? FEA? Trial and error? Eyeball? Excited to be here and learn more!
I used to work for an FEA software development company, so here’s my take on it
FEA and fatigue can go together quite well, if a few important criteria are met:
the loading situations are well known in their entirety
the material properties including elastic-plastic properties and some woehler charts are known for the specific shape/size of the object in question
the material is quasi-isotropic in its properties
the stress history (also thermal stresses) is known and/or can be reset
surface roughness and its influence
Basically, fatigue simulation assumes, that some kind of “damage” accumulates in the part each time it is stressed to a certain percentage of its yield. The more percent, the more damage. For the simulation to predict a point of failure as number of load cycles, all the above has to be taken into account.
Now, the first point is probably the strongest: we don’t really know all load cases that actually occur, neither do we know how “hard” the bike is going to be ridden and by whom. So it’s a bit of a crystal ball scenario to derive a fatigue simulation scenario from that in the first place.
The next point is a tricky one as well: Usually, Woehler charts are derived from round test specimen with a certain diameter. Every shape that deviates from that needs to be corrected somehow. There are algorithms within FEA software that can take that into account (for example through analysing the angle between neighboring elements and calculating the “sharpness” of a surface kink) but this is far from being an exact science, more of a dark art. Since all these calculations take place in logarithmic scales, it is quite easy to miscalculate by a few powers of ten…
To wrap things up (I could go on about this a lot longer but I feel this is not the right place for it… maybe I can start an FEA thread…), fatigue analysis in FEA works great for something like a conrod, where thousands of variations of very similar parts have been analysed already and contribute to the statistical security of the simulations. But for one-off parts that are built by hand, it is more than appropriate to go by the seat of your pants (and fail sometimes)
That’s cool. That somes up my understanding of the problems with using FEA as a tool, especially that first point you brought up. a mate of mine who is an engineer and I have been trying o figure out a practical set of values to put into a Solidworks model to help tune the wall thickness of the front end on my Enduro bikes. We found it hard to be happy with any of it and have gone back to the ‘gut feel’ method from doing this for a few years and having a couple of failuers. I’m not hunting grams so adding a bit more material where I feel it’s needed is a method that seems to work. I do wonder what would happen if someone got hurt from a failure though. Can of worms I think.
@Luniz82 , thanks for the thorough summary! My understanding is that once you throw lattice structures at FEA it all goes out the window too…
I’m really interested in this problem though and would love to learn more. If we as the 3D printing machine manufacturer can get out and drive some standards it would make everyone more comfortable with the technology. And yes, @DEVLINCC, if we get injuries from 3D part failures it’s going to push back the adoption years so we want everyone to be implementing this stuff as safely as possible.
