Alright, I sat down to do a write-up on this subject over the weekend and was tracking toward a post that was >3000 words long, so I decided to call it quits for the time being. It is a challenging topic to address adequately, in my opinion, because it involves technical aspects as well as issues of safety and responsibility. So, for now, here’s a few bullet points summarizing key considerations for designing additive parts. Many of these have already been suggested, but they’re worth re-emphasizing:
-
Unless something has gone horribly wrong, bike frames and parts fail due to fatigue. So, while you can use the yield strength, ultimate strength, and elongation provided in material datasheets to make basic comparisons between materials, they’re not going to give you much insight into fatigue performance. As a general rule, it is a good idea to use alloys in their intended applications. In the case of MS-1, I’d be reluctant to use a tool steel in a structural application unless I could generate data showing that its performance was equal to or better than something like SS 17-4, for example.
-
The fatigue performance for additive metals generally exhibits significant anisotropy (direction dependence), with the lowest strengths corresponding to loading across build planes (trying to pull layers of the print apart). Unfortunately for tubular bike parts, this corresponds to the direction that maximizes printability and space efficiency on the build plate, and minimizes distortion, so it is what we have to work with in most situations. Regardless of print orientation, the fatigue performance of additive metals is worse than their conventional counterparts.
-
The fatigue performance for additive metals also exhibits significant variability, which makes it hard to determine the correct design allowable and make comparisons between materials. This variability is also hard to characterize, especially across suppliers, because there is not a standard set of print parameters or a standard heat treatment that everyone uses. This variability is due to defects which result from the printing process (i.e. they are inherent to the technology and are unavoidable), and the particular sensitivity of fatigue performance to stress concentrations.
-
In my opinion, the variable fatigue performance exhibited by additive metals poses a challenge to people who are testing their parts – unless a part or frame passes the required ISO test with flying colors (for example, it makes it to 1,000,000 cycles when only 200,000 are required, not sure of the exact numbers in these specs) I would not feel comfortable counting that as a pass. Testing is definitively a good thing, so I’m not arguing against its usefulness, but if you’re testing a low number of parts made from material with properties that vary more than the standard metals used in the industry, different success criteria are justified.
-
I am also dubious about the efficacy of internal lattice structures. If the print orientation is optimized for the outside wall of a part, the internal lattice (or gyroid – applies to all infill patterns) will inevitably print in more extreme orientations in certain locations and increase the likelihood of generating a problematic defect. Infill structures also increase the unfinished surface area of the part, and likely create small stress concentrations where they intersect the walls. If these structures dramatically increase the part’s stiffness they may be worthwhile, but if I were designing a part I’d be inclined to add the extra material to the wall thickness instead. The same goes for internal ribs – internal ribs are most effective at reacting bending loads, but bike parts experience combined loading (bending, torsion, shear, tension and compression) and adding material to the wall instead will benefit the structure’s ability to react these other loads, too.
-
I’d like to see more people experiment with vibratory tumblers to finish their printed parts. Tumbling media that works on hard metals like titanium and stainless steel is available for purchase, so it should be an effective way to smooth surface irregularities. Other than material selection, surface roughness is the only variable affecting fatigue performance that builders have control over, so it is worth putting some effort into improving this area. Printed titanium will benefit most, but it is worthwhile for stainless steel parts, too.
-
I would be hesitant to stray from print fabs that have an established reputation in the cycling world unless I was able to work with a place that also serves more advanced industry. I credit Bastion, Prova, and Sturdy Cycles with bringing additive manufacturing to the drop-bar bike world, and each of them started by using Ram3D. Somehow Tom Sturdy has gotten comfortable enough with Ram3D’s processes to print entire forks, handlebars, and stems, but I don’t necessarily think this is an example that should be followed.
-
If I had limited ability to do useful stress analysis or testing of my parts, I would try to move my design as far in the direction of “goodness” as possible. So, I would do as much as possible to eliminate stress concentrations (small radii, large changes in part thickness or cross section over short lengths, etc.) and look to aluminum frames for inspiration when it comes to wall thickness and geometry. There’s a lot of nuance and detail to this discussion, which I may have time to develop further over the winter, but I can’t make any promises for now.
-
Finally, in my opinion, ride testing is useful but not good enough for parts that have catastrophic failure modes like stems, handlebars, and forks. The insidious thing about fatigue failures is that they happen suddenly and usually after long periods of time. So, if you do 500 miles on a prototype and call it good, you can really only sign off on your product for 500 miles of riding. The ISO tests are not perfect, but they do a decent job of solving this problem on timescales that are useful for product development cycles, and generally ensure that riders are safe and businesses retain good reputations.
Anyway, again, mostly reiterating what others have said but hopefully that is helpful!