This is Not a 3D Print Fail
Analysis of 3D Printed Titanium Handlebar Failure at Japan 2020 Olympics
It is not often that we get to see a publicly available deep dive into a scenario where a 3D printed component fails while in use but fortunately/unfortunately the complete failure of a 3D printed handlebar at the Japan 2020 Olympics has provided that opportunity.
Commissioned by AusCycling the full report investigates the events leading up to the process failure and resulting crash on August 2nd 2021 when Alex Porter of the Australian pursuit team’s 3D printed titanium ‘base bar’ fractured and experienced catastrophic failure.
The failure of the 3D printed base bar was probably one of the highest profile public 3D printed failures, the kind of event that might make many people wary of the quality of 3D printed components and question whether they are really ready for high performance functional use cases.
Thankfully, in this instance, this was not a failure due to the material properties of the titanium component, or ‘technically’ the design of the part, but on the specification of the load on the base bar when in use, and fatigue testing.
The report commissioned by AusCycling found:
Like most failures, there is no one cause for this event, there is however a prime cause which had more impact than any other…
In use, the additive manufactured Titanium Base Bar was exposed to riding and training forces some one-and-one half times the Australian Cycling Team Specified design and test forces.
The Titanium material specified was lighter and stronger than steel but more sensitive to fatigue damage. Fatigue testing Specified was reduced by the Australian Cycling Team from 200,000 to 50,000 cycles.
The 3D printed Base Bar mates to a machined steering fork and immediately forward of the most forward attachment bolt was a .29mm elevated area which raised the local stress. Even without this mismatch, the high rider forces would have precipitated a failure elsewhere on the Base Bar.
So how did this happen?
The Australian Pursuit team had adopted a particular starting posture on the Argon 18 bike frame that forced the fourth rider on the team, Alex Porter into a position where his knees were hitting the base bar due to his height. To avoid the contact the team proposed to move the arms forward 35mm (1.37 inch) to provide clearance and attempted to secure a modified bar from Argon 18 but could not be guaranteed the part in time due to supply chain issues.
3D Printing to the rescue, right?
Right…
The documentation of what happened next get’s a little blurry (one of the problems identified in the report) but basically it looks like after a year of inaction due to ‘organizational changes’ the Australian Institute of Sport somehow obtained 3D printed a part to test fit and clearance using a Markforged Carbon Fibre reinforced PLA based on a modified based on the Argon 18 design.
“It is not known how the Argon 18 Base Bar Computer Aided Design drawing was obtained or how it was modified to reflect the 35mm extension as the Performance Systems Manager responsible was not available for interview.”
Ok?
From there, once the fit was tested with the Carbon reinforced prototype the Australian Cycling Team sent a request to Bastion Advanced Engineering to design and manufacture the base bar by sending.
Computer Aided Design skin-only drawing and a test requirement: the skin drawing only specified a printed surface where the Base Bar mated with the machined aluminium surface on the top of the steering forks. Previously supplied Bastion Head Stems which mated to the steering fork were provided with a printed-only mating surface without any reported or documented mismatch.
The static force of 1000N cited in ISO 4210-5:2014 was applied to each Base Bar at production end and they all passed.
In discussions between the Australian Cycling Team and Bastion Advanced Engineering, the ISO 4210-5:2014 durability or fatigue requirement of 200,000 cycles was reduced by the Australian Cycling Teams Performance System Manager to 50,000 cycles reportedly on the basis that when installed, rider familiarisation and training would demonstrate acceptable durability: an acceptance that possible failure could occur in service.
Before we go further I want to make clear that in the report findings Bastion designed and manufactured to meet the requirements set forth by the Australian Cycling Team, and have a long history of designing and producing additively manufactured components and provided detailed documentation of their entire process in the report.
There was a a 0.29mm bulge on the front of the clamped Base Bar to steering forks mating surface which would raise the stress in that area and attract adjacent fatigue damage and the Base Bar to Steering Fork connecting Bolt Torque was set to 6Nm not the 12Nm torque value specified for the Argon 18 or 8Nm as recommended by The Australian Cycling Teams Bicycle Build Book.
The major problem was that the calculated loads were significantly higher than supplied load cases. The resultant stresses were up to 371% higher than with the supplied requirements. Worst Safety Factor of reduces from 2.97 to 0.65.
If you really want to deep dive into to all the details I do recommend you read the entire 171 page report that will strike fear into the heart of any engineer (especially the poor graduate who in the interview the report stated his ‘Most telling was his statement “ have tried to forget [my experience]”.
The importance of the correct specification in all design, not just DfAM, topology optimization and generative design.
Along with the baseline skepticism of the material (and fatigue) qualities of additively manufactured parts, there seems to also be an inherent distrust of topology optimization and generative design in a similar vein.
As we see from the example of the base bar, it may be more likely the specification of the engineering problem more than other aspects that can cause catastrophic failure.
In my opinion the hardest thing to get right in engineering is translating the real world in-service environment into a set of modelling assumptions for design & analysis. With numerical methods you can demonstrate that your design can satisfy the requirements you modelled, but those requirements you modelled might not reflect reality…
If we extend this to computational, generative and even simulation tools, the reality that each of these can only give you results as good as the assumptions and inputs you, the engineer enter into the system.
The tool is rarely to blame for bad design.
We need to approach engineering with the same rigor no matter what tool we use and cannot expect any software to provide answers if we do not ask the right questions.
Oh, and the Australian pursuit team won the Bronze medal in the end, the only medal the entire Australian cycling team won at the Japan Olympics, not quite on par with Steve Bradbury with Australia’s first ever winter olympic gold.