Ashribb22

@Giacomo77: Love the passion and the bike! I'd love to know your rationale on why was DP 490 chosen compared to say other toughened epoxy adhesives from 3M such as DP 460 or 420?

Applaud the effort! Making a frame is no easy task.

@vinay: I appreciate the interest and I love to learn and discuss about composites. You bring up and interesting point. I hadn’t considered the “free” fiber ends as a potential stress-concentration site for inter-laminar shear (I’d be interested in reading some papers on the quantifiable effects of that failure mode). Wrapping the fibers inward would certainly help mitigate that, but for this project we were constrained to stock tubing by budget. Just sourcing tubes with the desired layup schedules, properties, and wall thickness/OD/ID was very difficult to begin with. That said, I believe any effect here would be minimal for several reasons, both for general applications and my specific application. Firstly, and most broadly, the way such forces (tension, torsion, compression) are transferred into the bonded joint is inherently inefficient, as the material is not infinitely stiff and therefore will experience a non-uniform force governed by shear lag. Instead, the stress rises from the free end of the overlap to a maximum at the point where the load enters the adhesive and decays exponentially with respect to the distance from that point. That suggests the free ends at the end of the tube, due to the deformation of the material, will actually experience the lowest forces in the bonded area. On the other hand, shear lag also applies if you look at this from the perspective of wall thickness. In a single lap joint, that distance is twice as large as in a double overlap joint, so the single overlap experiences higher peak inter-laminar shear between the bonded layer and the next layer of fibers. So, yes my inner most ply of fiber ends at the largest distance from the start of the adhesive would experience more inter-laminar shear than a comparable length double lap shear joint. We compensated for this via longer joints, as I’ll explain later, for a few reasons. In our lugs we used a lip to mark full insertion, and when adhesive was applied and the tube inserted, the adhesive inevitably coated the end of the tube butted up and then bonded to the lip. This effectively tied the free fiber ends into the discontinuous carbon laminate of the lug. We also gave all of our tubes a low-angle chamfer at the ends to increase peel strength, and I imagine this also provides the added benefit of giving the free fiber ends more surface area to contact the adhesive with the lug’s outer wall and lip, further securing them in the joint. We used very long bond surfaces (longer than those seen in most lugged frames, like Atherton, or 90s Specialized). This choice was driven by a desire for stupid large safety factors, given limitations in our surface-preparation techniques, inevitable variation in bond gap, and the fact that bond failure would be catastrophic and sudden failure mode that would likely lead to injury. Because weight was not the primary goal of this project, adding a few extra grams per joint to ensure safety was a worthwhile tradeoff in my mind. That choice resulted in a very large inter-laminar contact area, failure of the outer plies shearing away from the inner plies would more likely indicate a poor decision in a tube that had insufficient properties leading to excessive inter-laminar stresses, rather than a weakness in the bonded joint itself. As you mentioned, yes, a comparable length double lap shear joint would have helped both to increase the surface area and reduce the effect of shear lag in the thickness of the wall. However, I’m not sure how well the article conveyed that we had originally planned to make the lugs from CNC'ed 6061 Al, and therefore we chose to use a single lap shear joint for DFM purposes. We already had to do significant work to redesign the lugs with thicker walls to get similar stiffness and reduce peel effects on the joints when pivoting to the discontinuous carbon lugs, and we didn’t find the time to also redesign all of the bonded joints to become double lap shear joints. Additionally, since the majority of the joints are carbon-to-carbon and we couldn't afford chemical etching even if we wanted to, mechanical abrasion and cleaning of the bond surfaces in a 130mm deep, ~2 mm gap double lap shear would have been extremely challenging. Atherton uses much shorter double lap shear joints that even if they have the same total surface area due to the overall shorter length means that due to the reduced shear lag the ends of the fibers would experience higher inter-laminar shear in theory. More practically speaking if your joint is pushing the laminate it’s bonded to fail due to inter-laminar shear (not a adhesive governed failure), and you ensure the laminate on its own chosen passes your requirements, I'd say the joint served its function in effectively transmitting the load. (not to say it can't be optimized for weight or some other factor) I speak confidently about this as I have done a lot of reading on composites for this project and genuinely find composites fascinating. But I am a second-year undergrad who still has got a lot to learn and hasn't taken any formal classes on classical laminate theory or composites, so I recognize I may be incorrect in my application of these concepts. Curious to hear your thoughts.

@WheelNut: Some details didn't make it into the article, we did attempted to source a jig to borrow. Obviously, a professional frame jig would have been ideal but it didn't pan out. As for making a jig ourselves, with a total budget of $1500 making a jig that is accurate and precise would have been a large portion of the budget (again this was a personal project with no external funding). While the joints didn't have a interference fit, the lugs had a tolerance that was based on maintaining a bond gap around 0.15-0.2mm radially. The tubes had fairly course tolerances on the OD from the factory so we decided to make the lugs with a tolerance of +- 0.1mm radially of the tubes nominal OD and then spending many hours sanding the tubes with fine sand paper and checking with a micrometer to get the bond gap we desired. Additionally we used 0.125mm bond line control beads mixed into the adhesive to help with geometric alignment and maintain an absolute minimum bond gap. If your interested this is the study that influenced us to use BLC beads. https://www.researchgate.net/publication/350915173_Evaluation_of_Bonding_Gap_Control_Methods_for_an_Epoxy_Adhesive_Joint_of_Carbon_Fiber_Tubes_and_Aluminum_Alloy_Inserts

@nnowak I touched on this in my response to @Will-narayan below.......

@vinay: The DT,TT, and CS are roll-wrapped tubes and the ST is a filament-wound tube. Each has myriad plies and fiber orientations but all tubes have 90 degree "hoop" plies to combat the radial crushing forces exerted by the lugs. Validated via Classical Laminate Theory hand calcs and FEA simulations.

Thanks for helping some random students from halfway across the globe!

Aadi here. I love reading all the guesses and the humor! The real technique is truly novel and quite interesting....and no, I'm not giving hints. If you want the reveal, you'll have to keep an eye on Fiber Form's (my venture), website, Instagram, or LinkedIn until the patent is filed. We are currently looking for beta customers so if your interested in custom carbon parts feel free to reach out. Insta: @fiberform_ , Website: https://tinyurl.com/Fiber-Form, LinkedIn: https://www.linkedin.com/company/fiber-form/

Should Read Fbolt= 440NM/.03M*

@rideordie35: Preface: I am a Physics and Engineering Student who has performed analysis of bike frames. I strongly recommend some basic hand calculations to take a look at how this would actually play out before making such harsh comments. Lets start with some worst-case/conservative assumptions: 1. Lever arm is 440mm (bolts are a shorter distance from the rear axle in reality) 2. Bolts are centered at the pivot point (clearly not the case in reality) 3.The rear axle experiences a static load of 2KN (Lower than ISO 4210 but on par with 16 CFR 1512 the minimum requirements for a bike to be sold in the US. Valid for a prototype project) 4. The chainstays are infinitely stiff and the entire load is transmitted from the axle to the bolts 5.Two Class 12.9 M5 bolts spaced ~30mm (no way to know fs) First, isolate each chainstay: Farm=Faxle/2, Farm=2000N/2=1000N. Then calculate the moment on the arm: Marm=Farm*L, Marm=1000N*0.44M=440NM The shear force per bolt if they are 30mm diametrically opposed. Fbolt=Marm/D, Fbolt= 440NM/.3M = 14,666N Now to find the theoretical shear capacity of a Class 12.9 M5 bolt. Class 12 means a UTS of 1200 MPA The cross sectional area of the bolt is A=Pi*r^2, A= 3.14*(2.5mm)^2= 19.63mm^2 ~ 20mm^2 The theoretical shear capacity can be calculated via: Fshear = 0.6*UTS*A, Fshear=0.6*1200N/mm^2*20mm^2 = 14,400N Shear capacity per bolt This puts the Shear Capacity at 14.40KN and the Shear Force as 14.66KN. Given the amount of assumptions made to get to this point these can be considered roughly equivalent. With all of those extremely conservative assumptions, a 2KN load doesn't leave room for a safety factor, but they could have been designing for a 1.5KN Load or even 1KN which would give room for a SF. Choosing a load to design around is THE most tricky aspect of bike design in my experience. I would suggest they increase the load they designed around and make relatively minor hardware alterations like larger diameter bolts and/or larger diametric opposition. Riding on gravity trails might be premature, but basic testing in a controlled environment is reasonable. It by no means lends credit the statement that its a "design that so far away from mechanically feasible that it discredits the education of its designers."