Tough carbon-fiber bike w/o vacuum, oven, autoclave ... Just Do It [yourself]
Updated: Feb 20
I want to tell here the story of how I repaired a bicycle aluminum frame with a carbon-fiber patch. This is because sometimes a few dollars, some "how to do it" research, and a couple of days of hard work, can save us a lot of money and, most importantly, give us a lot of practical experience. I hope you enjoy it.
As an avid cyclist, I have owned the same bike for the last 17 years. Other than the front carbon fork, the frame is made of an aluminum central triangle glued to a carbon rear end. A few months ago, after 17 years, the aluminum, which is a ductile material, had naturally “fatigued” and propagated a crack that started right behind the seat-post where the aluminum triangle meets the rear carbon end. Trying to weld the crack, I increased the damage by melting part of the aluminum frame (figure 1).
NOTE: I am very satisfied with the frame. I attribute the crack to natural fatigue along 17 years of use plus a couple of accidental big impacts during that period. Indeed, I wanted to fix the bicycle because I believe that, thanks to its well-made nature, it can still withstand many years ahead.
What is shown in figure 1 is the result of the following:
The vertical aluminum tube is very thin (1.5 mm thickness), and that is because it does not need to guarantee an exceptional rigidity since it usually hosts the seat-post within, which is a carbon-fiber cylinder of about 3 mm thickness. For that reason, as soon as I heated the tube to realize the weld (TIG welding), what was initially a little crack - originally started somewhere at the center of the grey patch in figure 1 - became a big hole of melted aluminum. I tried to patch things up by pouring some aluminum back and cover everything, that is the result shown in figure 1 where all holes are covered. As it might appear evident from the picture, that new aluminum patch was very soft and could not provide the required rigidity; it would probably open after a couple of rides and bumps since it was possible to move the patch by simply applying some pressure with the thumb.
I was about to give up and start looking for a new bike when I decided to give another shot to my old bike. I decided to try to repair the frame with some carbon fiber – even though the section I was repairing was made of aluminum; more on this below. The general process of carbon fiber depends a bit on the specific case (i.e. OEM manufacturing, repairing, etc.). In general carbon-fiber structures start with sheets or clothes of fibers which can be handled like actual tissue clothes, which are then impregnated with resin - nowadays, usually epoxy resin. Once placed on a mold or a part to be reinforced, pressure and heat must be applied: the former helps eliminate the resin in excess, while the latter helps cure (i.e. solidify) the resin. Pressure is often applied through a vacuum bag with a spongy layer within to absorb the resin in excess. Heat is usually applied through an industrial oven. In high-end applications, both pressure and heat are applied through an autoclave, which usually also involves pre-preg sheets (i.e. sheets of fibers already pre-impregnated with the right amount of resin, therefore more expensive and tougher to handle needing freezing storage, etc.). After the vacuum (i.e. pressure) is applied, fibers can adhere to each other and expel the resin in excess, with the remaining one curing and guarantying the packed structure of the fibers. The curing process may vary from a few hours to days, depending on the type of epoxy resin. Different resins cure at different temperatures and with different curing times. In general, the higher the curing temperature, the higher the temperature the final part can operate at without having the resin melting back to liquid.
There is the misconception of high curing temperatures always being a good and premium parameter (i.e. F1 car), however, that is not always the case. Every application should be designed properly. Higher curing temperatures leave higher residual stresses within the part after cooling, and it can lead to lower structural performances at lower temperatures. Therefore, the right curing process must be customized on the application.
Considering the tools at my disposal, I was fortunate enough to have only a couple of options:
Pressure: replace the vacuum bag with insulating tape tightened around the part.
Heat: replace the oven with a hand-made plastic bag which would be heated with a domestic electric heather.
Considerations on the two choices above:
1 - Even though I had a vacuum pump and the plastic bag to make a proper vacuum bag in a few minutes, creating a good vacuum around the part I was dealing with would not be easy - complex geometry with open ends. Moreover, tightened insulating tape around the part would be fairly easy since I was dealing with a positive geometry (i.e. convex surfaces) – in the case of negative surfaces (i.e. concave) vacuum bags are almost the only way to go.
2 - After the catalyst is added to the resin before impregnating the fibers, the epoxy resin I used would cure at room temperature in about 24 hours. That allows for a post-cure of about 2 hours at about 80 degrees C. Not having any oven big enough to fit the frame, I decided to accept about 50 degrees C as post-curing temperature and do it with the domestic heater. Even if I had had an oven at my disposal, I probably should have not reached 80 degrees C: being a repair, heating a part to 80 or 100 degrees could give unexpected results being the underlying structure aluminum, which was glued to carbon fiber parts, which had on top of everything my carbon-fiber new reinforcement … the structure had different thermal expansion coefficients which could result in strange behavior of the structure and unexpected residual stresses after cooling (e.g. aluminum transfers heat very well). Moreover, not knowing at what temperature the existing carbon fiber of the rear-end had been cured at - and which cannot be surpassed a second time not to melt the old and already cured resin - I was better not to reach the 80 degrees mark. In addition, the rear-end had probably been glued to the aluminum triangle with some bicomponent glue - think about it as the same resin we were impregnating the new fibers with – therefore, we had again the issue of melting the glued joint too.
Now the fun part. Carbon-fiber composites are highly anisotropic. We can think about fibers of carbon as ropes. The sheets that we work with are just made of bunches of ropes loosely woven together. While fibers are extremely tough along their length, with a little pull orthogonal to that direction, we could tear a dry carbon fiber sheet apart with no effort. In my case, I used two types of clothes: twill clothes (200 gr/m^2), in which the fibers are woven in the two main directions at 90 degrees (figure 2/a), and unidirectional clothes (125 gr/m^2) in which all the fibers are aligned in one unique direction (figure 2/b) – note, the sheets have some white stitching to help hold the fibers together; however, the actual fibers’ directions are highlighted by the red lines in the pictures.
My choices represent the result of a quick stress analysis. I had to guarantee both torsional and bending stiffness to the structure to balance the torsional and bending moment generated during the worst possible conditions, which we can think of as the moment in which the rider is pedaling off the saddle, figure 3. The result is a twisting torque in addition to a bending moment on the vertical tube of the central triangle (purple and red arrow in figure 4, left).
In general, those stresses propagate across round structures like the one we are dealing with, respectively as:
Two perpendicular tensions at 90 degrees between each other and at 45 degrees from the longitudinal axis of the tube (purple perpendicular arrows in figure 4, right).
Two parallel tensions at 0 degrees from the same axis (red parallel arrows in figure 4, right).
Figure 4: generated twisting torque and bending moment (left), and resulting tensions respectively at 90 degrees and parallel between each other (right)
I laid down the fibers in such a way that the 90-degree twill would resist all the tensions of figure 4. By aligning the fibers of the 90-degree twill with the tensions at 90 degrees generated by the twisting torque, I would allow the fibers to exactly match them. At the same time, the vectorially resulting and resisting force of the fibers would be exactly aligned with the parallel tensions given by the bending moment. Of course, fibers cannot provide any resistance against compression, which is provided from the structure as a whole … more on this probably on a following [engineering] part-2 of this post. Finally, the unidirectional fibers were used mainly to hold everything together as a lace. The actual final fibers’ layup is shown in figure 5.
Figure 5: fibers layup. In red are the directions of the 90-degree twills’ fibers, and in orange are the directions of the unidirectional sheets’ fibers. While the orange fibers resemble a 90-degree twill as well, they are in reality two different unidirectional sheets, placed, one in one direction and the other almost at 90 degrees; that configuration is just the result of trying to fit the four main cavities around the frame to hold it together.
The sheets were laid in such a way to have at each spot about 5 layers of clothes overlapping. The series in figures 6 show the pouring of the resin on the fibers and the laying out process.
At the end of the positioning phase, the part was taped and plastic ties were applied to hold the tape and prevent it from loosening with the heat provided during the post-curing phase (figure 7).
The part was left curing at room temperature for about 24 hrs, and then heated for 2 hours at about 50 degrees C through a plastic bag and a domestic heater (figure 8) – note, the plastic used for the bag and the bi-adhesive used to hold it together are specific for actual vacuum begging and they can withstand high temperatures.
In the end, I had to clean-cut the ends of the fibers, paint the junctures, and spray with transparent paint the carbon-fibers – a transparent coating is still recommended for protection against UV and similar. The result is shown in figure 9.
Final considerations: mistakes and what I would do differently
I am very satisfied with the result which allowed me to keep riding my bike, and most importantly save a few thousand $/EUR. The material I used add-up to less than 100 $/EUR (e.g. the carbon-fiber sheets were about 20 $/EUR, the resin was a comparable amount, and then some miscellaneous). The time the job took was something more than a weekend and it was mostly due to waiting time for curing and drying of the primer and of the glossy painting finally applied. However, I do have things that I would do differently. Here are the main improvements I would apply, with a big final consideration on building composites on aluminum.
1/4 - Finishing and aesthetics:
As it is possible to see from the pictures above, while the sheets and the fibers were carefully lined up according to the stress analysis described above, the result does not provide that visual precision we may be used to when it comes to high-end carbon-fiber composites. That is because I disregarded any esthetic concepts: I did not use any of those sheets of carbon-fiber specific for finishing and which are usually used in the industry, nor I cared about aligning the sheets of the final layer all in the same direction to provide visual order – I aligned the fibers according to the stress analysis till the last layer, while commercial products have the last layer orientated mainly according to aesthetic principles. After seeing the result, I would at least align the final sheets all in the same direction, even without using any specific cloth for the last layer.
2/4 - Creasing and mechanical performance:
Moreover, while the part I was repairing was not easy to cover with carbon-fiber sheets (not a regular surface like a panel) the result has too many creases. Those spots, other than issues with aesthetics, can compromise performances being spots where cracks can generate and where toughness is not guaranteed – resistance against tension not guaranteed by the fibers which are not stretched; the resistance is locally provided by the resin, which in turn has a tensile strength as low as 1/10 that of the fibers. Still, at least in the medium-term, I do not think they are going to cause structural problems considering five layers of sheets which for a repair are more than enough. Probably, I should have not applied too much tension through the tape, while I should have used mainly the plastic ties to apply a more uniform pressure simulating the constriction of a vacuum bag.
3/4 - Post-curing system:
Having to do everything all over again, I would also think about building a different heating bag for the post-curing phase. Even though I have already explained that surpassing my post-curing 50 degrees C could cause issues in my case and my solution was more than enough to heat the composite, my bag and heater system had issues of its own: temperature slow to rise, leaks causing temperature drops, etc... The post-cure heating phase is extremely delicate and the process must be precise to guarantee mechanical performance. Moreover, an heater is not the same of an oven because it does not really heat the air inside an ambient through an heating element like a resistance, rather, it tries to push in heated air and it needs an open end to push cold air out ... not immediate to properly balance. Thinking about doing a couple more of similar jobs, I would build an oven with some insulating panels and heating elements like light bulbs. Therefore, even sticking to my 50 degrees C, a slightly different do-it-yourself bagging and heating system would be advisable.
4/4 Carbon fiber on aluminum:
Finally, a big consideration on my carbon-fiber repair being done on an aluminum structure - I cannot tell yet whether changes are needed. The bond of carbon-fiber composites to aluminum is not easy to guarantee in general. Especially on new parts where the aluminum is likely to be oily from production, carbon-fibers and resins are likely not to stick. In my case, the complex structure helped a bit. The shape should be able to retain the carbon-fiber “box” around it even just through mechanical interference. However, I took an initial big bet: before applying any carbon-fiber, I decided not to remove entirely the old paint from the aluminum; I applied the wet fibers directly on the old painted aluminum after just a medium scratching with some sandpaper and some decreasing with alcohol. While I was initially heavily scratching the old paint to remove it entirely, I saw the paint was tougher than I expected. Having the paint come off the aluminum would imply removing a thick layer of aluminum since they had bonded almost entirely. I also consulted with a body-shop mechanic friend of mine who instructed me on specific industrial processes through which aluminum is painted from original manufacturers and which aim at creating almost a unique element without separation of surfaces … long story short, I decided to just create a rough surface with light sandpaper on the old paint on which I applied the new composite. Only time will tell whether I made the right choice.
For those interested, a part-2 of this post may follow with more technical notes and numbers on carbon-fibers, resins, design, and composites.