Invisible carbon damage?

Can someone explain to me in plain language how carbon fiber could sustain damage that is not visible on the surface? My understanding is that carbon fiber parts (rim brake tracks excluded) is really only vulnerable to impact damage that would, by definition, show evidence of itself on the surface, where the impact happened. Thanks.

1 Like

A well known carbon bike repairer in Melbourne (who is advertising his services in the same article) says it’s because of the possibility of delaminating.

To me it seems like a reasonable argument: if bike is metal it is just a layer of paint then metal all the way to the ID of the tube. no impact could deform the metal without it showing. A material that is multiple layers, and multiple materials with slightly different elasticity (epoxy and cloth,) might be struck or squeezed and the outer layer spring back but with some broken fibres or separated layers behind. How often this happens in practice I could not say.

Not all damage to carbon parts (frames, wheels, etc.) is caused by direct impact–if the part is twisted enough, such as sometimes happens in car accidents, where the car actually hit the rider, not the frame, but the frame still got slightly twisted, a crack can occur below the surface cover of paint, which over time, with use, will move around, loosening up enough to cause the paint coat to eventually break up along the fissure line–have seen this a few times in the shop.

It’s not just impacts that can damage it. Carbon fibre is a composite material, meaning it comprises different materials with different properties. These are formed with components like kevlar weaves and resin to glue it all together. The heterogeneous nature of the material means that stresses are not distributed equally through it. Particularly where there are weaknesses, e.g. voids in the forming process (gaps) - these can mean the forces are concentrated such that stresses are very high at certain points. Once the stress exceeds a threshold, you get damage such as cracks, which concentrate the stresses even more and that’s how cracks propagate. That doesn’t necessarily happen at the surface, meaning you can’t necessarily see it. The cracks don’t necessarily propagate at a steady rate, as the stresses increase as the crack grows to a point where it all gets too much and the crack grows rapidly. That’s catastrophic failure where the frame just gives up almost instantly.


There’s a recent interview with Ruckus Composites here. There are some actual photos of sub-surface delaminations that wouldn’t be visible to the eye.

1 Like

Ok, that’s great information from a materials perspective, and I appreciate people taking the time. It just seems to me that if this were an issue of true concern for everyday cyclists, it would render carbon fiber an inappropriate material for bike things. How would one even begin to suspect this invisible damage was present, without some testing apparatus like an X-Ray to scan the bike?

So if I am reading this correctly, a poorly manufactured frame or rim or whatever, with weak spots/voids could develop issues like this from normal use, potentially without ever being crashed or dropped or anything like that. Then, at some critical point, this issue gets bad enough that the frame just fails completely. This seems like a pretty scary scenario, but also a very unlikely scenario, based on how many aged carbon frames are on the road and still seem to be performing just fine?

It’s not unique to carbon fibre. Aluminium alloys can also fail catastrophically. High carbon steel apparently tends to hold on longer with visible/obvious signs of cracking.

No one’s saying it happens frequently. But it does happen. There are no guarantees in life. If you’re looking for proof that carbon fibre frames can’t fail without warning, you won’t find it. It’s unusual but it’s possible.

1 Like

Even a well manufactured carbon product can do that: Carbon fibre composites have much lower Weibull moduli than metals, meaning the spread of the probability distribution of failure is much wider.

1 Like

In practice it’s not worth fussing over. After falls or crashes you can do a tap test to listen for anything untowards. If you feel something is off then you want to inspect it, sure.

However, after riding CF for a while (and experiencing the odd crash or fall) you learn to trust the material. I’ve had a steel downtube snap with zero warning after less miles than my CF bike has - but the bike was held together by the top tube enough to come to a stop safely.


Failure happens but thanksfully not as often as we would be afraid You don’t want to be the one running the bad batch though and appart from getting your frame/parts scanned, it is impossible to know beforehand.

More common issues have been galvanic corrosion on frames with mixed material. It could be frames mixing alu and carbon or just some alu inserts, for example to provide threaded BB, or axle/pedals inserts on cranks. For example raceface carbon cranks were (are still?) known for failing at those areas, I don’t know of any owner who hasn’t got his crank replaced under warranty at least once.

Check out some of the videos from Luescher Teknik on YouTube. He does a good job of explaining and provide visuals.

1 Like

Can I probe you a bit to see what you mean?

The Weibull distribution is frequently used in survival analysis. The modulus is the parameter k in the parameterization on the Wikipedia article. Values of k under 1 indicate that the failure rate decreases over time, i.e. the probability of failure decreases with time. Values over 1 indicate that the failure rate increases over time.

Did you actually mean that with CF, the probability of failure decreases faster with time than with metal frames? Or something else? I think the spread in the probability of failure is not what you meant - by referring to the Weibull distribution, you are implying that each class of frame has a probability of failure described by the distribution and the parameters therein (there’s another parameter, lambda, and that plus k describe the mean and median lifespan).

Alternatively, did you mean that with CF frames, failures a long time after manufacture are still possible, just less common than with metal frames?

(It’s been some time since I’ve taken survival analysis, and I usually don’t use parametric models, so the above is not guaranteed to be 100% correct but it should be mostly on point.)

As I understand it, the Weibull distribution is used differently in different fields.

To quote the article you linked:

and then later

I was using the term in the second sense whilst I think you are referring to the first.