Yan 

Let me explain using an example: you are familiar with fridge magnets. The magnetic force is pulling the magnet sideways into the fridge. But wait, if the magnetic force is sideways and not upwards, then how can it counteract the downward force of gravity?

Answer: it is the friction between the magnet and the fridge that counteracts the downward force of gravity. Where does this friction come from? The friction comes from the sideways force pulling the magnet against the fridge. If the magnet were to be suddenly turned off, the friction force would instantly disappear and the magnet would immediately fall.

The spacer is very much structural. The spacer allows the screw to pull the entire system tightly into the fork. The resulting perpendicular friction is what holds the rack up. If the spacer disappears, then there is no tension at all. Friction gone. At this point all the pieces are just hanging out in space, and the only thing holding up all the weight is the wimpy screw itself. The screw is not engineered to serve this type of load and will fail very quickly. It will look like the photo below. The outside part of the screw will shear off at the surface of the fork, leaving a bit of unremovable screw permanently stuck inside your fork.

This type of failure is the most common type of failure in bicycle rack screws. The screw doesn't even have to be cantilevered out. Even in a regular rack installation without a spacer, the screw will shear like this if it vibrates itself loose. As soon as the friction disappears, the screw is on the way to failure. That's why you have to check your screws for tightness regularly.

You could trying buying a nylon nut or similar plastic object from Home Depot and shaving it to the correct wedge shape to act as a spacer.
https://www.homedepot.com/p/4-40-Nyl...4648/204274585

Alternatively, forks are only $100. Just get another one and be done with it. Make sure you match the existing fork's geometry if you don't want to change the bike handling.
https://surlybikes.com/parts/forks

smasha

The spacers that came with my stainless Tubus rack are stainless. So far as they need to withstand a compression force of the bolt, they are structural. I don't think I've ever seen a rack that came with plastic spacers.

Yeah, the nut is tightened against the bolt-head. I can tighten that pretty much to the rated limits of the bolt; I'm not worried about crushing the flange or the washers. That tension will keep the bolt from turning, and maintain a small clearance between the rack and the fork.

john m flores 

If the tension of the bolt is carrying the rack load, then try replacing the bolt with a high tension cable or a length of high load fishing line (looped multiple times if necessary) running through the aluminum spacer.

When those solutions fail, ask yourself what the difference between those alternatives and the bolt is? The answer is the sheer and bending strength of the bolt, properties that cables and wires - designed to act only in tension - don't have.

smasha

The Trek FX2 fork is on my radar, if I can't find a good solution with the Escape/ Cross City fork. I like aluminium.

As I understand your explanation, it doesn't even need a bolt. A cable, under tension, inside a spacer, could hold the parts securely. Release the tension, and it all falls apart.

Are you saying that even a bolt, in a similar configuration, will fail if it's not under lengthwise tension? Even if it's only 2mm of bolt that's not under tension?

About the forces... In terms of leverage forces on a bolt: More leverage is worse, less leverage is less worse. Ideally, there's zero leverage on the bolt, between the two surfaces being bolted together. This ideal requires a flush mating surface. Right?

If I had 7mm of naked/not-tensioned bolt between the rack flange and the fork, that would be kind of bad. Right?

If tension between the bolt-head and the nut effectively leaves only 2mm of naked/not-tensioned bolt sticking out of the fork, that's less bad than 7mm of naked/not-tensioned bolt sticking out of the fork. Right?

The questions, then are:

1- Is 2mm of naked/not-tensioned 304 stainless M5 bolt asking for trouble, in this application?

2- Could this be alleviated/resolved by hand-shaping a plastic spacer to sit flush with the fork? I would think the plastic would be to soft to have the same structural effect as a steel spacer.

3- Could a steel spacer (7mm long) be a good/ideal solution, if the face around the fork eyelet was faced/counterbored? Would this require a filleted counterbore? Would this be a bad idea, as it could compromise the structural integrity of the fork?

smasha

ChatGPT says it's ok, so it must be ok, right...?


With different incarnations of GPT, it's telling me the safe static load (accounting for cyclic loading and vibration) is anywhere between 144kg and >1500kg. I can't reasonably get it to estimate a safe static load <144kg. If that's close to reality, then 9kg is fine.

What I want to know, is if that is close to reality.

Yan 

A high tension cable would indeed work. It would need to be a magical fishing line with the same tensile strength as a 5mm diameter metal screw, but yes, if you found such a magically strong cable, the rack would work perfectly. The strength of the connection is coming from the friction between the mated surfaces. The assistance provided by the screw itself is secondary. Loosen your screw by a turn and then go on a tour. That screw will shear very quickly.

This high tech fiber would probably cost a lot more than a screw though. Plus you'd need a way to both fix and tighten it, so now you're back to using more metal. That's why we use screws instead.

Yan 

ChatGPT is a language model, it doesn't do math. It answers simple math correctly because at some point in the past it digested a bit of sample text that it is regurgitating back to you. Anything complex and it falls apart. You're not concerned about static load. You're concerned about dynamic load. Any time your bike hits a bump, that's a dynamic load. The formula for dynamic load is force = mass x acceleration. For example if your combined rack and luggage load is 10kg, and you hit a bump that gives you 4g of acceleration, then the dynamic load is f = 10kg x (9.8m/s/s x 4) = 392 Newtons = roughly 40 kgf; That 40kgf is shared (imperfectly) between 4 bolts.

Here is a discussion on bolt pre-tension:
https://www.eng-tips.com/viewthread.cfm?qid=438548

Summary:
"If the bolts are clamped down hard enough, plates are held by friction, and there is no shear load on the bolt. If the bolt has come loose or is deliberately assembled loosely, there is no friction, and the joint is held entirely by shear. There is a transition where there is some friction clamping and some shear, but for most joints, this is a very tiny range of nut movement. Would you really try to design for this state? This is mostly a failure condition. If a bolt is used properly it induces enough friction that the bolt is only in tension. Your observation is a good example of why bolts should be properly torqued."

I suggest you ride the rack as you have it. If your 9kg planned load ends up light enough and everything is ok, great. If it fails, you're back to where you started and haven't lost anything. I would not suggest modifying the fork in anyway if you wish to keep all your teeth.

john m flores 

The sheer strength of a low-grade M5 bolt is about 600#, so it'll be fine backed out a turn.

I'm out. Good luck to the OP.

Yan 

The failure mode is progressive cracking via fatigue, not instantaneous exceeding of ultimate shear resistance. Every thread on the bolt is a stress riser that initiates cracking. Here's an explainer of the mechanism of failure:
https://www.boltscience.com/pages/fa...e-of-bolts.pdf


"The crack can start at some existing defect, such as an inclusion in the metal, or at point of high stress, such as a notch. In the vast majority of applications, the most effective way to ensure that the bolt is fatigue resistant is to ensure that it is tightened sufficiently. A preloaded bolt in a typical joint sustains usually only around 5% (or less) of the applied loading (the remaining 95% reduces the clamp force acting on the joint). Due to this, a properly tightened bolt is highly resistant to fatigue loading. (A conn-rod bolt for example would rapidly fail if it were not tightened.) Because the alternating load is small, so is the alternating stress, usually well below the bolt's endurance limit. Fatigue failures, when they do occur, are frequently the result of inadequate tightening or loosening that can expose the bolt to bending stresses that it is not designed to sustain."

smasha

Yes and no. For the purposes of calculating stresses and forces, it's all about dynamic load. At some point, I need to translate that safe dynamic load into a static load, ie how heavy can I safely pack my pannier bags.

As for GPT, I've seen it give some incredibly stupid answers with confidence. In this case, it explained the calculations and how it made assumptions about cyclic loading and vibrations for bicycle applications, but I'm smart enough to know that I have no idea if those assumptions and calculations are correct. I was able to get answers that varied by more than an order of magnitude, but even on the low end, it says I'm fine with 9kg per bag.

In a few years, I may have a mechanical engineering AI app on my phone that can give good answers to this. For now, trying to get GPT's opinion(s) on this is as much about entertaining myself and trying to understand GPT as it is about finding the right answer. As you noted, GPT is a language model, not a mechanical engineering model, not a maths model, not a physics model. Next time, I'll ask GPT to write a poem about the problem and it'll probably be impressive.

As you've described the tension forces on a bolt, an ideal solution would be using a spacer that mates flush with both the inside of the rack flange and the outside of the fork eyelet. Clearly, the curved face of the fork eyelet is causing problems. In the pursuit of a better/stronger/safer solution, the question then becomes whether or not I can get a flush mating surface between the eyelet face and a spacer. Everything that I can think of, and everything that's been suggested, seems like at best it would not be flush (exacerbating bending forces under load and tension), and at worst it would cause the bolt to bend under tension. Actually, exacerbating bending forces under load and tension could probably be worse than just a bending force under tension.

smasha

Not sure what's worse... GPT's mechanical analysis or its poetry...

The pannier hangs upon the fork,
But how it's mounted is the talk.
Two methods vie for strength and might,
Each offering its own delight.

Method one, a bolt alone,
With washer and spacer to hold its own.
Tightened with care and torqued just right,
It holds the load with all its might.

Method two, a bolt and nut,
With washer and flange, a simple strut.
Tightened down with force and will,
It holds the weight and doesn't spill.

But which is stronger, which is best?
It's hard to say, put to the test.
The bolt alone or bolt and nut,
Both have merit, both have guts.

So ride with pride, and carry on,
With pannier mounted, and bolted strong.
The road ahead, with twists and turns,
And either way, the load will hold.

grumpus

Fair enough. I like steel bikes, I have a couple of aluminium framed bikes that I've built up from odd bits and pieces but they both have steel forks - I prefer the feel of the springy fork while the aluminium frames are rigid for reduced flex with heavy loads. Steel is great for modifying if you need extra braze-ons. ;-)
I don't think I've ever ridden a titanium bike. We have a titanium wheelchair, the frame has fatigue cracks. I may try welding it.

smasha

Not my area of expertise, but as I understand it, titanium needs to be welded in an oxygen-free environment.

grumpus

It does, but it shouldn't be too hard to create that - argon is sufficiently denser than air that it will sit for a while in an open container, so building a dam around the weld will keep the shielding gas in place while the metal cools. Along with ample post-flow and back-purging, that should be enough for a small joint like this, or at least half of it, then rotate and repeat. Not aerospace quality, but sufficient for this application.