Why Bolts Come Loose

Fenders, racks and other attachments can remain maintenance-free for 10,000+ miles, yet other bikes require tightening bolts on a regular basis. Why do some bolts stay tight, while others loosen quickly?

We have covered this in detail in Bicycle Quarterly’s article “Engineering a Bicycle” in Vol. 5, No. 4. The essential concept is simple: Vibrations of parts allow bolts and nuts to loosen. (At the end of the post is a link to a video that shows this in action.) The more parts you have that vibrate, the quicker your bolts loosen. Here is how to avoid this:

Do not use the same bolt to attach several things. The more “layers” you have, the more movement results, and the quicker the bolt will loosen. This is especially true for front racks that are attached to the posts of cantilever (or centerpull) brakes.

The rack above is more likely to come loose than the one below, which uses a bolt with a forward extension, onto which the rack is mounted.

Even simple metal washers increase the risk of bolts coming loose by adding another layer between bolt and the surface into which the bolt threads. Spacers are even worse, because they provide a longer lever arm for the forces that vibrate the bolt head.

Washers should be used only where they are needed, for example, to provide a smooth surface when a bolt head tightens onto paint or aluminum. Spacers should be avoided in most situations. (It is better to make your parts to fit together properly.)

All connections should be designed so that the bolts won’t tend to rotate. For example, the rack below attaches to the fork crown with straight tubes that are aligned to the vibrations of the rack (forward and backward).

If these connecting tubes were angled or even curved, then each vibration would tend to turn the bolt.

Another neat detail of this particular rack is the forked mount for the lower bolts that carry the weight of the load. This is the best possible attachment. Here is why:

With a simple mount (above), the surface underneath the bolt head and the surface underneath the nut move independently from each other. The bolts loosens as the parts vibrate.

With a forked mount (above), the two surfaces (under the bolt’s head and under the nut) no longer move against each other, so the bolt experiences much less force that could loosen it.

There is more to consider when designing good connections. And it’s true that even sub-optimal connections can work fine. The bottom line is simple: On a well-designed bike, there should be little need to re-tighten bolts.

Recently, Peter Meilstrup posted a link to video that shows how bolts loosen. I find it especially impressive to see how the nuts turn as they vibrate. And commonly used lockwashers actually make things worse! It is obvious that there is no substitute for good design. Click here to see the video.

About Jan Heine, Editor, Bicycle Quarterly

Spirited rides that zig-zag across mountain ranges. Bicycle Quarterly magazine and its sister company, Compass Cycles, that turns our research into high-performance components for real-world riders.
This entry was posted in Testing and Tech. Bookmark the permalink.

16 Responses to Why Bolts Come Loose

  1. Garth says:

    Split lock washers also tend to spread apart when tightened. They cannot be highly torqued.
    I wonder how wave-type lock washers compare.

    Interestingly, overtightening can be more detrimental than undertightening. Read Duane Ausherman’s information on his experience with BMW motorcycles. A good read: http://w6rec.com/duane/bmw/torque/index.htm

    Threaded bolts are easily taken for granted. I have come to think of them as similar to springs. When tightened, bolts stretch, thus applying clamping force.

  2. Jerry S. says:

    Good information! Thanks! I thought I knew a thing or two about fasteners but seems I had more to learn. Probably a good thing I’ve been Loctite-ing everything.

  3. Fred Blasdel says:

    Except that almost nothing on modern bikes uses nuts that can turn independently — only historical reenactment bikes like yours have unthreaded holes with unkeyed nuts.

    For the rest of us, most of our loosened bolted joints are because one of the parts rotated around the bolt the wrong way.

    • You are right, the worst scenario is the part rotating around the bolt, for example, a light that “stands” on its mount and thus tends to rotate, taking the bolt with it. A similar arrangement is the “R bolt” for attaching fenders, which means that each time the fender stay rotates a bit, it tends to turn the bolt.

      But that mechanism isn’t the only reason bolts come loose on many current bikes that we have tested for Bicycle Quarterly, and vibrations don’t need nuts to loosen bolts. All the nut does is adding another layer that vibrates. On a part that is directly threaded into the frame, the frame acts as the nut. The principle is the same: A bolt will come loose if vibrations cause it to bend as shown in the video. And the more layers you add, the quicker it loosens.

      For example, if you mount your fender to the seatstay bridge with a spacer between fender and the frame braze-on, the fender bolt is likely to come loose, even if it threads directly into the bridge. If you mount a rack onto the canti bolts, you also risk having the canti bolts come loose.

      I am not sure I like the term “historical reenactment bikes,” but in any case, I’d think if refers to classic racing bikes that aren’t used for racing, but ridden by people who like the aesthetic and feel of such bikes… (not that there is anything wrong with that.) It’s hard to see how a randonneur bike that is used for randonneuring is a reenactment in any way. It’s simply a tool, optimized for the purpose at hand.

      In any case, on my bikes, there are almost no unthreaded holes with bolts and nuts, except for the fenders, where they are required to pull the drawbolt through the hole. (That attachment is more secure than a threaded hole with an R clamp – see above.) Bolts and nuts more often are found on modern bikes that are retrofitted with fenders, racks and the like.

      • Fred Blasdel says:

        You had a new bike built around a Nivex derailleur. There ain’t much more historical reenactment than that, and I’m pretty sure when e-Ritchie coined it as a pejorative he was referring to your aesthetic.

        I think the correct solution to the fender stay problem is to abandon the historical Lefol-style hardware entirely. Berthoud gets it right by taking advantage of modern materials and machining, Honjo and VO get it wrong in their reenactment.

      • I can see that a Nivex derailleur, metal fenders and 650B tires may appear like nostalgic choices, but that is not why I use them. I chose the Nivex derailleur for very practical reasons: Better shifting (no return spring), constant chain tension, light weight and less risk of damage if the bike falls over. I wonder whether you consider the BB30 on a Calfee carbon bike a “reenactment” of the pressed-in bottom bracket bearings of Singer and Herse.

        I always though that e-Ritchie was talking about the bikes he makes when he used the term “reenactment bikes”: Even his “Newvex” lugs are “reenactments” of the classic Nervex style. I think they are neat, and I took it as e-Ritchie being self-deprecating… There is nothing wrong with riding a classic racing bike, even if you don’t race. It’s an aesthetic choice, and not a bad one at that.

        The Berthoud fender attachment simply is a copy of the method used by Jo Routens, no modern materials and machining needed. It’s a good design, but the classic drawbolts work very well, too. Berthoud’s “R clips” tend to loosen, because the plastic doesn’t allow you to tighten the bolts enough, and the fender stay is off-center, so you get the problem of the whole assembly rotating that you mentioned in your earlier comment.

      • Fred Blasdel says:

        Routens used stays with a flat spot at the center, drilled to use two M4 bolts to mate it directly to the fender?

        The metal R-clips can be tightened even less, and are inconsistent enough that some can’t be tightened enough to keep the stay from moving back and forth through them with a gentle tug. The cast + machined plastic that Berthoud uses there means that they consistently hold the stay and don’t need a washer.

        With the huge lever arm that the stay presents, the slight offset at the dropout is irrelevant. The way they attach at the fender and the massive rigidity increase of the stainless steel works in their favor at a scale orders of magnitude larger.

      • Routens used stays with a flat spot at the center, drilled to use two M4 bolts to mate it directly to the fender?

        Yes. Check the Routens bikes in The Golden Age of Handbuilt Bicycles, and you’ll see that attachment.

        The fenders on my Singer – aluminum and with standard drawbolts (Daruma) – have not come loose in six years of hard riding and more than 40,000 miles. I don’t see any need to change that system.

        The Routens/Berthoud method has the advantage to reduce toe overlap by keeping the stay close to the fender.

  4. Martin Visser says:

    Disagree with your “washers increase the risk of bolts coming loose “.
    IMO the opposite is true. If you run a square taper crank bolt without a washer the crank WILL loosen itself. That’s why there is always a washer there.

    ps, appreciate your blog.

    • You are right, there are places where washers are essential – as mentioned in the blog – and crank bolts are one of them. You need the washers because the bolt needs to turn on the aluminum crank as you draw the crank onto the spindle. The torque is much greater than the torque for a simple fastener. In place where washers are needed, it is best to make them part of the bolt, as René Herse, Ettore Bugatti and other makers did. That way, you get the benefits of the washer without the vibration between bolt/nut and washer.

      You also need washers on the inside of fenders, as you’d otherwise score the soft aluminum and cause stress risers as you tighten the nuts.

      Washers under bolts where none are needed, for example to attach a steel rack to the frame, or underneath the head of a cantilever brake bolt, do increase the risk of loosening.

  5. Martin says:

    My theory is you need a small amount of dynamic slippage between bolt -washer-part. A washer that is part of the bolt will not do the trick on a crankbolt IMO. Shimano has those and still use a washer IIRC.

    • The crankbolts with built-in washers work very well without extra washers. They cost more to make, though.

    • I looked into the crank bolts a bit more. Most modern crank bolts do have built-in washers. Obviously, the crank bolt has to turn on the aluminum surface, but you could do that without the washers. (If you look carefully at most hex bolts, you see that the corners are rounded, so that there is a smooth, round surface on the backside, and the sharp corners never touch the piece you are attaching.) The built-in washer serves mostly to make the bolt head smaller than the bearing surface of the bolt. On traditional crankbolts, that was needed to create space for the wrench, yet provide the larges possible bearing surface for the bolt. On modern bolts, it mostly serves to save weight and improve appearance.

      When you look at the brake attachments in the blog post, you see Mafac using a bolt with a very thin, large 12 mm head and no metal washer. Herse used his own bolt with a 12 mm built-in washer and a 10 mm head. Mechanically, they are the same. (The actual bolt design is a bit hard to see on the Rebour drawings, as Rebour was not focusing on that detail.)

  6. marmotte says:

    @Fred Blasdel: modern bikes versus historical reenactment bikes? Are you sure, this is really the right way to look at things. Why is “modern” always associated with “better”? There are scores of things that prove this assumption’s is not true (as would be the opposite, there’s a lot of old cr** around, too)

  7. Thomas G says:

    While i agree that much bolt-loosening can be averted by good design, sometimes you find yourself stuck with a loosening-prone bolt.
    Enter Nord-lock: http://www.nord-lock.com/default.asp?url=2.16.37
    I have successfully used Nord-Lock washers on my bike in critical locations for 3 years and had no loose bolts.
    Highly recommended for retrofitting on less than optimally designed bikes.

Comments are closed.