Re: chain type question

[in4]

I’ve always used 1/2 x3/32 KMC X1 chains on my MK2 Nomad. No issues at all.
Out of curiosity I went ChatGPT ing ( fatal time waster btw) and it was suggesting a 1/8 would last longer than a 3/32. If that were so I’m curious what the difference might be. Here’s the ChatGPT partial quote for reference:

With a Rohloff hub, bicycle chain longevity tends to be significantly better than on derailleur-based drivetrains because:
   •   The chain always runs straight (no cross-chaining).
   •   There’s less shifting stress.
   •   There’s typically better protection from dirt (especially on touring setups).

Expected Chain Life with a Rohloff Hub
   •   Typical range: 10,000–20,000 km
   •   Well-maintained chains (e.g. oil-lubed, kept clean): up to 25,000+ km
   •   In harsh conditions (e.g. grit, salt, or mud): as low as 8,000–10,000 km

The actual lifespan depends on:
   •   Maintenance (cleaning/lubing frequency)
   •   Chain quality (e.g. KMC X1 vs basic models)
   •   Riding conditions (gravel, rain, sand, winter salt)
   •   Load and torque (touring gear, rider weight, terrain)

Tips to Maximize Chain Life with Rohloff:
   •   Use a 1/8” chain (heavier duty and longer lasting than 3/32”).
   •   Consider chain tensioners or eccentric bottom brackets to maintain ideal tension.
   •   Use a chain wear checker (0.5% stretch for 1/8”, 0.75% for 3/32”) to avoid excess sprocket wear.
   •   Replace the chain before it’s excessively worn to extend the life of the rear sprocket and front chainring, which are harder to replace than the chain.

Would you like chain recommendations for a Rohloff-specific touring or bikepacking setup?

[mickeg]

I have never used any form of AI for anything, other than when I do a Google search then Google does use AI in the process.

So, are you asking a question here?  Are you asking if you should go to 1/8 inch single speed chain?

I have never looked into the KMC Rohloff sprockets, but I think they use a 1/8 inch chain.  So, for this reply I did a Google search to find more info on the KMC sprockets.
https://www.kmcchain.eu/e-bike-solutions/sprockets

I was surprised to learn that the KMC sprockets are aluminum, not steel.  I would expect much faster wear on the KMC sprockets for that reason.  So, if I was going to try a 1/8 inch chain, I would use the Rohloff sprockets.

I am surprised that the AI system did not find this page on the late Sheldon Brown website, I follow these recommendations for my Rohloff bike to extend chain and sprocket life.
https://www.sheldonbrown.com/chain-life.html

Most on this forum bought their Rohloff hub from Thorn/SJS.  I bought my frame and fork from Thorn, but I built up my bike from parts and bought my Rohloff elsewhere.  For that reason, most on this forum use a 17T sprocket (the Thorn standard) and I use a 16T sprocket which is the default sprocket when you buy a Rohloff hub from a retailer.  Thus, I might be the only one on this forum that uses a sprocket with an even number of teeth as suggested by the late Sheldon.

Another point that I would have expected the AI system to mention was that if you use larger chainrings and sprockets, that should also extend chain life as the wear occurs when the chain flexes when it comes off of the sprocket and straightens out, and again when the chain flexes as that the wraps onto the chainring.  In other words, the wear is on the part of the chain that is under load during the flexing on that chain.  Larger sprockets and chainrings result in smaller flex angles, thus less wear.  Also there is less tension on the chain when that occurs with larger sprockets and chainrings, thus less wear for a second reason.

It does look like the AI system used generic chain wear percentages for replacement.  I run my Rohloff chain to 1 percent elongation, but my derailleur bikes (most are eight speed drivetrains) I change the chain at 0.75 percent. 

So, I would say that the AI system failed rather badly.

Most of my distance is ridden on derailleur bikes, not my Rohloff bike, so my chain experience on my Rohloff bike does not really matter here.  I will actually take a chain that was worn out on a derailleur bike to 0.75 percent, then put that on the Rohloff bike to get some more wear out of that chain before I discard it.  At this time I have three chains that were worn to 0.75 percent elongation set aside to put on my Rohloff bike later.

[PH]

I’ve always used 1/2 x3/32 KMC X1 chains on my MK2 Nomad. No issues at all.

The pins are around a third longer, and it's those that wear.  All other things being equal, you might expect a third longer usage.  Though decent quality 1/8th chains are not as easy to source as 3/32th, there's a lot of cheap wide chains that are unlikely to have the same degree of pin hardening.  KMC's top 1/8th chain is 20% dearer than the narrower equivalent. Also for it to run as smooth, you'd also want to match it with a 1/8th sprocket and chainring, KMC do the sprocket, but it's aluminium and considerably dearer.  Likewise expect to pay more for the thicker chainring.  I can see how it might last longer, I'm doubtful it'd be better value.

AI has no intelligence of it's own, it's merely repackaging what's already available. Sometimes that's useful, other times you're better off looking for yourself.  if you do a google search for anything Rohloff, this forum comes up quite frequently, yet it only has a fairly small contributor base. So a lot of what AI knows it's learnt from the few of us!

[PH]

I am surprised that the AI system did not find this page on the late Sheldon Brown website, I follow these recommendations for my Rohloff bike to extend chain and sprocket life.
https://www.sheldonbrown.com/chain-life.html

We can only guess at the algorithms, but it's been pointed out elsewhere that the search bots weight the information by age.  Which seems a bit ridiculous in the context of anything as established as we're discussing here. 

[in4]

That’s interesting. I thought a heat map might have driven the results. AI is however still very much in learning mode.

[John Saxby]

Interesting exchange, guys - thanks.

Kudos, George, for invoking St Sheldon, and 👎 to AI's evident lack of curiosity and nous.  And, my Scottish ancestors would applaud your frugality in managing your chains 

[Andre Jute]

Kudos, George, for invoking St Sheldon, and 👎 to AI's evident lack of curiosity and nous.

Not to mention Sheldon's partner in the crime of actually knowing what he/they were talking about, the sometime Porsche engineer Jobst Brandt, developments of whose brakes are still used in Formula One. It would be an interesting exercise to ask Mr A. I. Google if tread on bicycle tyres is a good idea. My forecast is that AI will report the popular side's misconceptions and ignore the scientific evidence; that is basically what makes the crowd-sourced Wiki so unreliable: that crowdsourcing in both AI, which calls it "scraping", and Wiki leads to responses hugging the lowest common denominator of stupidity. With the added aggravation in AI that all that is overlaid on its programmer's bias set in concrete in the operational algorithms of any particular brand of AI.

Here's a possible instance. At least one respondent above in this thread believes that AI scrapes this forum disproportionately for reasons he doesn't give but which are easy to imagine: there's a lot of technical bicycle discussion here, and there's a huge readership in relation to a small group of active contributors, both factors just about everyone with and IQ bigger than his shoe size would build into an algorithm. We see a result of this in what AI fed back about chains: AI appears to think eccentric bottom brackets are important totally out of proportion to EBBs in the total population of bottom brackets, or even Rohloff-equipped bikes. Now, where else would AI get such an idea, except on the Thorn forum?

It'll be a good while before AI is ready for primetime, though that hasn't stopped Google from sending it out to be ridiculed.

[in4]

There’s a good story re the accuracy of AI that’s worth bearing in mind.
Someone asked AI how long it should take to wash a football shirt in a washing machine. AI came up with X minutes. The person then asked AI how long it would take to wash 11 football shirts in a washing machine. Apparently the answer was X x 11!