I'm not 100% convinced the figures are true on the real bike. Comparable definitely, but absolute values? Not convinced.
Reading the CeramicSpeed Web site, as U read it the measurement of friction losses are done on a Chain Efficiency Tester (CET) that loads up the chain in equal tension on both halves (top and bottom) running around 2 sprockets (doesn't say what size sprockets). So doesn't represent the real bike situation (other than on a tightly tensioned single speed) where the top of the chain is tensioned by the pedalling power and the bottom only tensioned by the derailleur spring - a tiny fraction of the force in the top part of the chain between chain ring and rear sprocket. Also doesn't reflect the losses from running over the derailleur jockey wheels.

If the installed tension in the CET is high, that will distort the results on an absolute value on-a-bike basis (Try over-tensioning a single speed track bike - and see how much stiffer it is to rotate the cranks !)

So this is one of the hardest things in my world to really accurately measure as an absolute value. The gold standard here is most definitely the system that Jason Smith at Friction Facts, now Ceramicspeed, built, and it uses two separate tests to get to an actual final number. Believe me, everything you can think of related to this test, Jason has thought through 1000 times more deeply.. and in some aspects your point about tension is hot the FTT actually functions.

He and I discuss this with some detail in this interview from Marginal Gains here: https://www.youtube.com/watch?v=wcMWqp_8FWM

While Ceramicspeed and SILCA are competitors in some spaces, we are also friends and I have immense respect for what he has done.. and can promise that his numbers are as good as they come.. there are a lot of brands spouting a bunch of nonsense, they are not one of them.

Having said all of that, your intuition here is correct.. there are so many variables that finding the absolute loss number is insanely hard. One of the big reasons that we do not openly publish loss data is that we don't do FTT testing, we've built a machine that really can only test delta's between chains or between lubricants.. so I can tell you with insane accuracy that this chain is 0.2 watt faster than that one.. but I can't 100% certain tell you what the absolute value of the loss is. For my numbers above, I actually started with absolute values created by Jason on his machine, and then calculated in all the deltas to get to all the other data!!

---

http://www.SILCA.cc
Check out my podcast, inside stories from more than 20 years of product and tech innovation from inside the Pro Peloton and Pro Triathlon worlds!
http://www.marginalgainspodcast.cc

Hope my comment wasn't too 'arsey' - I defo agree it's ever so tricky with so many variables, and measuring pretty small values.

Thanks for the data you've been able to post- really useful to make comparisons and decisions.

I've always understood the "default" powertrain loss to be ~3% for a well-maintained system, not "1-2%".

---

http://bikeblather.blogspot.com/

Certainly agree, and understand the rationale for the bundling, but still think a chain-only offering makes sense, both for those customers who have already purchased lube and those who may order more than one chain (e.g. for different bikes with different chain lengths). First world problems to be sure.

---

Amateur recreational hobbyist cyclist
https://www.strava.com/athletes/337152
https://vimeo.com/user11846099

Hi Tom, though this is older data the study by Martin et al Validation of Mathematical Model for Road Cycling they found that it was 2.3% so it is likely in the range 2 % +/- so 3 is a good conservative number. The paper does not provide any insight into the chain. I was thinking that this model might be able to tease out the number for chain loss. In the paper they compared an SRM to a Monark bike system and found the difference was 2.3% and felt it was due to the chain/ drive system loss. If one were to assume the only change in the system was the chain then one could maybe figure out the impact of the lubrication? What is interesting is they found that the loss was a constant regardless of the power input. So at 80 watts it was ca 2.3% and at 280 watts still ca 2.3%.

So I tend to agree I was just pulling a number from memory (hence the would "say") and so I agree 2-3 would have been a better statement and still makes the argument that chain loss and lubrication can be very important.

If one looks at the data Josh presented with a 9 watt loss for a white lightning (or even triflow at 7 w) chain vs a 2 watts for a super prep'd chaiin that is at a ca. 3-4x improvement.

So assuming say a loss of 2.3% at 280 is 6.4 watts then divide that by 3 and the outcome is an improvement of 4 watts. Does this sort of make sense?.

I'm pretty sure a Monark erg has a drivetrain like a fixie/track bike (i.e. fixed chainring/cog), so add in the additional losses of a derailleur-type transmission, and 3% is looking like the better number.

And yeah...the improvements discussed can, in total result in some pretty good power savings (relatively speaking).

---

http://bikeblather.blogspot.com/

So Tom what you are talking about is drive train losses, what Martin et al and I are talking about in the context of this thread is chain losses. So then if the chain and bottom bracket (since it is almost impossible to isolate the bb alone) loss per Martin is 2.3% then the improvement with the super efficient chain is actually less than 1% per my previous calculation if going from a regular lube to the super prepared chain assuming their chain was a typical lube at that time. So you are probably right for the complete drive train at 3% but for the chain it seems likely it is about 2% and could go down to 1% or better. So a Monark might get you to the chain loss and changing chains should be observable?

edit: I found another indication of an estimate of drive train loss as 2.4% per Dr. Rodger Kram. But the more one looks the more difficult it is to get at the true effect that a chain has on the system much as Josh said above. Another interesting thing that seems to be hard for me to understand is the efficiency gets better the more watts input. That indicates that the chain power loss is not a simple % over the power range but varies and gets less with more input. So yeah it ain't simple it seems..... but drive train losses of 3% can be reduced with a number of things and obviously chain lubrication is a big one? and may be the best cost /performance option vs the cost of most of the other things like ceramic bears, over sized pulleys with low friction bearings etc.

No I think he's still talking about chain loss. With a derailleur present, the chain has to wrap around four time instead of two, even if the pulley friction is zero. Not only that, the derailleur pulleys require greater articulation since they're smaller. So while they aren't under tension, they may still contribute significantly.

ceramic speed say clearly that they have measured the relative loss at pulleys, part of the loss is articulation and part is bearings. Ceramic speed say even staying with 11 tooth pulleys better bearings can save 1.3watts. They also say 2.4 watts for over sized pulleys... So 3% loss due to drive train, but the hard part is how much of that 3% is due to the chain?

My response was about the comment regarding Monark, not Silca