Rohloff v. Derailleurs: Efficiency: The Fine Detail

[JohnR]

It shows why I'm so keen on Gear 11 that my whole transmission is planned around it and the overdrive gears come into use here in the Rome of West Cork (very hilly!) only on the downhills.

I had previously read (probably here) that 11th gear is the most efficient as it's direct drive so I targeted my gearing to be 11th when on the level as this should be the average terrain. That leaves 3 higher gears for downhill after which it's a matter of pedalling faster. However, depending on what the wind is doing I can be in any gear between 9th and 13th on the level. Is 4th the most efficient gear when in low range?

[martinf]

Is 4th the most efficient gear when in low range?

Yes.

Having looked at the gear efficiency charts I decided that I probably wouldn't be able to notice the efficiency difference between the top 7 ratios.

So I haven't bothered to try and get 11th gear as my most-used level-road gear. I just went for the lowest "permitted" combination that would also fit a Chainglider at the time I bought my Raven Tour in 2012, which was 38x16. I hardly use the top two gears and the lowest "permitted" combination is much lower now,  so I may go to 38x19 at some time in the future.

[Andre Jute]

A useful graphic comparison of the efficiency of the two gear trains inside the Rolloff box is Figure 6 on the original's p14 bottom left of
https://www.hupi.org/HParchive/PDF/hp55/hp55p11-15.pdf 
Figure 6: SPEEDHUB 500/14 efficiency comparison. Gears 8-14 shifted to the left to compare with gears 1-7.

It looks like the upper seven gears make pretty much a higher stack of the lower seven with 2, 4, 6, 8, 9, 11, 13 as the outstanding (sorry!) gears, if all you want to consider is efficiency. I also had mechanical silence in mind in choosing my transmission, and it is likely that direct drive, by simply not wearing the other gears, aids longevity (a bit irrelevant with a Rohloff).

Also notable: Herr Rohloff confirms, on top of the middle column on the first page of the link above, that dragging seals need running in, and that they will make a difference under 200W. I can't say I remember it ever bothering me, and not because I pull over 200W all the time.

[JohnR]

I've finally got round to reading the paper (I had previously incorrectly thought it might be in German) and all is revealed. Actually, almost all, as they didn't test the scenario of a well-used and filthy drivetrain.

In real life I don't think it's possible to tell the efficiency difference between 4th (one set of gears) and either 3rd or 5th (3 sets of gears). I just use whatever feels best for the conditions and that's one of the joys of the Rohloff hub - being able to easily shift a gear when going up a hill without lots of gnashing of teeth or loss of momentum.

[Andre Jute]

Yes. that's why I qualified the observation with "if all you're interested in is efficiency".

In practice I can tell you when I'm in gear 11 because I change up to 14 and then back three notches, but for the rest I have no idea.

That paper also contains another table, perhaps more useful than the efficiency table, which lists the distance travelled for one revolution of the pedals in each of the gears.
https://www.hupi.org/HParchive/PDF/hp55/hp55p11-15.pdf
Figure 4 on p14 of the original
If you're the kind of scientific cyclist who had a laminated table of half steps on your handlebars in derailleur days, I should think that table would be your bible, because on the Rohloff, with its even gearing steps, the fact that there are not even steps in forward motion per pedals-revolution for different gears subsumes both utility and efficiency into a single number, very handy once you put your head around it or make a laminated table to go on your handlebars.

Of course, a Rohloff makes it easy to change into only the gears that are the most efficient, 2, 4, 6, 8, 9, 11, 13. I tried it once for a 22km ride but it reminded me too much of a wedding march in 7/11 time I once heard an organist with a sense of humor play in concert (no bride will stand still for being made to look handicapped on her wedding day -- excepting reportedly the composer's bride, "a jolly gal game for anything"). By the time I returned home, it was a very old joke.

[John Saxby]

I just use whatever feels best for the conditions

Yep.  I bought the Rohloff eight years ago 'cos I'd had it with derailleurs & haven't regretted it for a moment.  I rarely bother looking at the gear-number indicator, either--half the time, shadows prevent me from seeing it anyway.

In the lower register, I find 4 is the quietest, so take that as a proxy for efficiency, with 3 being a close second (as it were).

in the high register, 8th always feels/sounds nice, after the hill relents. Then, I suppose 11th is quietest, but again, it doesn't really matter: I just choose whatever I need to maintain a decent cadence.

[JohnR]

In practice I can tell you when I'm in gear 11 because I change up to 14 and then back three notches, but for the rest I have no idea.

Is your hub so well run in that you can't hear the difference in sound when shifting between 8th and 7th (or vice versa)?

[Moronic]

A useful graphic comparison of the efficiency of the two gear trains inside the Rolloff box is Figure 6 on the original's p14 bottom left of
https://www.hupi.org/HParchive/PDF/hp55/hp55p11-15.pdf 
Figure 6: SPEEDHUB 500/14 efficiency comparison. Gears 8-14 shifted to the left to compare with gears 1-7.

I took the opportunity to have another look at that discussion, having read it quite a few years ago. I see more meaning in it now that I have a Rohloff to ride.

I think part of the burden of the Rohloff response was that efficiency is complex enough that simple tests are likely to mislead.

I'm reminded of the rolling resistance conversation promoted by Jan Heine at Rene Herse. He has argued that measuring the resistance of a tyre to rolling on a steel drum misses a big part of the resistance experienced in real riding, and also it ignores contributions a tyre's suspension performance might make to rider fatigue. His conclusion is that 50mm tyres with supple walls are best for most of us most of the time, and more and more people are finding that conclusion compelling.

When I've been out on my Rohloff Mercury, I've been observing to myself that it feels fast, which is what matters most to me, even though I might well be slower on it than, say, a similar Club Tour with derailleurs. The thought that I might be slower has been founded entirely on my prior reading of efficiency testing.

Rereading he Rohloff piece reminds me that I missed a lot last time. To put it another way, it helps me reconcile the narrow-dimension test results with my experience. I feel like I'm significantly faster point to point on the Rohloff Merc than on my prior bike, which has derailleurs. I don't time myself, so have  relied for that on my sense of ease, and have acknowledged that might mislead. I'm thinking now that it doesn't mislead. Of course efficiency isn't just about friction under load. Stress-generated fatigue is relevant, as is capacity to match load to cadence. And it will vary enormously with conditions. What is most efficient in a group on a smooth, open and mainly flat road might be least efficient solo on my cycle path network.

Rohloff is right to point broadly to such considerations in his article.

[Andre Jute]

In practice I can tell you when I'm in gear 11 because I change up to 14 and then back three notches, but for the rest I have no idea.

Is your hub so well run in that you can't hear the difference in sound when shifting between 8th and 7th (or vice versa)?

It's raining cats and dogs here, John, so I can't take my bike out to discover the answer to your question. Short answer: I don't know. I'll have to ride the bike on some quiet piece of level road and pay attention to know for sure.

When my Rohloff was new, I put Tippex (white typewriter corrector fluid) on gears 8 and 11, but it has long since worn away. In fact, in 10K kilometers and change, some part of the numbers have worn from the rotary control. I hardly ever look at it, and find gear 11 by going to 14 and stepping back three notches, burp.

You should also understand that when I say a Rohloff is silent, or mine is anyway, I'm referring to gear 11. Theoretically all other gears should be somewhat louder or in the lower gears at least more audible because you aren't riding as fast and generating as much wind. But I've long since ceased listening to the gearbox because all the gears require a) a special arrangement of riding on the most dangerous and fastest piece of road around here, because it is smooth and has well-defined white and yellow lines to kill the noise of my 60mm Big Apples, which are the loudest thing about my bike, including the Rohloff, but is so busy it can only be done in the middle of the night, and b) I have to listen carefully because there isn't much in it.

On the other hand, I've been a classical music critic all my life, so I'm very aware of noises off, and I design and build my own single-ended tube hi-fi to be played through electrostatic loudspeakers, so extraneous noise is the second-greatest obsession with me, next to micro-vibrations in my hands (a writer is a manual laborer: he operates a keyboard). So, if there was something to hear in the last several thousand kilometers, I would have heard it, and stopped to investigate.

Also, see above what John Saxon says about "feels/sounds" -- yeah, I know, Herr Rohloff, an upright German engineer, is cringing! -- but it happens to be an exact description: You just know when your Rohloff HGB is in a happy place and you stop turning the control.

By the way, I haven't adjusted the rotary control on my bike in the slightest in over 10K. People who get a ride on my bike are amazed at how loose I leave it. I don't see the point in adjusting it when the transmission appears happy. It's just a different paradigm to a derailleur whose efficiency and even operation depends on fine adjustments.

***
I don't think we know for sure when a Rohloff is run in. There'll be a natural variation because it is a piece of hand-assembled intricate engineering built light for what it offers in convenience and longevity, with the assembler having some leeway for judgement calls. So different Rohloff HGBs will run in at different rates, possibly significantly different rates because the service life is so long.

I felt that mine was a bit looser at 2K and there was another bigger step at about 5-6K, that's kilometers, where it felt looser again and some tiny noises that I knew to listen for in gears other than 11 had gone away. I'd originally thought that the Rohloff would never have the smooth gearchange of a Shimano Nexus, which is a thing of wonder if it is set up right; but about there I changed my mind. The next significant thing that happened was that at 8500km I replaced half-worn Big Apple Lites with new ones, and I noticed in the middle of listening to the rougher ride of the minimal tread on the Big Apples that of gears 7 and 8, eight appeared to have gone silent. As I say above, about 2K further on, I can't tell you anything about gear 7 without riding the bike under ideal testing conditions. Maybe I can tell a difference, maybe I can't.

Sorry to be so longwinded about telling you I can't tell you, but you'll arrive at the same place with your Rohloff eventually.

[Andre Jute]

Stress-generated fatigue is relevant, as is capacity to match load to cadence.

I'm big on the biggest balloons your bike can take, and with the softest sidewalls available, which is where the comfort comes from even with stiff anti-puncture bands. My favorite bike doesn't even have a suspended fork, nor a suspension seat post, just the Big Apples and the sprung Brooks saddle.

I also think that if I had a decent cadence (I don't, I'm a masher), and concentrated on being in the right gear in the Rohloff HGB, as my automatic Trek Smover is always in precisely the right gear, the Rohloff could have come out on top in my longterm tests described above in this thread.

The greatest advance on the present Rohloff box would not be a lighter version but an electronic stepper motor and control for it, plus a torque and speed sensor, to replace the Rohloff EXT box (or work with it) and handlebar rotary control, turning it into an automatic gear change; it's irrelevant whether the connection is by cable or radio or Bluetooth. With a gear range of over 500% and 14 effective gears (near enough the same number of effective gears on a derailleur bike), the Rohloff would own the hills.

[PH]

I think part of the burden of the Rohloff response was that efficiency is complex enough that simple tests are likely to mislead.

I think that's in in a nutshell, as the thorough Rohloff testing showed the conclusions are restricted to a narrow set of parameters.  I haven't re-read the results for at least ten years, they were not universally accepted, several well respected cycling engineers criticised the mythology and conclusions.
The idea that anyone can come up with something more conclusive is.. well laughable, if they could I'm sure Rohloff would be happy to review and publish, but I'm not holding my breath.
As I said earlier

Even if you could come up with a number to demonstrate that, it would only be your number and would likely vary day to day and ride to ride. it's then largely down to personal preference

martinf's testing is closer to reality, it provides everything he needs to know, it's interesting to hear about it, but they're right to acknowledge it isn't scientific, or at least what the parameters are. They can come up with the conclusions they're looking for, any claim that it applies to anything outside that testing is conjecture.
If someone's hub feels like riding through treacle then a zillion data points don't prove otherwise.

[martinf]

martinf's testing is closer to reality, it provides everything he needs to know, it's interesting to hear about it, but they're right to acknowledge it isn't scientific, or at least what the parameters are. They can come up with the conclusions they're looking for, any claim that it applies to anything outside that testing is conjecture.
If someone's hub feels like riding through treacle then a zillion data points don't prove otherwise.

In 2004 I had a commute of about 44 km for the round trip. To make things interesting, I used all the bikes I owned and compared the average times, eliminating days with strong winds that would introduce a lot of bias.

Conclusions over a total of 3391 km:

Position is very important, drop bar bikes performed consistently better than flat bar or roadster bar bikes (expected). 2.5 km/h difference for the same Moulton bike fitted with drops or roadster bars.

Tyres are important. Huge difference for (on-road) speed between knobbly mountain bike tyres and slicks. Significant differences for some brands/models of tyres (expected). Later on I was surprised by the small drop in efficiency between my Nokian studded tyres and my more usual Marathon Supreme slicks, I was expecting a much more important efficiency hit, especially as the studs make a lot of noise.

Gears seem not be particularly important, with one exception. Very little difference between two bikes with similar (flat) bar positions, similar tyres, one with 21 speed derailleur and the other with 5 speed hub gears (not expected). Later tests on my old mountain bike when I replaced the 21 speed derailleur with an 8-speed Nexus Premium hub confirmed this, again contrary to my expectations.
My Nexus 7 hub gear was, however, significantly less efficient than other gearing systems. (not expected, especially as the hub feels smooth). This Nexus 7-speed improved significantly after I cleaned out the original grease and replaced it with a less viscous lubricant (invalidating the guarantee in the process), but was still significantly slower than the S5/2 it replaced.

Fastest bike was my 700C derailleur geared lightweight (expected). Second fastest was a Moulton Stowaway with drop bars and 7-speed derailleur gears. IMO slower because of the tyres (not so good as the ones on the 700C bike) and the less efficient small sprockets needed with small wheels. Maybe also the small wheels, in real-life conditions small wheels deal less well with road imperfections (potholes, etc.).

Slowest bike was my old ATB with knobbly tyres. Same bike with flat bars and slicks was much faster, equal third with my old utility bike, also with flat bars and (reasonably efficient) Marathon Tour tyres but with an S5/2 hub gear.

But, if I add the time spent on maintenance, the hub gear systems come out as clear winners, as I spend far less time cleaning and lubricating the transmission. Note that this is only true if the hub gear is reliable - I had an unfortunate experience with unreliable Sturmey Archer Sprinter 5-speed hubs that were more hassle than they were worth. One of the selling points of the Rohloff is it's supposed reliability (I haven't yet done enough kilometers to validate or invalidate this claim). Where wide range isn't important, Nexus 8 Premium hubs seem a reasonable compromise between price and longevity, not yet had any problems with any of mine, despite abuse on the 2 visitor bikes.

At that time, I didn't have a Rohloff, so for touring in really hilly areas a derailleur system was mandatory. I have since got a couple of Rohloffs for my large wheel bikes. Even if they are a couple of percent les efficient than the derailleur systems they replace, when I factor in the maintanance time I will still be ahead. Especially after fitting Chaingliders. 

I was very sceptical about the Chainglider chaincase, reckoning that something that justs sits on the chain with no other support must generate a lot of friction. So I bashed my old utility bike 4 times round a 25 km circuit before fitting a Chainglider, then another four times after fitting the thing. I was surprised to find no significant speed difference, despite the non-optimal setup with a relatively thick 1/8" chainring. I am now a Chainglider convert, all but one of the large-wheel bikes I maintain have Chaingliders, the exceptions are a visitor bike with verical dropouts and the 3 family Bromptons.


 

[PH]

In 2004 I had a commute of about 44 km for the round trip. To make things interesting, I used all the bikes I owned and compared the average times, eliminating days with strong winds that would introduce a lot of bias.

Interesting reading.  Some universal truths like position and tyres and a demonstration of just how many variables there are.
We should also note how speed is only one indication of efficiency.  We could make it the primary one, but only with a measured input. I've tried a HR monitor as a poor man's alternative, but it's too influenced by other factors so not accurate enough.  We also have to consider that such data only applies to that testing, OK we can speculate how much wider we consider it relevant, and within reason we'd probably be right, but it would be far from conclusive.  Acceleration often doesn't get adequately taken into account, people who don't do a lot of urban riding don't always grasp how much energy is used pulling away from a stop.  It's the reason so many cyclists don't stop at red lights if they can avoid it, there's very little time advantage, it's a big saving in effort (No I'm not endorsing it!).  This consumption shouldn't surprise anyone, just compare car MPG  between urban and motorway driving.
Then, we can only use what we have.  We tend to have a preferred cadence, particularly on an uninterrupted segment, such as a long flat road or hill with steady gradient.  You could raise or lower the gearing by some amount and you'd still naturally fall into the same cadence at a slightly higher or lower speed.  I don't know the extent of this and it undoubtedly varies between riders, I know I can change tyres equivalent to a 1.5% gearing difference and the result on the flat is a corresponding change of speed.  There is a cost of course, though the energy difference is probably lost in the other factors. Any comparative efficiency testing would require identical ratios.  it's the reason I'm not a huge fan of the Shimano 8's, those big gaps in the middle often means maintaining my preferred cadence requires more effort or accepting a lower speed, this is likely to happen without realising it. Strava says my fastest times over a segment that includes a short steep bridge are on my 3spd Brompton!  It's nothing to do with the bikes efficiency, I don't have much choice - push hard or get of and walk... 
...variables, so many of them, which I think is where I came in, it isn't possible to isolate them outside of a laboratory and inside one doesn't reflect the riding experience... make your own mind up.   

[martinf]

it's the reason I'm not a huge fan of the Shimano 8's, those big gaps in the middle often means maintaining my preferred cadence requires more effort or accepting a lower speed, this is likely to happen without realising it.

For my own use, I don't mind gear systems with big gaps when doing short rides. I used Sturmey Archer and SRAM 3-speed hubs for urban trips for quite a long time, these have a gap of about 33% between gears. Pushing this to extremes, I used a wide range 2-speed derailleur system on my "light" Brompton for quite a while with 12/19 sprockets and 58% gap, used like a single-speed but with a bail-out gear for hills. Quite usable for short urban trips, and the minimal gearing made this Brompton significantly easier to carry.

It is different for me on longer rides. With a 3-speed hub I often wanted a gear in between the 2nd gear and 3rd gear. The gaps on an S5/2 five-speed are slightly closer at 18%, 27 %, 27 %, and 18 %, this made a significant difference for me and I have done a lot of distance rides with this system.

I experimented with close ratio derailleur cassettes, 5 sprockets 13 to 17 with one-tooth increments, then 2 bigger sprockets. On the close range part of the sprocket spread the gaps were between 6 and 8 %, but I generally jumped 2 or 3 sprockets, so concluded I didn't really need such fine tuning.

The Rohloff with it's approximately 14% gaps is OK for me, on the upper part of the range I never find myself looking for in-between gears and on the lower part of the range I could happily live with much bigger gaps.

Now that I have been spoilt by the Rohloff, I do notice the 27% gap between 3rd and 4th on my S5/2 equipped bikes.
And with the Nexus 8 premium hubs I notice the 22% gap between 4th and 5th, but not the identical gap between 1st and 2nd.

In both cases, it is only on longer rides, and is only a minor nuisance. For me the important advantage of the Rohloff is the very wide range, which has enabled me to get rid of derailleurs for rides in very hilly areas. 

[Andre Jute]

In practice I can tell you when I'm in gear 11 because I change up to 14 and then back three notches, but for the rest I have no idea.

Is your hub so well run in that you can't hear the difference in sound when shifting between 8th and 7th (or vice versa)?

Okay. The main road I billed as being smooth enough to conduct listening tests in the middle of the night, is no longer smooth, and a piece of smooth newly made road only a few hundred yards from my house carries heavy and loud traffic. So I improvised.

My house is old, with very thick walls, so not a lot of sound gets past multiple glazes and heavy drapes. I turned the bike upside down and changed the gears while turning the motor. The whir of the motor was the loudest thing, not very loud. I also tried turning the pedals by hand but the rustle of my clothes was drowning out the effects I was trying to listen for.

At something over 10,000 kilometers, my Rohloff is silent for all practical purposes in all gears, but it is less silent in gears 7 and 8, and between those two, I would say 7 is less silent than 8. There's not much in it, and certainly nothing you will hear over the sound of tires on the road and the rush of air through the spokes.

There's really not much to say about the sound of silence unless you're a neurotic; I've spent my life in silent rooms and I rather like the sound of silence; if I were ever unlucky enough to lose my bike with the Rohloff, I'd just order the same again, and break the Rohloff in, because it does run in -- it's just a more extended process than someone new to the Rohloff perfectly reasonably expects.

Sorry I can't give you a precise number for your gearbox running in, John. But, if you read carefully, you'll discover that people with new Rohloffs are a lot more confident in their claims of their Rohloffs running in than the older hands, whose predictions one after the other required to be extended. The one verity is that your Rohloff will become smoother with time and miles. Ride more often, ride further.

***
It would be interesting to learn whether a new NuVinci, which is a CVT, a continuously (stepless) variable transmission, is more silent than the Rohloff, or whether the NuVinci too breaks in and thereby becomes more silent.