Relative efficiency of different commute bike setups

Question

My interest is in long distance commuting (e.g. more than 50 km a day) and the relative efficiencies of different commute bike setups (e.g. drop bars vs flat bars, fenders vs no fenders, panniers vs backpack, etc).

As such, does anyone know if this type of information has been pulled together (e.g. magazine article, web page)? I am looking for hard hard numbers (e.g. wattage or relative difference in wattage) that document real world differences associated with how one sets up their commute bike.

Why I even care

Background

Last year I had my commute bike stole. When I replaced it, I kept most of the build identical (e.g. drop bars, 2x racks, fenders, dynamo hub, disc brakes), except I also moved to an internal hub. When commuting distances, I notice this bike is slower. On the top end I think I lose anywhere from 5 - 10 km/hr. This bothered me so much I ended up building another commute bike, but went in the complete opposite direction (classic steel road, with low profile fenders and one low profile rack). This second bike is so much faster I typically use it on club rides and put the carbon crew to shame.

While there is no question the second bike is more efficient (32 vs 42 km/hr top flatland cruising speed), the builds are also night and day apart so I am racking my brain to point the finger at the main culprits behind this efficiency difference.

Crazy Plan

If this type of info isn't out there, I would like to run a series of experiments to understand the differences. I would have a designated time trial, record the power needed to cruise at a given set of speeds. I would then slowly strip the bikes down and repeat. The hope would be to determine the relative differences in efficiency different build decisions produce and by extension how much time difference there would be in a 50 km ride.

However before I start this, I would want to make sure this type of experiment hasn't been done before.

---

NOTE - There may be scepticism over how 'doable' this type of experiment is. If the type of information I am looking for doesn't already exist and there is sufficient interest, I am willing to draft up my planned methodology and post online for review (perhaps at a different and more appropriate venue). I do have expertise in both statistics and experimental design so I believe I would be up to the task.

Answer 1

Jan Heine performed some wind tunnel tests of "Real World Aerodynamics" a few years ago. A link to a blog post (and the results published in Bicycle Quarterly) can be found here. Those tests cover only one component (the aero drag component) of commuter-type bicycles vs. "racing" bikes.

If you want to make your own apples-to-apples comparisons of performance potential between any number of bikes of any type there are five pieces of information you need to know:

1. how much power you produce on that particular bike;
2. how much (you and) the bike weighs;
3. the Crr for that bike;
4. your CdA on that bike; and
5. how much of your power is lost in the drivetrain.

With those five bits of information you can compare the performance of a road racer to a TT bike to a penny farthing to a velomobile to a MTB; on a hill, on the flat, on a descent, on a velodrome. Of course, those five variables won't tell you if the bike fits, or if it handles well, or if it has a way to carry your groceries and laptop and change of clothing, but as far as performance goes, that's all you need to know.

A good bathroom scale will help you measure 2). Since you have a power meter, there are field test methods and protocols that will help you determine 3) and 4).

1) explores whether you can produce more power on one bike than another, because of positioning (for example, many riders find they produce lower power in a TT position than in a traditional road racing position) or crank length or gearing (for example, some riders find they produce less power on a single-speed than on a multi-geared bike).

Finally, 5) measures losses in the drive train, such as in a fixed-gear or with an internally-geared hub like the Alfine. In general, measuring losses in the drive train can be expensive because you need a way to measure power at each end of the drive train. See Kyle and Berto here, for an example of a rig used to measure losses. A similar rig was used by Spicer here. However, if you have a crank-based power meter (and, if you're using an Alfine then I suspect you must), it is possible to use the crank-based power meter to measure power at one end of the drive train and then use "virtual elevation" methods to measure the power at the driving wheel. A detailed description of the protocol is long but a quick description is to climb the same hill on a calm day with different configurations. Since the hill is the same and the conditions are calm you know that the "virtual elevation" should match the true elevation profile of the hill; so if you can measure power at the crank then differences in observed VE must be the result of differences in losses in the drive train.