On Friday, April 21, 2017 at 8:31:27 PM UTC-7, Jeff Liebermann wrote:
On Fri, 21 Apr 2017 10:40:30 -0700 (PDT), Doug Landau
wrote:

So you are saying that losses from routing the lower run of
the chain through tensioners and idlers are negligible?

Worse. I was ignoring the losses from tensioners, idlers, grease, pin
rotation caused by a slight chain droop, inertial loading from a
longer and thus heavier chain, chain acceleration lag, etc. I assumed
that the original question was about a simple power transmission
system, not the complex mess that such systems inevitably evolve into.
Something more like this test fixture and a practical bicycle:
http://cdn.mos.bikeradar.imdserve.com/images/news/2012/11/06/1352163122826-1476emv18vmdi-630-80.jpg
The additional losses can be tested separately and included later.

You're correct that tensioners, idlers, etc are important.

I'm not saying they are I'm just wondering. It doesn't look like the losses from the derailer on an upright is going to be significant. What I noticed with my rickshaw was that the chain was so long it was hard to keep off the ground due to it's weight. I tried using a derailer as a tensioner, and mounted it upside-down underneath the cab,so the spring would lift it up, near the middle of the run. To my surprise the spring wasn't anywhere near stiff enough. Not even enough to keep the pulleys engaged, let alone the chain held up above level, let alone taught and not flapping. It's what it was going to take to take up the slack that I thought might be non-negligible.