Hi,
I'm totally new to this forum, having stumbled across it whilst looking for advanced driving courses. However, when I saw this thread I thought I might be able to contribute...
There are two sides to the OPs original question. Chris has pointed out that in an actual road test, the theory held true. However, with other posters questioning the blanket 90% rule, there seems to be a good discussion going on.
The OP showed a graph of in-gear torque vs. rpm.
Taking this a little further, with some more complete data, here are two graphs; The first is plainly torque and power vs engine speed, the second, is tractive effort vs. vehicle velocity.
The data is from a Honda 600cc motorcycle, hence the high max rpm. However, the following points still hold true..
The tractive effort is effectively a torque curve, adjust for gearing, drag, lift, wheel size and a few other parameters. Plus a rudimentary tyre model - nothing fancy like a pacejka model, just a simple curve of grip vs. normal load.
Essentially the maximum available forward thrust of the vehicle will be found by the curve which follows the tops of all the other curves, shown in red on the lower graph.
The optimal gear shift points (green dots) can be seen to get closer and closer together towards the higher velocities.
What you are seeing is Chris's experience, that shifting out at the maximum end of the range is costing you forward momentum.
However, looking back at the power and torque curves you can see that for the first (and almost first two gears in this case),it pays to stay in gear for as long as practically possible - due to the massive torque delivered by these first two ratios. But as the speed increases, the shift point moves further back in the range.
This is down to a couple of factors.
Firstly, at higher speeds (where the shift point moves back towards the peak power), "power", or the rate at which work can be done is the most important factor. However, in the early phases of acceleration the amount of work which can be done, or "torque", is the most important, as you have to get the large mass of the vehicle moving.
Secondly, modern vehicles tend to have split gear ratios with 1-3 being acceleration and 4-5, or 6, being town or fast road cruising gears for fuel economy. In this case, the early gears on the tractive effort graph would move left and up slightly, and the later gears would move right and down, exacerbating the effect Chris is talking about.
So...
Basically, I think what Chris was experiencing is what I've described here but without data for an individual vehicle it's impossible to say where the "optimal" shift point in the rev range would be, especially as it moves around with each new gear!
Cheers,
Ed