Another idea here involved a pedal-powered flywheel,
and
an annotation suggested that gearing would be needed
to
help get the flywheel up to speed.

Gearing, however, is a clunky solution. There is an
eleganter way.

The MaxCo. Self-gearer Peddlable Flywheel has a
relatively
lightweight
carbon-fibre rim. Five heavy lead weights
are
mounted on sprung arms, so that they sit inside the ring.

At low speeds, the spring arms keep the weights close to
the hub of the flywheel, so that they have little
rotational
inertia. Starting the flywheel is therefore relatively
easy.

As the speed builds, so the springs of the arms are
compressed, and the weights swing out closer to the rim.
At full speed, the weights are hard up against the rim,
contributing the greatest possible moment of inertia and
allowing the flywheel to store the greatest possible
amount of energy.

Of course, some energy is wasted in compressing the
springs but, as you ease off the pedals and the flywheel
slows down, this energy is recovered and contributes to
the residual spin of the flywheel.

Flywheel physicshttps://en.wikipedi...ki/Flywheel#Physics It appears that the variable-inertia flywheel was considered by one Sr. L. da Vinci... [8th of 7, May 11 2016]

I think this might be brilliant, but sort of the opposite of
useful for a normal flywheel. In an engine, you need the
mass at low RPM, if you could find a way to reverse the
effect... Anyhow, it lead me to staring at diagrams of dual
mass flywheels and pendulum style centrifugal absorbers.
I'm now confused and want a BMW diesel all at the same
time.

//In an engine, you need the mass at low RPM, if you
could find a way to reverse the effect... //

I haven't looked at the mechanics of it, but a gut
feeling for the physics tells me that you can't do it
automatically using simple levers and whatnot.
However, in a car you can afford to have a lot of
mechanicky stuff to dynamically adjust the flywheel's
moment of inertia.

//With the right spring tension, once spun up, the flywheel
will continue to spin at a constant rate, despite losing
energy, which sounds useful//

V.V.V useful for, say, maintaining exactly 50 Hz* during
power generation. In fact, put a viscous coupling on the
front end and within a given range the speed OR torque of
your mechanical power source aren't important.