Or "why acceleration occurs when a passive generating output coil in an open magnetic system is short circuited or placed under a higher than nominal load".

This explanation is drawn from conventional electronics and focuses on the impedance characteristics of a coil wound on a ferromagnetic core. The driving motor in this explanation can be any sort of electrical motor. Passive generator output coils in an open magnetic system is the main topic of this explanatory review.

First, here's a list of simplified descriptions of events as they normally occur within a permanent magnet electrical alternator/generator system, where the magnets are moving with the rotor and the generator output coils are stationery.

1. A motor (of any kind) turns the generator rotor. The rotor has permanent magnets embedded within it. 
2. The magnets of the generator rotor move past the passive generator pick-up coils, and as they do, they create a varying magnetic pressure upon the cores/coils.
3. This varying magnetic pressure results in a varying voltage in the coils. (It also results in eddy currents in the core – but I'm going to ignore them for this explanation) 
4. If there is a load on the generating coil, current will begin to flow.
5. This coil current will, in turn, create its own magnetic field in the core.
6. The magnetic polarity of the core field will now act in opposition to the varying magnetic field of the passing magnets.
7. This opposition will cause a breaking effect, the magnitude of which is related to the amount of coil current and hence the induced magnetic field strength of the core.

It seems pretty straightforward, but the current in the loaded coil which produces the magnetic field of the core in opposition to the motion of the magnets, does not arise instantaneously. There will be a time lag which is dependant on the impedance and reluctance of the coil/core. To understand the following explanation, it is necessary to view the moving magnets as the actual AC power supply, and to treat the generating coil as an inductive load which, when coupled to an external resistive load, is in series with that external resistive load. It is also necessary to understand how the current in the inductor reacts to changes in the resistive load, because the coil current is responsible for the counter mmf (oppositional magnetic field), which is normally associated with the breaking effect of the generator load, in accordance with Lenz's law as it applies to generating systems.

An inductive coil has a constant resistance to a given DC voltage (provided excessive current is not creating high temperatures).
It also exhibits an inductive reactance (impedance), which arises in response to connection to a Pulsed DC or AC power signal.
Furthermore, it also exhibits some capacitance due to the accumulated inter-winding capacity between each full turn of a coil winding, and therefore, it also posseses a small amount of capacitive reactance.

Capacitive reactance in a coil is opposite in vector and response to inductive reactance. As frequencies rise (rotor speed increase), capacitive reactance decreases, while inductive reactance increases. Capacitive reactance and inductive reactance cancel each other due to vector opposition. But the value of capacitance and therefore capacitive reactance (Xc) is usually negligable compared to the value of inductance and it's XL in a coil with a ferromagnetic core. Consequently, the coil will still exhibit an overall inductance and inductive reactance (XL), after the vector sum of XL and Xc have cancelled out.

The amount of inductive and capacitive reactance (impedance) is determined by the frequency of the AC or pulsed DC and is not constant.
E.G., the impedance Z of a coil at 500 hz will be roughly twice the impedance Z than at 250 hz, yet the dc resistance will be the same.

The actual total impedance of a coil (Z) is the square root of the sum of the resistance (R) squared plus the (inductive reactance (XL) minus the capacitive reactance) squared. The actual total impedance of a circuit that comprises a coil in series with an external resistive load and or external inductors and capacitors in series, is determined by the same formula. The total resistance of a series Coil, resistor and capacitor circuit comprises the internal resistance of the coil plus the external resistance/s of the load.

Here are some links to a good site, explaining impedance Z in an AC or pulsed DC circuit.

http://hyperphysics.phy-astr.gsu.edu/Hbase/electric/imped.html#c1
http://hyperphysics.phy-astr.gsu.edu/Hbase/electric/impcom.html#c1

1*. Regarding the reluctance characteristics of a ferromagnetic core:

When a generating coil is connected to a load, there is a magnetic field produced around the wire that forms the coil (which is wound on a ferromagnetic core), and changes to this magnetic field are in phase with the current in the wire. But the reluctance of the ferromagnetic core causes changes to it's own magnetic flux to lag the changes in current produced by the coil, and also to lag the changing magnetic field of any passing moving rotor magnet. If flux induced into an inductive ferromagnetic core instantly reached the levels dictated by the current and/or passing magnet, and then instantly demagnitised when current ceases, and then instantly reached opposite polarity flux levels dictated by current in the opposite direction, then there would be no such thing as a hysterisis loop or BH curve/s. We would have perfect inductors. But ferromagnetic materials do not instantly change their magnetic flux in direct relationship with either current changes or inducing magnetic field changes, instead, they always lag behind to a degree which is determined by the reluctance / permeability characteristics of the core.

Now combine the natural ferromagnetic core lag characteristics with the total Z impedance characteristics of the coil/core combination, and together, there is an ample amount of already accepted electrical theory to account for the acceleration as being a result of negating both the core drag and the coil induced counter mmf.

2*. Regarding the coil's Z impedance and it's relationship with induced current phase and counter mmf, see the phasor diagrams at the site linked below:

http://hyperphysics.phy-astr.gsu.edu/Hbase/electric/phase.html

The voltage (electromotive potential- emp) in the coils is produced by the changing magnetic field strength from the passing magnets (magneto motive force – mmf) of the rotor, and will manifest as current (electromotive force – emf) in the coils, when there is a load connected. This load forms part of the impedance (Z) triangle of RLZ in the phasor diagrams shown at the site link above. When you increase the load (lower the resistance) the phase angle increases.

Counter mmf (counter magneto motive force) is produced by the current (emf) in the coil, and it also arises out of phase with the coil current that is producing it. Conventionally speaking, counter mmf is said to be 180 degrees out of phase with the inducing mmf and is therefore oppositional to the inducing mmf. In reality no inductor is perfect, and the phase is more likely to be between 170 – 179 degrees out of phase depending on the inductors characteristics. Bear in mind, that in a perfect inductor this phase opposition is theoretically a 180 degrees polarity vector difference, but a zero degree difference with respect to time. The time phase is where all the changes occur in this acceleration anomoly!

When there is an incrementally increasing load placed on the coil, the phase angle of the coil current (time lag) increases as the resistance of the load decreases towards S/C. The resulting counter mmf phase angle of the core (time lag) also then changes with respect to the original inducing mmf of the passing magnets. As the counter mmf approaches 90 degrees out of phase (time lag) with the inducing MMF it also approaches physical vector neutrality and thus zero opposition. Since The counter mmf is a product of both the current phase in the coil and the degree of magnetic phase lag due to reluctance of the core material, then the opposition normally produced by the coil current and core drag are nullified together. At short circuit, with maximum current and counter mmf phase angle change, the coil/core appear to magnetically "disappear" with respect to the rotor, and so the rotor accelerates.

Because the motor / drive coils experience less opposition, due to less magnetic drag being placed upon the rotor by the generator coils, it accelerates, resulting in an increased motor back emf and motor coil impedance.

The increase in motor back emf and motor drive coil impedance results in a decreased current input from the supply, and a lower motor power consumption.

The (motor) rotor speed, combined with the number of magnets on the rotor, determines coil current frequency, and plays a very important initial role, because the inductive reactance (XL) of the generating coil increases with frequency. As a consequence, the current phase angle will be greater for a higher frequency coil output than a lower frequency coil output, into the same given load. The acceleration effect will occur at a lower rotor rpm when using high impedance pick up coils than the rpm required when using low impedance pick up coils.

Because the coils I showed on page 10 of my article are very low impedance coils, then a high rotor speed is required for the acceleration effect to occur. I indicated that high speed was desirable for that particular set of coils, but didn't explain why. I chose low impedance coils for the experiment shown because the majority of working generators that are in use, such as car alternators, are low impedance generators, with low voltage but high current output capability. (I was trying as near as possible to compare apples with apples.)

Below is an animation of inductor current phase changes occurring due to 1. Varying load. 2. Varying increased frequency and 3. Varying both load and frequency.
Note that because capacitive reactance is usually negligible, for the sake of ease, (mine) it is not included in the simplified visual representation below.

In my experiments with incremental loads placed on the output of the coils, the amount of acceleration increased non linearly with the load change. This actually makes sense to me, (within the context of this whole explanation) because as the load increases, it forces the current toward a critical phase angle, (where the drag and counter mmf are significantly diminished with respect to the rotor), the rotor (motor) begins to spin faster with less drag to oppose it, which in turn increases the power output frequency, which in turn increases the current phase angle. This cascading effect contributes to the rotor acceleration at each interval of increased load beyond the critical loading/frequency point, until maximum phase change occurs. (Theoretically, a 90 degree max phase angle at short circuit). Fig 3 above shows the cascading effect when both load and frequency are increased.

Put simply, IMHO, the acceleration effect is the result of negating oppositional forces associated with the generator core/current, and not the addition of extra energy into the system. This negation occurs due to a phase shift in the coil current and core counter-mmf, as a result of increasing frequency and /or higher than nominal output loads up to and including a short circuit.

Now I ask myself – why don't conventional closed system generators act like this, and accelerate under higher than nominal loading or short circuit? Can they be made to act in the same manner.? After all, even with Lenz's law applying, the coils of a conventional generator are bound by the same set of other accepted electrical rules which govern the Z impedance characteristics of an inductive power supply connected to a load.! And if they can't be made to have the same characteristics- then why not ?

Is it really a useful anomoly anyway, or just a curious thorn in the side of conventional wisdom.??

Hmmmmm! Another can of worms maybe….. I just love worms!……. Hoptoad