A frame challenge.
From the cited article
...or else they end up being a kind of perpetual motion machine. This
is due to the fact that applying constant force results in constant
acceleration, which means that the object’s kinetic energy increases
quadratically over time, whereas the energy input increases linearly.
As a result, the object’s kinetic energy will exceed the energy input,
and (if this energy is collected via deceleration) there would be a
net gain in energy.
If the kinetic energy of the system (what you are using to rotate the generator shaft) exceeds the energy that you are putting into it (the output from the generator) it is readily apparent that what you have is a perpetual motion machine on steroids. You are getting work out of an apparatus without either putting external energy into it or without using up the internal energy of the system. It seems to me that this device is intended to use externally supplied energy to produce thrust, not to produce more energy than was put into it. An interesting concept that came out in following links suggested by the cited article is that additional energy MAY be somehow coming from harnessing zero-point energy.
TL:DR Your requested system is akin to trying to hook up a generator to a motor, and keep the motor spinning in perpetuity without ever having to supply more energy. I posit that there has to be some as yet unknown limitation on the kinetic energy of the system.
What is of interest to science fiction, however, is the following passage from a link suggested by the cited article, Modelling the Pioneer anomaly as modified inertia
One possibility for a model of inertia is that of Haisch et al.
(1994) who proposed that an accelerated object feels a magnetic
Lorentz force through its interaction with a zero point field (ZPF)
similar to the Unruh field (Unruh, 1976). This force is given by F
= −Γω ^2 cha/> 2πc^2 where Γ is the AbrahamLorentz damping constant of the parton being oscillated, ¯h is the reduced Planck constant, ωc
is the Compton scale of the parton below which the oscillations of
the ZPF have no effect on it, c is the speed of light, and a is
acceleration. Haisch et al. (1994) showed that this force behaves
like inertia. One objection to a modification of inertia is that it
violates the equivalence principle, which has recently been tested to
an accuracy of 10−13 kg by Baessler et al (1999). However, this
principle has not been tested at the low accelerations seen by the
Pioneer craft or by stars at the edges of galaxies.
(See original for an accurate rendition of the equations)
Note of particular interest is the qualification '...has not been tested at the low accelerations...'. Perhaps we are entering an era where, in order to continue expanding the thickness of the physics textbook, we will have to look at modifying many, if not all, of our physics equations with denominator terms that in effect amount to enforcing the 'Law of Diminishing Returns', just like the relativistic equation for adding velocities has modified the Galilean equation by adding a 'diminishing denominator' term. For instance, the equation F=ma has been empirically supported at accelerations normally experienced in our relativistic reference frame, but is there a need to add a qualifying denominator for the leading and trailing edge of the curve? Is it possible that we will have to recognize that 'nature abhors a straight-line curve'?
EDIT To clarify, my frame challenge is the assumption that this engine will, indeed, provide more kinetic energy velocity than the energy put into it, and thus qualify as a 'perpetual motion machine'. As long as it does not cross the threshold of 'the thrust leads to a velocity that produces kinetic energy greater than the total input thrust energy' it is not in violation of any laws. There COULD be factors at play in the engine that follow all of the rules, with some modifications to the science that we don't yet know are required, just like Newtons Laws required modifications due to relativistic factors. Relativity did not throw out the physics textbook, it just made it thicker. For instance, the Alcubierre Drive is, in essence, a drive that only requires energy, not reaction mass, for propulsion, and I know of no reference that even remotely suggests the Alcubierre Drive is a perpetual motion machine. It breaks no rules, it just requires a thicker physics textbook. If this drive does, indeed, work, it is not proof that it is a result of it being a perpetual motion machine.
ALL drives, no matter how they work, are bound by the limitation of kinetic energy to thrust input, so no drive can supply a constant acceleration beyond this limit. Yet thrust drives do work, despite this limitation, and all of the techniques that purport to turn them into perpetual motion machines (and there are a lot of them in science fiction) are based on either false assumptions or misunderstood physics. These methods mostly have in common some method that allows this thrust-energy-to-kinetic-energy barrier to be crossed.
TL:DR
This question is based on the assumption that this drive is indeed a perpetual motion machine that does cross that barrier, and not due to some as-yet-not-understood factor that requires a thicker physics textbook, not a completely re-written textbook.
