"Repmag" Part I — a Net Energy Output is Modelled

"Switching" permanent magnets

Although a permanent magnet cannot be switched on or off, its flux can be temporarily diverted as required, to the extent that it will no longer attract or repel another adjacent magnet.

In the next few posts I'll examine a particular case utilizing this principle.

"Repmag" permanent magnet motor

Here is an image of one of my own ideas for a permanent magnet motor, which does give some net energy output, according to the modelling I've done on it so far.

There are twelve identical 25mm IR × 62.5mm OR × 6mm × 20º arc shaped NdFeB35 permanent magnets, all magnetised in the same direction, through the z-axis (red). That means they will repel each other when placed close together, in the absence of any magnetic shielding. However, a steel magnetic shield is placed over half of the array as shown, with a small (0.25mm) airgap above and below the magnets.

The magnets rotate clockwise, when seen from above, about the z-axis. They close up against each other within the shield, and spread apart outside the shield. Although, as can be seen, the magnets' flux density is greatly increased when they have entered the shield, the flux lines will then be perpendicular to the magnet and shield surfaces. Very little flux should return around the straight sides of the magnets. (Recall that it is this flux that causes the mutual repulsion outside of the shield).

So, it would be reasonable to expect the magnets to separate outside the shield with more mutual repulsive force, and to close up within the shield with less mutual repulsive force. Also, because it is more separated from its neighbours, a magnet about to enter the shield should have a slightly higher overall flux density than one that has just left it, which would also help the desired movement.

Analysis in a silux model

I decided to analyse this idea by firstly finding all forces and torques on all the magnets at 2.5º intervals (of "spider" rotation — see below), and then assigning these quantities into the silux model shown below.

Repmag — silux model

This silux model has the magnets (black) and the steel shield (blue — only for illustration) as before. The magnets pivot around the left fixpoint. Each magnet has a roller attached to it, which slides in a spider (red), rotating around the right fixpoint, which is offset by 15mm. This ensures that the magnets remain correctly located at all times.

The model was started with a spider rotational speed of exactly 10 radians/second, with magnet forces and torques newly assigned from the magnetostatic modelling every 2.5º as discussed above.

Results from silux model for spider rotational speed and energy,
derived from the force and torque data from magnetostatic analysis

Results

The spider does at first accelerate, but then it decelerates, and accelerates again, finishing at 10.0705 radians/second after 30º, which is the end of one cycle of operation.

(Some further details: In the model the combined magnet and roller mass was an almost negligible 0.0482kg each, but the spider was made deliberately heavy, at m = 10kg and I = 0.0375kg-m², so that it would act as a flywheel. Therefore its energy (½Iω²) increased from 1.875 to 1.90153J in 0.05236 seconds, i.e. a power output of 0.507 watts.)

Conclusion

The analysis shows an increase in energy and power, which is more than the nominal 1% solution error used in the magnetostatic modelling.

If the magnets themselves, at 0.0258kg each, are considered to be the only "active" mass in the model, with everything else including the arbitrarily heavy spider excluded, this particular model indicates that 0.507/(12 × 0.0258) = 1.638 watts per kilogram of active mass could be achieved. This is better than the minimum figure of one watt per kilogram at which I'd consider building a physical prototype.

So, while no single model like this could be considered decisive, it does seem that further work on this idea would be worthwhile. 

Obviously further modelling is required. I have done some initial work on the mechanical design for a physical prototype for this idea, which I'll discuss next.

A Magnetic-Gravitational Motor

Introduction

This is another quite old idea, which could have been designed as a purely mechanical device. But it is probably easier to understand initially as a combined magnetic-gravitational device.

Fig 1. Weights suspended by springs in a wheel

Introductory design

Suppose a wheel is made as shown in Figure 1, where a number of weights (red) are suspended by tension springs (blue) between the rim and the center of a wheel centered at point P. Then the locus of the weights will be very approximately a circle (purple) centered at a lower point, O.

Obviously the weights in this wheel will be balanced overall, and so the wheel will have no tendency to turn.

Fig 2. A fixed magnet repels magnetic, spring-suspended weights,
on one side of the wheel only

A magnetic modification

Now suppose the weights have been made as magnets, and another fixed, radially magnetised repelling, semicircular magnet (orange) is added on one side of the wheel, as shown in Figure 2. Then the locus of the magnet-weights will be as shown by the purple curve in Figure 2, i.e. they will be raised just before and after they have crossed the vertical centerline.

In this design, with the magnet-weights suspended by springs as they are, it would be reasonable to expect that the repulsive force on each moving magnet tending to turn the wheel back as it rises just before top center would not be very much greater than the similar repulsive force on each magnet tending to turn the wheel forwards as it rises just after bottom center.

So, on the one hand, the magnet-weights fall through a greater height on the descending side, and this should cause the wheel to turn. On the other hand, it could be argued that as the descending magnet-weights complete their fall, there will only be a net exchange of gravitational potential energy to spring stored energy (where there was no such net exchange before), which will be enough to ensure that the wheel won't turn.

Resonance

However, something else can also be done:— wherever weights are combined with springs, mass-spring resonance can be made to occur.

Modelling — summary

In Figure 1, assume the weights are 0.5 kg each, at approximately 0.125m radius from O. The springs between the weights and the rim, and the center P of the wheel are:—

Unstretched length Lo = 0.05m
Max. stretched length Ln = 0.4m
Max. force Fn = 11.445N
Spring constant k = 32.7N/m

(In Figure 1, the weights are shown in their correct positions for the above data, for a stationary or very slowly turning wheel of 0.25m radius and distance OP of 0.05m).

With the wheel stationary, we then add a force radially outwards from O of 8.177N on the top weight, and allow it to rise to its natural position for this, at very nearly 0.25m above O. (This is similar to Figure 2, but with greater displacement). We then remove this artificial force, and rotate the wheel clockwise at a steady 8.0888 rad/s. Then (radial) centrifugal force of 8.177N will replace the artificial force. Then, as the wheel rotates, approximately one half-cycle of mass-spring resonance per half-revolution of the wheel will occur. At the bottom of its fall each weight has dropped to slightly over 0.323m below O, i.e. well below its natural position for a 8.177N radial force. (And so its spring to P could then raise it to well above that natural radial position on the ascending side).

No energy gain

However, I have not found any overall energy gain from this approach. The old question "Where does the (extra) energy come from?" applies here. Even with mass-spring resonance occurring, there is just an exchange of energy between different forms, i.e. kinetic, potential, spring stored energy, and rotational energy of the wheel. As for forces of magnetic repulsion, in this device the magnets could just as well have been replaced by mechanical components, e.g. compression springs.

Can we be more creative?

So, we'll have to be more creative about how we use permanent magnets, if we hope to gain any net energy from them. More on this next.

Modified Bowman Motor

Primary reference: 

http://www.freeenergynews.com/Directory/MagneticMotors/Bowman/index.html

In 1954 Lee Bowman invented a three-rotor permanent magnet motor using Alnico magnets, which was claimed to have worked as a perpetual motion, as shown below:—

The original Bowman motor

A modified version

After a quick look, I had not thought this motor interesting enough to investigate further, but a friend who was also interested in magnet motors eventually persuaded me to build a physical prototype of a modified version of it — see below. (This was a long time before I was able to do any computer modelling of devices like this).

(Most of) my modified Bowman motor

Unlike the original, this prototype used NdFeB35 magnets, and had only two rotors geared together, of equal size. The magnets on each rotor were arranged to attract those on the other rotor. The axle with the pinion on it originally passed through another piece of framework and carried a flywheel (these two items not shown).

As with the original Bowman motor, a separate fixed "actuator" magnet was arranged to repel the magnets on one rotor. The intention was that when a pair of rotor magnets crossed each other, the repelling actuator magnet would reduce the attractive forces between them at the end of the crossing, as they started to move apart again. (It would do that by reducing the flux density of the rotor magnet that it was repelling).

It is almost needless to say that this prototype did not work.

I had originally intended to add at least a second actuator magnet below the one shown, which would have doubled the energy output — had there been any — because each rotor magnet crosses with its partner on the other rotor twice per revolution. I never did that, since there was no detectable energy imbalance at all with the single actuator magnet.

Conclusion

Even though a version of the Bowman motor might possibly be made to work, at very low power output, with highly non-linear magnets (like Alnico, as used in the original version), I don't think it worthwhile to investigate this concept any further.