The Kozeka Principle Part II

Magnets with non-linear demagnetisation curves

So far I had only looked at neodymium-iron-boron magnets, which have linear demagnetisation curves. Would magnets with non-linear curves do better?

Repulsion force graphs for two 96 × 60 × 8 Feroba2 magnets (upper),
and for two 50 × 50 × 150 Feroba2 magnets (lower)

Here are two sets of results for repelling Feroba2 ferrite magnets. (Note that for all these repulsion graphs, repelling forces are now defined as positive. This was done for easier comparison with attraction graphs for similar pairs of magnets).

Once again there is negligible energy difference between horizontal and vertical repulsion.

A significant energy gain — but probably only a "one-off"

Repulsion force graph for two 50 × 50 × 150 Alnico5 magnets

Here are results for a large pair of Alnico5 magnets (150mm high), repelling. At last, here is another potentially interesting feature:— there is a significant energy difference. The energy gained by horizontal repulsion is over 5% higher than that expended against vertical repulsion. However, I'm fairly sure that what is really happening here is that the modelling program is showing only a "one-off" energy gain. In practice, repelling alnico magnets like these would be permanently demagnetised to a significant extent during the first time they were placed close together, but the program is not taking account of the already largely demagnetised state of the magnets, which is what would exist for all cycles after the first one.


Conclusions

1. It is possible that with further work, a very low-power permanent magnet motor could be made using magnets with very non-linear demagnetisation curves (such as Alnico), perhaps even with the magnets significantly demagnetised. The Bowman motor is a possible example, which I'll look at next time.

2. Since, unfortunately, I cannot see any way of avoiding the energy "well" previously discussed in Part I, it is not worthwhile for me to investigate the Kozeka idea any further.