If you can replicate what you report above under controlled conditions then you can retire for life next year. Really.
The overwhelming opinion by all reputable test houses or suitably competent careful and honest individuals is that for combustion improvement by treating the **fuel* with magnets you'd be just as well off rubbing on snake oil - or drinking it. Really.
Magnets are the universal cure / improver / energy source / you name it.
I am not saying that you are dishonest OR that you did not think you saw the result that you report.
But, it is overwhelmingly likely that you have missed something crucial that affects the result.
Having this happen is a part of everyday engineering testing. Spotting it is the trick.
For many months CERN (who should be as real as they come) were reporting faster than light transit speeds for some particles. They checked everything they could think of scrupulously before they reported their results. In due course a non FTL explanation was found. If CERN can manage this, so can the rest of us.
I have had factory engineers give me written reports containing data that showed processes were running at > 100% efficient when there was no suggestion or expectation that this was the case. I knew with certainty that something was wrong with their data. They didn't know.
I know this is not the answer you wanted and I also know that some people will assure you that you can treat fuel "magnetically" and improve combustion efficiently. And, it MAY be true. But "everyday Science" says that it's "extremely unlikely" and the necessary path for you is to
Replicate your results, then
Have somebody else replicate your results and then, if it still seems to be working,
Arrange "double blind" testing so that there is no way to fool yourself.
Note that use of magnets to affect a combustion process in some way (rather than pretreating the fuel) is potentially more likely to be valid - still unlikely enough that very great care is needed, but ionised gases can be affected by magnetic fields so there is some basis for possible changes.