Normally to reduce Rdc value in a transformer/inductor you put several windings with the same Number N of turns in parallel. But what would happen if these had different number of windings? Could I get winding numbers that are not round? where is my mistake?
If you connected windings in parallel, that had different numbers of turns, then that would create a 'shorted turns' situation, which is generally bad.
You can regard the parallel connection of an N turn and an M turn winding as being made of several things in series
(a) A winding of (N-M) turns, with a voltage to match
(b) A there and back connection of 2*min(N,M) turns, with no inductance
(c) A total resistance made of (N+M) turns of wire
Why no inductance in the there and back turns? They are coupled, and the inductance of one cancels out that of the other (almost completely, there's some leakage inductance, which is usually negligible).
How bad it gets depends on how many turns. If you had a 1 turn and a 2 turn winding connected, then very bad. A large current would be driven through the windings. The transformer would probably overheat and die. It's not much better to paralleling a 0 turn and 1 turn winding, and this is obviously a 'shorted turn'.
If you had a 1000 turn and a 1001 turn winding in parallel, then there might be enough resistance in the windings to limit the current that flowed to so little that you might not even notice, beyond maybe thinking that the transformer was 'a bit lossy'.
The voltage across the paralleled winding is a weighted average of the voltage of each individual winding (weighted by the residual resistance of the windings). But there are better ways to control the turns ratio of a transformer.
Interestingly, there is a transformer type in which we do this, the adjustable auto-transformer. The output brush that contacts the winding makes contact with several different adjacent wires, so the brush is shorting several turns together. This is one of the reasons an adjustable transformer has a carbon brush, so that its resistance helps to limit the current that flows. As a result, the adjustable transformer is not quite as efficient or ideal as a conventional transformer, but it's used for convenience, not efficiency.
But what would happen if these had different number of windings?
On a transformer, there is only one Volts per Turn factor: \$V_t/N = A_e \cdot dB/dt\$.
So, for example, paralleling a 20-turn winding and a 100-turn winding is nothing different than paralleling a 20V voltage source and a 100V voltage source.
If one winding was 10 turns and the other was (say) 11 turns you would get a circulating current that could cause a significant problem. Think about a perfect 1:1 transformer. Whatever voltage you put on the primary comes out of the secondary and, if you wanted, you could connect the primary and secondary together and there wouldn't be a problem (phasing matters of course).
This is exactly the same scenario as two parallel windings. You can regard one as a secondary and the other as a primary and there's never a problem when you short equal turn windings together but, if one has fewer turns then you might get a large internal circulating current and the transformer could burn.