If $g(fg)^x(B-D)∩g(fg)^y(B-D)≠∅$ let a belong to the intersection of the two sets. Then there are u, v in B-D such that $g(fg)^x(u)=g(fg)^y(v)=a$. Then $(fg)^x(u)=(fg)^y(v)$ because we can multiply each side of the first equality by $g^{-1}$, which is 1-1, and get the second equality. But then $(fg)^{-y}(fg)^x(u)=(fg)^{-y}(fg)^y(v)$ and thus
$(fg)^{x-y}(u)=v$. But fg(u) is not in B-D and so also every $(fg)^n(u)$ is not in D-B for every natural number n (because fg is 1-1 and thus it preserves the disjointedness of sets), but v is in B-D so we get a contradiction.
Incidentally, (fg)^x is a (commonly used) misnomer because fg need not be applied again to get to the xth image of a member form B-D.