I've seen it exactly once (in a numerical analysis journal). Authors were blacked out. This was undermined almost immediately when the authors wrote in the introduction "... following previous work of the present authors [1]...".

Even if all these clues could be effectively removed, the nature of the research work, tools used, writing style, all unambiguously indicate authorship to someone in the field. I think for this reason double-blind review is not popular.

Which is not to say that it would be great if it were possible...