There is no good reason to simply copy large portions of an existing publication to make a new. By rewriting everything, you have the opportunity to rephrase your thoughts and most likely produce a better paper, at least from a writing point of view. Some journals definietly look at plagiarism and in this case self-plagiarism and the only way to avoid it is to rewrite. I think the benefits of rewriting clearly outweighs the possible time-saving by just copy-pasting.
Just to provide an example. I have written tens of papers about the same natural physical object that I have studied. In each paper, I have to provide a descritpion of the locality and characteristics and not once have this been copied. It is in fact interesting to see how many dofferent ways the same (dull) information can be conveyed.
As for a percentage of new material, I really would advice against relying on a numbr. It is possible this is a silent understanding in some fields but in general, only new insights and conclusion should warrant a new publication. Another year of data or something similar to that is in itself not enough. The paper might be published but such publication strategies are not looked upon favourably in the long term.
So, I strongly recomend rewriting each paper and not rely on copy-pasting. I am convinced you will develop as an author and also generate better papers by doing so. consider each paper a separate new entity.