This question goes a bit beyond openLCA and is more LCA methodology; if one of your systems produces more byproduct than the baseline scenario, I think it is up to you. You only need to ensure that the benefit provided by all compared products (as quantified) is identical, but maybe, since there is always some decision in defining processes used for system expansion, you are better off crediting only the "surplus byproduct" (if I understand the question correctly).
Best wishes,
Andreas