Adapting state-space reduction techniques to match steady-state responses