As a result of the complexity of activated sludge systems, most biological water treatment plants have been designed by scale-up, based upon the empirical knowledge gained from pilot plant experiments and simulations. Due to a lack of applicable design methodologies that are able to handle the complexity of the activated sludge process models, systematic process synthesis issues have not received much attention to date. We apply recently developed stochastic optimisation-based synthesis technology for reaction/separation networks to the activated sludge process design problem known as the single sludge denitrification, which aims at performing simultaneous oxidation of the organic matter and complete removal of organic and ammonia nitrogen. The study aims at acquiring design knowledge from the biochemical reactor network with sludge separation and recycles in order to identify design directions towards improved process performances as well as to reveal the scope for improvement. The study reveals designs and the principles behind these that allow for the development of activated sludge processes with significantly enhanced denitrification performance as compared with the conventional, sequential sludge system designs.