Aiming at reducing their emissions, wastewater treatment plants (WWTP) seek to reduce their energy consumption, where a large amount is used for the aeration. The case plant, Grindsted WWTP uses an alternating aeration strategy, with a common air supply system facilitating the process in four aeration tanks and thus making optimisation challenging. In this work, a nonlinear model of the air supply system is designed, in which multiple key parameters are estimated by data-driven optimization. Subsequently, a model-based control strategy for scheduling of compressors and desired airflow is proposed, to save energy without compromising the aeration performance. The strategy is based upon partly static- partly dynamic models of the compressors, describing their efficiency in terms of system head and volumetric airflow rate. The simulation study uses real plant data and shows great potential for improvement of energy efficiency, regardless of the aeration pattern in any of the four process tanks, and furthermore contributes to a reduction in compressor restarts per day. The proposed method is applicable to other WWTP with multiple compressors in the air supply system, as this study is conducted using first principle models validated on data from the daily operation.