The results of a collaborative study on the operation of a steam production network are presented in this paper. An overdesigned utility network is first modelled, based on plant measurements, where the network equipment is modelled through sets of conventional processing blocks. Data reconciliation methods are applied to determine design parameters, as turbine efficiencies, pump efficiencies etc. A multiperiod optimization framework is then employed to determine network operation and energy management schedules for variable demand levels and for uncertain process parameters.