The complexity of thermodynamic processes — large-scale systems, numerous operating modes, numerous system interactions, etc. does not allow operators to take into account all system parameters. In order to help the operators in their tasks, integrating the decision-making tools into the supervision system is necessary. This paper deals with a multi-modelling approach for supervision using functional analysis, bond graph modelling and structural analysis. In order to show that they are complementary in the design of supervision systems, the different methods are applied to a steam generator installation. It is shown how each contributes to help the operators in their supervisory task.