Model Based Projective Monitoring of Process Unit Performances under Uncertainties

The performance monitoring has the fundamental aim to analyze the operative efficiency of single unit operations or overall process sections, especially for lowering costs, increasing profitability and improving the supply chain management in terms of production sites planning and maintenance scheduling. In any case, the industrial state of the art is based on approximated techniques, such as the extrapolation of future behaviour by using plant data historian and/or adopting a (non)linear regression of measurements for predicting the trend of future efficiencies. Instead, as reported in recent literature, a more careful and accurate approach concerns the utilization of detailed first-principles model: the present paper discusses and applies the projective performance monitoring approach, under stochastic disturbances and based on mathematical models. It shows, as preliminary results, comparisons between the expected behaviour obtained by a conventional approximated technique and the prediction of rigorous model based performances.