The research of a novel soft sensor technology for chemical reactors based on thermodynamic models opens up previously unavailable optimization possibilities in chemical production processes and is the starting point of this project. The problem in CAT-SAVE is that inside continuous chemical reactors where phase transitions occur during the reaction, thermodynamic states and compositions of the process stream are subject to such strong local and temporal changes that it is currently not possible to determine the states within the chemical reactor based on measurable inlet and outlet conditions via soft sensors. This involves a considerable risk with regard to product quality and process reliability. In particular, exothermic processes often have temperature peaks inside a reactor, which destroy catalysts and components or adversely affect their service life. It therefore requires a new approach for the process industry in the field of soft sensors, which makes it possible to accurately predict the reaction progress in the apparatus based on exact thermodynamic models (locally and temporally).

Since chemical production engineering often involves multicomponent mixtures with up to more than one hundred individual components, the modelling for the soft sensor system within the reactor is very complex since the phase equilibrium of the multicomponent mixture must be modelled with the progress of the reaction. Attached to this is the following objective:

• In the CAT-SAVE project, a soft-sensing method is to be explored on the basis of a real chemical reactor of a petrochemical production plant, which allows precise prediction of pressure, temperature, phase states and compositions of a process stream inside a multiphase reactor with time and place. The real process sensor technology required for this only supplies the process parameters of the input and output currents.
• Real-time visualization of the key Parameters
• Destructive temperature peaks and unwanted by-product formation should be avoided
• Plant throughouts should be precisely optimized under inclusion of material holding time and catalysts life time
• This should increase the throughput-related catalyst life cycle by up to 10% in the reference plant
• The method should have highest reproducibility in other processes

Planned results: Simulation model of the reference unit condensing unit and algorithm for determining the wall temperature; The most realistic simulation model possible; Implementation of the interface between process database and CAT-SAVE model and simulation of the application in real operating parameters; Value criteria of the performance Standards;