Sensitivity analysis for the thermal stability criteria of hydrogen peroxide

Chemical reactors carrying on exothermic processes may undergo to runaway reactions. To prevent from this hazard, Early Warning Detection Systems (EWDSs) can be used in industry, because they allow the on-line detection at an early stage of the runaway. The stability criterion of Hub and Jones is frequently implemented in EWDSs. Despite its simplicity during the detection step (the criterion is based just on measurements of the temperatures inside the reactor and jacket), the effectiveness in distinguishing between dangerous and non-dangerous occurrence is strongly affected by the presence of noise in the monitored signals. Furthermore, the numerical methodology for the calculation of the derivative of the measured signals may be of great importance for the definition of alarm conditions. In this paper, the sensitivity analysis of the performances of Hub and Jones criterion with respect to the Savitzky and Golay smoothing filter degree is discussed. The analysis is applied to experimental data on the decomposition of hydrogen peroxide 35%wt carried out in a Thermal Screening Unit.