Data reconciliation—An industrial case study

Abstract

In this paper the application of data reconciliation to an industrial pyrolysis reactor is discussed. The reconciliation problem was developed around simplified mass and energy balances, and was solved via successive linearisation, although true linear and non-linear methods were briefly explored. The approach was tested using plant data collected at regular intervals over a full operational cycle of the reactor. Presence of gross errors in the data set was determined using a global test, and these were detected and rectified using serial elimination. An independent analysis of the process measurements also identified the source and magnitude of the gross errors and these agreed with results obtained form the data reconciliation. The overall heat transfer coefficient, one of the operating parameters of the pyrolysis reactor, calculated using reconciled data showed a trend consistent with plant experience and could be used to determine better regeneration cycle time of the reactor.