[5E2] A data-driven dynamic process modelling and optimisation framework for condition monitoring of a liquefied gas refrigeration unit in South Pars petrochemical processing plant
S Rahbarimanesh¹, A Nejat² and A Rahbarimanesh³
¹University of British Columbia, Canada
²University of Tehran, Iran
³University of Manchester, UK
A commonly reported issue in the modern maintenance procedures of refrigeration units in gas processing plants is the lack of suitable process control and condition monitoring tools that can properly diagnose the unit faulty behaviours at off-design operations in real-time sense and then deploy necessary self-adjusting measures. Such problems, on the other hand, most often cannot be predicted and/or resolved through available dynamic process simulator packages, which is mainly because of the 'general-purpose' functionality of these packages, beside the constraining assumptions employed in developing their numerical calculations. The present study aims to address a malfunctioning refrigeration unit in the currently operating South Pars (SP) liquefied petroleum gas (LPG) plant by putting forward a data-driven condition monitoring framework that is able to effectively predict and resolve real-time faulty behaviour of the unit through customising its dynamic process simulators – based on the actual history of operation provided by the unit embedded measurement tools – and then coupling the output of the simulators to the plant central control unit for planned optimisation purposes. Given the high level of adaptivity of the proposed framework, it can ultimately be utilised as a complementary means for improving the condition monitoring procedures at a broad range of relevant industrial processing plants.
