Application of real time optimisation for fatty acid fractionation process

Abstract

This thesis discusses the application of Real Time Optimization (RTO) in improving process plant profitability. The RTO cycle consisting of five major components, namely, plant model in steady state and dynamic modes, steady state detection, data reconciliation, gross error detection and economic optimisation routines were developed and tested on a selected base–case operating condition of a fatty acid fractionation (FAF) process. The cycle of RTO implementation began with collection of selected process data from the plant, represented by a dynamic simulation model developed using HYSYS.PlantTM version 2.4. The measured data were then evaluated by the steady state detection mechanism to ascertain that the process had reached steady state operating condition prior to the evaluation by the data reconciliation and gross error detection stages. Following these data validation phases, the search for optimal operating conditions was executed by the HYSYS.PlantTM optimiser, facilitated by the steady state model of the plant. Successful implementation with profit improvement of 5.61% over the base-case condition was obtained. Larger profitability was difficult to realise due to tight constraints imposed on this low pressure fractionation plant. The RTO scheme was then tested for robustness by introducing four types of process uncertainties. These were the variation in product prices, measurement noise, leakage in process streams and process disturbances. In all cases, errors introduced by these uncertainties were successfully detected and rectified and successful process optimizations were obtained. The results obtained in this study proved the capability of the RTO scheme in improving the profitability of process plant operation.