Neuro-genetic approach for optimization of the water flowrates distribution on a hydrogen sulphide cooling system

Abstract

Efficient utilities usage and enhanced heat transfer are imperative in todays’ industrial and technological processes. However, there are several facilities in the nickel industry not upgraded yet. This research studied a hydrogen sulphide production plant with limited heat exchange capacity. Within this context, a neuro-genetic procedure was proposed for optimization of the water flowrates distribution on a hydrogen sulphide gas coolers system. It relied on Genetic Algorithms, combined with an improved ɛ-NTU model for simulation of jacketed shell-and-tube heat exchangers. Artificial Neural Networks were furtherly applied to correlate the optimum water flowrates to predictive variables. The heat transfer incremental was estimated from 3695 to 10514 W, while reduction of the gas exit temperature was projected between 2.9 and 9.8 K. Calculated heat recovery averaged 12.44 %. The optimized water distribution scheme have improved the system energy performance. Applied concept of fixed network and unvaried overall feed water flowrate avoided the additional costs related to topology modifications. A technological solution was provided, consisting on installation of automatic valves and programmable flow control-loops linked to a PLC.