Energetic Minimization of Liquefied Natural Gas Single Nitrogen Expander Process Using Real Coded Genetic Algorithm

Abstract

LNG process optimization using Genetic Algorithms was investigated and compared with knowledge-based search algorithm implemented on the same process with the same objective function. The aim was to investigate the effectiveness of such algorithm in contrast to Genetic Algorithms. Scrupulous attention was given to simulating the same process as previous research using HYSYS^®. The simulation software was connected to the C++ GA library (GALib) via Component Object Model (COM) Technology. Steady State, Incremental and Deme Genetic Algorithm implementations were tried out and the Deme Genetic Algorithm was found to be superior to other implementations. Mutation and crossover operators were changed exponentially throughout the GA run. The results show 27% reduction in specific power consumption when compared to the optimum case obtained by earlier research. This proves the superiority of Genetic Algorithms over Knowledge based search algorithms suggested by earlier research.