Prediction of Natural Gas Z-Factor by Using ANN Method

Abstract

Abstract-Calculation of the natural gas thermodynamic properties is one of the most important topics in natural gas industry. To calculate the natural gas thermodynamic properties, the compressibility factor must first be calculated. Accordingly, the aim of current investigation is to calculate the natural gas compressibility factor in different conditions with high accuracy. AGA8 EOS is one of the well-known and accurate EOSs in natural gas industry which historically is used for determining natural gas properties. Although, AGA8 equation of state could predict compressibility factor precisely but has a few limitations. In current investigation, an Artificial Neural Network (ANN) is used to overcome this limitation and also speed up calculations. The use of the ANN systems for predicting natural gas properties is a novel method in this field which limited researches have been done around it. The inputs of the ANN system are temperature, pressure and mole fraction of the natural gas components and the output is compressibility factor. For training the ANN system, more than 20000 datasets are collected from AGA8 EOS. This data number is higher than all previous studies in this field and guarantees the high accuracy of the calculations. The results show that the errors of the ANN predictions are less 1% at temperature and pressures that describes in AGA8 validation range which proves the accuracy of ANN model.