(6aa) Application of New Explicit Correlation and Neural Network-Based Models for an Efficient Prediction of Natural Gas Compressibility Factor

This study presents two new methods for calculating the compressibility factor (z-factor) of natural gases. The first method, explicit correlation, is an efficient empirical model to calculate this important parameter. In this approach, 4158 data points obtained from the Standing and Katz z-factor chart have been used. The second method is based on Artificial Neural Networks (ANN), in which a 2:5:5:1 structure is used as an optimum network to predict Z-factor. In this approach, the proposed ANN model for predicting the Z-factor behaviour is detailed. This method employs a back-propagation algorithm, which is very effective in predicting model nonlinear variables.

In this case study, the z-factor data of a gas sample is calculated by different existing methods. Afterwards, the calculated data are graphically compared to experimental Z-factor data. In this research a new correlation for rapid estimation of the z-factor for sweet gases is studied. This correlation is developed based on 4158 points from the Standing and Katz Z-factor chart. The advantage of the developed correlation is that it is explicit in Z and thus does not require an iterative solution unlike other methods. The accuracy of the proposed Z-factor correlation has been compared to other correlations. The comparison demonstrates the superiority of the new correlation over other correlations used to calculate the Z- factor of natural gases. The results also show that the proposed ANN can be properly trained for the purpose of estimating the compressibility factor with an acceptable accuracy.