A new correlation based on artificial neural networks for predicting the natural gas compressibility factor

摘要

In this study a new correlation of natural gas compressibility factor based on theory of Mohammadikhah-Mohebbi-Abolghasemi's equation of state (MMA EOS) is developed using an artificial neural network. In MMA EOS, the compressibility factor as a function of M-factor (BP/RT) is expressed. An artificial neural network (ANN) is designed in which the M-factor, reduced temperature, and reduced pressure are selected as input variables, whereas the natural gas compressibility factor is selected as output. Then, a new correlation based on the weights of ANN is obtained. Results of this correlation are compared with some other equations and experimental data. Proposed correlation for 597 data points has an average absolute deviation (AAD%) of 0. 6% and a correlation coefficient (R ~2 value) of 0.9999.