Application of a new ANN For prediction of Thermodynamic Conditions of Natural Gas Hydrate Formation

Determination of hydrate formation conditions is of the utmost interest for the petroleum and process engineers. And petroleum industry spends millions dollars to combat the formation of hydrate. In this work four empirical correlations proposed by Berge, Motiee, Sloan and Hammerschmidt were evaluated with a new artificial neural network (ANN) method. These correlations are based on gas gravity method has served the gas processing industry well, as an initial estimate for a long period of time. ANN is applied for more than 400 rows of data in the range of 30-74 F, 50-4200 psia and 0.6-1 for temperature, pressure and specific gravity, respectively. In the ANN method, 70% of samples are used for network's training and 30% are chosen for data testing and validation. Also, 6 types of statistical error analyses is used to evaluate the performance and the accuracy of the correlations for estimating natural gas hydrate formation to guide designers and operators in selecting the best correlations for their particular applications. Approximately in all comparisons, the average error of ANN Method is less than the other correlations.

Keywords:

artificial neural network , ANN , natural gas hydrate , hydrate formation , Gas Gravity