Evolving a robust modeling tool for prediction of natural gas hydrate formation conditions

• Predicting gas hydrate formation for a wide range of gas mixtures using neural network.
• Selecting best NN structure by checking the performance through optimization processes.
• A detailed comparison between ANN predictions and the results of correlations and models.
• Excellent agreement between experimental data and predicted values is observed.

Natural gas is a very important energy source. The production, processing and transportation of natural gas can be affected significantly by gas hydrates. Pipeline blockages due to hydrate formation causes operational problems and a decrease in production performance. This paper presents an improved artificial neural network (ANN) method to predict the hydrate formation temperature (HFT) for a wide range of gas mixtures. A new approach was used to define the variables for formation of a hydrate structure according to each species presented in natural gas mixtures. This approach resulted in a strong network with a precise prediction, especially in the case of sour gases.This study also presents a detailed comparison of the results predicted by this ANN model with those of other correlations and thermodynamics-based models for an estimation of the HFT. The results showed that the proposed ANN model predictions are in much better agreement with the experimental data than the existing models and correlations. Finally, outlier detection was performed on the entire data set to identify any defective measurements of the experimental data.