Simulación y validación de un proceso de deshidratación de gas natural por absorción empleando trietilenglicol

Abstract

The present work was developed in order to simulate and validate a dehydration process of natural gas by absorption, using Triethylene glycol (TEG) with the help of the DWSIM software. This process served as the basis for the creation of an Artificial Neural Network (ANN) with the ability to predict the main outputs of the process. For the creation of the ANN, a sensitivity analysis was developed, by which the inputs and outputs of the process were determined. 100 simulations were performed to obtain a set of 100 data pairs with the determined inputs and outputs. The software used was MATLAB and a network was designed with seven hidden neurons that were tested with the Bayesian regularization algorithm. The mean square error (MSE) obtained in the test phase was 5.3824e-04 with a regression coefficient of 0.99746. For the network validation, a statistical analysis was performed with a set of 20 additional data, taking into account the same variables used for the design. It was concluded that with the results obtained there was no significant difference between simulated values and predicted values with a 95% reliability, in addition, the created ANN is apt to predict the output variables that were determined. It is recommended to perform a correct processing of the data that will be used to create an ANN in order to minimize the presence of atypical data that can reduce the predictive capacity of the net.