Development and testing of an artificial neural network model for predicting bottomhole pressure in vertical multiphase flow

Abstract

A Neural Network technique has been used successfully for developing a model for predicting bottomhole pressure in vertical multiphase flow in oil wells. The new model has been developed using the most robust learning algorithm (back-propagation). A total number of 206 data sets; collected from Middle East fields; has been used in developing the model. The data used for developing the model covers an oil rate from 280 to 19618 BPD, water cut up to 44.8%,and gas oil ratios up to 675.5 SCF/STB. A ratio of 3:1:1 between training, validation, and testing sets yielded the best training/testing performance. The best available empirical correlations and mechanistic models have been tested against the data and the developed model. Graphical and statistical tools have been utilized for the sake of comparing the performance of the new model and other empirical correlations and mechanistic models. Thorough verifications indicated that the new developed model outperforms all tested empirical correlations and mechanistic models in terms of highest correlation coefficient, lowest average absolute percent error, lowest standard deviation, lowest maximum error, and lowest root mean square error. The new developed model results can only be used within the range of used data; hence care should be taken if other data beyond this limit is implemented.