Undersaturated Oil Compressibility Prediction for Mishrif Reservoir in the Southern Iraqi Oil Fields Using Artificial Neural Network | ||
| Journal of Petroleum Research & Studies | ||
| Article 1, Volume 377, Issue 10, December 2014, Pages 54-72 | ||
| Authors | ||
| Emad A. Fakher; Dr. Hussain A. Baker; Dunia Al-Shamma | ||
| Abstract | ||
| Coefficient of isothermal oil compressibility is required in transient fluid flow problems, extension of fluid properties from values at the bubble point pressure to higher pressures of interest and in material balance calculations [1, 2]. Coefficient of isothermal oil compressibility is a measure of the fractional change in volume as pressure is changed at constant temperature [3]. Coefficients of isothermal oil compressibility are usually obtained from reservoir fluid analysis. Reservoir fluid analysis is an expensive and time consuming operation that is not always available when the volumetric properties of reservoir fluids are needed. For this reason correlations have been developed and are being developed for predicting fluid properties including the coefficient of isothermal oil compressibility. This paper presents an application of Artificial Neural Network (ANN) methods for estimation of isothermal oil compressibility for the Mishrif reservoir oils among of commonly available field data, according to the fact that this method is useful when relationships of parameters are too complicated. The method is proposed as a more effective prognostic tool than are currently available procedures. In this study, back propagation (BPN) network was used to develop an ANN model to predict isothermal oil compressibility. A three layer feed-forward network has been selected which has the best correlation coefficient in testing the models. The new model of undersaturated oil compressibility in which the first layer consists of five neurons representing the input values of pressure above bubble point, solution gas oil ratio at bubble point, oil gravity API, reservoir temperature and gas specific gravity. The second (hidden) layer contains 12 neurons, and the third layer contains one neuron representing the output values of undersaturated oil compressibility. It was found that the new model estimate undersaturated oil compressibility for Mishrif reservoir crudes in the southern Iraqi oil fields much better than the published ones. The present model predicts undersaturated oil compressibility with an average absolute relative error of 3.86% for the testing data set. | ||
|
Statistics Article View: 120 PDF Download: 26 |
||