BOTTOM-HOLE FLOWING PRESSURE CALCULATION IN DEVIATED MULTIPHASE FLOW GAS WELLS USING ARTIFICIAL NEURAL NETWORK (ANN) - A CASE STUDY IN THE TUNU GAS FIELD, TOTAL EP INDONESIE

摘要

TUNU is a mature gas field located in the Mahakam Delta of East Kalimantan, Indonesia that has been producing since 1990. Currently, there are ± 360 active producing wells with a total production of ± 600 MMSCFD. Well monitoring (SGS, FGS, PLT, etc.) is an important job in the Tunu field. In 2015, there are ± 850 jobs performing static and flowing gradient surveys. The objective of this well intervention is to understand the bottom hole flowing and shut in pressure (BHFP/BHSP). This huge operation is impacting significantly on cost and safety (barge mobilization, rig-up/down, etc). BHFP can be estimated using vertical flow correlations that are available in production engineering software. Among correlations available, not all of them are fit or relevant with actual conditions. The available correlations are acceptable for high gas rate wells while they give high errors (> 25%) at low gas rate (< 2 MMSCFD), whereas in TUNU today 64% of them are flowing at this condition. Therefore, another approach such as Artificial Neural Network (ANN) is required to calculate BHFP. The actual BHFP data were gathered and then imported into the ANN models. The model training inputs used are gas rate, condensate gas ratio (CGR), water gas ratio (WGR), tubing diameter, measured depth (MD) and true vertical depth (TVD). In this paper, ANN techniques were applied to predict BHFP and proved to have better prediction and performance. Back-propagation (BP) method is used in building the neural network to modify the fitting to achieve higher prediction accuracy and broaden the prediction range. The result is then compared to the existing correlations. ANN has proven to give a better result (error ± 4.32%) than the existing correlations in prediction of BHFP in deviated gas wells.