A novel study of the gas lift process using an integrated production/injection system using artificial neural network ap
- PDF / 3,671,445 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 7 Downloads / 174 Views
ORIGINAL ARTICLE
A novel study of the gas lift process using an integrated production/ injection system using artificial neural network approach Farshad Abdollahi1 · Seyednooroldin Hosseini1 · Maziar Sabet2 · Seyyed Hamid Esmaeili‑Faraj3 · Fatemeh Amiri4 Received: 6 May 2020 / Accepted: 1 September 2020 © Springer Nature Switzerland AG 2020
Abstract The gas lift process is an applicable method in the oil and gas industry and is employed to increase the production rate and recovery from depleted reservoirs. So, examining the operating parameters and their functionality seems essential for the operating companies. In the current work, an integrated model is proposed to define the whole system by taking into account the effect of several parameters such as the separator pressure, wells’ bottom-hole pressure, and the amount of gas specified to the injection wells. The results demonstrate that the proposed integrated system leads to more accurate and reliable results since they consider and monitor both the surface and sub-surface entries at the same time. Hence, naturally flowing wells as well as those equipped with the gas lift equipment have been modeled by PROSPER while REVEAL helped build the reservoir model and GAP created the surface facilities. In each section, the system is modeled using different software which among them RESOLVE acted as a base where each component could be coupled with one another. In the other section of the study, an artificial neural network (ANN) was used to model the data combined with the BOX–BENKEN design of experiment method. The findings reveal an opposite behavior shown by the two parameters of bottom-hole pressure and separator pressure upon an objective function defined as cumulative oil production. In addition, the findings show that among the examined parameters the bottom-hole pressure introduces more complicated nature. Overall, according to the obtained results, it seems that it is possible to model the system using the artificial neural network with the minimum average absolute relative deviation percent of (AARD %) of 3% and 2% during the training and testing subsets. Keywords Gas lift · Artificial neural network · Integrated system · BOX–BENKEN · Petroleum experts
Introduction * Seyednooroldin Hosseini [email protected] Seyyed Hamid Esmaeili‑Faraj [email protected] Fatemeh Amiri [email protected] 1
EOR Research Center, Department of Petroleum Engineering, Omidiyeh Branch, Islamic Azad University, Omidiyeh, Iran
2
Petroleum and Chemical Engineering Programme, Universiti Teknologi Brunei (UTB), Bndar Seri Begawan, Darussalam, Brunei
3
Department of Material and Chemical Engineering, Shahrood University of Technology, Shahroud, Iran
4
Department of Petroleum Engineering, Islamic Azad University (I.a.U), Masjed‑Soleiman Branch, Masjed‑Soleiman, Iran
As the reservoir pressure falls, leaving not sufficient natural energy for the fluids to be produced on their own, a new source of energy is required to assist the transportation of oil into
Data Loading...
