Arisni, Nadya (2025) Optimisasi Injeksi Air pada Lapangan Minyak dengan Waterflood untuk Meningkatkan Perolehan Minyak menggunakan Physics Informed Neural Networks dan Algoritma Optimisasi Stokastik. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5009211079-Undergraduate _Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5009211079-Undergraduate _Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy |
Abstract
Waterflooding merupakan teknik yang umum digunakan untuk menjaga tekanan reservoir dan meningkatkan perolehan minyak. Namun, distribusi laju injeksi yang tidak optimal sering menyebabkan inefisiensi dan menurunnya profitabilitas. Penelitian ini mengembangkan model berbasis Capacitance Resistance Model–Physics Informed Neural Network (CRM-PINN) untuk menangkap hubungan non-linear berbasis fisika antara laju injeksi air dan laju produksi fluida. Model ini dilatih menggunakan data historis harian dari Oktober 2013 hingga Oktober 2016 sebanyak 1127 time step. CRM-PINN menunjukkan akurasi tinggi dengan Normalized Root Mean Square Error (NRMSE) di bawah 20% pada seluruh sumur produksi. Model ini kemudian diintegrasikan dengan metode optimisasi untuk memaksimalkan total produksi fluida melalui penentuan distribusi injeksi optimal pada empat sumur injeksi. Dua skenario dijalankan berdasarkan Voidage Replacement Ratio (VRR) 0,8 dan 0,9, dengan batasan total dan individu. Lima algoritma optimisasi stokastik digunakan: Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Duelist Algorithm (DA), Rain Water Algorithm (RWA), dan Killer Whale Algorithm (KWA). Hasilnya, seluruh algoritma meningkatkan revenue, dengan peningkatan tertinggi oleh DA (133%) dan KWA (101%) pada skenario VRR 0,9. Hal ini membuktikan bahwa CRM-PINN efektif untuk prediksi dan perencanaan injeksi waterflood.
======================================================================================================================================
Waterflooding is widely used to maintain reservoir pressure and enhance oil recovery. However, suboptimal injection rate distribution often causes inefficiencies and reduced field profitability. This study developed a predictive model based on the Capacitance Resistance Model–Physics Informed Neural Network (CRM-PINN) to capture the nonlinear, physics-based relationship between water injection and fluid production rates. The model was trained using daily historical data from October 2013 to October 2016 (1127 time steps) and showed high accuracy, with Normalized Root Mean Square Error (NRMSE) below 20% across all production wells. The trained CRM-PINN was integrated into an optimization framework to maximize total fluid production by determining the optimal injection distribution across four injection wells. Two scenarios were evaluated using Voidage Replacement Ratio (VRR) values of 0.8 and 0.9, subject to individual and total injection constraints. Five stochastic optimization algorithms were applied: Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Duelist Algorithm (DA), Rain Water Algorithm (RWA), and Killer Whale Algorithm (KWA). Results showed that all algorithms improved revenue compared to the pre-optimization case, with the highest increases from DA (133%) and KWA (101%) under the VRR 0.9 scenario. These results demonstrate CRM-PINN’s effectiveness as a predictive and decision-support tool for optimizing waterflooding strategies.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Capacitance Resistance Model, Physics-Informed Neural Network, Optimisasi Injeksi, Produksi Minyak, Waterflood, Capacitance Resistance Model, Injection Optimization, Oil Production, Physics-Informed Neural Network, Waterflood |
| Subjects: | Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. |
| Divisions: | Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis |
| Depositing User: | Nadya Putri Arisni |
| Date Deposited: | 05 Aug 2025 08:48 |
| Last Modified: | 05 Aug 2025 08:48 |
| URI: | http://repository.its.ac.id/id/eprint/127530 |
Actions (login required)
 |
View Item |