Doan Huy Hien
Doan Huy Hien has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Bui Viet Dung
Vietnam Petroleum Institute, Hanoi 100000, Vietnam
3 shared publications - Pham Quy Ngoc
Vietnam Petroleum Institute, Hanoi 100000, Vietnam
3 shared publications - Pham Huy Giao
Asian Institute of Technology
2 shared publications Bui Viet Dung
Vietnam Petroleum Institute, 167 Trung Kinh Street, Hanoi, Vietnam
1 shared publication- Dinh Duc Huy
Vietnam Petroleum Institute, Hanoi 100000, Vietnam
1 shared publication - Do Van Chanh
Vietnam Petroleum Institute, Hanoi 100000, Vietnam
1 shared publication
Latest Publications
- Open Access
- 30 September 2025
- Published by Vietnam Petroleum Institute in Petrovietnam Journal
- Vol. 3, 13-21
SỬ DỤNG THUẬT TOÁN MẠNG NEURAL NHÂN TẠO TRONG TỐI ƯU BƠM ÉP NƯỚC CHO MỎ BẠCH HỔ
Sản lượng dầu ngoài khơi Việt Nam được khai thác chủ yếu từ vỉa chứa móng Bạch Hổ, nơi có chế độ dòng chảy rất phức tạp do sự phân bố không gian bất đồng nhất của các thuộc tính vật lý thạch học như độ rỗng, độ thấm và độ bão hòa nước. Do đó, phương pháp dự báo sản lượng dầu truyền thống dựa trên mô phỏng vỉa chứa thường thiếu chính xác hoặc đòi hỏi nhiều công sức và thời gian để tối ưu hóa các thông số thủy động lực học. Sự phát triển gần đây của các thuật toán học máy (machine learning - ML) sẽ giúp dự đoán được sản lượng dầu từ lưu lượng bơm ép nước của từng giếng bơm nhanh hơn và đáng tin cậy hơn. Một khi có thể dự đoán sản lượng dầu bằng phương pháp học máy, việc tối ưu hóa quá trình bơm ép nước có thể được triển khai bằng nhiều thuật toán tối ưu hóa khác nhau.Trong nghiên cứu này, sử dụng thuật toán mạng neural nhân tạo (ANN) đã cho kết quả tốt nhất: hệ số tương quan giữa các giá trị dự đoán và giá trị thực tế lần lượt là 0,98 và 0,95 đối với tập dữ liệu huấn luyện và kiểm tra. Sau đó, thuật toán tối ưu Gauss-Newton được áp dụng để tìm ra lưu lượng bơm ép nước tối ưu nhất cho từng giếng bơm nhằm tăng sản lượng dầu. Kết quả cho thấy sản lượng dầu khai thác tăng trung bình 1,5% khi sử dụng các lược đồ bơm ép mới được tối ưu hóa.
- 30 September 2025
- Published by Vietnam Petroleum Institute in Petrovietnam Journal
- Vol. 3, 13-21
Open AccessSỬ DỤNG THUẬT TOÁN MẠNG NEURAL NHÂN TẠO TRONG TỐI ƯU BƠM ÉP NƯỚC CHO MỎ BẠCH HỔ
Sản lượng dầu ngoài khơi Việt Nam được khai thác chủ yếu từ vỉa chứa móng Bạch Hổ, nơi có chế độ dòng chảy rất phức tạp do sự phân bố không gian bất đồng nhất của các thuộc tính vật lý thạch học như độ rỗng, độ thấm và độ bão hòa nước. Do đó, phương pháp dự báo sản lượng dầu truyền thống dựa trên mô phỏng vỉa chứa thường thiếu chính xác hoặc đòi hỏi nhiều công sức và thời gian để tối ưu hóa các thông số thủy động lực học. Sự phát triển gần đây của các thuật toán học máy (machine learning - ML) sẽ giúp dự đoán được sản lượng dầu từ lưu lượng bơm ép nước của từng giếng bơm nhanh hơn và đáng tin cậy hơn. Một khi có thể dự đoán sản lượng dầu bằng phương pháp học máy, việc tối ưu hóa quá trình bơm ép nước có thể được triển khai bằng nhiều thuật toán tối ưu hóa khác nhau.Trong nghiên cứu này, sử dụng thuật toán mạng neural nhân tạo (ANN) đã cho kết quả tốt nhất: hệ số tương quan giữa các giá trị dự đoán và giá trị thực tế lần lượt là 0,98 và 0,95 đối với tập dữ liệu huấn luyện và kiểm tra. Sau đó, thuật toán tối ưu Gauss-Newton được áp dụng để tìm ra lưu lượng bơm ép nước tối ưu nhất cho từng giếng bơm nhằm tăng sản lượng dầu. Kết quả cho thấy sản lượng dầu khai thác tăng trung bình 1,5% khi sử dụng các lược đồ bơm ép mới được tối ưu hóa.
- 11 July 2024
- Published by Springer Nature
- p. 2887-2894
Seabed Morphological Characterization and Shallow Geohazards at Offshore Binh Thuan Wind Farm Sites, Southern Vietnam Shelf
- 11 July 2024
- Published by Springer Nature
- p. 2887-2894
Seabed Morphological Characterization and Shallow Geohazards at Offshore Binh Thuan Wind Farm Sites, Southern Vietnam Shelf
- Open Access
Hydraulic Flow Unit Classification and Prediction Using Machine Learning Techniques: A Case Study from the Nam Con Son Basin, Offshore Vietnam
The test study area is the Miocene reservoir of Nam Con Son Basin, offshore Vietnam. In the study we used unsupervised learning to automatically cluster hydraulic flow units (HU) based on flow zone indicators (FZI) in a core plug dataset. Then we applied supervised learning to predict HU by combining core and well log data. We tested several machine learning algorithms. In the first phase, we derived hydraulic flow unit clustering of porosity and permeability of core data using unsupervised machine learning methods such as Ward’s, K mean, Self-Organize Map (SOM) and Fuzzy C mean (FCM). Then we applied supervised machine learning methods including Artificial Neural Networks (ANN), Support Vector Machines (SVM), Boosted Tree (BT) and Random Forest (RF). We combined both core and log data to predict HU logs for the full well section of the wells without core data. We used four wells with six logs (GR, DT, NPHI, LLD, LSS and RHOB) and 578 cores from the Miocene reservoir to train, validate and test the data. Our goal was to show that the correct combination of cores and well logs data would provide reservoir engineers with a tool for HU classification and estimation of permeability in a continuous geological profile. Our research showed that machine learning effectively boosts the prediction of permeability, reduces uncertainty in reservoir modeling, and improves project economics.
Open AccessHydraulic Flow Unit Classification and Prediction Using Machine Learning Techniques: A Case Study from the Nam Con Son Basin, Offshore Vietnam
The test study area is the Miocene reservoir of Nam Con Son Basin, offshore Vietnam. In the study we used unsupervised learning to automatically cluster hydraulic flow units (HU) based on flow zone indicators (FZI) in a core plug dataset. Then we applied supervised learning to predict HU by combining core and well log data. We tested several machine learning algorithms. In the first phase, we derived hydraulic flow unit clustering of porosity and permeability of core data using unsupervised machine learning methods such as Ward’s, K mean, Self-Organize Map (SOM) and Fuzzy C mean (FCM). Then we applied supervised machine learning methods including Artificial Neural Networks (ANN), Support Vector Machines (SVM), Boosted Tree (BT) and Random Forest (RF). We combined both core and log data to predict HU logs for the full well section of the wells without core data. We used four wells with six logs (GR, DT, NPHI, LLD, LSS and RHOB) and 578 cores from the Miocene reservoir to train, validate and test the data. Our goal was to show that the correct combination of cores and well logs data would provide reservoir engineers with a tool for HU classification and estimation of permeability in a continuous geological profile. Our research showed that machine learning effectively boosts the prediction of permeability, reduces uncertainty in reservoir modeling, and improves project economics.
- Open Access
Numerical Simulation of Low Salinity Water Flooding on Core Samples for an Oil Reservoir in the Nam Con Son Basin, Vietnam
Low-salinity water flooding (LSWF) is environment-friendly and operates similarly to conventional waterflooding without the need for synthetic chemical materials. The application of LSWF makes sense in Vietnam as HC production has steadily declined since 2002, and the majority of main oil fields have become near mature and mature fields. In the next years, Enhanced Oil Recovery (EOR) should be a top priority for Petro Vietnam to boost its oil production, for which the key issue is how to select a suitable EOR technology. In this study, LSWF of the Lower Miocene sand using low salinity water from Lower Oligocene sand was investigated. Previously at the Ruby field in the Cuu Long Basin, an LSWF feasibility study was carried out based on a conventional core flooding experiment, which is time-consuming and costly. This study targets the Chim Sao field in the Nam Con Son Basin, for which a cheaper and faster assessing method is required. As a result, a numerical code written in Matlab was developed and successfully validated with the core flooding experiment results obtained at the Ruby field. The LSWF simulation was conducted using the multiple ion-exchange mechanisms (MIE), and the results obtained showed an increase in the oil recovery factor by 2.19% for the Lower Miocene Sand. Another important outcome of this study is the innovative proposal and successful simulation to use the abundant low salinity water from the underlying Lower Oligocene sand as a natural LSW source to inject into the Lower Miocene oil reservoir that can be a decisive factor to help apply LSWF in practice on a wide scale not only for Chim Sao but also other similar oil fields in southern offshore Vietnam.
Open AccessNumerical Simulation of Low Salinity Water Flooding on Core Samples for an Oil Reservoir in the Nam Con Son Basin, Vietnam
Low-salinity water flooding (LSWF) is environment-friendly and operates similarly to conventional waterflooding without the need for synthetic chemical materials. The application of LSWF makes sense in Vietnam as HC production has steadily declined since 2002, and the majority of main oil fields have become near mature and mature fields. In the next years, Enhanced Oil Recovery (EOR) should be a top priority for Petro Vietnam to boost its oil production, for which the key issue is how to select a suitable EOR technology. In this study, LSWF of the Lower Miocene sand using low salinity water from Lower Oligocene sand was investigated. Previously at the Ruby field in the Cuu Long Basin, an LSWF feasibility study was carried out based on a conventional core flooding experiment, which is time-consuming and costly. This study targets the Chim Sao field in the Nam Con Son Basin, for which a cheaper and faster assessing method is required. As a result, a numerical code written in Matlab was developed and successfully validated with the core flooding experiment results obtained at the Ruby field. The LSWF simulation was conducted using the multiple ion-exchange mechanisms (MIE), and the results obtained showed an increase in the oil recovery factor by 2.19% for the Lower Miocene Sand. Another important outcome of this study is the innovative proposal and successful simulation to use the abundant low salinity water from the underlying Lower Oligocene sand as a natural LSW source to inject into the Lower Miocene oil reservoir that can be a decisive factor to help apply LSWF in practice on a wide scale not only for Chim Sao but also other similar oil fields in southern offshore Vietnam.
- 1 November 2018
- Published by Australian Society of Exploration Geophysicists in Exploration Geophysics
- Vol. 49 (6) , 833-843
Sensitivity evaluation of a seismic interpolation algorithm
Sparse seismic acquisition is a new trend in seismic exploration, as it costs much less than conventional methods. To maintain the initial resolution of the seismic image, we propose several ways to sample data irregularly but periodically. These were tested by decimating the synthetic data, then interpolating, imaging and inversion. At every processing step, we quantified the effect of interpolation by comparing the results with those from the fully sampled data. Once the numerical test suggested the best decimation scheme, we were able to proceed to test the real dataset. This test confirmed that sparse acquisition using 60% of the available data is feasible.
- 1 November 2018
- Published by Australian Society of Exploration Geophysicists in Exploration Geophysics
- Vol. 49 (6) , 833-843
Sensitivity evaluation of a seismic interpolation algorithm
Sparse seismic acquisition is a new trend in seismic exploration, as it costs much less than conventional methods. To maintain the initial resolution of the seismic image, we propose several ways to sample data irregularly but periodically. These were tested by decimating the synthetic data, then interpolating, imaging and inversion. At every processing step, we quantified the effect of interpolation by comparing the results with those from the fully sampled data. Once the numerical test suggested the best decimation scheme, we were able to proceed to test the real dataset. This test confirmed that sparse acquisition using 60% of the available data is feasible.