Application of decision tree, artificial neural networks, and adaptive neuro-fuzzy inference system on predicting lost circulation: A case study from Marun oil field

Sabah, Mohammad; Talebkeikhah, Mohsen; Agin, Farough; Talebkeikhah, Farzaneh; Hasheminasab, Erfan

doi:10.1016/j.petrol.2019.02.045

Cited by 83 publications

(36 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Model performance such as classification accuracy, precision, recall, F1-score, false alarm rate and missed detection rate achieved was similar for the above architectures as shown in Table 5- The capability of a simple ANN architecture, as can be seen in Figure 5-11, to model the available experimental data and deliver on the task of kick detection makes it attractive. Most papers in the literature use much more complex structures [23], [33], [39]- [41]. We concluded not to proceed with a complex architecture for the current study after trying several complex and simple architecture.…”

Section: Comparing Simple and Complex Neural Network Architecturesmentioning

confidence: 96%

“…A closer look reveals that DT was more accurate relative to other models since a large amount of variables including northing, easting, depth, weight on bit, hole size, pump pressure, pump rate, shear stress, viscosity, drilling meterage, drilling time, gel strength, solid percent from retort test, formation type, bit rotational speed, drilling mud pressure, pore pressure and formation fracture pressure were used as input into the models. It was also stated in [23] that ANN followed by ANFIS produced better results if a small number of variables is considered.…”

Section: Surface Parameter Monitoring For Kick Detectionmentioning

confidence: 99%

“…Kicks could be induced in a zone above the drilling bit as a result of lost circulation which dampens the equivalent circulating density of the circulating fluid. Prediction models such as Decision Tree (DT), Artificial Neural Networks (ANNs), Adaptive Neuro-Fuzzy Inference System (ANFIS) and Genetic Algorithm -Multi Layer Perceptron (GA-MLP) were employed in [23] to predict the severity of a lost circulation using field data from Marun oil field in Iran. It was concluded that due to the values of regression coefficient for all models, they could be accurate for predicting amount of lost circulation fluid.…”

Section: Surface Parameter Monitoring For Kick Detectionmentioning

confidence: 99%

“…There are several machine learning algorithms such as logistic regression, decision tree [23] that could suffice for kick detection. However, they may not be easily tuned to capture complexities as the process dynamics evolves rapidly.…”

Section: Data From Lds Kick Experimentsmentioning

confidence: 99%

“…The overall methodology is summarized in Figure 5-1. The author was further motivated to look for simple ANN models that would be sufficient for the kick detection problem because numerous papers report the use of complex neural networks to solve the problem without justifying the need for such complex solutions [23], [33], [39]- [41]. In many cases, the researchers arbitrarily select a complex neural network architecture, e.g.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Supervised data-driven approach to early kick detection during drilling operation

Muojeke

Venkatesan

Khan

2020

Journal of Petroleum Science and Engineering

View full text Add to dashboard Cite

My profound gratitude goes to God Almighty for the gift of life, good health and connecting me to be refined by renowned supervisors. I would like to express my sincere gratitude to my supervisors, Dr. Ramachandran Venkatesan and Dr. Faisal Khan for their continuous support, insightful comments, and for sharpening my research skills from various perspectives. My sincere appreciation goes to Dr. Stephen Butt, who provided me the opportunity to join his team with access to the Drilling Technology Laboratory and research facilities. I would like to thank team member Nhat Minh for his excellent team play, Augustine Osarogiagbon, DTL team members Mohammed Said, Abdelsalam Abugharara, Jeronimo De Moura, Idris Sule, Olalere Oloruntobi and Justin Royce for assistance during the design, development of experimental setup and conducting the experiments. I would also like to thank Technical Services team Dennis Cramm, David Snook, Ken, Aaron and Shawn Organ of the Concrete laboratory. I would like to thank my family: my parents, siblings, godparents, and friends for supporting me throughout my research and my life in general. Special appreciation to Archbishop Martin Currie, Fr. Joe Mroz, SJ, Presentation Sisters and the Sisters of Mercy for immense spiritual support. I also appreciate the Public Service Alliance of Canada for rich extracurricular learning and volunteering experience which included an oversea trip as a delegate on a Social Justice mission to Guatemala and participation in conferences and convention. I thankfully acknowledge the financial support provided by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Canada Research Chair (Tier I) Program in Offshore Safety and Risk Engineering.

show abstract

Section: Comparing Simple and Complex Neural Network Architecturesmentioning

confidence: 96%

Section: Surface Parameter Monitoring For Kick Detectionmentioning

confidence: 99%

Section: Surface Parameter Monitoring For Kick Detectionmentioning

confidence: 99%

Section: Data From Lds Kick Experimentsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Supervised data-driven approach to early kick detection during drilling operation

Muojeke

Venkatesan

Khan

2020

Journal of Petroleum Science and Engineering

View full text Add to dashboard Cite

show abstract

Research on spectral method of lost circulation layer located by transient pressure wave

Wang

Zhu

Lei

et al. 2023

Energy Science & Engineering

View full text Add to dashboard Cite

Leakage in the wellbore annulus during drilling operations can affect normal production operations, resulting in a severe waste of resources and economic loss, so it is crucial to adopt a fast and effective leak identification method for subsequent plugging operations. For the complex problem of judging the location of the leakage layer, we proposed the method of excitation pressure wave to identify the location of the leakage layer. By analyzing the transfer of pressure waves within the annular pipe system and the pressure head response spectrum, the leak's location is identified based on the location of the resonance point and the change in resonance amplitude. The pressure wave signal contains too much noise. The variational mode decomposition (VMD) algorithm and the Hilbert joint spectrum were used to extract the main frequency components to reconstruct the signal to achieve the denoising effect. On this basis, the reconstructed signal is processed by fast Fourier transform (FFT) to obtain the pressure wave response spectrum, analyze the frequency domain features, and then determine the location of the leakage layer. The experimental results verify that: ① When a leak occurs in the wellbore annulus, the pressure wave will generate additional resonance points in the frequency domain due to the presence of the leak point. ② The combination of the VMD algorithm, FFT, and Hilbert joint spectrum can effectively remove the noise of the pressure wave signal. ③ The method effectively avoids the difficulty of identifying negative pressure waves in the time domain analysis of pressure wave signals. It can effectively locate the leaky layer in the frequency domain analysis. It is concluded that the principle of the method is feasible and has practical significance for field application.

show abstract

Advanced Porosity Prediction in Heterogeneous Oil Reservoirs: Using Novel Machine Learning and Deep Learning Techniques

Al-Khafaji,

Meng,

Yahya

et al. 2024

Springer Series in Geomechanics and Geoengineering

View full text Add to dashboard Cite

Application of decision tree, artificial neural networks, and adaptive neuro-fuzzy inference system on predicting lost circulation: A case study from Marun oil field

Cited by 83 publications

References 43 publications

Supervised data-driven approach to early kick detection during drilling operation

Supervised data-driven approach to early kick detection during drilling operation

Research on spectral method of lost circulation layer located by transient pressure wave

Advanced Porosity Prediction in Heterogeneous Oil Reservoirs: Using Novel Machine Learning and Deep Learning Techniques

Contact Info

Product

Resources

About