Baaba-Yakuub Abdul-Muumin Yakuub Yemboni

Integrated Field Development Project – Vic Bilh Oil Field Evaluation
Kwame Nkrumah University of Science and Technology, Department of Petroleum Engineering

• Conducted an extensive evaluation of both reservoir and well deliverability for the Vic Bilh field, located in the southern Aquitaine Basin, France.
• Utilized PROSPER software to perform well modeling and determine the natural operating flow rate, resulting in an oil production rate of 22,933 STB/day at a reservoir pressure of 3,500 psi.
• Carried out sensitivity analyses on tubing size, reservoir pressure, and water cut, leading to the selection of an optimal tubing size of 5 inches for the well design, which maximized flow rates.
• Performed seismic surveys and reservoir modeling, integrating seismic and well data to construct a detailed static reservoir model, aiding in identifying hydrocarbon traps and reservoir heterogeneity.
• Estimated productivity index (PI) at 120.6 STB/day/psi and an absolute open flow (AOF) potential of 261,639.1 STB/day.
• Applied Monte Carlo methods for probabilistic estimation of stock-tank oil initially in place (STOIIP), yielding P50, P90, and P10 values of 2075, 704, and 3745 MMstb respectively.

Comparative Study of Machine Learning Algorithms for Predicting Oil Formation Volume Factor (Bo)
Kwame Nkrumah University of Science and Technology, Department of Petroleum Engineering

• Collaborated in a team of seven to conduct a comprehensive study comparing machine learning models (ANN, RF, DT, SVR, GB) with empirical correlations for predicting the Oil Formation Volume Factor (Bo).
• Processed and analyzed a dataset consisting of 221 observations, which included key reservoir and fluid properties such as Rs (563.95), Temperature (225.31°F), and Oil Specific Gravity (0.829).
• Utilized feature scaling and correlation analyses (Pearson and Spearman) to enhance data quality and model performance.
• Developed and evaluated machine learning models, achieving the best performance with the Artificial Neural Network (ANN) model, with a mean squared error (MSE) of 0.0011, an R² of 0.985, and an average absolute percentage deviation (AAPD) of 4.20%, significantly improving prediction accuracy compared to traditional empirical correlations.
• Applied k-fold cross-validation to ensure robust model evaluation, reducing the potential for overfitting and improving generalization across diverse datasets.