Abstract:
Machine Learning (ML) techniques are more and more applied in hydrocarbon
exploration and production (E&P) in general, and in petrophysics in particular. In
this research, a Genetic Algorithm (GA) code was developed in Python language to
analyze the fracture porosity of a Fractured Granite Basement (FGB) reservoir,
which is difficult to calculate due to the reservoir heterogeneity caused by fracture
networks. The study well was in the Cuu long basin, Vietnam. The steps of GA code
development include defining the GA and evaluation functions, calculating fracture
porosity, training and generating new population as well as printing and plotting
the results of the models. For main GA functions, the Multiple Linear Regression
(MLR) and Root Mean Square Error (RMSE) formulas were used. The best model
was evaluated based on the least total prediction error, cost and execution time. The
fracture porosity was first calculated by a conventional method and further used to
train the GA models, among which the GA model consisting of 1080-training data
with 100 population showed the best performance.
