Rate of penetration prediction in drilling wells from the Hassi Messaoud oil field (SE Algeria): Use of artificial intelligence techniques and environmental implications

Algeria, an oil-rich country, also hosts one of the largest groundwater systems in the world. To protect water resources from pollution in oil-drilling locations, the monitoring of drilling parameters is extremely important, as it represents a powerful key to detect mud losses. The present study aimed to estimate ROP, one of the parameters that provide information on the losses of mud in offset wells from the Hassi Messaoud oilfield using two artificial intelligence methods (ANN and SVM) based on real-time surface drilling parameters. The results indicated that ANN was the most accurate model with the best performance parameters (R2 = 0.9456, RMSE = 0.1101, and MAE = 0.0223) and confirmed the robustness of the AI models in predicting ROP. The developed model enables good control of the ROP in drilling new wells that can be used to predict mud losses and reduce the risk of groundwater contamination.