A new intelligent optimization method for drilling parameters of extended reach wells based on mechanical specific energy and machine learning

Extended reach wells (ERWs) can efficiently develop offshore satellite oilfields, reduce development costs and improve economic benefits. However, owing to the complex geological conditions, it is difficult to determine the drilling parameters of extended reach drilling, which greatly restricts the rate of penetration (ROP) and increases the cost of drilling operations. In this paper, a new intelligent optimization method for drilling parameters of ERWs based on mechanical specific energy (MSE) and machine learning is proposed. Unlike conventional approaches, this method combines an ensemble regression (ER) model for predicting ROP with the non-dominated sorting genetic algorithm-II (NSGA-II) to optimize multiple objectives, including MSE, ROP, and unit footage cost (UFC). The results show that through the intelligent optimization of drilling parameters for extended reach drilling wells in Block M of Bohai Oilfield, the two decision variables of the weight on bit (WOB) and rotations per minute (RPM) are increased, MSE is constantly converging or even equal to the confined compressive strength (CCS) of the rock, UFC is reduced by nearly 51.57%, and ROP is increased by approximately 31.88%. The findings demonstrate the effectiveness of the approach in enhancing drilling efficiency and reducing operational costs, offering an innovative solution for the optimization of drilling parameters in ERWs.