Prediction and Optimization of ROP Based on Geological and Engineering Parameters Fusion Model

The main lithology of the Shahejie Formation reservoir is mudstone and dark sandy mudstone, and the rate of penetration (ROP) is generally low when drilling through this formation, significantly impacting the drilling cycle and costs. To address this issue, a geologically and engineering-integrated ROP prediction and optimization model is proposed. This model consists of two parts: ROP prediction and ROP optimization. The prediction model leverages geological and engineering data to establish an auxiliary same-well ROP prediction model based on adjacent well data. After completing the ROP prediction, a feature contribution coefficient is defined to quantify the influence of various features on the final result. This coefficient allows for both an interpretation of the predicted results and the identification of key controllable parameters that significantly impact ROP. For these key controllable parameters, the ROP optimization model uses a grid search optimization algorithm to identify the optimal parameter combination, thereby improving ROP. Results from applying the prediction and optimization model show an average ROP increase of 6.34% in test wells, with the three parameters contributing most to the prediction results being gamma values, weight on bit, and bit drilling time. This ROP prediction and optimization model effectively integrates geological and engineering factors, achieving high-accuracy ROP predictions and substantial ROP improvements, providing valuable guidance for ROP enhancement in two development wells where it has been applied.