Optimization of Cementing Displacement Efficiency Based on Machine Learning Ensemble Model

Abstract

Increasing displacement efficiency during cementing has been identified as a crucial measure for enhancing cementing quality. The correctness of evaluating displacement efficiency using the Fluent simulation method was verified based on field logging data. On this basis, Fluent numerical simulation results were utilized as the dataset Prediction models for displacement efficiency were developed based on machine learning techniques, employing Random Forest, Extreme Decision Tree, Artificial Neural Network, and Support Vector Machine algorithms to explore the prediction accuracy of each model. Furthermore, the Stacking method was applied to combine models with high prediction accuracy, and the prediction accuracy of the ensemble model, along with the influence weights of various cementing parameters on displacement efficiency, was analyzed. Additionally, cementing parameters were optimized based on the Stacking prediction model using the L-BFGS, simulated annealing, and gradient descent algorithms to achieve optimal displacement efficiency. High computational accuracy was demonstrated by the Random Forest and Extreme Decision Tree algorithms, and further improvement in prediction accuracy was achieved through the ensemble model developed from these two algorithms. The L-BFGS algorithm was found to perform excellently in predicting displacement efficiency in both wide and narrow annular gaps, resulting in improvements of 3.74% and 5.46%, respectively, compared to the original method. Furthermore, the deviation between the predicted results and the actual simulated values was controlled within 3.5%, indicating a high degree of accuracy and reliability. This research provides a novel theoretical approach for the optimization of cementing parameters and the prediction of displacement efficiency.

Keywords: Cementing, Machine learning, numerical simulation, Displacement efficiency, ensemble model