Application of Nanomaterials and Machine Learning to Evaluate Compressive Strength of Oil Well Cement Based on Experimental Data

Abstract

Well cementing in the oil and gas industries plays a fundamental role in long-term energy production. Maintaining wellbore integrity for a longer duration is a key challenge facing the energy sector. The use of nanotechnology and machine learning (ML) in enhancing and evaluating well integrity is gaining attention. This paper presents compressive strength (CS) prediction using a hybrid random forest linear model (HRFLM) and nanomaterial supplement for durable well integrity, deploying experimental data. The experimental analyses of uniaxial compressive strength (UCS), high-temperature, high-pressure (HTHP) test, and scanning electronic microscopy (SEM) were conducted. Experimental findings revealed that the addition of MWCNTs significantly enhanced the CS. The average UCS for 0.2% MWCNT reinforcement after curing for 28 days was higher at 62.79 MPa compared to 58.16 MPa for ordinary Portland cement (OPC) as a result of the nucleation reaction of the MWCNT microstructure and compatibility with the C−S−H bond. SEM shows the needle-like structure, which confirmed the presence of MWCNT and revealed that multi-wall CNTs can enhance the CS of cement. ML findings affirmed that HRFLM statistically outperformed Xgboost, RF, and support vector machine (SVM) in predicting CS, giving an R-square (R2) of 0.967, root mean square error (RMSE) of 0.469, and a mean absolute error (MSE) of 0.242, followed by Xgboost ranked secondly with an R2 of 0.958, RMSE of 1.287, and MSE of 1.674; RF with an R2 of 0.936, RMSE of 1.613, and MSE of 2.581; and SVM with an R2 of 0.889, RMSE of 2.109, and MSE of 4.043, respectively, during training. The comparative findings from experimental analysis and ML disclose the potential of HRFLM in offering accurate and high computational efficiency in terms of CS prediction, which can be established as an effective methodology for the evaluation of compressive strength in the oil and gas industry.