Effect of particle size distribution on the class G oil well cement properties: Experimental measurement and intelligent modelling

Adequate thickening time is essential for effective cement placement in oil wells, facilitating distribution and pumpability. Sufficient compressive strength is crucial to maintain the cement sheath's integrity, preventing fluid migration and ensuring long-term stability in the wellbore. This study conducted experiments on the particle size of HSR Kerman class G oil well cement using an HPHT consistometer and laser diffraction analyzer. Furthermore, the influences of the fineness of oil well cement particles on thickening time, compressive strength at two different temperatures (38 and 60 degrees C), and free fluid were investigated. Afterward, two robust machine learning models were built to predict cement fineness's effect on the parameters mentioned. Multilayer Perceptron (MLP) and Radian Basis Function (RBF) neural networks were used for modeling and these models were integrated into a single model namely committee machine intelligence system (CMIS). Based on the findings, it appears that slurries containing smaller particle sizes tend to thicken at a quicker rate. In contrast, as the Blaine (specific surface area of cement particles) decreases, the thickening time of cement increases. Additionally, reducing the specific surface area of cement particles leads to an increase in compressive strength. However, based on laser diffraction particle size analyzer apparatus outcomes, cement's compressive strength increases as cement particles shrink. Moreover, laboratory results showed that by growing the specific surface area of cement particles, the amount of free fluid in cement was reduced, but the laser diffraction analyzer depicted that the smaller particles of the cement had less amount of free fluid. In addition, the proposed CMIS models for thickening time, free fluid, and compressive strength at 38 degrees C and 60 degrees C indicate acceptable AAPRE values of 0.403%, 0.51%, 0.022%, and 0.075%, respectively. Finally, sensitivity analysis revealed that specific surface area had a positive effect on compressive strength, and contradictory had a negative impact on thickening time and free fluid.