@article{d8674dc0e6b54bdd90cb90a82fc89969,
title = "Optimization of wellbore cement sheath resilience using nano and microscale reinforcement: A statistical approach using Design of Experiments",
abstract = "In this study, silica (SiO2) nanoparticles were deposited on the surfaces of micro-synthetic polypropylene (PP) fibers through the sol-gel process and combined with Alumina Nanofibers (ANF's) to enhance the cement composite mechanical properties. Cement samples were cured at 82.2 °C with 20.68 MPa for 24 h to emulate wellbore conditions. Mechanical properties were experimentally tested and modeled through the design of experiments (DOE). Treated PP fibers did not contribute to the ultimate strength mainly due to their inability to resist stresses before cracks appear. Ultimate modulus of elasticity (MOE) and Poisson's Ratio were synergistically improved on the nano and microscale levels. This is due to cement samples remaining in the elastic region and the flexibilities of the fibers. 0.15% ANF and 0.09% PP fibers by weight of cement (BWOC) were considered the optimum values after performing the multi-objective optimization analysis. Derived models were statistically significant and useful for predictions.",
author = "Phillip McElroy and Seyedhossein Emadibaladehi and Marshall Watson",
note = "Funding Information: Fig. 18 displays the normal plot of residuals. Multiple regression assumes the residual values (difference between observed and predicted values) are normally distributed. The data points are closely aligned along the inclined straight line (representing normal distribution), which indicates the experimental data is normally distributed (Wang et al., 2015). Fig. 19 displays the model predicted (inclined straight line) vs. actual values (data points), which displays adequate model precision. Fig. 20 displays the standardized residuals vs. the model predicted line. The experimental data was uniformly distributed and did not follow a specific pattern, thereby exhibiting homoscedasticity (constant variance). This validates the significance of the model and indicates there is no systematic error within the system (Wang et al., 2016). Fig. 21 displays the perturbation plot for the developed response model. The perturbation plot helps to compare the effects of both factors at a particular point in the design space; with “A” representing ANF and “B” representing PP fibers. The reference point is set at the midpoint (coded “0”) and the response is plotted by changing one factor over its range while holding the other factor constant. A curvature or steep slope in a factor shows that the response is sensitive to that factor as a relatively flat line shows insensitivity (Khed et al., 2020). The results support the previous observations, which indicate ANF has a larger influence on the response than PP fibers. It is worth noting that after the optimum value is reached for each fiber, there is a decrease in the mechanical properties as the fiber dosage is increased. This is due to nanofiber clustering, which decreases cement composite mechanical properties (Noordin and Liew, 2010). Essentially, the results of the ANOVA and RSM model analysis have been statistically validated to suggest that the developed formulations have high predictive accuracy. The compressive, tensile and MOE models also possess similar results. Publisher Copyright: {\textcopyright} 2020",
year = "2021",
month = may,
language = "English",
journal = "Journal of Petroleum Science and Engineering",
publisher = "Journal of Petroleum Science and Engineering",
}