Comparative study of the effect of additives in water-based drilling mud on filtrate volume, along with presentation of an artificial neural network estimator model

Fluid loss during drilling operations is one of the primary causes of formation damage and reduced well productivity. Therefore, minimizing drilling mud filtrate volume is essential for improving drilling efficiency and maintaining wellbore integrity. In this study, the effects of three additives, namely rice husk, natural gum (apricot gum), and xanthan gum, on the filtrate volume of water-based drilling mud were experimentally investigated. Static filtration tests were conducted using additive concentrations of ۰.۵, ۱, ۲, and ۳ gr and filtration times ranging from ۰.۲۵ to ۷.۵ min. The results indicated that increasing the additive concentration consistently reduced filtrate volume for all tested additives. For all additives, adding ۰.۵ g caused the largest incremental decrease in filtrate volume, while further increases in concentration produced diminishing reductions. Xanthan gum exhibited the highest filtration-control efficiency, followed by apricot gum and rice husk. At the maximum additive concentration, filtrate volume was reduced by ۶۰.۰%, ۸۴.۰%, and ۸۸.۴% for rice husk, apricot gum, and xanthan gum, respectively, compared with the base mud. To estimate filtrate volume under arbitrary additive concentrations and filtration times, a two-layer feedforward artificial neural network (ANN) with ۱۰ neurons in the hidden layer was developed. The ANN demonstrated great predictive capability, achieving overall correlation coefficients (R) of ۰.۹۹۹۷۹, ۰.۹۹۹۸۸, and ۰.۹۹۹۹۱ for rice husk, apricot gum, and xanthan gum datasets, respectively. The corresponding minimum mean squared errors (MSEs) were ۱.۵۲ × ۱۰⁻⁵, ۱.۰۸ × ۱۰⁻⁵, and ۸.۵۶ × ۱۰⁻⁶, while the average prediction errors were ۰.۵۹%, ۱.۲۷%, and ۱.۸۶%, respectively. These results confirm the effectiveness of the investigated additives for fluid-loss control and demonstrate the reliability of the proposed ANN model for rapid filtrate-volume prediction.

کلیدواژه ها:

Rice Husk ، Natural/Apricot gum ، Xanthan gum ، Drilling mud filtration ، Feedforward artificial neural network