A Polycyclic Side-Chain Filtration Reducer for Water-Based Drilling Fluids: Experimental Evaluation and Artificial Neural Network–Based Apparent Viscosity Prediction

Summary Fluid loss control under harsh conditions remains a major challenge in the field of drilling fluid. In this work, a polycyclic side-chain polymer (ASAA) was synthesized and characterized by Fourier transform infrared spectra (FTIR) and thermogravimetric analysis (TGA), which was evaluated as filtration control additive for water-based drilling fluids (WDFs) under high-temperature and high-salinity conditions. The rheological and filtration performance of bentonite-based mud (BM) with additional ASAA were explored over a range of temperatures and sodium chloride (NaCl) concentrations, and the dynamical stability was examined via the multilight scattering method. Mechanism studies involving FTIR, adsorption capacity, zeta potential, particle-size distribution (PSD), and scanning electron microscopy (SEM) of ASAA-treated BM (ASAA-BM) were systematically investigated in detail. The result suggested that the additional ASAA significantly improved the tolerance of BM, with a filtration loss of 6.2 mL under saturated salt contamination after aging at 200°C. It was revealed that ASAA strongly adsorbed on bentonite even under high-temperature and saturated salt conditions, resulting in the enhanced particle dispersion and the formation of dense, low-permeability filter cakes. In addition, the rheological behavior of WDFs was influenced by additives; thus, an artificial neural network (ANN) model was developed to predict the apparent viscosity of drilling fluids with varying additive and salt contents under different aging temperatures. Trained with fivefold cross-validation, the model achieved high accuracy of the test set (Pearson correlation coefficient R = 0.9884) and exhibited strong agreement between predicted and experimental values. These findings indicated that ASAA was a promising additive for WDFs in harsh drilling environments, and the ANN model provided a credible tool for predicting and optimizing WDF performance, demonstrating its potential as a practical “mud-laboratory assistant” for drilling fluid design and optimization.