Prediction of Stress Overshoot in Thixotropic Xanthan Gum 4 wt.%/Hydroxyethyl Cellulose 4wt.% Blends: Integrating Rheological Characterization and Restart-Flow Modeling

Restarting the flow of yield-stress and thixotropic fluids remains a critical challenge in deepwater operations, where gel-like microstructures develop during static periods. This study investigates the solid–liquid transition of aqueous biopolymer systems formulated with 4 wt.% xanthan gum, 4 wt.% hydroxyethyl cellulose, and a 1:3 blend in brine (286 g/L). Steady-state flow curves and restart-flow tests were conducted at shear rates between 0.001 and 10 s⁻¹ following controlled aging times of 60, 600, and 1,800 s. The XG solution exhibits elastoviscoplastic behavior due to the presence of yield stress ( τ y =20.7±0.55[Pa]), whereas the HEC solution displayed viscoelastic characteristics. The blend also exhibited viscoelastic behavior, on the restart flow the blend is characterized by a well-defined stress overshoot at a critical strain of 1.82. The overshoot magnitude increased with shear rate, showing no dependence on aging time. Longer rest periods decreased the pre-yield elastic shear modulus without affecting the maximum overshoot stress. The experimental dataset was used to train artificial neural networks to predict the complete stress response and overshoot stress as functions of shear rate and aging time. The optimized model achieved R² ≥ 0.99 with prediction errors below 1.3 [Pa] across all conditions. Statistical analysis of the mean absolute error indicated increased structural heterogeneity at longer aging times. The proposed framework provides a reliable predictive tool for restart-flow behavior in thixotropic yield-stress fluids relevant to pre-salt drilling operations.