CFD-DEM modeling of fluid-driven fracture induced by temperature-dependent polymer injection

This study numerically investigates fracture initiation and propagation during polymer-based solution injection under varying thermal conditions. A coupled computational fluid dynamics and discrete element method (CFD-DEM) framework is used to model non-Newtonian fluid flow through a granular medium. The rheology of shear-thinning fluids and fluid-particle heat transfer are modeled with temperature-dependent power-law parameters. The current model is validated by comparing fracture propagation behavior and peak pressures against the similar numerical study. The adequacy of the fluid-particle heat transfer model is confirmed by comparing the results with an analytical approach. The simulation results show that polymer concentration significantly influences fracturing behavior. Less concentrated, lower-viscosity fluids are more likely to create linear fracture paths with enhanced fluid infiltration. In contrast, fluids with higher polymer concentrations and viscosities tend to produce wider fractures characterized by greater particle displacement. An increase in the fluid temperature injected into the cooler medium leads to a reduction of fracture size for the 0.4 % (w/w) XG solution, while the 0.6 % (w/w) XG solution tends to form more linear fracture tips. At sufficiently elevated medium temperatures, the injection of cooler fluids prevents fracture initiation for both concentrations. Lower-viscosity cases, dominated by infiltration, reflect broader thermal transitions in particle temperature distribution, whereas higher-viscosity cases, characterized by particle displacement, exhibit narrower transition regions along fracture boundaries. A fracture initiation criterion for shear-thinning fluids is proposed based on the dimensionless parameters Π₁ and τ₂. Fracture occurs when Π₁ > 73 and τ₂ > 3.58 × 10⁻⁹. The 0.4 % solution exhibits lower thermal sensitivity with relatively minimal variations in the dimensionless parameters, while the 0.6 % solution shows a greater response to temperature changes, reflected in broader variations of these parameters.