Flow and Geometry Control the Onset of Jamming in Fractures with High Solid-Fraction Fluids

Fluids containing a large fraction of suspended solids are common in the subsurface. Examples include fluids used for environmental remediation, hydraulic fracturing fluids and magma. These fluid-solid mixtures behave as non-Newtonian fluids where interactions between fluid, suspended solids, and pore walls can lead to jamming of the suspended solids. Jamming causes the velocity of the solid to decrease locally to zero causing a rapid decrease in permeability as the fluid is forced to flow through the pore space within the immobilized solid. Here we present results from experiments that quantify the flow of non-Newtonian suspensions in an analog parallel-plate fracture (transparent 15cm x 15cm with ~3-mm aperture) and explore the dependence of jamming on flow conditions, fracture geometry, and the action of gravity. We used guar gum mixed with water (0.75%) as the fluid and added 50% by volume of crushed silica (< 300μm). Flow rates ranged from 0.2ml/min to 6.0ml/min, cell orientation varied from horizontal to vertical (bottom to top) flow and a transducer provided continuous measurement of differential pressure across the cell. A strobed LED panel backlit the cell and a high-resolution CCD camera captured frequent (0.2 Hz) images during all experiments. Particle image velocimetry (PIV) yielded measurements of the evolving velocity field during experiments (see Figure). In the vertical orientation during the initial period of high flow rate, outflow decreased rapidly and the differential pressure increased indicating jamming within the cell. Subsequent efforts to flush solids from the cell suggested that jamming occurred at the inlet of the cell. This was likely due to settling of solids within the flow field indicating that the time scale associated with settling was shorter than the time scale of advection through the cell. In the horizontal orientation, localized jamming occurred at the lowest flow rate in a region near the outlet. This suggests that when settling and advection are not competing, localized regions of low velocity within the cell trigger jamming. Subsequent increases in flow rate did not remobilize the jammed region demonstrating that the jamming process is at least hysteretic, if not irreversible. Light absorbance image from a sub-region adjacent to the outlet of the fracture for a flow rate of 0.2 ml/min. Flow was from right to left. The PIV-derived streamlines show divergence of the flow field around the jammed region adjacent to the outlet. Velocity profiles from the inlet and outlet of this region clearly show that the velocity field is near-uniform at the inlet but that jamming leads to a region along the outlet with zero velocity.