LATTICE BOLTZMANN SIMULATIONS COMPARED TO NUCLEAR MAGNETIC RESONANCE MEASUREMENTS OF MACROPORE VELOCITIES IN AN EPOXY-RESIN MODEL OF A STRATIFORM ICHNOGENIC GROUNDWATER FLOW ZONE IN THE KARST BISCAYNE AQUIFER

Whether flow in stratiform ichnogenic groundwater flow zones in the karst-carbonate Biscayne Aquifer of southern Florida is Darcian is a topic of debate. Measurement of flow velocities with a ‘stimulated-echo', phase-difference nuclear magnetic resonance imaging (NMRI) technique and simulations with Lattice Boltzmann (LB) methods are compared and investigated as a means of characterizing flow. The flow zone rocks are characterized by cm-scale, well-connected maropores related to an Ophiomorpha ichnofabric, macroporosities as high as 64%, and permeabilities orders of magnitude above the normal range of laboratory-derived measurements of carbonate rock samples. Non-Darcian flow in these zones is likely in hydraulically-stressed areas, e.g., near wellfields.

A 0.1-m diameter by 0.25-m long epoxy-resin model that mimics a cylindrical drilled core was created from thresholded computed tomographic (CT) images using 3-D printing techniques. CT resolution is 0.8 mm/pixel along the longer z-axis and 0.271 mm along the other axes. Only macropore flow was analyzed in the epoxy and LB models representative of this dual-porosity (matrix and macroporous touching-vug porosity) aquifer. The NMRI resolution is 1.18 mm/pixel along the z-axis and 3.1 mm/pixel along x and y. The region of distortion-free NMRI data is 0.09 m along the z axis of the epoxy model. NMRI velocity measurements made at fluxes of 0.00013 and 0.00025 m/s suggest that flow at these fluxes is laminar.

A LB model was used to simulate flow in the core rendering. Reynolds numbers matching the experimental conditions were used to ensure hydrodynamic similarity of the LB simulation and NMRI results. LB simulates a fluid more compressible than water; results are presented as mass flux vectors to eliminate the compressibility effect on velocity magnitudes.

Comparison of velocity magnitudes obtained from NMRI measurements and LB simulations indicates reasonable agreement, though these preliminary NMRI results are much coarser than the LB simulations. The lower resolution of the NMRI data may also result in a less-than-satisfactory agreement of detailed velocity vector fields. NMRI and LB approaches should ultimately agree and represent true field velocities and thereby contribute to improved understanding of flow in macroporous karst.