Tests of subsurface storage of freshwater at Hialeah, Dade County, Florida, and numerical simulation of the salinity of recovered water

Abstract

Injection and observation wells were drilled in late 1974 for the purpose of conducting tests of storage and recovery of potable water in the brackish Upper Floridan aquifer. Three tests, involving storage and recovery cycles of varying volumes and storage period lengths, were performed between July 1975 and January 1980. Recovery was by natural artesian flow, and recovery efficiencies were 32.9, 47.8, and 38.5 percent. Wellbore plugging occurred during the injection stages, but injectivity was restored by periodic 2- to 3-hour backflushes at the natural artesian flow rate. An interval of shelly limestone between 1,015 and 1,050 feet below land surface contained the flow zone. Data from an analysis of 18 spinner flowmeter logs indicated that the principal part of the flow zone extended from 1,024 to 1,036 feet below land surface and that minor amounts of flow occurred to a depth of about 1,047 feet. A neutron porosity log indicated the bulk porosity of both the flow zone and confining layers to be 35 percent. Chloride and dissolved-solids concentrations of water in the flow zone were 1,200 and 2,700 milligrams per liter, respectively. A three-dimensional, finite-difference flow and solute-transport code was used to simulate pressure data measured during an aquifer test and observed salinity increases in recovered water during storage and recovery cycles. The aquifer test conducted in February 1975 was simulated by using a hydraulic conductivity estimate of 800 feet per day and a rock compressibility estimate of 0.0000400 (pound per square inch)-1. The equivalent transmissivity and storage coefficients were 9,600 cubic feet per day per square foot times foot of aquifer thickness and 7.8×10-5, respectively. Simulation of observed salinity increases during the three recoveries required dispersivities of 65 feet, a molecular diffusivity of 0.0002 foot squared per day, and a regional pore velocity of 260 feet per year. Central differencing in space and time was used for the solute-transport computations as well as an experimental method of computing vertical dispersion that used a scaling factor of 0.013. Additional simulations of the aquifer-test data and recovery salinities were obtained based on assumptions that (1) the flow zone was 21 feet thick, (2) flow-zone effective porosity was 20 percent, and (3) flow-zone hydraulic conductivity was bipolar anisotropic by a ratio of 10:1. The four sets of simulation values were used in model runs in which 10 years of annual injection, storage, and recovery cycles were simulated. Computed recovery efficiencies increased from 40 percent in the first year to 68 percent in later cycles. The high regional pore velocity required for model calibration substantially limited the recovery efficiency achieved in later cycles.