In May 1980, the University of Minnesota began a project to evaluate the feasibility of storing heated (150 degrees Celsius (°C) water in the deep (180 to 240 meters (m)) Franconia-Ironton-Galesville aquifer and later recovering it for space heating. The Aquifer Thermal-Energy Storage (ATES) system was a doublet-well design in which the injection and withdrawal wells were spaced approximately 250 m apart. High-temperature water from the university's steam-generation facilities supplied heat for injection. Water was pumped from one of the wells through a heat exchanger, where heat was added or removed. Water then was injected back into the aquifer through the other well. The experimental plan for testing the ATES system consisted of a series of short-term hot-water injection, storage, and withdrawal cycles. Each cycle was 24 days long, and each injection, storage, and withdrawal step of the cycle was 8 days.

The Franconia-Ironton-Galesville aquifer is a consolidated sandstone, approximately 60 m thick, the top of which is approximately 180 m below the land surface. It is confined above by the St. Lawrence Formation a dolomitic sandstone 8 m thick and below by the Eau Claire Formation a shale 30 m thick. Initial hydraulic testing with inflatable packers indicated that the aquifer has four hydraulic zones with distinctly different values of relative horizontal hydraulic conductivity. The thickness of each zone was determined by correlating data from geophysical logs, core samples, and the inflatable-packer tests.

A comprehensive network for data collection, storage, and analysis was designed to monitor temperature and pressure changes during the ATES test cycles. A total of 22 pressure transducers and 56 thermocouples monitored pressures and temperatures in the aquifer and in the upper and lower confining units.

Temperature and pressure measurements were collected in observation well nests at distances of approximately 7 and 14 m from the production wells. All pressure and temperature data were transmitted through buried cables to a central data logger, where the measurements were viewed independently or stored on computer magnetic tape for later analysis. Interactive computer programs were available to display data stored on magnetic tapes as individual measurements or as plots of pressure and temperature versus time.

Analyses of step-drawdown and constant-discharge aquifer tests indicate that the Franconia-Ironton-Galesville aquifer is anisotropic in the horizontal plane. Major and minor transmissivities are 101.5 and 44.6 m²/d (square meters per day), respectively, for the Ironton and Galesville Sandstones and 40.0 and 24.0 m²/d, respectively, for the upper part of the Franconia Formation. The average transmissivity of the entire Franconia-Ironton-Galesville aquifer is about 98 m²/d. Effective porosity ranges from 0.25 to 0.31, and the average storage coefficient is 4.5x10^-5.

Two computer models were constructed to simulate the movement of ground water and heat. The first was a nonisothermal, isotropic, single-phase, radial, ground-water flow and thermal-energy-transport model that was constructed to examine the sensitivity of model results to various hydraulic and thermal properties. The model also was used to study the potential for buoyancy flow within the aquifer and the effect of various cyclic injection and withdrawal schemes on the relative thermal efficiency of the aquifer. The second model was a threedimensional ground-water flow and thermal-energy-transport model that was constructed to incorporate the anisotropy of the aquifer.

In the first model, the sensitivity analysis assumed 8 days of injection of 150°C water at 18.9 liters per second (L/s), 8 days of storage, and 8 days of withdrawal of hot water at 18.9 L/s. The analysis indicates that, for practical ranges of hydraulic and thermal properties, the ratio of horizontal to vertical hydraulic conductivity is the least important property and thermal dispersivity is the most important property used to compute temperature and aquifer thermal efficiency.

Buoyancy flow was examined for several values of hydraulic conductivity and ratios of horizontal to vertical hydraulic conductivities. For the assumed base values of hydraulic and thermal properties, buoyancy flow was negligible. The greatest simulated buoyancy flow resulted from simulations in which horizontal hydraulic conductivity was increased to ten times the base value and in which the vertical hydraulic conductivity was set equal to the horizontal hydraulic conductivity.

The effects of various injection and withdrawal rates and durations on computed values of aquifer relative-thermal efficiency and final well-bore temperature were studied for five 1-year hypothetical test cycles of injection and withdrawal. The least efficient scheme was 8 months injection of 150°C water at 18.9 L/s and 4 months of withdrawal of hot water at 18.9 L/s. The most efficient scheme was obtained with 6 months of injection of 150°C water at 18.9 L/s and 6 months of withdrawal of hot water at 37.8 L/s. The hypothetical simulations indicate that the calibrated model of the doublet-well system would be a valuable tool for use by the university in selecting a highly efficient system operation.

In the second model, analytical solutions of anisotropic hydraulic flow around the doublet-well system were obtained to provide fluid-flux boundary conditions around the heat-injection well in three dimensions. This information simplified simulation of the doublet-well system because only the heat injection well needed to be simulated.

This second model was calibrated with data from an 8-day ambienttemperature injection test at 18.9 L/s. Boundary-flux conditions were examined for nonisothermal conditions by simulating 8 days of injection of 150°C water at 18.9 L/s.

Results of simulations using both models indicate that the fluxboundary conditions are adequate for simulations of short-term heatinjection testing.