Pump testing in heterogeneous aquifers

Abstract

A finite-difference model of an infinite horizontal non-leaky aquifer was used to simulate a constant-rate pump test in a heterogeneous aquifer; the values of transmissivity specified for the nodes of the finite-difference grid were chosen at random from a uniform distribution over the interval 0–0.1 m 2/min and the ratio transmissivity/storativity was kept constant. The spacing of the finite-difference grid was 10 m in the vicinity of the pumped well at the center of the grid and became progressively wider towards the boundaries of the square area. The boundaries were located at a distance of 23 km from the well. Simulated time-drawdown curves for 624 nodes within 1,000 m of the well were analyzed using the Theis equation for drawdown in an infinite homogeneous aquifer and a least-squares technique to obtain the best values of transmissivity and storativity. Calculated transmissivities and storativities were almost normally distributed in a narrow range about the average of the values assigned to the nodes. Statistically, the pump tests yielded increasingly consistent results both as the observation node was removed farther and farther from the pumped well and as the points on the time-drawdown curve selected for the analysis represented later and later times. It is concluded that, at least for the type of hypothetical transmissivity distribution utilized, drawdown curves will not clearly reveal the inhomogeneous nature of the aquifer. Thus standard analytical aquifer-test techniques can be safely applied in such circumstances.