Evaluation of drawdown curves derived from multiple well aquifer tests in heterogeneous environments

Abstract

Aquifer coefficients derived from nonsteady-state, multiple well, aquifer tests in laterally heterogeneous environments often have uncertain meaning. Drawdown at observation wells reflects the removal of water from storage in the aquifer and transient refraction of ground water pathlines during the evolution of a non-symmetrical cone of depression. These effects are masked within observation well drawdown data such that “good” Theis (1935) type curve matches often result. Transmissivity and storativity values derived from independent drawdown curves plotted as drawdown versus time (t) or drawdown versus time/distance² (t/r²) usually differ from observation well to observation well. These aquifer coefficients often are considered to represent some type of average of the materials between and/or about the pumping well and the observation wells. Simulations of two multiple well aquifer tests with simple, arbitrary distributions of block heterogeneities suggest that transmissivity (T) and storativity values derived from independent drawdown curves by the Theis (1935) method generally increase with distance from the pumping well. This apparent scale effect is related to the force-fitting of earlytime drawdown data to the steep portion of the Theis type curve without sufficient late-time drawdown data to constrain vertical shifting of the drawdown data relative to the type curve.

Log-log plots of drawdown versus t/r² for multiple well aquifer tests form families of curves that are characteristic of the distribution of observation wells and the degree of heterogeneity within the cone of depression. Separation between discrete drawdown curves within a family provides a qualitative measure of the degree of heterogeneity within the cone of depression. All of the drawdown curves within a family converge on a single curve at large values of t/r². A composite analysis of all of the drawdown data within the family yields an estimate of the average T within the cone of depression. Analysis of discrete drawdown curves as integral members of the family of curves provides a means to constrain type curve matches and minimizes force-fitting if drawdown data are defined for large values of t/r² for at least one well. The constrained type curve matches provide more reasonable estimates for T near individual observation wells than analysis of drawdown curves independently.