Efficient Method for Calculating Hydraulic Conductivity from Pneumatic Slug Tests

Abstract:

Pneumatic slug tests are widely used in characterizing the hydraulic conductivity of aquifers. In comparison to a manual slug test wherein the water level is measured using a water level tape, pneumatic slug tests are especially useful when the water level recovery is very fast (<10 sec) in a high hydraulic conductivity (K) aquifer (>10^-4 m/s) and when the recovery is very slow (<10^-7 m/s). The submerged pressure transducer monitors pressure changes at intervals of fractions of a second and for longer recoveries no personnel are required to make repeated measurements. A pneumatic slug test begins with pressurizing the well at the well head using an air pump followed by several minutes waiting for the pressure in the well to equalize with the pressure outside the well screen. In semi-confined settings this equalization may take >5 minutes. In lower K media it’s not always feasible to wait until the well fully recovers before making the next replicate measurement. Therefore, it would greatly reduce the time needed to make replicate measurements if these waiting times could be reduced. Here we present a method using non-linear least squares regression on a portion of the recovery curve to simultaneously fit 3 parameters used to determine K from a slug tests using the Hvorslev method. The advantage of this approach is that waiting for the well to reach static head between replicate measurements is not required. This is because the regression fits static head (H) from the shape of only part the recovery curve. We compare the resulting K values from this new method to values obtained from manually measured static heads for triplicate measurements on 50 wells. The well’s settings ranged from unconfined to semi-confined and K ranged from 10^-3 to 10^-5 m/s. The new method gave identical results. We performed the same comparison on a subset 16 wells using data collected in half the time, where only part of the recovery curves were measured before starting the next replicate. This also resulted in near identical K values. Sensitivity analyses indicated that only about 30% of the recovery curve needs to be measured to accurately determine K. This method saves time in investigations where large numbers of monitoring wells are involved and/or where wells are placed in low K media.