Validation of non-Darcian flow effects in slug tests conducted in fractured rock boreholes

SummaryA series of rising and falling head slug tests with different initial applied head differentials (ΔHo) were conducted in open fractured dolostone and sandstone boreholes using straddle packers isolating specific depth intervals (1.5 m length) to examine the influence of non-Darcian flow. The open holes were developed and inspected using video and acoustic televiewing (ATV) to ensure that evidence of skin effects due to drilling were absent. The transmissivity (T) values obtained from both the rising and falling head slug tests were very similar at low initial applied head; however, the T values were progressively smaller at larger ΔHo, suggesting error due to non-Darcian flow. Non-Darcian flow behavior was confirmed by constant head step tests conducted in the same test intervals where the injection rate (Q) vs. applied head (dH) relationship became non-linear at relatively low injection rates, and the non-Darcian data also resulted in lower T values. For a series of slug tests conducted at different ΔHo, non-Darcian flow effects gradually increased as ΔHo increased, consistent with the trends for constant head step tests conducted in the same test intervals. To maintain Darcian flow conditions in the fractured dolostone and sandstone tested in this study, ΔHo must be kept small, generally less than 0.2 m. This study demonstrates that by conducting both “stepped” slug tests and constant head step tests, the Darcian flow assumption for both types of tests can be rigorously validated. However, when only slug tests are conducted, it is necessary to conduct a series of “stepped” slug tests, including tests with small applied head differentials, to avoid errors due to non-Darcian flow.

► Agreement between more than one type of hydraulic test increases the confidence in T.
► Constant head step tests can be used to validate the Darcian flow assumption for slug tests.
► A method is proposed that validates the two main assumptions in the Hvorslev model.