In-situ falling-head test for hydraulic conductivity: Evaluation in layered sediments of an analysis derived for homogenous sediments

• Published in: Journal of hydrology (Amsterdam) 2016-08, Vol.539, p.319-329
• Main Authors: Burnette, Matthew C., Genereux, David P., Birgand, François
• Format: Article
• Language: English
• Subjects: Computer simulation; Fluid flow; Freshwater; Groundwater–surface water interaction; Hydraulic conductivity; Hydraulics; Mathematical models; Numerical modeling; Permeameter; Permeameters; Sand; Sediments; Streambed; Tanks
• ISSN: 0022-1694; 1879-2707
• DOI: 10.1016/j.jhydrol.2016.05.030

•Hvorslev falling-head analysis may yield accurate estimates of K in layered sediments.•Error in K was large only when the permeameter base was at the top of a low K layer.•Numerical simulations and lab results agree, and can help interpret field results. The hydraulic conductivity (K) of streambeds is a critical variable controlling interaction of groundwater and surface water. The Hvorslev analysis for estimating K from falling-head test data has been widely used since the 1950s, but its performance in layered sandy sediments common in streams and lakes has not previously been examined. Our numerical simulations and laboratory experiments show that the Hvorslev analysis yields accurate K values in both homogenous sediment (for which the analysis was originally derived) and layered deposits with low-K sand over high-K sand. K from the Hvorslev analysis deviated significantly from true K only when two conditions were present together: (1) high-K sand was present over low-K sand, and (2) the bottom of the permeameter in which K was measured was at or very near the interface between high-K and low-K. When this combination of conditions exists, simulation and laboratory sand tank results show that in-situ Hvorslev K underestimates the true K of the sediment within a permeameter, because the falling-head test is affected by low-K sediment outside of (below the bottom of) the permeameter. In simulation results, the maximum underestimation (occurring when the bottom of the permeameter was at the interface of high K over low K) was by a factor of 0.91, 0.59, and 0.12 when the high-K to low-K ratio was 2, 10, and 100, respectively. In laboratory sand tank experiments, the underestimation was by a factor of about 0.83 when the high-K to low-K ratio was 2.3. Also, this underestimation of K by the Hvorslev analysis was about the same whether the underlying low-K layer was 2cm or 174cm thick (1% or 87% of the domain thickness). Numerical model simulations were useful in the interpretation of in-situ field K profiles at streambed sites with layering; specifically, scaling the model results to the maximum measured K at the top of the field K profiles helped constrain the likely ratio of high K to low K at field locations with layered heterogeneity. Vertical K values are important in field studies of groundwater–surface water interaction, and the Hvorslev analysis can be a useful tool, even in layered media, when applied carefully.