Optimizing Hydraulic Conductivity Estimation from In Situ Measurements Using Artificial Data

ABSTRACT Determination of the hydraulic conductivity-void ratio function (k-e) of sedimenting soft soils can be performed using in situ density and pore-pressure measurements in column tests, centrifuge, and in pilot or field deposits without resorting to back-calculation using numerical methods. The disadvantage of back-calculation is that it requires knowledge of the compressibility curve, which is known to change over time in soft soils. Direct determination of k-e has been employed in previous studies, essentially applying Darcy’s law to find k from fluxes and gradients determined by various interpolation strategies of the known density and pore-pressure profiles. This technique is sometimes called the instantaneous profile method (IPM). This paper examines this technique rigorously through both (1) generating artificial data from a large-strain consolidation model and applying the IPM to this artificial data to attempt to recover the consolidation properties inputted into the model, and (2) by comparison with known case studies in the literature. This exercise provides clear requirements and limits on data needs as well as data interpolation methods. It was found that a substantial reduction in the scatter of the estimated k-e values was achieved when a local dependency of k on void ratio was assumed only for the variation of k across an IPM element: k∼eb. This assumption of local variation did not inhibit IPM from regenerating the global hydraulic conductivity function, even if the power of this global function was very different from the assumed form of the local variance. Screening out fluxes generated by measurement noise also improved the results.