Electrical conductivity of partially saturated porous media containing clay: An improved formulation

Authors

  • A. K. Greve,

    Corresponding author
    • National Centre for Groundwater Research and Training, Connected Waters Initiative, University of New South Wales, Sydney, New South Wales, Australia
    Search for more papers by this author
  • H. Roshan,

    1. National Centre for Groundwater Research and Training, Connected Waters Initiative, University of New South Wales, Sydney, New South Wales, Australia
    Search for more papers by this author
  • B. F. J. Kelly,

    1. National Centre for Groundwater Research and Training, Connected Waters Initiative, University of New South Wales, Sydney, New South Wales, Australia
    Search for more papers by this author
  • R. I. Acworth

    1. National Centre for Groundwater Research and Training, Connected Waters Initiative, University of New South Wales, Sydney, New South Wales, Australia
    Search for more papers by this author

Corresponding author: A. K. Greve, National Centre for Groundwater Research and Training, Connected Waters Initiative, University of New South Wales, Sydney NSW 2052, Australia. (annakgreve@web.de)

Abstract

[1] Prediction of the bulk DC electrical conductivity of partially saturated porous media containing clay has remained a significant challenge. A variety of models have been proposed; however, no single model has yet been shown to be entirely satisfactory. In this paper, we propose an improved formulation to account for partial saturation by extending an earlier proposed effective medium formulation. In order to test the performance of the proposed formulation, 28 data sets are extracted from the literature and analysed by our model. The results are then compared with the results from the original effective medium formulation, the Waxman and Smits model as well as the original and recently extended Volume Average (VA) model. The new formulation shows a strong improvement compared to the original formulation with an average R-square of 0.996 for 28 data sets compared to 0.506 for the original formulation. Detailed comparison of the five fitted models shows that the new formulation and the extended VA model both perform very well and in a similar fashion. Simplifying assumptions made during the fitting process of the models allow the application of these two models with data commonly measured in the field.

Ancillary