| Popis: |
This thesis documents modifications and improvements to the McNeal-Ezlit Generalised Clay Disaggregation (GCD) model. This semiempirical model provides soil specific predictions of soil behavioural response to water quality, quantified by measurement of the saturated hydraulic conductivity (KS). The aim was to identify limitations and shortcomings currently present in the model, and to modify or propose improvements to the model. Concretely, this thesis provides a simplification of the underlying model equation which is evaluated on 26 soils for which the fitted parameters of the model have previously been published. It is shown that this improves overall transparency and interpretability of the model without significant modification to the model outputs. The present algorithm for transforming measured KS data to fit to the model is evaluated and shown to be flawed. Alternatives are investigated and a revised method using a single reference datum per soil data set is proposed. The GCD model does not presently account for variability as part of its measurement process. A novel method of z-score normalisation of pre-treatment data is proposed to estimate soil-packing related variability in the model. In undertaking this evaluation, the validity of the method by which conductivity measurements are made for fitting this model is questioned with respect to the concentration of pretreatment, and equilibrium conditions for obtaining suitable measurement data. The improved GCD model is then tested for its ability to compare two soils or distinct soil conditions for site comparison or change measurement. It is shown that there is no mechanism within the model to make direct comparisons between two soils or soil states, and alternatives are proposed and demonstrated in a preliminary sense. Through doing so it is found that salt-dominated soil behavioural responses patterns described by the GCD are dynamic and not static properties. |
