Evaluation, extension, and application of a full field light transmission technique for the investigation of hysteresis in thin homogeneous sand slabs.

Abstract

The objective of this dissertation was to show that significant errors in hydraulic conductivity occur when hysteresis is ignored. A light transmission technique was tested, extended, and then used to collect data on the unsaturated properties of five narrow grain size distribution sands. The data was reduced to produce a family of scale similar pressure/saturation curves and then used to challenge the ability of three numeric hysteretic models to accurately reproduce the experimentally measured permeability. The light transmission technique uses a uniform, controlled, diffuse light source, a transparent slab test cell (100 to 10000 cm² or greater) filled with translucent silica sand, and a digital imaging system. Five sands were chosen with differing mean grain size but the same distribution about the mean. Each of the sands were packed homogeneously and cycled through a series of filling and draining sequences. The light transmission technique was evaluated by comparing gravimetric inflow and outflow to that predicted by image analysis using several proposed saturation equations. It was found that the technique is limited in its ability to quantitatively resolve small-scale "heterogeneity" contained within the transmitted light field. The utility of this method is emphasized, however, by its ability to obtain on the order of 2000 x 2000 points with over 4000 levels of saturation for high resolution shuttered imaging systems. The in situ pressure/saturation data generated by this light transmission technique was compared to that obtained using standard column methods and described with the equations of Brooks and Corey and van Genuchten using the computer model RETC. The results show that the optical technique exhibited increased accuracy in comparison to the standard column method. The ability of three capillary hysteresis models; Scott's (1983), Mualem's modified dependent domain (1984a), and Hogarth's (1988), to predict the hysteretic curves was then examined. The models were modified to put them on an equal comparison basis and then implemented with the same minimum input data. The results revealed that small errors in predicted moisture content are amplified when computing the relative hydraulic conductivity values as a function of saturation. None of the models distinguished itself in terms of predictive accuracy.