The principal objective of this project was to develop a rapid, robust, and reliable tool for in situ measurement of hydraulic properties in heterogeneous, anisotropic, variably saturated, porous media. To this end, technical objectives were identified, including:
The different components would ultimately be integrated into a single multipurpose tool for deployment using direct push technology, providing a means to estimate in situ hydraulic properties from downhole measurements.
To obtain the data necessary to meet these objectives, a combination of numerical, laboratory, and field studies were designed and implemented to characterize sediment properties and relate them back to those that could be easily measured with direct push technology. A team of researchers developed direct push, in situ sensors that could be deployed individually or as part of an integrated tool. The sensors included (1) a tension (subatmospheric pressure) permeameter for measuring variable relative permeability as a function of saturation and (2) a microscopic imaging system for in situ imaging of sediments from which grain-size characteristics and independent estimates of hydraulic properties would be derived using digital photogrammetry.
A critical project component was modification of the existing SCAPS/GeoVIS system to function as dual field-of-view video camera system for subsurface soil imaging. This required modification of the single-camera GeoVis system to enable simultaneous capturing of images at different magnification factors and a more complete description of the grain size distribution curve. One camera captures grain size data in the silt-sized fraction and the other captures information in the sand-sized fraction. These two levels of magnification provide fields of view ranging from 2 to 20 mm diagonal.
Digital images of sediments were converted into particle size distributions and their moments using the Pixel-Vernier, a state-of-the-art imaging tool consisting of a suite of photogrammetric algorithms that combine markers-controlled watershed algorithm with minimum-distance clustering to solve the segmentation problem. The segmentation algorithm decomposes the image into separate particle regions, which are used to derive several geometric attributes for each particle. These were used to estimate the particle size distribution and their relevant statistics. Particle size distributions were then used with a packing model to estimate porosity and saturated hydraulic conductivity, which have the added benefit of constraining hydraulic conductivities derived from the borehole permeameter measurements. The approach proved successful in characterizing a diverse set of materials including soil, computer-derived complex digital patterns, and sediment images from the Mars surface.
In laboratory tests to calibrate the instruments and models, sediments were separated in 1φ fractions and characterized to determine particle shape, porosity, and hydraulic conductivity. Binary mixtures of coarse and fine fractions were prepared with the fine fraction ranging from 0 to 100%. Measurements of particle shape, porosity, and hydraulic conductivity were made on end members and mixtures. Results show that particles in all size classes are aspherical and that end-member porosity increases with decreasing particle diameter in the sand fraction and smaller. In the gravel fraction, end-member porosity initially increased with increasing diameter before becoming constant. A model for the incomplete mixing of aspherical particles proved successful in predicting porosities and conductivities. Both parameters decreased to a minimum at a critical content of smaller-sized particle “fines,” confirming the importance of the fractional concentrations of each component.
A subatmospheric borehole permeameter proved successful in overcoming limitations in subsurface conductivity measurements. Use of a subatmospheric pressure permeameter reduced the effect of macropores and fissures on matrix flow. For the analysis of data, a new method using the Brooks-Corey hydraulic model was developed to solve the steady-state infiltration equation. Results from numerical simulations compared well with laboratory experiments. Field measurements were also in good agreement with independent measurements although the saturated hydraulic conductivity was slightly underestimated. With the successful application of the packing model to the prediction of porosity and saturated hydraulic conductivity, the saturated hydraulic conductivity can be easily constrained in the permeameter analysis.
This approach will lead to a better understanding of the measurements available for permeability characterization and help to establish the validity of permeability predictions in untested intervals based on measurements like grain-size distributions that are easier to make. The data obtained using the direct push in situ sensors will be invaluable in the design and evaluation of remedial systems and the prediction of future contaminant migration.