Air Drying of Unsaturated Soil
In order to interpret the results of any soil test it is essential to know the internal state of the sample. It is commonly assumed that the soil sample is uniform. In practice it is known that any tests involving the flow of water into or out of the soil sample must inevitably lead to an internally non-uniform state. This is because the flow of water requires a gradient of pore water pressure and as a result there will be internal variations of water content and stress state in the soil. In order to keep the soil sample as uniform as possible it is common practice in saturated soil testing to restrict loading/displacement rates. For unsaturated soils the same problems exist but are made worse by the presence of air in the soil sample.
Understanding the air drying process is of great importance for filter paper measurements of suction, where the sample is usually left exposed for around two hours and then the filter paper is placed on the sample and left until equilibrium is achieved. Although the air-drying process looks very simple, the process is highly complex involving coupled thermo-hydro-mechanical phenomena that take place during process of evaporation of the sample. The sample after compaction is suddenly put in contact with an exterior environment which contains a mixture of water vapour and dry air at atmospheric pressure but at a lower relative humidity than that initially existing in the sample. Hence an initial thermodynamic imbalance occurs between the external vapour concentration and that within the sample. To restore thermodynamic equilibrium, the sample exchanges water vapour with the exterior environment. Consequently, the vapour pressure within the sample decreases. For the liquid and vapour to remain in isothermal equilibrium within the sample, the liquid pressure also decreases. This requires the water to evaporate. Since the pressure decreases and the pressure of the external environment remains unchanged, the capillary forces (suction) increase causing the sample to shrink. Due to non-uniform shrinkage of the sample, tensile stresses are generated and this can lead to cracking in plastic soils. The spatial variation of suction in a drying soil sample is mainly dependent on the soil water characteristic curve, the hydraulic conductivity, environmental conditions, initial water content and the boundary conditions.
The STOMP simulator, which models both liquid and vapour flow under isothermal and non-isothermal conditions, was used to study the effects of different types of boundary condition on the distribution of matrix suction in a drying sample under transient and steady state conditions. The analysis uses a finite difference method to represent the sample. Three sets of isothermal analyses are reported in this paper. Three different boundary assumptions were made: top drying, top and bottom drying and drying on all boundaries. In all analyses where boundaries are allowed to evaporate atmospheric conditions are prescribed. The air pressure was set to 101.325 kPa and the relative humidity was 35%.
The geometry of the sample was modelled as a cylinder 100 mm in height and 50 mm in diameter (250 grid points; radial spacing 2.5 mm, vertical spacing 4.0 mm). Due to symmetry in both the horizontal and vertical directions only a quarter of the sample has been represented by an axi-symmetric mesh. Fig. 1 shows the measured changes during three days of the sample drying test with predictions for the idealised isothermal air drying test. It can be seen that agreement between measured and predicted weight of the sample during drying process is quite good. It is interesting to note that fig. 1 indicates equilibrium conditions were achieved around 2 days of drying.
The first simulation was of "top drying". At the beginning of simulation, there is liquid water migration towards the bottom of the sample because of gravitational body forces and water vapour migration to the top surface because of the lower relative humidity at the top boundary. After 30 days the problem is still in a transient state with drying continuing. Both the aqueous and gas pressures increases with depth at the end of the simulations and the effects of vapour pressure lowering keep the top node from drying to the residual saturation. The soil moisture distribution patterns (fig. 2) predicted by the simulator show transient conditions.
The second simulation was of "top and bottom" drying. This simulation reached steady state conditions after 30 days. During the transient portion of the simulation the bottom of the sample remains wetter than the top, because of gravitational body forces causing water to migrate downwards. After 30 days the soil has dried to equilibrium conditions slightly above the residual saturation because of vapour pressure lowering. The representative time-histories for "top and bottom" drying demonstrating the effect of boundary are shown in fig. 3. Comparing fig. 2 and 3, it is observed that after t=20 days the degree of saturation at the mid-point is 0.466 in the case of "top" drying whereas 0.113 for "top+bottom" drying. The difference between these two simulations can be mainly accounted for by a H2 effect where H is drainage path length. In the case of "top+bottom" drying, the drainage path is a half of the drainage path length for "top" drying simulation.
The drying through all boundary simulations reached equilibrium conditions after only five days. As in the previous simulations, the water initially migrated toward the bottom of the sample due to gravitational body forces causing a skewed saturation profile. As drying proceeds capillary forces become stronger making a non-uniform distribution of soil moisture (fig. 4). The comparison between fig. 2, 3 and 4 demonstrates the importance of the influence of boundary conditions on suction measurement using the filter paper method.
Cabarkapa, Zeljko, and Mark White. 2002. "Air-Drying Processes in Unsaturated Soil." Environmental Geomechanics.