Simulating the water content and temperature changes in an experimental embankment using meteorological data

An experimental embankment was constructed in Rouen, France. The construction was completed in December 2004. The first objective of this experiment is to investigate the influence of climatic changes on the soil response such as changes in water content and temperature as well as the induced vertical and horizontal displacements; and the second objective is to investigate the soil response under water flooding from the base of embankment. In this study, the changes in temperature, volumetric water content and suction along the central axis have been analysed using a one-dimensional model and based on the meteorological data obtained in the field. Comparisons made between the measurements and simulations have shown the relevance of the method adopted, provided that suitable boundary conditions and soil parameters are taken into consideration. Moreover, both the simulation and field monitoring showed that climatic effects are limited to a shallow depth, which results from the low permeability of the compacted fill.

Research highlights
► An experimental embankment was constructed in Rouen, France. Significant instrumentation was installed, including a meteorological station on the embankment surface, soil temperature, soil volumetric water content, soil negative water pressure (suction) and soil vertical and horizontal displacements, in various depths. This enabled the analysis of the interaction between atmosphere and embankment soil using the first period monitoring data over about 9 months prior to water flooding from the base.
► According to the embankment design, soil A28 was designed to be loosely compacted at 80% optimum in order to emphasise its collapsible behaviour upon wetting; by contrast, soil SNEC was densely compacted at 95% optimum in order to expose its swelling behaviour. The results of density control showed that the actual densities were far from the designed values. This reflects the common difficulty encountered in soil compaction work in field conditions.
► Because of the low permeability of the two constitutive compacted soils, the effect of evaporation and precipitation on soil volumetric water content was limited to a certain depth. Indeed, the field data showed that even at the first measurement level for volumetric water content (layer 14 at 34 cm depth), only a fluctuation smaller than 2% was recorded. According to the simulation results, the climatic effect on volumetric water content was limited to 10 cm depth.
► In terms of changes in suction or negative pore pressure, it appears that all depths were affected. This suggests that suction was more sensitive to climatic changes than volumetric water content. Nevertheless, the field data and the simulation results showed that the most significant variations were in the near surface area: the variations were very limited in the soil deeper than 10 cm.
► As far as the temperature changes are concerned, it appears that during 24 hours, it was limited to 34 cm depth. But this limit depth changed over a longer period: if a five-month period is considered, the soil temperature changes involved all the depths.
► Energy analysis method has been found to be an efficient method for the definition of the soil-atmosphere boundary condition. Indeed, only the easily measured meteorological data were required to predict changes in suction, water content and temperature. In practice, this kind of prediction has obviously significant economical and environmental impacts because direct monitoring of the suction and the water content are known to be difficult.
► Comparison of simulation results with field measurements shows that an approach combining the soil-atmosphere interaction analysis and a coupled heat-water flow model can be used to calculate the water content, temperature and suction of the soil at any time. As all numerical modelling work, the main difficulty is the determination of model parameters. In this work, for the determination of roughness length z0, a calibration was carried out using the first five-day field data. For the initial volumetric water content at the embankment surface, an estimated value of 20% was given. The simulation results showed that the numerical approach adopted is able to auto-regulate itself over a short time.
► It must be mentioned however that the adopted one-dimensional model represents a simplistic method because the geometry of the experimental embankment. It seems to be valid for the central axis. For a full geometry analysis, a two-dimensional or three-dimensional model is necessary. Note also that the work performed has benefited from the relatively rich data; when making an analysis of a real geotechnical embankment, the boundary conditions are certainly more complicated and less available data are expected; as a result, the analysis would be more difficult.