Modified soil suction models to enhance the predictions of seasonal surface movement in expansive soils

<p>Expansive soils that exhibit shrink-swell behaviour is known to cause significant damage to structures and its components. Studies in the United States indicate that expansive soils cause $US15 billion damage every year to houses which is more than double of the damage caused by natural disasters such as floods, earthquakes, cyclones, and landsides (Jones and Jefferson, 2012). Richards et al. (1983) estimated that 20% of surface soils in Australia could be classified as moderate to highly expansive soils with an imposed damage potential ranging from minor cracking to irreparable damage to buildings. Recently, reports of major damage to 5000 residential houses post extreme weather conditions have been reported in the state of Victoria, Australia. Extreme weather and seasonal variations naturally influence the presence of moisture in soils, and expansive soils shrink and swell to such changes in soil moisture. Therefore, it is imperative to gain an understanding of the influential parameters causing these volume changes in soils and model these effects for an efficient design of structural components. Per AS 2870, these influential parameters determine the characteristic surface movement. Site class is defined based on the characteristic surface movement, and footing design will be based on the site class.</p> <p>The Australian Standard (AS 2870) addresses the above issues mainly based on the following: i) classification of a site based on Thornthwaite Moisture Index (TMI) [TMI supposedly addresses the climatic influence on footing design], ii) the design depth of soil Hs that can be obtained from TMI iii) the variation of soil suction “∆pF” (i.e. the plausible variation of `total suction¿ at a specific point in the soil in pico Farad units) modelled as a linearly varying profile over the design depth "Hs", and iv) the design vertical movement (ys) at a site calculated using soil Instability Index (Ipt) obtained through laboratory testing and v) an appropriate the design of foundations is recommended based on the calculated ys and site classification. The above-simplified explanation of the code procedure encompasses several complex mechanisms as demonstrated in the literature. The most notables are the non-linear or near-random variation of soil suction that is sensitive to seasonal influences and the calculations of TMI from different approaches that leads to uncertainties in the estimations. Literature also identifies that soil suction measurements are sporadic in nature, given its difficulty with field measurements. Given these limitations, a comprehensive research program has been undertaken to address the above issues in two folds: i) The variation of soil suction measured at a case study site over a two-year period on daily basis using sensor networks and the data analysed using a modified approach to the solution of 1D partial differential equations (PDE) and ii) the effect of climatic influences modelled using water balancing techniques rather than TMI. Results from the above two approaches favourably compare with the site observations paving the way for more realistic predictions of soil movements at a given site. Further details of the research activities and the results are provided below.</p> <p>Three Watermark<small>tm</small> soil sensors that can measure changes in soil suction were placed at depths ranging from 0.3 m below ground up to 1.5 m. Daily data obtained were used to analyse the suction variations on monthly basis. This measured variation of suction combined with the laboratory testing of the soil for Instability Index (Ipt) were used to calculate the ys displacement at the site as noted in AS2870. These displacements over time were compared with actual site measurements monitored using high-tech levelling instruments. Suction data (obtained in the first year) were analysed using a novel approach to the solution of the 1D diffusion PDE to obtain the diffusion and advection coefficients for the soil. It is interesting to note that the diffusion coefficient remained the same throughout the year, whereas the changes in advection were of several orders in magnitude. This finding suggests that the existing code approach based only on the diffusion parameter may indeed provide inaccurate results over long time. Again, these calibrated parameters were used to forecast suction variation for the second year and verified with the actual measured suction data. A MATLAB program was developed to complement the research approach. Model predictions were also compared with field data from other research projects and with the literature where appropriate and found to be satisfactory.</p> <p>In parallel to the above research work, the variation of TMI over a twelve-year period obtained from the literature was compared with the variation of available water at a given site using water balancing techniques. Patterns of this water balance variation with that of suction variation were observed to be consistent. Further analyses of these patterns during dry and wet periods were undertaken to calculate the sinusoidal wave of vertical displacement ys over time. This approach allows a generalization of the patterns and the ability to model variations of soil suction and surface movement over long term. The approaches noted above are codified into simple Excel sheet programs for efficient use by practicing engineers. In particular, the above approaches provide credibility to the estimation of surface displacements both in the short term and long term. Further improvements and validations of the above approaches are recommended in the concluding parts of the thesis.</p>