THERMODYNAMIC APPROACH TO COMPUTATIONAL MODELING OF CHEMICALLY STABILIZED SOILS

Civil engineers have extensively used chemical stabilizers to change the physical, chemical and engineering properties of clayey soils for infrastructure development. Various types of chemical stabilizers have been developed to treat clayey soils. Depending on the chemical composition of the stabilizer, they are divided into two categories (1) Non-calcium based stabilizers such as hydrogen ion stabilizer (HIS) (2) Calcium based stabilizers such as lime, cement, fly ash, etc. The application of thermodynamic models to simulate the reaction between the stabilizer and soil minerals can help to understand the stabilization mechanism. In addition, the models can predict the formation of reaction products including deleterious products such as ettringite. This dissertation proposes the application of thermodynamic models to simulate the stabilization reaction between the soil minerals and the chemical stabilizer. The dissertation is divided into three parts that investigates application of thermodynamic models for chemically stabilized soils. The first part shows the application of two thermodynamic equilibrium models (Visual MINTEQ and Geochemist’s Workbench) to simulate the reaction between the shrink-swell prone clay mineral smectite and HIS. In addition, the model predicted partial dissolution of smectite and release of Al3+ ions from the octahedral layer. The modeling results helped to establish a stabilization mechanism, which is described as a reduction in the shrink-swell potential by adsorption of Al3+ ions on the surface of smectite. X-Ray diffraction (XRD), Fouriertransform infrared spectroscopy (FTIR) and cyclic swell tests were used to validate the results. The second part presents a case study where fluidized bed ash (calcium based stabilizer) was used to stabilize a subgrade soil causing structural distress due to excessive ettringite formation. Thermodynamic modeling code GEMS-PSI was used to evaluate the occurrence of ettringite and, more specifically, to quantify the extent of ettringite formation. The model is shown to be 91% reliable in its ability to predict ettringite (qualitatively). The quantitative evaluations of the mass percentage of ettringite formation are shown to have a mean error of 12.4% with a standard deviation (SD) of 12.8%. The results from thermodynamic model were calibrated to account for the assumptions that reduced the mean error to 0.12% (SD 4.3%). The third part investigates the application of thermodynamic models to predict ettringite synthesized from Ca-Al-SO4 suspensions. The objective of the study was to predict ettringite without calibration. This was achieved using material characterization methods such as XRD, differential thermogravimetric analysis (DTA), and scanning electron microscopy. Five suspensions with different stoichiometric ratios of Ca(OH2) to Al2(SO4)3 · 18 H2O were used to thermodynamically and experimentally evaluate ettringite formation. Qualitatively, the model predicted ettringite as a stable phase for 3 out of 5 samples, which was also experimentally verified. Quantitatively, the models mean prediction error was close to 4%. The results indicate the application of thermodynamic model to predict ettringite formation