Dispersive soils have caused failure of many slopes and earth fills due to external and internal erosion. This study aims to investigate various factors used for identification of dispersivity and to develop some new approaches for the prediction of dispersivity of clays. To achieve this purpose, physical and index properties, as well as degree of dispersivity of 29 clay samples taken from five different locations in and around the city of Ankara were determined. Various statistical prediction models were used for prediction of new dispersivity classes obtained by weighting ranking method. It was determined that dispersivity classes obtained from physical and chemical dispersivity tests performed on the same clay samples using distilled water were different from each other. In addition, crumb and pinhole tests were performed by using test waters with varying TDS values on five selected samples to find the impact of water chemistry on dispersivity. It is concluded from all dispersivity tests that total dissolved salts (TDS) values and sodium percentage (SP) remarkably affect the degree of dispersivity, and the use of these two parameters give more reliable results for the determination of dispersivity. By considering all these facts and to predict the most reliable dispersivity class, all dispersivity classes obtained from physical and chemical dispersivity tests were reevaluated by a weighted ranking system, and new dispersivity classes were assigned. In order to estimate these new dispersivity classes, various statistical models were established by using results of chemical analysis of pore water of clay samples. For this purpose, prediction models including soft computing methods such as decision tree and logistic regression are used and most reliable prediction models having the highest prediction performance are suggested.

The strength of geological materials is a fundamental property used in the design of civil engineering works; including projects constructed in complex geological mixtures or fragmented rocks such as mélanges, fault rocks, coarse pyroclastic rocks, breccias and sheared serpentines. These and other, often chaotic, mechanically and/or spatially heterogeneous rock masses are composed of relatively strong rock blocks surrounded by weaker matrix rocks. These common rock mixtures, known as bimrocks (block-in-matrix-rocks) or bimsoils (when the matrix material is soil-like) are very difficult to evaluate. It is almost impossible to recover high quality, undisturbed drill core samples or to prepare laboratory specimens perform laboratory studies and evaluate geomechanical parameters such as cohesion, internal friction angle and uniaxial compressive strength from these complex mixtures. There is sparse literature describing empirical and laboratory studies on strength of bimrocks. Hence, this study was devised to develop a preliminary Bim strength criterion, a generalized conceptual empirical approach for predicting the overall strength of unwelded bimrocks and bimsoils. The approach considered fundamental rules of rock and soil mechanics and the reported mechanical behavior of bimrocks as a function of volumetric block proportion (VBP), as reported in literature. In addition, cores of artificial bimrock were prepared in the laboratory for uniaxial and triaxial compression testing. Empirical equations useful for predicting the strength of bimrocks were devised, which depend on practical charts and input parameters, such as parameter “A”, defined to relate the contact strength between matrix and blocks. The predictive equations were calibrated by using the literature database. The predictive performance of the preliminary Bim strength criterion for bimrocks and bimsoils was checked by using the established database, and was found to have high predictability, although erring slightly on the conservative side.

The uniaxial compressive strength of rock material (UCS) is one of the fundamental input parameters for engineering applications to be constructed on/in rock masses such as deep slopes, tunnels and dams. However, preparation of the high quality cores for laboratory studies is generally difficult for some types of rock such as laminated and/or fragmented rock material. To overcome this difficulty empirical prediction models were developed by considering some input parameters. Geological mixtures composed of rock blocks surrounded by weak matrix material are known as Block-In-Matrix-Rock (Bimrock) in literature. Agglomerate is a special type of Bimrock, which is composed of andesite fragments surrounded by tuff matrix and it is an example of Volcanic Bimrock. Preparation of core samples for experimental studies from agglomerate is problematic due to the strength contrast between andesite rock fragments and tuff matrix. To overcome these difficulties, some prediction tools have been studied by regression analyses in the literature. In this study, Artificial Neural Network (ANN) as a prediction tool was used to construct a model for prediction of overall UCS of Volcanic Bimrock. While Volumetric Block Proportion (VBP), Volumetric Block Count (VBC) and fractal dimensions (1 and 2 dimensional) were selected as input parameters, normalized overall uniaxal strength of agglomerate to uniaxal compressive strength of tuff matrix is output parameter. Fractal geometry has been used as popular method to define irregular shapes as a quantity in literature. The boundary strength between an-desite fragments and tuff matrix is also sensitive to fragment shape and surface roughness of andesite fragments. Therefore fractal dimensions were selected as input parameters to incorporate this effect on boundary strength. While previously developed computer code FRACRUN was used to determine average fractal dimension of andesite fragments in agglomerate cores, previously developed computer code ANNES was used for ANN based model construction. In addition, similar to Volumetric Joint Count (Jv) which is widely used in rock mass characterization, Volumetric Block Count (VBC) was defined as another input parameter for determination of Bimrock UCS considering some of studies about performed in literature. The highest prediction performance was obtained from the model which considers Volumetric Block Proportion (VBP), Volumetric Block Count (VBC) and 1D fractal dimension as inputs.

It is almost impossible to prepare representative cores of rock masses including discontinuities patterns for laboratory studies. To overcome these difficulties, researchers have focused on developing empirical equations for estimating of the stress–strain behavior of a rock mass, including measurements of the discontinuity patterns. As can be seen in the literature, the uniaxial compressive strength value of rock mass (UCSRM) can be estimated by reducing the uniaxial compressive strength of intact rock material (UCSi) based on the quality of a rock mass, represented by variables such as Rock Mass Rating (RMR), Geological Strength Index (GSI) and Q value. For this reason, a unique reducing curve form empirical equation has limited application and generally, cannot be applied to all kind of rock masses from particularly soft to hard rock masses. In this study, a new general empirical approach is constructed to estimate the strength of rock masses of varying hardness. The new empirical equations have been calibrated using data from five slope failures and four sets of uniaxial compressive strength data of rock masses. In the new empirical equations, the UCSi is considered not only to be a scale parameter used in the strength reduction but also used to adjust the degree of strength reduction in conjunction with elastic modulus of the rock material (Ei). The disturbance factor on the rock mass is taken into consideration by two separate reduction factors applied to the Structure Rating (SR) to capture increasing joint density, and to the s and mb parameters of the Hoek–Brown criterion, to decrease the degree of interlocking. Hence, non-interlocked (cohesionless under zero normal stress) rock masses such as spoil piles can also be modeled in the new empirical approach.

Clays have long been used in waste-confinement processes. However, there are not sufficient data in the literature presenting their sustainability under the effect of leachate water. As a case study, Ankara clay was chosen due to its suggested utilization as a landfill liner by previous researchers. Its sustainability under the effect of local leachate water, however, was left in question by earlier studies. In order to determine long-term effects of leachate water, samples were taken at three locations: Golbasi, Middle East Technical University, and Cigdem District. The disturbed samples were compacted with leachate water at optimum water content and cured for 1 and 4 months. The compressibility of the compacted Ankara clay was determined to be low, and curing with leachate water did not cause significant changes in the coefficient of hydraulic conductivity. In all locations, cohesion and internal frictional angle showed a decreasing trend after curing with leachate water. Also, cured samples showed lower Methylene Blue Value and Cation Exchange Capacity than natural samples. In addition, the mass of loss after freezing-thawing cycles for cured samples was higher than natural clay but lower than distilled water-compacted samples. The highest measured pH values were obtained from leachate-cured samples. Ankara clay meets the requirements of the Republic of Turkey Ministry of Environment and Forestry and the U. S. Environmental Protection Agency. The properties of the clay do not change significantly enough to outrange the requirements under the effect of leachate water. It has, however, a higher pH and mass of loss after freezing and thawing cycles than suggested values for clay liners presented in previous studies.

Working in complex grounds have always been difficult for engineering geologists because of the heterogenous nature of such geo-materials, which results in different behaviors under stresses. The caliches outcropping in the Adana basin and its close vicinity are selected as the study material because of their highly complex nature and also understanding their mechanical behaviour and collapse potential is too difficult, although their aerial extent is large in the region. For this reason, investigation of the collapse potential and mechanical behaviour of the caliches by applying new approaches besides the conventional in situ and laboratory tests is the purpose of the study. The study includes five main stages such as measurements of caliche profiles, in situ tests (plate loading), sampling, shooting photographs for photoanalyses and laboratory studies. Four different levels such as hard pan; silty, sandy layer; gravelly, blocky layer and clayey level are described for the caliches employed. The hard pan level is a weak rock with an average uniaxial compressive strength of 11.89 MPa while the others have typical soil characteristics. A series of plate loading tests are applied on the blocky, gravelly level of the caliche to determine the modulus of elasticity. The modulus of elasticity and the allowable bearing capacity are determined between 28.6–65.3 and 1.5–2.0 MPa, respectively. To determine the grain size distribution curve, in addition to sieve analyses, a photoanalysis technique is also applied and a combination procedure between the results from both sieve analyses and photoanalyses is introduced and the grain size curves for the blocky, gravelly level of the caliche are obtained. According to the results of collapse potential index tests performed on the samples collected from 20 locations of the study area, the soft pan level of the caliche has slight to moderate degree of collapse indices. In the final stage, various simple and statistically meaningful empirical equations are proposed for the indirect determination of the collapse index by employing simple paramateres.

The modulus of elasticity of intact rocks is widely used in many rock engineering projects such as tunnels, slopes, foundations etc. as an input parameter. It is also used in the evaluations of deformation modulus of rock mass based on some empirical models. However, determination of this parameter from laboratory tests requires high-quality core samples and sophisticated testing equipments. Considering this difficulty, the use of empirical models to obtain this parameter has been attractive in rock engineering practice. In rock mechanics literature, some empirical relations exist between modulus of elasticity and other rock properties, such as uniaxial compressive strength (UCS), unit weight (gamma), Schmidt hammer, point load index and petrographic composition. The main deficiency of the existing empirical relations is that they either ignore the rock type or use limited rock types. To eliminate these deficiencies, total of 239 UCS, unit weight, tensile strength (Brazilian) and modulus of elasticity (E;) data were collected from the literature. Besides, total of 80 tests were performed on the greywacke and agglomerate core samples in this study. A total of 319 data set representing 37 different rock types were used throughout the analyses. To assign the rock type, the m; constant of the Hoek-Brown criterion was also considered. The UCS and the tensile strength data pairs, and the Hoek-Brown equation for intact rock were used to calculate the m; constant for each rock sample. In the first stage of regression analyses, a series of simple regressions were performed to define type and significance degree of relations between the independent parameters and modulus of elasticity. The simple regression uses one input and one output parameters. Due to this limitation, single value can be obtained from the simple regression based equation for modulus of elasticity depending on the input parameter. In other words, two different rocks may have same input parameter, although their elasticity moduli values are different. Considering this limitation, a series of simple regression were also considered by using combined parameter which include two or more input parameters. Finally, two prediction charts were prepared on the based on the empirical equations having coefficient of correlations of 0.863 and 0.872, respectively.