A research report submitted to the Faculty of Engineering and the Built Environment, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the Degree of Master of Science in Engineering, 2016 / Previously undermined areas pose a significant challenge to mining by opencast due to the risk of sinkhole occurrence. In order to optimise reserve utilisation as well as safeguard personnel and equipment there was need to develop a “Sinkhole Prediction Model” to assist in the prediction of areas prone to sinkhole formation. The aim of this research therefore was to develop a “Sinkhole prediction tool” with a view to pre-identifying areas of potential sinkhole hazard to inform better controls to assist in mining these areas safely. This was done utilising the current Hill (1996) caving height method culminating in the development of a hazard index model dividing the mining zones into high and low hazard. These areas were colour coded Red (High hazard) and Green (Low Hazard). The “Sinkhole Prediction Model” evolved to include over hundred sinkhole incidences that were statistically analysed to firm up on the robustness of the Prediction Model capabilities. The Hill (1996) caving height formula was discounted after the statistical analysis indicated that a good prediction model lies in the interrogation of site specific data. The outcome of the work conducted in this research report indicated a 97% correlation between the refined “Sinkhole Prediction Model” and the actual sinkhole occurrence at the Anglo American case study area (Mine X). Various refinements inclusive of lithological assessments, blast and drilling reconciliations as well as the implementation of the roughening up quality audits led to the implementation of a robust sinkhole management process that has managed to consistently assist in safeguarding equipment and personnel thus allowing for coal extraction optimisation in areas that could have been written off due to the sinkhole hazard. This risk can only be eliminated by mining the areas with the sinkhole risk. Currently the method is being impacted by significant roughening up cost incurred in a drive to make the areas safe to allow for coal extraction. The roughening up process on average costs R3.5 million per sinkhole and is a function of the number of sinkholes found, which translates to an equivalent cost of R7 / sales tonne. The current sinkhole prediction model being employed in deficient in that it cannot pinpoint the actual location of the void in the area previously undermined by bord and pillar and this is a great limitation of this report. Various geophysical techniques were pursued to assist in the precise identification of the actual sinkhole spatially. This process was aimed to reduce the roughening up cost (entire block stabilisation) as opposed to targeted sinkhole excavation and stabilisation. This process proved futile as the void identification systems are highly incapable of identifying the voids / iv sinkholes spatially (x, y and z coordinates) to assist targeted sinkhole treatment as a result of the following:  System inability to penetrate areas comprised of highly conductive strata such as clays.  Inability to distinguish between the underground voids and geological anomalies such as dykes.  Not suitable for penetrating wet strata.  Impacted by noise interference from mining machinery. The major result of this research is the establishment of a site specific “Sinkhole Prediction Model” that can generate hazard plans in real time thus informing the management on areas associated with a potential sinkhole hazard. The hazard plans can be generated timely and decisions made to facilitate safe coal extraction in areas of high sinkhole hazard. This has culminated in a robust sinkhole management process within the group that has managed to eliminate the risk of personnel and equipment exposure at Mine X. The roughening up process is accepted as the primary sinkhole mitigation or rehabilitation process with the need to work towards reducing the roughening up costs through development of the tool capable of precisely identifying the voids routinely to facilitate targeted rehabilitation. Significant research is required in this area as the mining environment is comprised of strata that currently cannot support the use of real time void identification to facilitate targeted void identification and rehabilitation. There is also merit in the future to formulate the database capable of assisting in the prediction of sinkholes in the Witbank coalfield as well as assist in robust management of mining boundaries across the different mining houses. The system implemented at Mine X is currently being deployed to other operations in the group where modification will be made to match the site specific conditions. Future research into understanding the sinkhole occurrence dynamics is quite crucial if targeted rehabilitation is to be achieved for cost reduction and mining sustainability. A combination of the understanding of the sinkhole occurrence driving mechanisms in conjunction with use of modelling packages such as ELFEN (a hybrid Modelling) tool will go a long way in enhancing the development of precise sinkhole prediction point in space.

The purpose of this study is to determine the most suitable remediation techniques via engineering geological assessment of the landslide that occurred during the construction of Artvin-Savsat Junction - Meydancik Provincial Road at Km: 12+200 in an active landslide area. For this purpose, the geotechnical parameters of the mobilized geological material which is colluvium along the sliding surface were determined by back analyses of the landslide at three geological sections. The landslide were then modeled along the most representative section of the study area by considering the landslide mechanism, the parameters determined from the geotechnical investigations, the size of the landslide and the location of the slip circle. In addition, pseudostatic stability analyses were performed comprising the earthquake potential of the site. The most suitable slope remediation technique was determined to be a combination of surface and subsurface conditions. A static analysis of the landslide shall also be performed through utilizing finite element analyses. The static analyses were compared with the inclinometer readings in the field to verify the direction of the movement. Consequently, shear strength parameters were specified as c = 0 kPa and f = 10&deg / for the landslide material and pre-stressed anchoring and rock buttressing were considered as a remediation method.

The purpose of this study is to evaluate and to compare the results of bivariate statistical analysis conducted with three different data sets in Geographical Information Systems (GIS) based landslide susceptibility mapping applied to the Devrek region. The data sets are created from the seed cells of crowns and flanks, only crowns, and only flanks of the landslides by using 10 different parameters of the study area. To increase the data dependency of the analysis, all parameter maps are classified into equal frequency classes based directly on the percentile divisions of each seed cells data set. The resultant maps of the landslide susceptibility analysis indicate that all data sets produce acceptable results. In each seed cell data set analysis, elevation, lithology, slope, aspect and drainage density parameters are found to be the most contributing factors in landslide occurrences. The results of the three data sets are compared by Seed Cell Area Index (SCAI). This comparison shows that the crowns data set produces the most accurate and successful landslide susceptibility map of the study area.

Koyulhisar settlement area is located on the northern flank of Kelkit valley which is seismically active and landslide-prone area. The settlement area was adversely affected from active landslides and some of the houses were evacuated. The purpose of this thesis is to delineate areas where slope instability exists within the Koyulhisar settlement area, and to investigate an active landslide by means of field observations, drilling, sampling, field and laboratory testing, and in-situ monitoring using inclinometer. Based on the field studies, it is observed that flyschoidal sequence as bedrock and colluvium consisting clay and silt with some gravel are the main lithological units exposed in the study area. Two landslide affected areas are identified, the one investigated due to its adverse effect to some important governmental buildings, has a non-circular failure surface due to the existence of the flyschoidal sequence below the colluvium. The inclinometer measurements reveal that the displacements are local and their velocities are generally less than 14 mm/year indicating that the landslide is an extremely slow landslide. In addition, high groundwater table is observed as one of the major parameters in occurrence of landslide.

Prediction of ground settlements have always been a big challenge for the engineers that are responsible for the design of subway tunnel projects. Since ground settlement is a crucial concept directly affecting the successfulness of a project, it must be taken seriously and should be accurately estimated. Consolidation settlements in the close proximity of Ulus-Ke&ccedil / i&ouml / ren Subway project due to groundwater drainage is the focus of this study. In this sense, the necessary data about the project characteristics and the site conditions were collected thru project descriptions and the geotechnical investigations conducted at the project site. Utilizing the generated database analytical calculations were carried out to predict the settlements. Upon completion of this stage of analysis several of the locations were numerically modeled for further investigation. Numerical analysis was conducted at four sections by using Plaxis, to determine the amount of expected displacements and the resulting groundwater situation. Despite of the differences between these two methods the resulting settlement estimations displayed consistency.

Thesis (Ph.D.)--University of Wollongong, 2009. / Typescript. Includes disc in back pocket. Includes bibliographical references: leaf 264-272.

M.Sc. (Geology) / There was a possibility that the rockbursts experienced in the faulty Klerksdorp Goldfield area could be related to the numerous faults, fractures and dykes occurring in the area. Until the end of 1979 it was impossible to establish with any degree of certainty which discontinuity was associated with most of the seismic events, as the location network was too inaccurate at that stage to determine such associations. The accuracy of the network was subsequently improved to 30 m in all directions in a specific area of the macro-network. Twenty seven percent of the seismic events which located in this area, occurred less than 30 m from the faults, 21 percent were less than 30 m from the dykes, while 42 percent occurred less than 30 m from dykes and faults. The remainder ( 10 percent) were located more than 30 m from any of the abovementioned discontinuities.

Arrival times from intermediate-depth (110-150 km) earthquakes within the region of flat subduction beneath central Peru provide constraints on the geometry and velocity structure of the subducting Nazca plate. Hypocenters for these events, which are beneath the sub-andean and eastern Peruvian basins, were determined using a best-fitting onedimensional velocity-depth model with a 15-station digitally-recording network deployed in the epicentral region. For that model, P-wave travel times to coastal stations, about 6° trenchward, exhibit negative residuals of up to 4 seconds and have considerably more complexity than arrivals at the network stations. The residuals at coastal stations are conjectured to result from travel paths with long segments in the colder, higher velocity subducting plate. Travel time anomalies were modeled by 3-D raytracing. Computed ray paths show that travel times to coastal stations for the eastern Peru events can be satisfactorilymodeled if velocities relative to the surrounding mantle are 6% lower within the uppermost slab (a 6 km thick layer composed of basaltic oceanic crust) and 8% higher within the cold peridotitic layer (which must be at least 44 km thick). Raytracing runs for this plate model show that "shadow zones" can occur if the source-slab-receiver geometry results in seismic rays passing through regions in which the slab undergoes significant changes in slope. Such geometries exist for seismic waves propagating to some coastal stations from sources located beneath the eastern Peruvian basin. Observed first-arrival times for such cases do in fact have less negative residuals than those for geometries which allow for \direct\ paths. Modeling such arrivals as trapped mode propagation through the high-velocity part of the plate produces arrival times consistent with those observed. / Master of Science

No description available.

Three hybrid moment tensor inversion methods were developed for seismic sources originating from a small source region. These techniques attempt to compensate for various types of systematic error (or noise) that influence seismograms recorded in the underground environment in order to achieve an accurate and robust measure of the seismic moment tensor. The term 'hybrid' was used to distinguish between the relative method proposed by Dahm (1995) and the methods developed in this thesis. The hybrid methods were essentially weighting schemes designed to enhance the accuracy of the computed moment tensors by decreasing the influence of any low quality observations, to damp (or amplify) any signals that have been overestimated (or underestimated) due to local site effects, and to correct for raypath focussing or defocussing that results from inhomogeneities in the rockmass. The weighting or correction applied to a particular observation was derived from the residuals determined when observed data were compared with corresponding theoretical data (for a particular geophone site, sensor orientation and wave phase) and were calculated using a cluster of events rather than a single event. The first and second weighting schemes were indirectly related to the mean and the median of the residuals where the residuals were defined as the ratio of the theoretical to observed data. In the third scheme, the residuals were defined as the difference between the observed and theoretical data and the weights were based on the distance of a data point (measured in standard deviations) from the mean residual. In each of the weighting schemes, the correction was applied iteratively until the standard error of the least-squares solution (normalised to the scalar seismic moment) was a minimum. The schemes were non-linear because new weights were calculated for each iteration. A number of stability tests using synthetic data were carried out to quantify the source resolving capabilities of the hybrid methods under various extreme conditions. The synthetic events were pure double-couple sources having identical fault-plane orientations, and differing only in rake. This similarity in the mechanisms was chosen because the waveforms of tightly grouped events recorded underground often show high degrees of similarity. For each test, the results computed using the three hybrid methods were compared with one another and with those computed using the single event, absolute method and two relative methods (with and without a reference mechanism). In the noise-free situation, it was found that the relative method without reference mechanism showed the highest resolution of mechanisms, provided that the coverage of the focal sphere was not too sparse (> 3 stations). The hybrid method using a median correction was found to be the most robust of all the methods tested in the most extreme case of poor coverage (2 stations) of the focal sphere. When increasing levels of pseudo-random noise were applied to the data, the absolute moment tensor inversion method, the hybrid method using a median correction, and the hybrid method using a weighted mean correction all showed similar robustness and stability in extreme configurations concerning network coverage of the focal sphere and noise level. When increasing levels of systematic noise were added to the data, the hybrid methods using a median correction and weighted mean correction were found to exhibit similar robustness and stability in extreme configurations concerning network coverage of the focal sphere and systematic noise. In all situations investigated, these two hybrid methods outperformed the relative and absolute methods. The hybrid moment tensor inversion methods using a median and weighted mean correction were applied to a cluster of 14 events, having remarkably similar waveforms, recorded at Oryx Gold Mine. For comparative purposes, the absolute method was also applied. The inputs to the inversion methods consisted of the spectral plateaus of both P- and S-waves at frequencies below the comer frequency of the time-integrated displacement traces. The polarities of dominant motion were used as an additional constraint and were determined from cross-correlation of observed with synthetic P- or S-waves. The solutions computed using the hybrid moment tensor inversion using a median correction displayed a distinct improvement after the iterative residual correction procedure was applied. The radiation patterns and faultplane solutions showed a high degree of similarity, and are probably more accurate reflections of reality than those computed using the absolute moment tensor inversion methods. These observations are very encouraging and point towards the method's potential for use as a standard processing tool for mine seismicity. The implications of this work are a better understanding of the focal mechanisms of seismic events induced by mining activities, ultimately leading to improved safety underground.