Over recent years, new methods have been developed to integrate uncertainty into the optimization of life-of-mine (LOM) production planning. This thesis makes use of two stochastic optimization methods: simulated annealing (SA) and stochastic integer programming (SIP); which are implemented in the context of the requirements of mining applications through the use of stochastic simulation to model uncertainty. For the case of SA, the second chapter of the thesis documents the case of a copper deposit where ten simulated realizations are sufficient to provide stable LOM optimization results. In addition, the study shows that the selected true optimal pit limits are larger than the ones derived through conventional optimization. Stochastically optimized pit limits are found to be about 17% larger, in terms of total tonnage, than the conventional (deterministic) optimal pit limits. The difference adds one year of mining and approximately 10% of additional net present value (NPV) when compared to the NPV of conventional optimal pit limits and a production schedule generated stochastically with the same SA algorithm. In the third chapter of the thesis, the SIP based optimizer is used with the purpose of integrating uncertainty into the process of pushback design. Results show the sensitivity of the NPV to the design of starting and intermediate pushbacks, as well as the pushback design at the bottom of the pit. The new approach yielded an increment of ~30% in the NPV when compared to the conventional approach. The differences reported are due to the different scheduling patterns, the waste mining rate and an extension of the pit limits which yielded an extra ~5.5 thousand tonnes of metal. / Depuis quelques années, de nouvelles méthodes ont été développées pour intégrer l'incertitude dans l'optimisation de la planification de la production de la vie-de-mine i.e. life-of-mine (LOM). Cette thèse se sert de deux méthodes d'optimisation stochastique : recuit simulé (RS) et programmation en nombres entiers stochastique (SIP); les deux méthodes sont programmées dans le cadre des besoins des applications d'exploitation de la simulation stochastique et modélisation d'incertitude. Pour le cas de RS, le deuxième chapitre de la thèse décris le cas d'un dépôt de cuivre où dix réalisations simulées sont suffisantes pour fournir des résultats stables d'optimisation de LOM. En outre, l'étude prouve que les véritables limites optimales choisies de mine sont plus grandes que celles dérivées par l'optimisation conventionnelle. Des limites stochastiquement optimisées de mine s'avèrent environ 17% plus grandes, en termes de tonnage total, que les limites optimales (déterministes) conventionnelles de mine. La différence ajoute un an d'exploitation et approximativement 10% de valeur nette additionnel (NPV) une fois comparée au NPV des limites optimales conventionnelles de mine et une cédule de production produit stochastiquement avec le même algorithme de RS. Dans le troisième chapitre de la thèse, l'optimiseur basé sur SIP est utilisé en vue d'intégration de l'inceritude dans le processus de la conception de fosses emboîtées. Les résultats montrent la sensibilité du NPV à la conception de fosses emboîtées de commencement et intermédiaires aussi bien que la conception de la fosse emboîtée du fond de la mine. La nouvelle approche a produit une augmentation de ~30% dans le NPV une fois comparée à l'approche conventionnelle. Les différences rapportées sont dues aux différents cédules de production, du taux de décapage du stérile et d'une extension des limites de la mine qui ont produit ~5.5 mille tonnes supplémentaires

Multiple-point simulation (MPS) methods have been developed over the past decade as a mean to generate stochastic simulations while reproducing complex geological patterns, such as high-grade depositional veins, groups of high-grade lentil shaped orebodies, or the spatial geometries and patterns of diamond-bearing kimberlite pipes. This thesis compares two MPS methods by modelling the geology of a diamond pipe located at the Ekati mine, NWT, Canada. The single normal equation simulation algorithm SNESIM, which captures different patterns from a training image (TI), and the filter simulation algorithm FILTERSIM, which classifies the patterns founded on the TI, are considered in this study. Both methods are used to generate stochastic simulations of a four-category geology model containing crater, diatreme, xenoliths and host rocks. Soft information about the location of the host rock is also used. Both MPS methods reasonably reproduced the geometry of the pipe, as per the TI used (crater and diatreme rock units); however, the methods differed in the proportion and location of the xenolith bodies within the pipe. The validation of the simulated results provided by the above methods shows a reasonable reproduction of the data proportions for all geological units considered; the validation of spatial statistics, however, shows that although simulated realizations from both methods reasonably reproduce the fourth order spatial statistics of the TI, they do not reproduce well the same spatial statistics of the available data (when these differ from the TI). An interesting observation is that SNESIM better imitates the shape of the pipe, whereas FILTERSIM reaches a better reproduction of the xenolith bodies. / Les méthodes de simulation multi-points ont été développées au cours de la dernière décennie afin de générer des simulations stochastiques reproduisant des structures géologiques complexes, comme par exemple des dépôts en veines à hautes teneurs, des groupes de formations lenticulaires à hautes teneurs, ou les géométries spatiales et structures de gisements diamantifères de kimberlite.Cette thèse compare deux méthodes multi-points pour modéliser la géologie d'un gisement diamantifère situé dans la mine Ekati, dans les Territoires du Nord-Ouest du Canada. Les algorithmes de simulation SNESIM, capturant les structures d'une image d'entraînement, et FILTERSIM, classant les structures trouvées sur une image d'entraînement, sont considérés dans cette étude et utilisés pour générer des simulations stochastiques d'un modèle à quatre catégories : cratère, diatrème, xénolites et roches hôtes. On utilise aussi des informations qualitatives sur la position de la roche hôte. Les deux méthodes reproduisent la géométrie du gisement de manière satisfaisante, conformément à l'image d'entraînement utilisée (le cratère et les diatrèmes); cependant, les méthodes donnent des résultats différents quant à la proportion et la position des corps de xénolite dans le gisement. La validation des résultats des simulations par les méthodes mentionnées ci-dessus montre une reproduction raisonnable des proportions pour chaque catégorie considérée, mais la validation des statistiques spatiales montre que, même si les simulations produites par les deux méthodes reproduisent les statistiques d'ordre quatre de l'image d'entraînement, elles ne reproduisent pas celles des données disponibles (lorsque ces statistiques diffèrent de celles de l'image d'entraînement). Il est intéressant de noter que SNESIM reproduit mieux la forme du gisement, tandis que FILTERSIM permet une me

This work develops a backfill design procedure aimed to facilitate the optimization of an available mine material in order to meet target objectives in a particular mining role. This required the compilation and analysis of data on fill usage, established design procedures, physical and geomechanical properties, testing techniques and procedures, as well as behavioural modelling methods. / A backfill classification system is proposed based upon size distribution. A series of design equations are presented which relate to this system. These equations represent the means by which backfill geomechanical behaviour can be related to physical properties. This is considered to be fundamental to an effective backfill design procedure. Derivation of the equations has been based upon analysis of data from a program of laboratory and in situ testing conducted in ten operating Canadian mines by the author, together with other published work. / The in situ testing required the development of a pressuremeter testing procedure novel to underground mining. The theoretical basis for the employment of pressuremeter data has been examined and behavioral equations have been developed to describe the deformation and stress history during a backfill material test. In addition two new equations have been developed for the analysis of pressuremeter data. The in situ data collected has been correlated with laboratory derived geomechanical data for the same backfill materials. / The geomechanical properties associated with the proposed backfill classifications have also been related to their influence on backfill behaviour in three mine backfill roles: free standing stability during pillar recovery in bulk mining methods; dynamic interaction with stope walls in rockburst prone ground; and ability to reduce stresses in highly stressed rock masses. This work has been based on new and established modelling methods and aims to provide insight into the effectiveness of the backfill classes in these roles of growing practical significance.

In order to develop a philosophy and set of methods which would allow mines to be operated more effectively, a study was initiated to determine how time-based methods used in manufacturing could be adapted to mining. The study was done in collaboration with Inco Ltd. A review was made of both lean manufacturing and time-based competition methods. In addition, continuous programs such as Total Quality Management and Total Preventive Maintenance were studied. This study showed that time based principles, cellular mining, and total preventive maintenance had the most promise for adoption in mining. Simulations of three different mining methods under different scenarios and cost analyses were done to demonstrate the viability of these methods. The conclusions were the following. (1) Time based principles lead to greater profitability, and thus, should be used to drive mine planning and production. (2) Cellular mining leads to more efficient mining production. (3) The adoption of a comprehensive Total Preventive Maintenance program is needed to improve equipment availability and to complement cellular mining.

This thesis proposes a morphological complexity index for use in classifying orebodies. Formulated and tailored for application on underground steep clipping narrow base metal orebodies from Inco Limited's Manitoba Division, the complexity index was proved transferable to the base metal deposits of Inco Limited's Ontario Division. Thus it appears that the index could be applied to various styles of mineralization and orebody morphology. / The complexity classification was designed to aid in geologic control and subsequently improve mining method performance. Motivated by an industry wide move from selective mining to bulk mining, to increase productivity and improve safety. It is intended that development and application of an orebody complexity index will increase the understanding of geology and prevent the sacrifice of selectivity, minimizing dilution and ore loss. With the advent of automation, simplification in mining is necessary and thus more intelligent design and control vital. This thesis offers one means by which MDPPC (Mine Design, Planning and Production Control) could be further integrated with geological understanding to achieve such an end. / The thesis provides explanation of how such a complexity index can be used to understand mining method performance and be used for more successful mine design.

Acid mine drainage (AMD) is one of the largest problems facing the mining of base metals in Canada today. It results in the leaching of toxic heavy metals from waste rocks and tailings into the environment. Solidification/stabilization is a process whereby hazardous wastes are chemically stabilized and their handling properties improved. The objective of the project was to stabilize two tailings obtained from base metal mines in Quebec by adding varying proportions of lime and fly ash to them. The fixing capabilities of the two additives were tested by a modified Toxicity Characteristic Leaching Procedure (TCLP) test after 1, 14 and 35 days of curing. Mineralogical changes were monitored by the x-ray diffraction (XRD) analysis of 6 selected samples. / Results indicated the capability of lime-fly ash binders in the immobilization of heavy metals. XRD analysis showed the formation of gypsum and the gradual decline in pyrite content in most of the samples. The mineral ettringite was not detected, probably due to the relatively low pH of the samples and a deficiency in reactive aluminum. Hence, the results suggest the existence of other phases, possibly amorphous calcium silicates, which were responsible for the reduction in leachability.

Canadian hard rock mine extraction practices have commonly created shallow stopes, 12% of which have caved to surface, from instabilities originating from stope hangingwalls, crowns or footwalls. To date, however, mine operators have applied few of the available data gathering and design tools to strike a balance between maximum economic excavation dimension and stope viability. The preference has been to use personal mining experience. / Several common rock mass environments are surveyed as well as the various ways in which shallow stopes have failed. It has been found that these rock masses develop gravity-induced movement in the form of plug failures, ravelling of rock blocks, strata failures, chimneying disintegration, and rock mass block caving. / New failure-specific analytical equations are developed here for these common failure mechanisms. They address the mechanics of the failure process and incorporate the capability to arrive at the ultimate failure outline, for comparison to the location of the bedrock surface. / This research has shown that: plug failures occur along steep, uninterrupted discontinuities bounding large blocks. Plug failure potential reduces substantially with confining compressive stress, discontinuity inclination, the absence of low friction surfaces and shearing of intact rock interrupting the discontinuity. Ravelling requires little peripheral confining stress for stabilization and prevention of block falls or slides. However, inherent conditions such as shallow dipping or vertically dipping joints can cause block falls to develop to surface. Low confining stresses, resulting from multiple stope extraction in orthogonal horizontal directions, would offer conditions suitable for plug and ravelling failures. Strata failures are caused by excessive stope spans, but the limited loading received from above strata is such, that the failure cavity created is of limited vertical extent, some 25% of the stope width. Chimneying disintegration occurs in weak rock masses with low cohesion, over narrow openings that can be as deep as 275 m. The onset of chimneying disintegration can be created by compressive stresses, but develops as a result of mobilization of the rock mass by gravity in active shear. Block caving requires large spans to develop, and stabilizing could be overcome from arching stresses overcoming bulk arching strength. Controlling instability elements are tabulated for these failure mechanisms. A limit equilibrium correlation between span and cohesion for chimneying disintegration is presented, and the controlling limits between the occurrence of chimneying disintegration and block caving is discussed. Ravelling and chimneying disintegration are the most expected failure mechanisms for shallow stopes of hard rock mines. Although failure of the shallow stope may start around its periphery, stope failure to surface would likely occur in or close to the surface crown pillar. / Although the analytical equations developed require input of in situ stresses defined by numerical modelling in order to yield a precise answer, conventional numerical modelling or empirical methods are shown to be unable to predict stope failures as the analytical equations have. A step-by-step stability analysis procedure is presented, incorporating rockmass environment, expected failure mechanism(s), and applicable data gathering and anlytical methods. (Abstract shortened by UMI.)

Rockbursts may cause damages to underground openings and to equipment, and constitute a major hazard to the safety of mine workers. One method that can be used to evaluate if there is a rockburst potential is to compare the stiffness of the failed rock with that of the surrounding rock mass. This method has been applied successfully in the past to rockbursts involving fracturing of the rock mass. / This thesis deals with the development of a similar approach for rockbursts involving a violent slip along major geological discontinuities. To evaluate the post-peak shear stiffness of a discontinuity, a new non-linear constitutive model for rock joint was developed. This model is based on two exponential formulations expressing the two phenomena taking part in the shearing process: friction resistance along surfaces and shearing of asperities. Compared with test results, the model showed a correlation factor (R2) of 0.90. The model was then implemented in an existing boundary element code to evaluate the interaction between underground openings and nearby geological discontinuities. Verification of the implementation was done by reproducing direct shear tests on a discontinuity. Parametric analyses were performed on the new model that highlighted the most important parameters. Methods to obtain the different stiffnesses involved in the violent slip process were developed. Examples of applications were given to illustrate the proposed methods. / Finally, an alternative method to evaluate the fault-slip rockburst potential was developed. This new method relies on a linear analysis and the calculation of a new index called the Out-of-Balance Index or OBI. The OBI showed some agreement with the stiffness approach.

A nonlinear finite element model, called MINE-PAL, has been developed for elasto-plastic, large deformation analysis of underground and surface mining excavations. The model accounts for both material nonlinearity due to yielding and strain-hardening and strain-softening effects and geometric nonlinearity due to large deformations. The finite element equations are formulated using incremental plasticity theory with Total Lagrangian approach where the Kirchhoff stress tensor and Green strain increment tensor are used. The model is developed for use on microcomputer system and is designed to perform 2-D nonlinear analysis of mine excavations in multiple-material domain, thus predicting the distribution of elastic, strain-softening, strain-hardening and residual strength zones. The model accommodates any arbitrary stope geometry and it allows for a variety of mining-type loading conditions in situ stresses, gravity loading and boundary tractions. Special provision has been made to permit the simulation of backfill as in situ stress-free material. Stress-space strain-softening is invoked into model using the concept of damage parameter in the incremental stress-strain relation. / In addition to model verification and sensitivity analysis, two numerical modelling case studies have been undertaken. The first case study is of a narrow vein mine with backfill which is located in the province of Karnchanaburi, Thailand. The second case study is about a room-and-pillar design of a potash deposit in the province of Chaiyaphum, Thailand.

This thesis proposed a petrofabric classification of the rock mass domains and rock units encountered at the Birchtree Mine in the Thompson Nickel Belt (TNB), Manitoba. A program of field and laboratory studies has enabled the classification to be related to both geological and physical-mechanical properties. The classification has been aimed to advance TNB mining, mineral exploration and processing research. / The thesis also considers the genesis and nature of both inherent and mining-induced rock mass damage at the Birchtree mine. An inherent rock mass damage index, D$ sb{ rm I}$, has been derived, based upon the rock unit intact strength, together with the quantity and conditions of any meso- and macro-structure present. D$ sb{ rm I}$ was closely related to rock mass domains. The physical characteristics of rock mass damage in mine developments induced both through drill-and-blast and roadheader mining were also monitored at the Birchtree Mine. This permitted the formation of a mining-induced rock mass damage index, D$ sb{ rm M}$. This was based upon decrease in the intact strength, half cast/tool marks, normalized scaling time, drift condition rating, and orientation of local structure. The type and intensity of mining-induced damage was found to be clearly related to the nature of the rock mass units and domains, as evident in their inherent damage. / Two types of mining-induced damage mechanism were evident at Birchtree. Firstly, in the inherently weak rock mass ($ rm D sb{I}10$), such as the Metasedimentary Domain, the rock units are foliated with wider spaced joints. More energy is evidently required to fragment the rock mass through what was observed to be predominantly intact rock breakage controlled by foliation. These two damage mechanisms were also clearly evident from analysis of drift and fragmentation morphology survey data. The damage mechanism associated with both forms of mining appeared to be similar, although the intensity of machine-induced damage was significantly less. Rock mass damage related to ground stress redistribution was also observed. / This thesis was motivated by the need to understand, control and minimize mining-induced damage, particularly in serpentized ultramafic domains which will host most of the future TNB mining at depth. It thus concludes by reviewing a proposed strategy for rock mass damage sensing and control in underground development and stoping, based upon the experience at Birchtree.