Search Documents

By I. D. Nock, R. M. Parkin

Paper presented at a meeting on: Artificial intelligence in the minerals sector, held in Nottingham, UK, 20 April 1993 (original title: Knowledge-based condition monitoring of cone crushers). A large proportion of the electricity used by the quarrying industry is consumed in the comminution and sizing process. In most cases aggregate processing plants operate with minimum control features resulting in limited operational efficiency. Improved control is required to increase operational efficiency with concomitant savings associated with reduction in re-circulating load and improved resource utilisation. Development work has suggested that a combination of adaptive control and process condition monitoring could improve control of comminution plants. Many commercial process control and condition monitoring systems are available, but none were deemed suitable due to the domain specific characteristics of the application. An assessment of the particular characteristics of cone crushing concluded that the application was best suited to the use of small domain specific process control combined with bespoke condition monitoring. The processing power of the control system for the project is a function not of the academic requirement, but rather the economic and practical constraints imposed by the application. As a result of the constraints a high-end PC system has been designed in preference to a workstation. Within the research project work packages are sub-divided into novel wear sensor development, knowledge based control/condition monitoring and the investigation and mathematical modelling of the wear mechanisms inherent in cone crushing. Numerous strategies were considered to implement the control system outlined. The system currently under development is a modular design which utilises intelligent novel sensors to provide a variety of data inputs

Apr 1, 1994

By R. D. Stewart

Paper presented at the conference: Risk assessment in the extractive industries, held in Exeter, UK, 23-24 March 1994, and first published London: IMM, 1994, preprint volume, 18pp., 5 refs. A seismic risk assessment method has been developed for application to rockburst prone deep-level mining sites which have sufficient seismic coverage. The technique utilises several seismic hazard parameters simultaneously and assesses the seismic risk on a daily basis according to the individual levels of these parameters. Critical levels are determined from investigation of the long-term behaviour of the parameters in relation to recorded potentially damaging seismic events. The method has been applied to a data set comprising 15 months' worth of seismic data, with encouraging results

Aug 1, 1995

By S. Henley

Paper presented at a meeting on: Artificial intelligence in the minerals sector, held in Nottingham, UK, 20 April 1993 (original title: The database: a pre-requisite for large-scale use of AI in mining). Artificial intelligence is a term which has been widely applied to different techniques including rule-based systems using inductive and deductive logic algorithms, neural networks, learning systems and genetic algorithms. In principle it should be applied to adaptive statistical techniques. Some of these applications have been coded in languages which are specifically suited to the novel requirements of AI but are not easy to integrate with more traditional computing methods. Efforts have been made to bridge the gaps, particularly with database management systems and such efforts at integration should provide the key to future success in the application of AI. A field which has been termed "knowledge engineering" cannot be credible as such if it fails to provide easy access to raw materials of "knowledge" i.e. the data residing in databases. In the mining industry there are a number of significant areas in which AI could usefully be applied throughout the industry, but its advance is slow because of the lack or inefficiency of such database links. Current development of object-oriented database systems for geology and mining holds out hope for such integration in the foreseeable future

Apr 1, 1994

By J. A. Hudson, D. L. Millar

Paper presented at a meeting on: Artificial intelligence in the minerals sector, held in Nottingham, UK, 20 April 1993 (original title: Rock engineering system performance monitoring using neural networks.) Synthetic and analytical approaches to objective-based design of rock engineering projects are discussed. Rock masses are considered as complex dynamic systems with components that concurrently interact with each other and their environment. The existence of a rock engineering system canon (i.e. a set of systems governing rules) is proposed. Its possible components are highlighted and an application is demonstrated using a "rock mass automaton" designed to simulate dynamic heat distribution in rock. In the example given, the canon implements the basic result from probability theory applied to the thermodynamics of solids. Rock engineering system performance is reduced to the mapping of the output of a rockmass to varying input (design) options, within the state space of the systems model. The Kohonen Self-Organising Map provides an extremely useful analytical tool to reduce the hyper-dimensionality of rock engineering system behaviour. Further, the hyper-dimensional state vector is mapped "quasiconformally" such that the neural map tends to spatially organise similar input state vectors, facilitating interpretation of rock engineering system behaviour (performance monitoring). Work is reported on the development and validation of the automation as a basis for an internal representation of the rock mass bahaviour. The results are reported of simulations compensating for the uncertainty of the exact mechanisms operating within the rock mass, using additive noise with the input state vector in the training stages of the self-organising map implementation. Initial work on adopting Grossberg's Adaptive Resonance Theory neural network architecture for classification of rock engineering systems behaviour is also presented

Apr 1, 1994

By P. J. Rowsell, M. D. Waller

Paper presented at a meeting on: Artificial intelligence in the minerals sector, held in Nottingham, UK, 20 April 1993. The applications of artificial intelligence systems in the upstream petroleum industry are reviewed. Particular emphasis is placed on current and potential applications in drilling engineering. Several recent advances in drilling technology have led to a large increase in the quality and volume of data available. This combined with the escalating cost, particularly of off-shore operations, is testing the traditional reliance on the skill and judgement of rig operators. AI systems offer enormous scope in enhancing and aiding the performance of these operations

Apr 1, 1994

By M. K. G. Whateley, W. Hatton

Paper presented at the conference: Risk assessment in the extractive industries, held in Exeter, UK, 23-24 March 1994, and first published London: IMM, 1994, preprint volume, 20pp., 31 refs. Estimates of coal tonnage on UK open cast mining sites often vary from the recoverable tonnage. This is because the estimates depend on the geological variability of the coal measures. Geological parameters, such as thickness, quality and disturbance of the seam, have a direct control upon this variance. Risk analysis techniques have been applied using commercially available software to combine each parameter's variability into a cumulative variance, resulting in a tonnage distribution. Sensitivity modelling has also been used to identify the geological parameters which have the largest influence on the variance of the simulated tonnage distribution. An understanding of the relationship between the tonnage and geological variability has helped to focus exploration more efficiently. A cost to risk relationship can now be monitored, helping management to avoid negative consequences on site

Aug 1, 1995

By P. A. Davies

Paper presented at the conference: Risk assessment in the extractive industries, held in Exeter, UK, 23-24 March 1994, and first published London: IMM, 1994, preprint volume, 11pp., 15 refs. Hazards such as vibration, dust emission and fragment projection can result in an area called the Danger Zone being set around a blasting site. Traditionally Danger Zones have been set solely on the basis of consequences (e.g. based on the maximum distance of rock projection or expected ground vibration plus a defined safety margin). Advances in risk assessment and its growing acceptance as a safety management tool have led to the setting of Danger Zones based on both consequences and frequency of occurrence. This allows distances to be optimised on the basis of acceptable risk levels rather than the potential consequences of infrequent events. The setting of a flyrock Danger Zone is used to demonstrate a risk-based approach compared with a traditional consequence approach

Aug 1, 1995

By M. Bustillo, S. Ordonez

"Paper presented at the conference Economic geology in Europe and beyond II, models for mineral deposits in sedimentary basins, held in Keyworth, UK, 13-14 April 1994. The Mississippi Valley-type deposits occur in massive, homogeneous dolostones from which diagenesis has obliterated all primary sedimentary structures. The mineralisation forms irregular, lens-shaped stratabound bodies, mostly in the lower part of the dolostones. They are composed exclusively of sphalerite, galena, pyrite, marcasite and barite, with white sparry dolomite a common cementing phase. Mineral zoning can be recognised from pyrite and marcasite in the east to barite in the west. Sphalerite, the most common sulphide, occurs basically as colloform and banded crusts and galena as isolated cubic or skeletal crystals, mainly in the centre of the banded botryoidal sphalerite. The combined observations of the deposits' geology, geochemistry and ore petrology suggest a three-stage genetic model: carbonate sedimentation in a shallow-water environment; dolomitisation, probably after compaction, by fluids of connate origin and lateral evolution; and mineralisation by migrating solutions, perhaps related to oil-field brines, that originated in deep zones of the basin during diagenesis of the basinal mudrocks in the NE"

Apr 1, 1995

By B. Denby

Paper presented at a meeting on: Artificial intelligence in the minerals sector, held in Nottingham, UK, 20 April 1993. The design and scheduling of open pit operations are consistent problems encountered by the surface mining industry. Numerous techniques and algorithms have been developed to assist the mining engineer in the creation of optimum development strategies. Virtually all the practical techniques that have been produced have concentrated on one of the aspects (limit definition or scheduling) at the expense of the other. It has long been recognised, however, that the two activities are intrinsically linked and thus solving either problem independently of the other can only ever produce sub-optimal results as a whole. This has been termed the "circular reasoning problem". It has also been recognised that, for all practical-sized situations, solving both problems simultaneously is virtually impossible using conventional techniques due to the massive size of the search space involved. A technique is presented which is based on a self-learning genetic algorithm, which provides a combined pit limit and extraction schedule simultaneously in order to maximise the NPV (net present value) of the extraction. The theory is developed in two-dimensions and a number of examples of its application are presented. It is proposed that this approach is a highly efficient, heuristic technique offering the potential to transcend the circular reasoning difficulties of separate pit limit and extraction scheduling systems. Extension of the technique into three-dimensions and its application to realistic problem sizes are discussed

Apr 1, 1994

By Xuejing Xie, Tianxiang Ren

Paper presented at the 28th International Geological Congress held in Washington DC, July 1989 (International geochemical mapping sessions). Standardised stream-sediment sampling procedures have been used in mountainous and hilly terrains. Special sampling procedures were developed for application in remote environments, such as high mountains, desert, semi-desert, karst and loess-covered regions. The diversified analytical methods used in more than 20 analytical laboratories participating in the project have been unified into two multi-element, multi-method analytical systems. A unified data monitoring procedure was developed based on primary and secondary standard samples. To date the project has covered about 4 000 000 km2 and the aim is to complete the coverage within the present decade

May 1, 1991

By N. Krauland, A. Nystrom, M. Board, S. Sandstrom

Several deep Swedish cut-and-fill mines experience severe ground control problems, including roof collapse and wall sloughing, as a result of variable rock quality and high in situ stresses. A project to examine the problem included two main areas. The first was prediction of mining conditions at depth and the adaptation of cut-and-fill mining to such conditions; the second was the assessment and improvement of support methods at present and future depths. Simple, elastic, mine-scale numerical models were used to optimise stope sequencing and more detailed, stope-scale models to simulate the failure processes around a stope and the interaction of rock mass and support. Kristineberg was chosen for testing and calibration of the predictive methods because of its long history of severe ground conditions and the large base of observational and monitoring data available for comparison. Horizontal cut and fill was found technically suitable to depths beyond 1 000 m and economically viable if ore value covers support costs. Stability can be achieved with the present support technology down to 1 000 m but the 7-strand 15.2 mm cables have poor deformation properties (5% failure strain) compared with rebar rock bolts (17%). Reducing the intervals between levels from 100 m to 50 m will cut support costs and improve production reliability and capacity. Simulation of sill-pillar mining indicated that conditions would be most severe when pillar height was reduced to 15-20 m; mining of a sill pillar has been carried out successfully to the last 3 m, whose extraction is planned for the near future

Dec 1, 1995

Paper presented at the UK minerals industry conference and published in: Minerals extraction towards the millennium. UK minerals industry conference, 3-5 April 1995, University of Leeds. London: Institution of Mining and Metallurgy, 1995. The various types of risks that arise in minerals industry projects are usually financial, technical, or environmental, and the key to valid risk analysis is complete integration of all risks, both within and between categories. Geographical information systems (GIS) and geostatistical simulation methods can provide a means of integrating data and conducting realistic risk analyses. Risk can be quantified but there are limitations to numerical approaches to risk assessment. Realistic quantification requires adequate models of the processes that give rise to risk. Risk perception must also be taken into account when conducting risk assessment and reporting the results, especially in public forums. The modelling and visualisation capabilities of a combined GIS/simulation system can provide the means of dealing with perception in risk analysis

Jun 19, 1905

By A. J. Davidson

A Cu-conducting solid electrolyte was used to develop the sensor. Instead of measuring the potential across the electrolyte from the reference material to the Cu-containing phase, a potential was applied and the stable current measured. This gave reasonable readings for copper-tin alloys, which agreed with the theoretical calculations, but was less satisfactory in the case of mattes because the surface area of electrolyte presented to the matte phase varied from sensor to sensor. To overcome this problem the potential was applied, the current was measured and then reversed and remeasured and the ratio of the currents was calculated. The current ratio is shown to be independent of the resistance of the electrolyte and the surface area of the electrolyte offered to the melt, but strongly dependent on the activity of the copper

Jun 21, 1905

By A. P. Whittlestone, J. D. Johnson, R. J. Pine, M. E. Rogers

standpoint. The effect on efficiency of collisions

Apr 1, 1995

By R. M. Cooper, H. K. Herbert, D. N. Phillips

A study was carried out to determine the effects of adding CaCO3, Fe, Al and CaSi2 to the ladle, rather than to the raw materials, in order to increase the concentration of Ca, Fe and Al in the final product to meet customer specifications at Simcoa's Kemerton plant in Western Australia. LA-ICP-MS, SEM-EDXRA and ICP-AES analysis of samples and controls showed that the technique is suitable for controlling minor element concentrations. Iron was successfully added, while calcium silicide with aluminium proved better than limestone with aluminium for controlling Ca and Al concentrations in the silicon

Jun 19, 1905

By M. Gallego, R. Oyarzun, L. Ortega

Evidence of metallogenic activity in an area dominated by Upper Precambrian and Palaeozoic metasediments was partially explained in the 1980s when drilling and a geophysical survey detected hidden intrusions, but questions remained to be answered about the nature of the fluids. Sb mineralisation at El Juncalon vein-type deposit was selected for geological, geochemical and fluid inclusion studies as a model with a relatively simple quartz-stibnite-pyrite paragenesis. Many of the characteristics were those of Chinese-type vein Sb deposits, which can be regarded as atypical of the epithermal type as they are hosted by clastic rocks

Aug 1, 1995

By G. S. Pauley

Paper presented at a meeting on: Artificial intelligence in the minerals sector, held in Nottingham, UK, 20 April 1993 (original title: Case-based reasoning can be a catalyst for improvement). A new computer software technology known as Case Based Reasoning (CBR) can be used to produce small decision support systems that are economically viable. This is because the knowledge that they contain can be owned and maintained by their users. A CBR tool has been used to perform a pilot study with Anglesey Aluminium Metal Ltd. The results are encouraging. They suggest that a combination of CBR and participative design and implementation techniques (such as Checkland's Soft System Methodology) can achieve sustainable improvement in mineral processing functions. An explanation is presented of why traditional, rule based systems are often an ineffective way to implement support systems. It is shown that CBR may provide an appropriate technology solution. Methods of achieving sustainable benefit are indicated. The pilot study with Anglesey Aluminium Metal is used as a practical illustration of the development of a CBR application. A review of potential application areas and some observations about the implementation are presented

Apr 1, 1994

By G. Kill

Although other energy commodities have developed successful futures contracts, long-term contracts still dominate the coal market. To establish whether coal markets will develop in the same way, coal price and volume data are tested against a set of criteria considered to represent some of the factors contributing to the development of a successful futures contract. Coal meets only some of the criteria: market players are subject to little risk in quite a stable market, have strong preferences for the specific nature of long-term contracts and profit from product uniqueness, each trade being individually negotiated. Potential structural changes in the western European market are not felt to be sufficient to create demand for a coal futures contract, but an empirical model has been developed that could form the basis of a further study to predict the potential for success

Aug 1, 1994

By J. A. Finch, D. Lin

The robustness of the large-sections correction procedure, Hill's fast approximation, Barbery's correction and the PARGEN correction was examined by comparing their ability to reconstruct the true liberation distribution using sectioning data from computer-generated spheres. The results indicate that the flexibility of a correction procedure is related to the assumptions it makes about the continuity of the liberation distribution, locked-particle characteristics and ore texture. Overall, the large-sections procedure appeared to be the most generally applicable. The Hill fast approximation performed well, but this was probably because it partially incorporates the sphere model; the performance of the PARGEN and Barbery corrections varied from case to case.

Dec 1, 1995

By M. W. Robotham, H Wang

removal efficiency is analysed from a stochastic

Apr 1, 1995