Search Documents

By Gordon W. Hines

This paper, dealing with network planning in an industrial complex, discusses the type and size of project that can be handled on a manual basis. After a review of the fundamentals of network planning, and specifically of the Critical Path Method, the author analyses several different types of projects, each with its own characteristics. Some can be handled in conjunction with a computer, but others can be dealt with equally well without one. Among the projects discussed are a blast furnace reline, the installation of a billet mill flying shear, and routine maintenance-type operations.

Jan 1, 1967

By J. Partyka

Adequate ventilation is a main concern if Lockerby Mine near Sudbury is to be expended. This study investigate a two-fold problem: (1) the new pressure and volume requirements of the main fans due to the theoretical. deepening of the mine, and (2) the effect of the natural ventilation pressure on the main fans. Field data was analyzed with the aid of a newly developed network simulation program, VNETPC Version 3. 0. The density change throughout the mine is accounted for by performing all computations at reference density. Results of the investigation indicate that the existing fan system must be supported by three to five additional underground fans as well, the density effect is significant. If neglected, it will greatly alter the simulation results

Jan 1, 1991

By H. Hamza, M. R. MacIver, M. Pawlik, L. Malin

"Size and shape are important properties of aggregates formed in engineering processes and natural systems. Measurement of aggregate size or shape is a challenge when the solids concentration is beyond the gel point, ?? > ??g, when the aggregates are in contact and it is difficult to distinguish or segment individual particles in the network. This paper describes the use of a micro-computed tomography (micro-CT) method for the calculation of the aggregate volume fraction, ??, and a characteristic aggregate length, ??a, based on a 3D chord length measurement algorithm. These parameters can be used to quantify changes in aggregate and pore size during a compressive yield stress test to help understand the impact of flocculation and network structure on dewatering. INTRODUCTION In a thickener, flocculation is often used as a pre-treatment step and the final solids content of the underflow depends on the network strength and permeability of the bed formed by settled aggregates (Usher, Spehar, and Scales 2009). This paper describes the measurement and analysis of micro-computed tomography (micro-CT) of flocculated mineral sediments to obtain a characteristic length, ??a, of the aggregate and pore phases during a staged compressive yield stress experiment. This parameter may be used as an indicator of changes in aggregate and pore size to quantify material behaviour in solid-liquid separation processes that depend on self-weight compression or applied pressure. This analysis was carried out to help understand the effect of flocculation on aggregate structure and the thickening process as part of a broader research on the improved solid liquid separation of tailings produced in the oil sands industry.Aggregate properties The solids phase of an aggregated or flocculated suspension will eventually settle – if left undisturbed – under the influence of gravity. As settling of a flocculated suspension continues, the solids volume concentration, ??, may increase to the gel point or gelation threshold (Landman and White 1994), ??g, above which the aggregates form a network of touching particles, or space-filling structure (Kranenburg 1994). In the sediment bed, the aggregates are surrounded by fluid and occupy a partial volume fraction, ??, according to"

Jan 1, 2018

By J. F. Archibald

This paper reviews the formation of the Canadian Mining Education Council (CMEC) and its terms of reference. CMEC considers that Canada?s mining schools represent a distinct competitive advantage to the Canadian Mining Industry. The core objective of CMEC is to reinforce the links between the constituent schools and their individual Industry Advisory Committees. The latter traditionally provide the schools with a great deal of guidance on a range of planning issues, including recruitment, curriculum, research and infrastructure. Beyond those bounds CMEC intends to improve its ability to network with the industry and government. It is concerned overall with the objective to raise the quality and level of its educational and research contributions to meet the needs of industry and society.

May 1, 2002

By F. B. Foley

A description of tests made by a committee of the Division of Engineering of the National Research Council to determine whether velocity of impact affects the formation of Neumann bands. FOREWORD Not infrequently, in the case of the failure of a metal structure, such as a bridge, tank, airplane, gun carriage, etc., a doubt arises whether the failure was due, among other causes, to the effect of an overburden or of an explosive maliciously or accidentally exploded in contact with it. As few means of discriminating between the effects of these causes have been known and these means were of a rough qualitative character, it was surprising to find in the record of an inquiry into the failure of a structure, the statement that, by observations on the Neumann bands in the structure, it was possible to determine the speed of breakage in the metal and to distinguish between a fracture caused by an explosion and one caused by an ordinary mechanical strain. Such a criterion, if well proved, was of value to the explosives experts of the Bureau of Mines, the Ordnance Department of the United States Army and the Bureau of Ordnance of the United States Navy, but search showed no record of it in the literature nor did any of the metallurgists and metallographers consulted know of any data on which such a criterion could be based having been assembled and critically studied. However, it was universally held that there was promise in the idea and that there existed unique facilities at the Bureau of Mines Experiment Station at Pittsburgh for the preparation, under definitely known conditions, of material for study by metal-lographers, and that material of value might also be obtained from the proving grounds of the Army and Navy. Therefore, with the approval of the Director of the Bureau of Mines and the Chiefs of Ordnance of the War and Navy Departments, the Chairman of the Division of Engineering of the National Research Council, on April 12, 1919, appointed a Committee on the Investigation of Neumann Bands, which has since been engaged in this study and which, after some slight change in its personnel, consists now of the following: Dr. William Campbell, Advisory Metallurgist, Bureau of Ordnance, Navy Department. F. B. Foley, Metallurgist, Bureau of Mines. S P. Howell, Explosives Engineer, Bureau of Mines. Comdr. O. M. Hustvedt, U. S. N., Bureau of Ordnance, Navy Department. Dr. Charles E. Munroe, Chief Explosives Chemist, Bureau of Mines.

Jan 9, 1922

By KeBler H-W, Seeliger A

Vibration analysis for diagnosis of machines is a powerful instrument for condition monitoring. Especially for diagnosis of antifriction bearings various equipment and techniques have been developed and are available. The envelope analysis is the most reliable method for bearing diagnosis. Different bearing failures lead to different patterns of the envelope spectrum. Checking the amplitudes of characteristic bearing frequencies is mostly not sufficient for a reliable diagnosis. Therefore human experts compare the pattern of the spectrum with typical patterns of bearing defects. In this paper a diagnosis system is presented which imitates this action of human experts. Using a neural net the system not only distinguishes typical bearing defects but also detects unbalance, beating of disengaged machine parts and other kinds of machine failures.

Jan 1, 1995

By D. H. Rubisov

The neural networks modelling approach is one of the most elegant black-box type methods to model industrial processes. In this paper, this approach is applied to model the performance of the CEZ zinc electrowinning plant. The model that was built was found to exhibit an excellent predictive ability for voltage and current density. The model was also used to investigate the sensitivity of the process to various key variables such as temperature, current density, zinc and sulphuric acid concentration, etc. Optimal regimes for zinc production were identified from the power consumption standpoint.

Jan 1, 1996

By G. F. Friel, J. C. Edwards

A series of seven underground-coal-mine fire experiments was conducted in the Safety Re-search Coal Mine at the National Institute for Occupational Safety and Health, Pittsburgh Research Laboratory. Coal and styrene-butadiene-rubber conveyor belting were burned upwind of two sensor stations, 18 m and 148 m from the fire source. Exhaust from a diesel locomotive flowed over the fire sources in six of the tests. Metal-oxide-semiconductor (MOS), CO, and optical-path-smoke sensors were positioned at both stations and found to be an optimum set of sensors for the fire discriminations. A representative set of 7,679 samples of CO data and data from the smoke and diesel-exhaust MOS sensors were used as inputs to train a neural network (NN). By testing 42,538 data samples from the seven experiments, all fires were detected by the NN within 9.67 min from the onset of significant changes in the MOS voltages without any false alarms.

By S. Katz, V. Stanek, K. L. Moore, E. D. Larsen, D. E. Clark, P. E. King

"Cupola melting is a complex physical and chemical process having a number of nonlinearities which make conventional process control difficult. In addition, sensors for both inputs and outputs of interest to the foundry operator are often difficult or impractical to apply in a production situation. The present work is an effort to control the process using both intelligent and conventional control techniques as appropriate. Neural networks are used as an inverse process model in a cascade, feedforward controller in conjunction with a finite difference model of the cupola. A separate implementation of the neural net serves as a stand-alone visualization tool; it can operate on a laptop computer and is a useful tool for the cupola operator. IntroductionThe goal of the present work is to develop a controller to improve the efficiency of cupola melting. Although cupola melting as a process has a long history, automatic control has been elusive because the process has been poorly understood. The purpose of this four-year program, which has been underway for about six months, is to develop a controller for the cupola process using intelligent and conventional control methods.The Cupola ProcessThe cupola furnace [1,2] is one of the oldest processes for making cast iron, and is still the dominant method. It takes the form of a water-cooled vertical cylinder, as large as 4 m in diameter and 20 m tall. The charge consists of scrap metal, alloying ingredients, limestone, and coke, which serves as fuel; it is composed by weight and fed through doors near the top of the cupola. Air, usually heated and often enriched with oxygen, is blown in near the bottom through tuyeres extending a short distance into the interior of the cupola. As the coke is consumed, the charge drops and melts. A continuous flow of iron emerges from the bottom of the furnace, at rates as high as 100 tons per hour, and. is transferred to casting lines in the foundry. The exhaust gases are drawn from the top of the cupola, where they are processed to meet pollution standards."

Jan 1, 1994

By L. A. Puchkov, Mesentsev. V. K., I. O. Temkin

The problem of increasing underground mine ventilation efficiency is still of top priority. The complication of geological conditions with the depth of underground mining processes and reinforcement of requirements to technological and ecological safety are extorting the necessity of solving this complex problem with using of new information technologies. There exist different approaches to the ventilation problem. One of them, which the authors adhere to, is based on the concept of monitoring (under special mining and geological conditions-automatic control) of mining atmosphere and operative re-distribution of underground mining air among consumer , in accordance with the criterion of optimality [ 1.3 1. ] The set of tasks which should be decided in the frame of con- crete ventilation system depends on the peculiarities of deposit, technological schemes and so on. Nevertheless, some ventilation tasks such as: defining of ventilation network parameters; finding of dependence between these parametrs; operative redistribution of air streems, could be interpreted as universal ones. Main aspects of practical solving of different ventilation problems and their classification were investigated and considered in detail before [I ,2 ]. Some years ago we have began our investigations to develope quite simple and effective methods for decision of ventilation tasks on the base of neural network technology. It is well known that ventilation net, formed by various mine workings, is the main component of underground mining ventilation system. In general, network air distribution may be described by some functional F(Q,R,H), where Q- is a vector of air consumptions; H- is a vector of depressions; R- is a vector of airdynamic resistences. (Q,R,H) - are the informative entities , characterising every branch of the net. There exist some fundamental tasks which should be decided during the process of control of mining ventilation: calculating of natural air distribution Q = Q,{R, H,}, or Q = Q2(R, Q?}, where H, - is the maximal depression of the ventilation net, Q, - is the air consumption in the ventilation shaft; identifying of airdynamic resistance for some branches on the base either of air-depression removal or operative monitoring: [ ]

Jan 1, 1996

By C. Cho

One problem with the application of neural networks to the multivariable control of mineral and extractive processing is deciding whether and how best to use them. The objective was to compare neural network control to more conventional strategies and find any advantages in terms of set point tracking, rise time, settling time, disturbance rejection, and other criteria. The procedure was to develop neural network controllers using both historic plant data and simulation models. We tried various neural network control patterns including both inverse and direct neural network plant models. These were compared to state space controllers that were, by nature, linear. For grinding and leaching circuits, a nonlinear neural network-based model predictive control strategy was superior to a state space-based linear quadratic Gaussian controller. We also learned the importance of incorporating state space into neural networks by making them recurrent--feeding certain output state variables into input nodes in the neural network. We concluded that neural network controllers can have better disturbance rejection, set point tracking, rise time, settling time, lower set point overshoot, and can be more foolproof and easy to implement in complex, multivariable plants.

Jan 1, 1996

By Darryl Amick, Keith A. Prisbrey, Lee E. Plansky

The problem addressed was to develop automated image analysis of metal welds to replace unreliable and costly 100% visual inspection. The objective was to evaluate neural networks combined with features from gray level surfaces. Photomicrographs of the welded seam in Zircaloy-4 channels used in nuclear fuel rod assemblies were examined. The methodology used was to select a set of morphological feature coefficients from Bessel Fourier morphological feature coefficients extracted from digitized photomicrographs. The selected morphological feature coefficients were then used to train a neural network. The approach was verified initially by successfully training a neural network using the known, secondary chemistry of the alloy welds. The results were good. The Bessel Fourier coefficients captured sufficient information from the metallographic images to distinguish weld quality. This approach has the combined advantages of compressing the images through Bessel Fourier coefficients and give the ability to rapidly process large amounts of data using neural networks.

Jan 1, 1993

By S. L. Jung, T. L. Nichols, K. Prisbrey

The problem was to evaluate the on-line detection of rock properties through acoustic emission sensors. Acoustic emission sensors, when attached to crushers, grinders, and roof bolters, are robust, economical, and relatively maintenance free. Our objective was to relate acoustic emissions to rock fracture mechanisms, including micropore closures, linear elastic deformations, and crack propagation. We evaluated acoustic emissions from steel jacketed rock cores during indentation tests. In addition we attached sensors to the rotating shaft of a rock bolter drill. A back propagation neural network "learned" rock fracture characteristics and rock type with 99% classification accuracy. A neural network's calibration and computational abilities suggest better acoustic emission sensor capabilities.

Jan 1, 1993

By J. T. Gepford, K. A. Prisbrey, T. Scott, M. Spangler

Using neural networks for plant control is challenging. The objective was to show how to make a neural network effective, even with noise, lag times, and unmeasured disturbances in the plant. The procedure was to evaluate the neural network in a laboratory flotation vision system and in a plant refractory gold autoclave system. Although the resulting neural network easily conquered simple laboratory data, the noisy plant data was much more difficult, and adaptive linear models were equally effective. We concluded that the neural network needs support by combining it with other tools found in the best control packages.

Jan 1, 2000

The relationship between neural network (NN) performance and the parameters that form the architecture of the neural network is complicated. This paper studies an issue that is still controversial—the effect of the number of neurons and number of hidden layers on NN performance. This paper tests if the above parameters affect network performance when the NN is trained according to the quick stop strategy. It was determined that the number of neurons did not have any effect on the NN performance for a single layer or a double layer network. It was also found that the training error and calibration error could be poor indicators of the prediction correlation coefficient.

Jan 1, 2003

By Walter K. Utt

The process of drilling and bolting the roof is currently one of the most dangerous jobs in underground mining, resulting in about 1,000 accidents with injuries each year in the United States. To increase the safety of underground miners, researchers froth the Spokane Research Laboratory of- the National Institute for Occupational Safety and Health are applying neural network technology to rite classification of mine roof strata in terms of relative strength. In this project, the feasibility of using a monitoring system on a roof drill to assess the integrity of a mine roof and warn a roof drill operator when a weak layer is encountered is being studied. Using measurements taken while a layer is being drilled, one can convert the data to suitably scaled features and classify the strength of the layer with a neural network. The feasibility of using a drill monitoring system to estimate the strength of successive layers of rock was demonstrated in the laboratory.

By R. Vance, C. Cho, N. Mardi, Z. Qian, K. Prisbrey

One problem with the application of neural networks to the multivariable control of mineral and extractive processes is determining whether and how to use them. The objective of this investigation was to compare neural network control to more conventional strategies and to determine if there are any advantages in using neural network control in terms of set-point tracking, rise time, settling time, disturbance rejection and other criteria. The procedure involved developing neural network controllers using both historical plant data and simulation models. Various control patterns were tried, including both inverse and direct neural network plant models. These were compared to state space controllers that are, by nature, linear. For grinding and leaching circuits, a nonlinear neural network-based model predictive control strategy was superior to a state space-based linear quadratic gaussian controller. The investigation pointed out the importance of incorporating state space into neural networks by making them recurrent, i.e., feeding certain output state variables into input nodes in the neural network. It was concluded that neural network controllers can have better disturbance rejection, set-point tracking, rise time, settling time and lower set-point overshoot, and it was also concluded that neural network controllers can be more reliable and easy to implement in complex, multivariable plants.

Jan 1, 1998

By Lynn B. Hales, Randy A. Ynchausti

Introduction There continues to be a large economic opportunity associated with improving process performance through advanced process control systems within the minerals industry. The basis for this opportunity lies in the inherently low efficiencies of many of the unit operations found in mineral processing plants as well as the continual variation that is present in the ores being processed. The ageless target of such control systems is improved recoveries and concentrate grades, higher processing rates and lower unit costs. Many plants are enjoying the benefits of modem, well functioning process control systems but few if any have achieved plant-wide optimization in an economical sense. As new process modeling and control technologies, such as neural networks, continue to evolve and develop, the potential for plant- wide optimization is becoming a closer reality. Neural network technology appears to have a significant potential in the general field of process control and control of mineral processing operations. Neural networks are an emerging technology that have been evolving out of the study of biological systems since the early 1940's. The first software simulation of neural computing networks was reported in 1956 by Nathanial Rochester of IBM Research during the first conference on artificial intelligence which was sponsored by the Rockefeller Foundation. Neural Networks Neural networks can be considered to be a class of computer program that perform their problem solving functions based on a rudimentary model of the human brain and how it solves problems. Traditional computer programs and even computers themselves solve problems or process data sequentially and in a deterministic manner. The human brain, on the other hand, processes data and solves problems in a massively parallel fashion. We think of computers normally in terms of having a single processing unit, sometimes a co-processor to assist in certain activities such as mathematical functions or communications, bytes of memory, processing speed and program/data storage size.

Jan 1, 1992

By R. H. Estrada-Ruiz

This work analyses a new approach to estimates bubble size distribution of froth surfaces using artificial neural networks (ANN).Also, the robustness of ANN to interpret images with illumination perturbations, produced by light problems or dirt attached to the window of the video camera is evaluated. The experimental work was carried out in a laboratory flotation column, instrumented with an image acquisition system. The images were processed making use of a perceptron model with a hidden layer, sigmoidal transfer function and unitary bias, and the ANN trained with a back propagation algorithm. The results of validation show that ANN are reliable for learning and producing generalized predictions of the froth mean bubble diameter and bubble size distribution, when the model is trained using a database that contains information on the illumination intensity. Keywords: image analysis, neural networks, bubble diameter, bubble size distribution.

Jan 1, 2009

By Wankun Wang, Fuchun Wang

Based on the study of artificial neural network, the neural model was established for the prediction of germanium extraction from zinc oxide dust by microwave alkaline roasting-water leaching. Alkali-material mass ratio, microwave heating temperature, liquid-solid ratio, aging time, leaching time and leaching temperature were the significant factors for the process. The results indicated that the neural network prediction model was reliable, and the forecast values fitted well with the actual experimental values. The model could be used to predict the regeneration experiments with high credibility and practical significance. The accuracy of convergence of the model reached 10−5.

Mar 1, 2018