Search Documents

By A. Patri, D. S. Nimaje

Wireless sensor networks and wireless communication systems have become indispensable in underground mines. Wireless sensor networks are being used for better real-time data acquisition from ground monitoring devices, gas sensors, and mining equipment, whereas wireless communication systems are needed for locating and communicating with workers. Conventional methods like wireline communication have proved to be ineffective in the event of mine hazards such as roof falls, fires etc. Before implementation of any wireless system, the variable path loss indices for different workplaces should be determined. This helps in better signal reception and sensor node localization, and also improves the method by which miners carrying the wireless devices are tracked. This paper proposes a novel method for determining the parameters of a suitable radio propagation model, which is illustrated with the results of a practical experiment carried out in an underground coal mine in southern India. The path loss indices, along with other essential parameters for accurate localization, have been determined using the XBee modules and ZigBee protocol at 2.4 GHz frequency.

Jan 1, 2015

By K. Lombard

"‘Roof support awaiting time’ (RSAT) is a term used at Goedehoop Colliery’s Simunye Shaft to describe the potential production time lost due to the continuous miner (CM) standing idle waiting for roof support to catch up. Investigations revealed that in 2013, Simunye Shaft had approximately1400 hours of RSAT, which suggests that the mine could have potentially produced an additional 280 000 t of coal. This project consisted of two parts. Firstly, the causes of the high RSAT and means to improve the situation were investigated. Secondly, as insisted by mine management, the CM cutting sequence was investigated as a possible cause of high RSAT. Machine-related challenges due to the roofbolter installing support too slowly, geological conditions (mostly hard roof conditions and slips),logistical challenges pertaining to the CM cutting sequence, man-related challenges related to operator fatigue, re-support, operator inexperience, and the absence of support targets were identified as main contributors to RSAT. Furthermore, results showed that the roofbolters in the sections at Simunye Shaft are slower than the CMs. A target of 28% reduction in RSAT was set. Experts from Kennametal and Fletcher were consulted to find solutions for the identified causes. In total, eight solutions for RSAT were identified, but the solution that contributed most significantly to reducing RSAT was to use hard roof drill bits as a standard product at Simunye Shaft. Calculations showed that by using hard roof drill bits, RSAT can be reduced by 43%, which is more than the specified 28% target. The cutting sequences at Kriel, Greenside, and Simunye Shaft, togetherwith three newly developed cutting sequences, were simulated using the UCMS (Underground Coal Mining Simulation) program. A re-aligningprinciple was incorporated into the newly developed cutting sequences toalign the cutting sequences to general support work and to reduce RSAT. A decision matrix revealed that a cutting sequence in which boxing takesplace in R3 (third road to the right of the belt road) and in which the re-aligning principle has been incorporated will be the best option for Simunye Shaft. The recommended cutting sequence will lead to a 5% increase in production.The work presented in this paper was carried out as partial fullfilment for the degree BEng (Mining Engineering)"

Jan 1, 2015

By G. M. Cresswell

The 2 360 t/h two module Phola coal preparation plant represents anew era of large plant in the Witbank coalfield designed to extend its life to 2020 and beyond. The plant uses the optimum processing model for Witbank coals, receiving a feed coal size of -50 mm, wet screening it into coarse and small coal fractions 50 x12 and 12 x0.63 mm before processing them in parallel through two primary DM cyclone sections, in order to make a 27.5 MJ/kg export floats product. The sinks are processed through a common high density DM cyclone section to produce a 21.5?23 MJ/kg thermal coal middlings product. The plant design incorporates the largest size DMS equipment used to date in South African coal preparation plants, i.e. 1 150 mm large diameter pump fed DM cyclones and 4.2 m width banana screens. The fine coal is conventionally processed by spirals, whereas the raw coal slimes are filtered using automated plate and frame filter presses and added to either the middlings or discards depending on its quality. The paper describes the database, innovative design concepts, construction, and commissioning.

Jan 1, 2010

By C. Righi, R. Carli

In modern steel continuous casting technologies mould fluxes are well-known to assure lubrication required by tribological system of mould-solidifying strand shell, and to provide crystalline phase solid film, contributing to control heat extraction between strand and mould wall. Actual cooling rates during normal continuous casting mould operations are far higher than critical cooling rate of the liquid slag. Indeed, DTA measurements point out that even relatively small cooling rates of 30?50 K/min are high enough to give exiguous formation of crystals. In this paper, contribution of devitrification in defining crystalline phase formation of industrial basic mould slag, used in continuous casting of crack-sensitive steel grades, has been investigated by means of non-isothermal experiments study. Devitrification has proved to play a major role in mould flux crystallization in a working environment explaining the higher amount of crystalline phase formed in plant sample in comparison with lab sample from DTA analysis with typical cooling rate of 5?20 K/min. The amount of crystalline phases formed ranks as follow: Cuspidine > Combeite > Gehlenite. Kinetic analysis of DTA non-isothermal experiments has provided kinetic parameters through Kissinger and Ozawa methods, as Avrami constant resulting to be n ?2 and activation energy of Cuspidine crystallization resulting to be EC= 315 kJ/mole. The effect of addition to the parent material of oxides as MnO, TiO2, ZrO2, Al2O3 and Fe2O3near to saturation point has proved to affect, in certain cases, crystallization behaviour of mold flux. In particular, MnO, TiO2, ZrO2oxides produce a relevant dump of crystallization of Combeite and Gehlenite phases, while Al2O3and Fe2O3do not appreciably affect crystallization. Key words: Mould flux, crystallization, devetrification, DTA non-isothermal analysis, Kissinger and Ozawa methods.

Jan 1, 2004

By M. Olalde, I. Watson

"The consequences of ineffective mine closure in South Africa are evident from the number of abandoned mines and operations on extended care-and-maintenance, the on-selling of mines to less well-resourced companies to close, and increasing illegal mining activities. However, the data to substantiate these observations and provide insight into the underlying issues has not been available. Through the Promotion of Access to Information Act, a list of mine closure certificates applied for between 2011 and 2016 and a list of certificates granted over the same period for all nine regions of South Africa was obtained. From the analysis of this data, we show that the mine closure system as implemented in South Africa is largely ineffective. Although closure certificates are being granted, these are for prospecting sites and small-scale mines, which have a relatively small environmental impact. No large mines of any environmental significance were relinquished over the period under review, with very few applying for closure certificates. Furthermore, the issuing of closure certificates varies significantly between regional offices, with the success rate for applications being generally low and issuing of certificates taking an extended period. IntroductionThere are a number of concerns with the mine closure process as practiced in South Africa, and these have largely been laid at the door of government. Alberts et al. (2017) stress that the legislation generally conforms to international best practice. However, the system is ‘complex and unwieldy’, involving various pieces of legislation and different government departments with overlapping requirements and different interpretations of the law. A further area of concern is the capacity and competence of government to implement legislation. The ‘shortage of relevant mine closure skills and knowledge within the regulator’ was identified by van Druten and Bekker (2017) as a key contributor to unsuccessful closure. Lack of capacity and resources within the regulator has also been highlighted by Botham, Kelso, and Annegarn (2011) and Milaras, McKay, and Ahmed (2014). A recent study of the mineral application process by Corruption Watch (2017) indicates that positions in the regulator have been frozen for many years, leading to a shortage of staff, and that unqualified individuals have been appointed. There is also the perceived reluctance of government to grant closure in order to limit transfer of the liability to the state and delay the inevitable job losses."

Jul 1, 2019

By H. F. Niemand, C. J. Postma, P. L. Crouse

"SynopsisThe separation of zirconium and hafnium is essential in the nuclear industry, since zirconium alloys for this application require hafnium concentrations of less than 100 ppm. The separation is, however, very difficult due to the numerous similarities in the chemical and physical properties of these two elements.Traditional methods for separation of zirconium and hafnium rely predominantly on wet chemical techniques, e.g. solvent extraction. In contrast to the traditional aqueous chloride systems, the AMI zirconium metal process developed by Necsa focuses on dry fluoride-based processes. Dry processes have the advantage of producing much less hazardous chemical waste.In the proposed AMI process, separation is effected by selective sublimation of the two tetrafluorides in an inert atmosphere under controlled conditions, and subsequent selective desublimation. Estimates are made for the sublimation rates of the two tetrafluorides based on the equilibrium vapour pressures. A sublimation model, based on the sublimation rates, was developed to determine if the concept of separation by sublimation and subsequent desublimation is theoretically possible.IntroductionZirconium requires several purification steps to conform to nuclear-grade specifications. Little information is available on the sublimation separation of Zr and Hf compounds, especially in the fluoride form, the majority of which deals with sublimation under vacuum conditions. On the industrial scale, only vacuum sublimation of ZrF4 has been reported. No records were found for the sublimation of ZrF4 in an inert atmosphere. Information on the sublimation rate of ZrF4 or HfF4 in an inert atmosphere is also limited. The rate is assumed to be dependent on several factors, of which temperature, area, and composition are considered the most important."

Jan 1, 2015

By A. McLean, M. Iwase, K. Wakimoto, K. Katogi

For many decades control strategies have been based upon samples taken from liquid steel, molten slag and gas phase as well as from solid products. Because of recent developments in the areas of dynamic control models, artificial intelligence and expert systems, there is an increasing incentive to develop diagnostic sensors, which permit in-situ rapid determinations of concentrations of various elements in molten steel. In this paper, an overview is given on recent developments in the areas of sensors for use in ferroalloy and steel production. Particular emphasis will be given on (i) innovative application of sensors for on-line determinations of dissolved elements, Si, P, N, H and Cr in molten iron and ferroalloys, (ii) evaluation of inclusions in liquid metal and (iii) rapid determinations of slag compositions.

Jan 1, 2004

By J. Krzyzanowska

Paper written on project work carried out in partial fulfilment of Bachelor of Engineering-Mining Venetia Mine, an open-pit diamond mining operation in the Limpopo Province of South Africa, currently has a mixed haul truck fleet consisting of Caterpillar 785B and C; 789C; modified 793D and 793D. Even in an ideal situation these trucks have different cycle times, which causes queuing at the loading area and their different speeds cause bunching on the ramps, leading to higher overall cycle times and lower productivity. This problem was identified but the root causes of the problem were not investigated and quantified. A time and motion study on the haul trucks was thus undertaken to measure actual cycle times and compare them to ideal cycle times as well as to observe any reasons for deviations. Ideal cycle times are partly evaluated on the basis of simulation and partly from production optimization expert input. Several areas that affect production were identified and these include: haul road conditions, the control room, dispatching programme and dispatching data management, as well as truck-shovel matching. The investigation was important in establishing control parameters for haul fleet operation since time spent queuing is production time lost, which defers waste tonnes to later in the life of mine, thus decreasing the tempo at which kimberlite is exposed.

Jan 1, 2007

By C. Lumley

The management of the Anglo Platinum (AP) mining technical systems (MTS) embarked upon a process of defining the strategy for the MTS department. The newly established department forms part of the greater Group Information Technology (IT). The strategic process took place within an environment that had witnessed a dramatic proliferation of IT implementations across the group within the last three years. The IT department has been structured in terms of the classical plan, build and run functions with the applications and integration streams running perpendicular across these three functions. The application stream is divided into commercial and mining technical systems with business systems integration the third stream. The mining technical systems department supports mine management to effectively gather, analyse and disseminate data relating to the ore reserve by integrating a number of strategically selected IT applications to ensure that the safety, legal and operational requirements are met. This service to business is rendered by management of a strategic outsource partner via tightly controlled service level agreements (SLAs). The framework used for creating a mining technical systems department that is empowered to add value to business was formulated using the AP value chain as both the starting point and guiding process to ensure alignment to the greater internal strategy of the business. Keywords: Platinum, Mine planning, Mining technical systems, Value chain.

Jan 1, 2003

By S. R. Sankaranarayanan, T. Sowmya

Slags and fluxes used in metallurgical processes are typically made up of oxides (especially silicates), with additions of fluoride and carbonaceous matter. Despite the efforts made towards understanding the role of slags in various processes, little information is available on the structure of these slags. Tools such as X-ray diffraction, spectroscopy, and electron microscopy could be used for this purpose. This paper is based on an ongoing spectroscopic and structural investigation of slags and related materials. Spectroscopic studies of various commercial slags (predominantly ferrous), ores, glass and related materials, covering a wide range of compositions, have been carried out to understand the structure. The infrared spectra were recorded, predominantly, using FTIR spectrometer (Bruker IFS 66U) spanning 4000 to 100 cm-1using the KBr pellet method. The resolution was typically 4 cm-1. The spectral outputs have been compiled in terms of bands present (corresponding to different bonds) and then compared with the typical bands expected for such bonds. Efforts are in progress to identify the trends with respect to the chemical composition?for example, the effects of SiO2, Al2O3, and non-bridging oxygens. Data obtained so far indicate some trends such as (i) a shift in the Si-O band (969 to 1099 cm-1) towards a higher wave number with increasing silica content (almost 2 to 92 weight percentage) and (ii) further strong shifts as a function of the alumina content. These are being further investigated structurally and in terms of chemical composition/parameters. Keywords: slags, silicates, structure, spectroscopic analysis, shifting of bands

Jan 1, 2004

By V. Labiouse, M. Jaboyedoff

GIS documents are of great interest for a quick and low-cost determination of areas endangered by rockfalls. They allow (i) the detection of potential instabilities from steep slopes and cliff areas, and (ii) the preliminary estimation of potential run-out areas by means of a so-called cone method. After the presentation of the tools used to compute these areas, the paper focuses on two methods developed for a preliminary hazard mapping, one at regional scale and the other at local scale. Mittels GIS-Dokumenten können von Steinschlag bedrohte Gebiete schnell und kostengünstig ermittelt werden. Die sogenannte Kegelmethode erlaubt die Erkennung potentiell instabiler Zonen von Steilhängen und Felswänden sowie eine erste Einschätzung der möglichen Reichweiten. Nach der Vorstellung der zur Berechnung der bedrohten Gebiete benötigten Hilfsmittel stellt der Artikel zwei Methoden zur vorläufigen Gefahrenkartierung in lokalem und regionalem Maßstab vor. Les SIG sont très intéressants pour une détermination rapide et peu onéreuse des zones affectées par des chutes de blocs. Ils permettent une détection des zones potentiellement instables et une estimation sommaire des zones de propagation des blocs par une méthode dite des cônes. Après la description des outils utilisés pour définir ces zones, l’article expose deux méthodologies élaborées pour un zonage préliminaire du danger, l’une à l’échelle régionale et l’autre à l’échelle locale.

Jan 1, 2003

By J. Ekengård, H. Srinath, M. Andersson, P. Jönsson, D. S. Sarma

The present paper deals with mixing between the metal and slag phase during the ladle refining process, including vacuum degassing. Dispersion of metal drops in ladle slags has been studied. The results show a larger quantity of metal droplets in ladle slag during the alloying and stirring phase than during the subsequent vacuum degassing phase. The thermodynamic and kinetic aspects of the interaction between metal drops and the slag matrix have been considered and probable reasons for the observed distribution discussed.

Jan 1, 2004

By W. A. Gericke, B. Olmstead, I. Mc Dougall, C. F. R. Smith

A finite element model was generated of a Søderberg electrode in a ferromanganese smelter. In this model, the casing and fins were modelled explicitly, and temperature-dependent material properties were included. A two-stage analysis procedure was used, the first part being a coupled thermal-electrical analysis using an applied DC current. As the casing material was non-magnetic, the DC current was deemed sufficiently accurate for this initial analysis. The proximity effect was not considered. Thermal boundary conditions based on assumed temperature distributions in the furnace were applied. Calculated temperatures agreed well with measured values. The joule heat calculated in this step was applied as a load to a coupled thermalstress analysis. The model was applied to a comparative study of alternative materials for the casing and fins. A number of different combinations of materials for the casing and fins were evaluated in a comparative study. The position of the baking zone was used as the criterion for evaluating the potential of each material combination. Based on this analysis, some material combinations were discarded as being entirely unsuitable, whilst the two most suitable combinations will be tested experimentally in the furnace. The stress analysis was performed on the most suitable option. The practical advantage of this analysis was that a number of what-if scenarios could be evaluated and discarded prior to embarking on expensive experimental work, thus limiting the amount of experimental effort required. A mathematical model allows for the comparison of options under controlled conditions, where parameters of interest can be varied individually, with all other parameters remaining the same, a condition which is difficult to achieve in an operating furnace.

Jan 1, 2004

The Annual General Meeting of the South African Institute of Mining and Metallurgy was held in Kelvin House on Wednesday, 24th August, 1977. Dr M. D. G. Salamon (President) was in the chair and declared the meeting open at 16h00. Obituaries The President: It is my sad duty to announce the death of the following members of the Institute : C. G. Evans, J. P. Rees, Life Fellows ; S. R. Bryant, H. MacConachie, M. Mrost, and A. W. van Zyl, Fellows; L. T. Campbell-Pitt, F. W. Collins, A. J. Duncan, I. G. Evans, F. W. Kleinschmidt, P. L. Vaughan, R. W. Wallace, and H. Wason, Members. As a mark of respect to the memory of the deceased and in sympathy with the bereaved, I would ask you to rise and observe a few moments' silence. Minutes The President: May we confirm the minutes of the General Meeting held on 23rd March, 1977, as published in the June issue of the Journal? Agreed.

Jan 1, 1978

By Z. Zhang, R. G. Reddy

Experimental and thermodynamic modelling studies on the structure of alkali borate melts and properties are discussed. A thermodynamic structure model for alkali borate melts is developed to describe the distribution of structural groups in the alkali borate melts as a function of composition. A qualitative interpretation for the boron anomaly in viscosity has been proposed, which regards the boron anomaly as a manifestation of the various antithetical effects caused by the introduction of alkali oxide into boric oxide. It has been found that the viscosity of alkali borate melts increases or decreases according to the changes in rigidity and spatial connectivity of the glass network structure, which is the function of temperature and composition. These concepts with some key examples are discussed in relation to their usefulness and practicality in industrial slags. Keywords: alkali borate melts; structural model; activities; structural groups; viscosity;

Jan 1, 2004

By A. M. Milev

A simulated rockburst was conducted underground at Kopanang Mine, previously called Vaal Reefs No. 9 Shaft. The rockburst was simulated by means of a large blast detonated in solid rock close to a crosscut sidewall. The simulated rockburst involved the design of the seismic source, seismic observations in the near and far field, high-speed video filming, a study of rock mass conditions such as fractures, joints, rock strength etc. Knowledge of the site conditions before and after the simulated rockburst was also gained. Some of the important findings are listed below: ?Two areas of damage were identified on the blasting wall: (i) an area of high-intensity damage in which ground velocities of 3.3 m/s were recorded by an accelerometer, which was subsequently ejected in a block of rock, (ii) an area of low-intensity damage where ground velocities of 1.6 m/s were recorded by an accelerometer which remained on the tunnel wall. ?High-speed filming revealed rock fragments being ejected from the wall at velocities in the range of 0.6 m/s to 2.5 m/s. The measurements were taken in the area of low-intensity damage. As planned, no gas pressure was directly involved in damaging the wall of the tunnel. ?The attenuation of peak particle velocities for the main blast in the near field (6 m to 30 m), as a function of distance R, was found to be proportional to R-1.7. ?The simulated rockburst was recorded as a local magnitude ML= 1.3 by the Klerksdorp regional seismic network. ?Peak particle velocities from nearby blasting, recorded on the tunnel wall after the simulated rockburst, were about six times greater then peak particle velocities recorded from the same source area before the simulated rockburst.

Jan 1, 2001

By Richard Peaker, Jens Lichter, Alexander V. Potapov

The recent combination of Computational Fluid Dynamics (CFD) and Discrete Element Modeling (DEM) can provide considerable insight into the performance of flotation cells. One area of interest from a fundamental equipment design perspective has been the influence of flotation cell size and the inflow rate on the metallurgical performance of the cell. One of the specific design objectives when increasing the size of flotation cells, or increasing the feed flow to the flotation cell, is to ensure that the feed stream flowing into the cell does not have an opportunity to bypass the mechanism. This paper describes the initial modeling efforts to quantify the inter-relationship of cell size, inflow rates, and feed and discharge box placement, on the retention time distribution of solids in the flotation cell, and the fraction of particles that bypass the mechanism on a size by size basis.

Jan 1, 2007

By D. S. Dihalu

A generalization of Gy?s model for the fundamental sampling error introduced the new ?parameter for the dependent selection of particles?, denoted as Cij. This allows for modeling deviations from the ideal situation where the selection of a pair of particles is composed of two independent selections. The generalized model potentially leads to more accurate variance estimates in the case of clustering of particles, differences in densities or sizes of the particles or repulsive inter-particle forces. A straightforward and practically applicable method is needed for the determination of this parameter for miscellaneous mixtures in industrial settings. In this contribution, the feasibility of using digital image analysis to determine this parameter Cij, will be demonstrated. Line transect sampling of a digital image was used to construct a transition probability matrix. A new algorithm to derive quantitative estimates for Cij will be presented and discussed. The applicability was verified by a photograph of zirconium silicate particles of sizes typical for industries dealing with pharmaceutical, food/feed, and environmental applications. Conditions affecting the practical applicability are identified and potential pitfalls will be discussed, including e.g. how a potential unrepresentative surface can affect the quality of the estimate of Cij.

Jan 1, 2010

By G. C. Smith

Process control applications in the minerals processing industry have grown extensively over the years due to the many benefits derived, both financially and operation-wise. Although control theory is widely known, conventional control does not address many of the problems posed by minerals processing circuits. Some of these problems include interactions between process units, large dead times, few and sometimes unreliable measurements of process variables, non-linear systems, disturbances entering the process, processes with varying dynamics, and many more. For the past 26 years Mintek has been active in the field of providing specific, customized solutions for the minerals processing industry. This paper focuses specifically on applications in milling and flotation control. The underlying control principles have been applied to a wide range of other processes including submerged arc furnaces. Advanced control application is implemented on a common advanced control platform, called the PlantStar. Much innovation has been brought to the fore in terms of adapting theoretical techniques and developing new methods to resolve specific milling and flotation circuit issues by using process control. The general approach adopted by Mintek when implementing advanced control is to enforce stabilization of the process as a solid foundation for optimization strategies to be built upon. In doing this, a robust control solution is presented, and the end result is flow and level stability, better product quality regulation, increase in throughput and a stable feed to downstream processes. The MillStar advanced milling control system ensures that the milling circuit is at all times easy to operate while achieving optimum product quality and throughput. Stabilization should include product size regulation and sump level control. Another important aspect of stabilization control is the handling of milling circuit constraints without introducing cyclical behaviour and with minimal disturbance to the final product quality. Process constraints may include factors such as process water limitations, cyclone feed density, and cyclone underflow angle. An optimization strategy should be developed to increase the mill throughput without affecting product quality or introducing mill overloads. The relationship between solids feed and power draw is examined and manipulated for optimization. The interactive nature of flotation circuits and the presence of recirculating streams tend to make satisfactory control almost impossible with conventional techniques such as PI (Proportional plus Integral) or PID (Proportional, Integral and Derivative) control. The FloatStar Level Stabilizer simultaneously considers all the flotation levels in a circuit by use of a multivariable algorithm, and takes compensating action in anticipation of disturbances. The FloatStar Flow Optimizer is an optimization strategy that controls the concentrate flow rate between various sections of a flotation circuit (e.g. Rougher ?Cleaner; Cleaner ?Recleaner). By controlling the concentrate flow between the sections, the mass pull, circulating loads and residence times of the circuit can be stabilized, thereby enhancing the performance of the flotation process. Optimizing grade and recovery are coupled with the dynamic optimization of circulating flows. The FloatStar Grade-Recovery Optimizer will maximize the grade in a section, whilst ensuring that recovery or any other process specific limits remain within predefined limits. With successful applications on several milling and flotation plants worldwide processing amongst others PGMs, gold, copper and nickel, Mintek has shown that it is a leader in milling and flotation control innovation

Jan 1, 2004

By T. Markus, K. Hilpert

Discharge vessels of advanced light sources are made of translucent polycrystalline Al2O3(PCA).These tubes contain salt mixtures, which essentially consist of metal halides. The salt mixture is present as a melt at the coldest spot of the vessel under operating conditions and vaporizes partly. The temperatures of the coldest spot range between 1200 K and 1300 K, depending on the lamp performance. The maximum wall temperatures range up to 1500 K and for special purposes up to1700 K. Important constituents of the melt are alkali halides AX (A = Na, Cs, X= Br, I) and rare earth metal halides LnX3(Ln = Dy, Ho, Tm). Severe corrosion attack on the wall material is observed in the PCA discharge vessels. The aim of this work is to elucidate the corrosion mechanism. Annealing experiments under well defined temperatures were carried out in order to understand the complex chemical corrosion processes. Two different annealing vessels were used, one for experiments under isothermal conditions with a temperature of 1473 K and one for experiments with a linear temperature gradient from 1450 K to 1600 K. The vessels were filled with salt mixtures of different compositions and annealed for 1000 h. The corrosion attack was characterised by XRD, analytical electron microscopy, high temperature mass spectrometry, and chemical analysis of the salt melts. The results of our investigations show that the corrosion attack is caused by vapour species and by the salt melt. The most severe corrosion is observed in vessels filled with mixtures of alkali halides and rare earth metal halides. No corrosion of the alumina is practically observed if pure alkali halides or pure rare earth metal halides are present in the annealing vessel. Keywords: thermodynamic, corrosion, lamps, metal halide salt

Jan 1, 2004