Search Documents

By A. K. Mukhopadhyay

[The mechanical status of a mine ventilation fan can be monitored effectively through vibration analysis. An expert system can be an effective tool in fault diagnosis for a variety of equipment. Computers have made it possible to implement an expert system to analyze the data for correct and appropriate diagnosis. In addition, vibration analysis can provide important information oil the condition of the machine. This paper discusses an approach for an expert system to determine on vibration analysis, practical problems that can occur with a fan. The frequency of the vibration is the only characteristics considered in the analysis, although other variables are discussed in the paper. Finally, a simple computer program, written in Turbo Prolog, is presented. A friendly query system, menu operation and a disk-resident separate knowledge base are a few of the salient features of this system.]

Jan 1, 1997

By C. R. Rinschler, G. R. S. Smith

A high-strength, industrial-grade membrane filter- cloth capable of high flow rates and sub-micrometric particle retention has been developed. A filtration system using this filter- combined with special filtration techniques and using tubular-filters makes possible the filtration of many difficult-to-jilter slurries. Small, discrete, slowly-settling particles are formed into filter cakes that when pulsed of the membrane filter cloth do not redisperse but settle rapidly to form a dense filter under- flow. A positive separation of solids from liquids is obtained.

Jan 1, 1991

By H. Nemutolluhi

Discussion by G.E. Agar In the April issue of Mining Engineering there is a technical note "New size laboratory ball mill for Bond Work Index determination" written by M. Nematollahi which is very confusing. My recollection is that a description of an experimental procedure should be such that someone skilled in the art should be able to reproduce the results. Applying this test to this particular article would according to me cause a failure to be registered. There are apparently three batch laboratory test procedures referred to in the article. One of these is the traditional Bond test with all its limitations such as dry grinding followed by exclusively dry screening, two attributes which fail to simulate conventional industrial grinding. There is a second procedure (NSBM) which the author is advocating as an improvement of the Bond procedure but from the description I doubt that anyone would be able to repeat the procedure. For instance, is the grinding to be done wet or dry, if wet at what pulp density, and how is the screening to he done, wet or dry or a combination? On top of that uncertainty the author refers to a simplified "Anaconda" test. No reference is provided for this procedure and in the text the results of the technique are referred to "Denver wet ball mill" results and say nothing of Anaconda results. Reply by H. Nematollahi I think that the method proposed to determine the Bond Work Index in this paper is clearly described in the paper as follows: •The method of grinding is dry (Ref. paragraph I, subject Method description); •The size of ball mill, the quantity and size of balls are noted (Ref. paragraph 1, subject Test equipment); and The test procedure is described in detail (Ref. paragraphs 1, 2 and 3, subject Test procedure). As it is noted in the paper, the test has been worked out under the same condition as Standard Bond Test with ball mill, so dry (screening should be dry also), but using a smaller ball mill and a smaller quantity of sample. Considering the above mentioned points, I believe that the method has been fully and clearly described and it ought to be repeatable as well. The Anaconda test is a method for determination of Work Index and could be done by any types of laboratory wet ball mill (Yan, R.F., et al., cf. References). I applied this method using a laboratory Denver ball mill (Ref. paragraph I subject Determining the Bond Work Index). So the results referred to Denver wet ball mill can be equated the same as Anaconda method.

Jan 1, 1996

By P. D. Rao, D. E. Walsh, W. W. McClintock

A column flotation system was designed to conduct release analyses on a very finely disseminated, stratiform copper ore from the Denali Copper Prospect in Alaska. Chalcopyrite is finely disseminated in a carbonate matrix, and pyrite is disseminated in both the chalcopyrite and the gangue. Grinding to -20 µm is required for adequate liberation. A 25-mm x 2.4-m glass flotation column was designed. The ore and its flotation tailings were repeatedly floated to obtain a rougher concentrate. The rougher concentrate was then repeatedly floated to produce release fractions. These fractions were analyzed for copper, iron, calcium and magnesium, and release curves were plotted. The procedure was replicated at each of two grinds, i.e., 99% and 88% passing 37 µm (400 mesh). The procedure should find application for flotation characterization and process selection. Bond grindability testing was also conducted.

Jan 1, 1996

By Pierre M. Gy

The final purpose of exploration is to esti- mate the grade and tonnage of the various qualities of ore contained in a given mineral deposit. This estimation cannot be directly carried out on the whole of the ore body. It usually requires four steps: Core Drilling Campaign The ore body is represented by a series of cores further divided into core sections of equal or unequal length. Core Splitting The core sections are longitudinally split into two halves: the first half is kept for various examinations and physical measurements, and stored for further reference; the second half (core sample) is used for determination of the chemical analysis. Core Sample Reduction The chemical analysis cannot be directly per- formed on the core samples. It requires the constitution of assay portions usually weighing from a few decigrams to one gram (more in precious metals assaying) and passing through a 100 or 150 mesh screen.

Jan 1, 1979

By S. D. Morrison, J. L. Watson

It is well known that the density and viscosity of a mill pulp have optimum levels for grinding efficiency, and sophisticated control loops are employed to control pulp density in industry. However, these control loops rely on a number of plant measurements, and often the input data will not permit effective determination and control of the mill pulp density. The use of a single remote pulp density measurement has obvious advantages, and it is suggested and demonstrated for laboratory batch grinding that mill noise measurements can indicate the effective density of the mill pulp. In addition, the mill noise is shown to indicate the actual effective pulp viscosity in the mill and hence permit the identification of the dynamic rheological regime existing in the mill. It is demonstrated that mill noise increases with grind time and, at any given grind time, with increasing pulp density in the pseudoplastic range. This increase in mill noise level with pulp density is interpreted as indicating a decreasing pulp viscosity. The results show that the transition from dilatant to pseudoplastic conditions within the mill, with increasing pulp density, can be identified by this increase in mill noise. When the pseudoplastic pulp started to exhibit yield characteris- tics, the mill noise fell dramatically. The technique of mill noise analysis is then used to investigate the effect of a grinding additive on the batch grinding of two ores. With a magnetite ore, the additive was shown to increase the mill noise, and hence decrease the pulp viscosity, in the pseudoplastic region and to increase the effective grinding. The molybdenite ore, however, did not show significant noise level changes and did not show improved grinding efficiency.

Jan 1, 1987

By M. J. Laurila

On-line analyzers have made possible the use of fuzzy logic control (FLC) for local control applications in coal preparation plants. This includes FLC of heavy media vessels, cyclones and froth flotation. These circuits can be controlled in real time in a feedback control loop with an on-line analyzer. Control strategies can be developed using statistical process control (SPC) techniques that are first applied offline in manual mode through the use of control charts. This process is then brought on line using FLC in conjunction with an on-line analyzer to automate the circuit. The advent of programmable logic controllers into preparation plants has enabled the implementation of regulatory control. As shown in Fig. 1, regulatory control is the first level of process control as applied in a processing facility. This involves: ? The logic sequences for start-up and shutdown procedures and all the necessary interlocks for safety purposes. ? The systems to control individual parameters and individual units. Various sensors provide information that is useful to plant operators, while some can be used as signals to simple control loops (Couch, 1996). This is the foundation of process control. It must be performed well if there is to be any success in supervisory control (level 2). Supervisory control involves overseeing and directing the operation of unit operations, or of the whole plant. This includes more complex loops with cascade control. And it may involve feed forward and feedback systems. There have been some applications of this kind of control to thickeners and flotation circuits in coal preparation plants. In addition, there are supervisory controls on some dense medium circuits and, in a few cases, on jigs. These are implemented with proportional integral- derivative (PID) controllers that are available in the PLC or distributed control system (DCS) operating soft ware. In a coal preparation plant, there might exist 200 to 400 PID loops. In a coal-pro¬cessing plant, this number is increased to 2,000 loops or more. It is important to use the correct algorithm or function to control a process variable to the point where automatic control shows a markedly decreased variability over manual mode. This is where fuzzy logic and neural networks have an advantage over traditional model-based expert systems. Because dynamic models of coal preparation unit operations do not exist, model-based expert systems that are prevalent in minerals processing cannot work. Not enough is known about the dynamics of the gravity separation processes employed in coal preparation. Therefore, an artificial intelligence (AI) system consisting of a subset of fuzzy logic, genetic algorithms and neural networks must be applied in coal preparation plants if a supervisory control system is to be successful across a range of feed characteristics and variable set points.

Jan 1, 1997

By S. Behl, S. Mathur

The viability of physico-chemical purification methods such as selective, flocculation for the separation of two components having similar chemical compositions and the same crystal structure but different surface properties is demonstrated. A loss of selectivity was observed in a dolomite-dolomite mixture when polyethylene oxide (PEO) (five million molecular weight) was used us the flocculant. Heteroflocculation was minimized using a low molecular weight PEO (MW =105) as a site blocking agent (SBA). Near complete separation was achieved at pH 10.5 at the 100% recovery level

Jan 1, 1996

By G. G. O. O. Uwadiale, M. A. Kollere, M. A. U. Nwoke

Birnin Gwari and Alawa graphite deposits were upgraded by flotation. Using 0.5 kg/t kerosene and 0.25 kg/t sodium silicate, flotation concentrates grading 90.7% and 88% Cg with recoveries of 86.2% and 84% were produced from Birnin Gwari and Alawa graphite samples, respectively. The combination of 0.5 kg/t kerosene, 0.2 kg/t pine oil and 0.25 kg/t sodium silicate yielded high-grade graphite (92.3% Cg) with a recovery of 83.9%. Flotation with refined rubber nut oil and kerosene, at a 1:1 ratio, produced a concentrate grade of 86.3% Cg with a recovery of 90.4%. With acid leaching of the flotation concentrate, a high-grade graphite concentrate containing 98% Cg was produced.

Jan 1, 1998

There has been some confusion over the selection of discount rates when performing stochastic simulations of a project's net present value (NPV) using risk- adjusted discounting techniques. Some argue that lower discount rates should be used when performing stochastic valuation simulations. This creates higher NPV values and improves the attractiveness of a project. Others argue that the discount rate should not be lowered for these simulations. This paper exposes the source of this confusion. It demonstrates that stochastic simulations do not remove the risk inherent in a project and, therefore-do not improve a project's value to the risk-averse investor. Risk-adjusted discount rates should, therefore, not be adjusted downward when performing stochastic NPV simulations. Stochastic simulations are, however, shown to he useful in several aspects of project valuation.

Jan 1, 1996

By Robert F. Jones

INTRODUCTION Upflow Solids Contact Clarification is a process long employed in the treatment of potable water, wastewater and municipal sewage. This process incorporated in a fiberglass conical structure seems to offer an economical solution to problems encountered in reduction of particulates in uranium effuents. CONCEPT OF UPFLOW CLARIFICATION Use of suspended solids contact units date back to the 1930's. These types of units produced by Infilco and the Permutit Company, among others, were used in the Lime-Soda softening process to precipitate calcium and magnesium salts from hard water. In general, the process operated as follows: (a) Raw water entered a central chamber where it was mixed with chemicals. The chamber had provisions for agitating and circulating the chemicals so that mixing and reaction zones were formed. (b) The water, precipitated chemicals, floc and turbidity flows to a clarification chamber (the Permutit precipitator used a truncated cone). In most cases the influent then passes upward through a blanket of suspended floc and other solids before it leaves the chamber at effluent weirs or launders.

Jan 1, 1984

By H. Nematollahi

The comminution laws The grindability of the ores or minerals is an important factor that could not be deduced from their mineralogical characteristics (Testut, 1958). Many theories have been proposed to determine the amount of work required in a comminution operation as a function of the nature of material, size and shape of the minerals before and after of comminution. The most important laws are proposed by Rittinger, Kick and Bond. According to Rittinger (1867), the work is directly proportional to the new surfaces produced during the comminution. Kick (1885) believes that the work needed to produce similar modification is proportional to the volume or mass of the material. Bond, during his 25 years of experience and research, reached to the conclusion that on the average the energy required for a known comminution is between the Rittinger's and Kick's laws (Bond, 1952). According to Bond's Third Theory, the work input is proportional to the new crack tip length produced during the particle breakage, and equals the work represented by the product minus that represented by the feed. The work input is given by the following equation: [111W = 1 Owi( FP FF(1)] where W is the The energy input to the mill, (kWh/st); F is the Feed size based on 80% passed, (µm); P is the Product size based on 80% passed, (µm); and W; is the Work Index, (kWh/st) Methods to determine the Bond Work Index Various methods have been proposed by Bond to determine the Work Index. There are some problems with the Standard Bond Grindability Test: •Requiring a special mill. •Requiring a relatively large amount of sample. •Being relatively time consuming and tedious. [ ] During the recent years, numerous attempts have been made to install faster and easier methods to determine the Work Index. A complete description of these methods is presented by Yap et al. (1982). Determining the Bond Work Index by a new size ball mill During the recent years, researchers at the University of Tehran determined the Bond Work Index, using the Anaconda simplified methods by its Denver 305- x 127-mm (12 x 5-in.) laboratory wet ball mill. The equation used to calculate the Work Index for the above mill containing 16.12 kg (35.5 lbs) of steel balls is as follow: [1 _ 1-1W;=0.54x [P [)(2)] In addition, an attempt has been made to determine the Bond Work Index similar to standard Bond ball mill procedure but by a different ball mill, smaller and in consequent portable and less laborious than the Bond ball mill. Method description Just as the standard Bond grindability test, this method is a locked-cycle dry grinding and screening process that is carried out until steady state condition is reached. Test equipment The test is conducted in a special ball mill, called the New Size Ball Mill (NSBM). The mill is scaled down with a coefficient of two-thirds from the standard Bond ball mill. It measures 200 x 200 mm (8- x 8-in.) inside (instead of 300 x 300 mm or 12 x 12 in. in Bond ball mill). Just like the Bond

Jan 1, 1995

By P. L. Padgett, J. C. Hower

Five coals blended at a Midwestern United States power station were subjected to detailed analysis of their Hardgrove grindability. The coals, representing four distinct coal beds, are a low-sulfur, high-volatile A bituminous Upper Elkhorn No. 3 coal (Pike County, Kentucky); a medium-sulfur, high-volatile A bituminous Pittsburgh coal (southwestern Pennsylvania); a low-sulfur, subbituminous coal from the northern Powder River Basin (Montana); and two low-sulfur, subbituminous Wyodak coals from two different mines located in the eastern Powder River Basin (Campbell County, Wyoming). All samples that were processed in the Hardgrove grindability tests were also analyzed for their maceral and microlithotype content. The high-vitrinite Pittsburgh coal and the relatively petrographically complex Upper Elkhorn No. 3 coal exhibited different grindability behavior. The petrographic compositions of the Pittsburgh raw feed, the 16 x 30-mesh fraction (HG! test fraction) and the -30-mesh fraction (HGI test reject) are similar, suggesting that the HGI test fraction typifies the whole feed. The eastern Kentucky test sample is not as representative of the whole feed, with the HGI test fraction having lower vitrinite amounts than the rejected -30-mesh fraction. The Powder River Basin coals are high in vitrinite content and show behavior similar to the Pittsburgh coal.

Jan 1, 1998

By D. J. Kennedy, W. G. Klinowski

Over the past few years the Cape Breton Coal Research Laboratory's staff designed, built, tested and proved the necessary hardware, software and methodology of tracer gas based surveying techniques applied in the field of mine ventilation. These methods are now used routinely in a variety of applications where traditional anemometric methods do not provide the required accuracies or cannot be effectively employed. Irregular or highly turbulent flows, leakages, geometrically complicated or highly obstructed airways are just a few examples where traditional anemometry can produce unacceptable errors in calculated ventilation parameters. Tracer gas based ventilation techniques have been used in many problematic situations where the standard ventilation methods are inadequate. One such problem, faced by many mine operators, is the emergency imposed by the ever present threat of a mine fire. An understanding of the propagation of smoke contaminants and the demonstrated integrity of appointed escape routes can save lives and property during a fire. That knowledge can be gained quickly and inexpensively by the application of tracer gas smoke simulation methods. The concept of controlled air recirculation has recently gained considerable attention in the mining industry. While mathematical models can be used to predict performance, the ultimate proof of a ventilation system's performance can best be provided by a tracer gas based simulation of contaminant levels. Fan performance tests and assessment of auxiliary ventilation systems are highly dependent on accurate definition of airflows; precise data can be easily obtained by tracer gas based volumetric surveys. This paper contains a brief description of the equipment and the methodology of tracer gas based ventilation techniques and presents typical examples of their use

Jan 1, 1991

By B. P. Boisen

INTRODUCTION The rapidity of onset, rate of increase, and magni¬tude of loads in an underground support system can be measured using load cells or pressure cells, whichever is appropriate to the type of support. These instruments should be installed at the various instrumented station locations immediately after excavation. If the loading of a member is of interest, the load should be measured, whereas if the deflection or strain of a member is of interest, the strain should be mea¬sured. It makes as little sense to use a load cell to allow computation of strain as it does to measure the strain and then back calculate the load. Tunnel steel set de¬signers today invariably rely on the early work of Karl Terzaghi, the basis for which is load, so load measure¬ment should be the main concern. The current trend of using strain gages on steel sup¬port systems stems from the inability by some to eval¬uate unusual loading conditions caused by uneven block¬ing. In fact, uneven blocking will obscure almost any attempt (whether with load cells or strain gages) to properly evaluate support performance. The load curve in Fig. I shows the development of a characteristic early peak load sometimes called the "rear abutment load" seen in many underground open¬ings. It is thought that the peak reflects the coupling of the rock and the support system, and that the magnitude will increase until some yielding, or minor failure, of the support system occurs. At that time, the slight deforma¬tion of the support system promotes the formation of minute shears in the opening walls, and these shears tend to distribute stress between the support system and the adjacent rock in proportion to the relative rigidity of the support elements and the rock. In normal rock, as a result of this stress redistribu¬tion, subsequent load magnitudes generally do not reach the magnitude of the early peak load. In physically unstable (squeezing) or chemically un¬stable (swelling) rock, however, the loads experienced after passage of the early peak load may in fact show a slow, continuous increase. One of the very important purposes of load instrumentation is to provide the means for recognizing such long-term adverse trends, thus enabling the proper remedial steps to be taken. Another purpose of load instrumentation is to pro¬vide a comparison between the magnitudes of the early peak load and the subsequent stable load. The ratio of these two loads is analogous to a safety factor, and may be used to evaluate the efficiency and economy of the support system design. Aside from considerations of economy, it may be well to design support systems which do not have excessively high ratios of early peak load to subsequent stable load. Should these ratios be exces¬sively high, the support system may be so rigid that the yield or failure associated with stress redistribution may occur with explosive violence. Load cells for use in mines, tunnels, and on con¬struction projects come in many forms. Almost all, however, employ the same procedures for installation, readout, etc. Therefore the following comments are almost universal in application. LOAD CELL INSTALLATION Load cells should be installed with bases parallel to the surfaces against which they bear. Care must be taken to orient the cells so that their signal cables are protected from accidental damage as a result of con¬struction, maintenance, or cleanup activities. Most electronic load cells are compensated for tem¬perature variations likely to be encountered during nor¬mal operations. However, if a large difference is an¬ticipated between the calibration temperature [21°C (70°F)] and the average operating temperature, the cells should be conditioned to the operating temperature for at least 8 hr prior to installation. This is to insure that the initial reading, made under no-load conditions prior to installation, provides a stable value to which subse¬quent measurements can be referred. Hydraulic load cells tend to be temperature sensitive and should be used with that in mind. Also, hydraulic load cells tend to be soft compared to electronic types and will sometimes allow movement to take place in a system intended to be semirigid. Furthermore, hydrau¬lic load cells tend to be difficult to read remotely. Care must be taken to use bearing plates on both sides of load cells which are sufficiently rigid and of high enough bearing capacity to prevent bending and crush¬ing under load. This is very important with tieback load cells (basically a ring of steel) which can easily dig into bearing plates. MAINTENANCE AND TROUBLESHOOTING Except for a direct hit by a miner's axe or flyrock from a blast, most hydraulic load cells are nearly in¬destructible and require little, if any, maintenance. Hy¬draulic oil on the bearing plates is a good indication of a leak and the need for corrective action. Field maintenance of electronic load cells involves protecting the instrument and signal cable from mechani¬cal damage and from unnecessary exposure to dirt and moisture, and recognizing and correcting damage and the effects of normal wear and tear.

Jan 1, 1982

By M. K. Pathak, A. K. Mukhopadhyay, D. C. Panigrahi

Changes in machinery operating and design parameters often result in unexpected vibrations or changes in the vibration patterns. Vibration in a system is an undesirable element that can cause machinery distress. Potential damage can be prevented by identifying the vibration level that is due to the changed operational parameters. This is necessary for the safe and reliable operation of plant equipment. This paper presents a documented study on vibration events that resulted from alterations in operating and design parameters in a mine ventilation fan and analyzes the causes by defining the sequence of events that led to its performance deficiency. This paper is based on experimental effort to study the vibration behavior of a ventilation fan under varying operating conditions.

Jan 1, 1999

By M. E. Lewellyn

A lower moisture content for a clean coal product means increased Btu content, or increased yields at the same Btu content, and lower operating costs. The conventional surfactant-based dewatering aids effeciently lower the moisture content but often the presence of residual surfactant in the filtrate and recycle water causes problems downstream. ACCOAL-DRI 1000, a surfactant-based dewatering aid, was developed to solve this problem by minimizing the downstream effects while efficiently lowering the moisture content. The theory of the use of surfactants in filtration will be discussed as well as the development and field evaluation of this new product.

Jan 1, 1988

By R. Larry Grayson

A menu-driven, linear-programming- based, user friendly computer program was developed to assist mine managers in determining optimal longwall panel dimensions for a given set of operating conditions. A sensitivity analysis must be performed by varying key mining and configurational parameters over an expected, practical range during the optimization process. Results from analysis of a base case mine indicated that no single combination of longwall panel length and face width is optimal for all operating situations. Each mine manager must therefore look closely at his operating situation and determine the proper dimensions to minimize total mining cost.

Jan 1, 1989

By Annan Cook

J.R. Atkinson, office engineer to Kennecott Copper Corp., summarized data on the Mineral Park area [Wallapai (Chloride) Mining District, Mohave County, Arizona, Fig. 1] and submitted it to the New York office in 1950. He recommended an examination. Weston Bourret, a Kennecott geologist, accompanied by B.S. Butler, the well known former US Geological Survey geologist, examined the larger of two hydrothermally altered areas in Mineral Park that same year. Bourret recommended that the property be optioned and mapped. Georges Ordonez, Kennecott's chief geologist, after a field examination, agreed with Bourret's conclusions At the end of the summer of 1951, Annan Cook (another Kennecott geologist) accompanied Ordonez to the property prior to starting a mapping program. At Ithaca Peak, the structure, alteration, and reddish-brown coloring in leached cavities-hematite, jarosite, and goethite with minor copper staining - made an interesting target. The Peak, which has since been mined, was a Laramide quartz monzonite porphyry grading into quartz porphyry, penetrating Precambrian rocks. Turquoise had been mined from several workings on the Peak. Surrounding the central area of alteration, but mainly to the northwest, there were many long-abandoned underground workings mined for zinc, lead, gold, silver, and copper. To the west of Ithaca Peak within the main area of alteration, surface evidence of a chum drill hole suggested interest in the disseminated mineralization probably prior to World War II. A topographic and geologic survey of the altered Ithaca Peak area was initiated by Cook in September 1951. He was assisted by Richard Vosburgh, field engineer, and by four geologists: Charles Goddard, III, Donlon LoBionda, Keith Martin, and A.T. Stone. When Cook returned to New York, Vosburgh was left in charge. Mapping continued and diamond drilling took place during the winter of 1951-1952. Drill sites were selected entirely on the basis of surface geology. Five drill holes, including a flat hole, located around the flanks of Ithaca Peak, averaged about 183 m (600 ft) in length. A sixth hole caved at less than 60 m (200 ft). The best values obtained were in the first hole (-45°): 90 m (300 ft) averaged 0.62% copper and 103 m (337 ft) averaged 0.041% molybdenum. In another hole, 43 m (142 ft) averaged 1.09% copper and 0:023% molybdenum. The values of copper and molybdenum in these two holes were generally higher than in the remaining holes, although molybdenum values were always around 0.024% or better. The New York office decided, however, to abandon the project even though the remaining field crew of three had recommended more drilling.

Jan 1, 1991

By G. H. Lewis, C. L. Krcil, N. S. Harding, D. E. Havekotte

INTRODUCTION Limestone injection work was completed in a 440-kW (1.5-million Btu per hour) pilot-scale combustor in preparation for a 50,000-kg/h (110,000-lb per hour) industrial steam boiler limestone injection test program. The main objectives were to: (1) identify the important variables that affect sulfur capture, (2) determine the maximum SO2 removal, and (3) determine the effect of limestone addition on furnace operability and heat transfer. A more detailed report is given by Lewis et al. (1985). The test program was divided into two phases. The objectives of Phase I were to determine the important variables that affect SO2 reduction and to investigate the effect of limestone addition on combustor operation and heat transfer surfaces. The objective of Phase II was to determine, using only the important variables identified during Phase I, the conditions that resulted in the greatest sulfur dioxide reduction. Phase I Test program Using one limestone, five variables were identified as having an effect on SO2 reduction. These are noted with their respective ranges: heat release rate [103 to 166 kW/m3 (10,000 to 16,000 Btu per hour-cu ft)], Ca/S molar ratio (0 to 3), flue gas oxygen concentration (3% to 5%), additive particle size [90% -150 µm (-100 mesh) and 90% -45 µm (-325 mesh)], and burner stoichiometry (SR = 0.75 and >1.15). A 2.3% sulfur northern West Virginia coal, pulverized to a nominal grind of 70% -75 µm (-200 mesh), was burned for the tests. The limestone used was obtained from a Maysville, KY, quarry and contained 51.75% (wt) CaO. An 11-run statistically designed program (plus three duplicates) was used to screen the five variables. The effect of limestone injection on the combustor heat transfer surfaces was determined by comparing the heat transfer rates through a slag panel and fouling probes during periods with and without limestone injection. Combustor operation was determined by noting any operating problems that occurred during the tests. Test procedure After furnace wall temperatures had stabilized at operating conditions, heat transfer surfaces (slag panel and fouling probes) and the convection section were blown clean, and a four-hour test was begun. The

Jan 1, 1987