Search Documents

MR. GIBSON (continuing): In connection with the coal from this mine, I will say that it is not a very dirty coal. The vein where most of the coal comes from is about eight feet thick, and two feet from the bottom there is a small boney band about two inches thick. The breaking shot has to be placed over the, band, therefore breaking the boney into fine particles which goes off with the slack. This coal is a very good domestic coal and the slack is in great demand for various power, plants, such as sugar plants, smelters and other industries, but where this, boney is broken up and goes into the slack, it is very injurious in effect and retards the sale of the slack. Therefore the Spring Canyon Coal Company went to the expense of' building this plant, costing over $250,000, just in order to get that, two inches of boney coal out of the rest of the vein. The market in, Utah is very critical, and in order to sell coal, the product must be thoroughly clean in every way and, as Mr. Murphy says, I believe they have succeeded in making a very good commercial product. In any event the Spring Canyon Coal Company put out more coal than any other one mine during the last year, 1924, they had the largest production. I thank, you. (Applause). PRESIDENT MARKS: With the assistance of Mr. Robinson and Mr. Miller, I am sorry you could not get the pictures a little better synchronized with the story. MR. F. C. MILLER (Chief Chemist The Colorado Fuel & Iron Company, Pueblo) : We had seven slides to show eighteen different views on the screen and we did remarkably well. (Laughter.) PRESIDENT MARKS: Is there : any further discussion on this paper? MR. G. B. JONES (Mining Inspector, Victor-American Fuel Co.) : I would like to ask a question on the

Jan 1, 1926

By Fred C. Bond

This article is interesting because it deals with the fundamentals of grinding in a novel manner. However, it is somewhat reminiscent of Mark Twain's famous evaluation of science as the production of pounds of conjecture from ounces of fact. Extrapolations abound, and some cases they are treated with a seriousness that is undeserved. Tests should undoubtedly be made of wet grinding at supercritical speeds, using smooth mill lining and mill charges of large grinding balls as well as pebbles. However, the performance predicted in the article needs more qualification than it has received.

Jan 5, 1958

By Katzer R. Klimpel, J. Sewell

Dr. Roussey Roussev The use of chemical additives on grinding is of significant importance considering the high energy consumption of grinding processes. My research indicates that the dispersant approach explains the effects, in this particular case of wet grinding. Further, the dispersant approach combined with torque measurements give the possibility of clarifying the mechanisms involved when using grinding aids. I find the conclusions derived from the data obtained to be correct and useful for grinding. In general, I agree with the explanations. However, I would like to discuss several additional points that I consider to be important, and in the same time to mention some of my own results from investigations similar to those presented in the study under discussion. The "bed volume concept" and the change of slurry viscosity explain adequately the data plotted on fig. 3, in conjunction with Figs. 1 and 2. However, the significant increase of the average torque in the case of 83% solid with XFS-4272 introduces the necessity of establishing the optimum of percent solid when dispersant is used. For example, the optimum percent of solid for different materials varies. One of the reasons is the material's mass density. On the other hand, it is necessary to determine the concentration of the chemical used. Because of this, it would be interesting to know how the cited concentration was selected. The supportive discussion and verification by Dr. Roussev, with results similar to that of the paper, is appreciated. I will try to answer the additional questions he has raised. To begin with, the mill was fitted with six lifters of approximately one-quarter inch height. There has been extensive lab and plant level work to understand downstream influences of using dispersants like XFS-4272*. In sulfide ore processing, which is the area in which most industrial scale testing has been performed with this approach, neutral to positive influences have been observed. For example, positive effects (e.g., increased recovery) have been noticed in flotation operations involving ores that have a tendency to produce excessive slimes upon grinding. (1) Also, the use of a dispersant has been shown to demonstrate a positive influence on classification by improving the sharpness of separation and preventing a decrease in d50 value that usually accompanies increasing flow rates to classifiers.(2) The biggest potential downstream problem is in thickening, which at the dosages involved, has not been a problem in plant scale usage. The commercial price of dispersants like XFS-4272 is quite variable depending on source, concentration, etc, but a typical cost of XFS-4272 to achieve a 10% increase in throughput in industrial mills has been between 20 and 35- US per ton of feed ore (1981).

Jan 12, 1982

By K. Hanumantha Rao, A. Nayak, K. S. Narasimhan, S. N. Mahapatra

B.A. Hancock It is not at all surprising that causticized potato and potato derived amylopectin starch solutions performed much better than their parent starches. Some preparation is required to rupture the starch granules to effect the polymeric adsorption and interparticle bridging necessary for selective flocculation. In laboratory work comparing the deslime performance of causticized and autoclave cooked laboratory corn starch solution preparations, it was found that higher deslime weight rejections, with attendant proportionally greater iron unit losses, occurred with the causticized starch. These results may be specific to the ore involved but they do suggest that cooking and causticizing cause different starch granule rupture and/or starch breakdown, which have an effect on desliming response. I calculated from the data in the article that the slimes product grades were high - 24.3% and 20.3% Fe when 53.7% and 54.5% Fe concentrate products were obtained, respectively, in Table 4, and 21% Fe with a 62.6% Fe concentrate in Table 5 - using the natural tailings sample, which had a head of 34.3% Fe. It may be advisable for the authors to consider different starch preparations in future investigations. The combination of upgrading and selectivity results presented in Table 4 are not as good as the authors suggest. The authors' claim that a system has been developed to produce saleable concentrates from the Kudremukh tailings is quite disconcerting. There are many hurdles yet to be crossed before commercial application of selective flocculation becomes possible because differences between the very small-scale laboratory tests conducted and commercial application are rather large. Among the many differences are varying circuit feed grades that will occur from use of tailings, the apparent face that much lower tailings grades will be encountered in practice (it is much easier to achieve a high concentrate grade with reasonable recoveries using 34.3% Fe tailings as in the study rather than 25.3% Fe tailings grades that the plant apparently averages), the hydraulic nature of the thickeners used in operations compared to the static system used in laboratory tests, the different size distributions that will be obtained from a plant closed grinding-classification circuit, and differences in water used in a plant operation and the laboratory. The authors wrote that it was necessary to overgrind to be sure that the coarse gangue would not settle with the iron oxide floccules. This situation is likely to be exaggerated in commercial operations where it is assumed cyclones would be used for classification. Because cyclone classification is greatly influenced by particle densities, there will probably be an even greater difference in size between the iron and gangue particles in the plant, which would make the gangue slightly coarser still in relation to the iron. This would make the selective flocculation-desliming separations using the procedure employed by the authors even more difficult and, using the dispersant system the authors employed, greater overgrinding would be required. To grind finer to minimize the coarse gangue in the flocculated iron oxides is quite inefficient and appears not to broach the problem. The actual problem appears to be insufficient dispersion of the ground pulp. In this situation, addition of a dispersant would likely be required to attain a sufficiently high pulp dispersion level to efficiently effect a selective flocculation-desliming separation. Although the very coarse particles would still have a tendency to settle with the floccules, it probably would be found unnecessary to overgrind as much as indicated. Use of an optimum combination of dispersant and pH modifying reagents may also significantly improve the selectivity of desliming. Additionally, although it is possible that sufficient dispersion may be obtained by pH control alone in some situations, it is quite probable that added dispersity was obtained in the reported work from using distilled water. It is research experience that distilled water enhances dispersion. In commercial operations it may not be expected that sufficient dispersion will be obtained by pH control alone, unless the water used in the process is by nature quite dispersive. Overall, a change in the Kudremukh tailings dispersant scheme appears necessary where a dispersant is used in conjunction with a pH modifying reagent. With this change, different dispersion-flocculation responses will result that would have to be further evaluated. Therefore, it is still an open question whether an efficient and effective selective flocculation separation using Kudremukh tailings may be obtained that will produce saleable concentrates.

Jan 9, 1986

T. T. READ, BUREAU OF MINES: Mr. McAuliffe has asked a very practical question. Expressed in another way, the question is: If we have decided to use rock dust to prevent coal dust from exploding, have we got to take rock, that is, solid material,, break it up and finally grind it into dust (and fine grinding is very expensive), or can we take material that is already in a pretty fine state of subdivision and thereby save the expense of pulverizing it? Most of the preceding discussion has been from the view-point of practical experience, and I should like to suggest that we have not had enough practical experience in rock dusting to be able to answer this question, from the viewpoint of experience, with the degree of definiteness that Mr. McAuliffe desires. With your permission I will discuss the matter briefly from the theoretical standpoint, because after all the theoretical considerations, insofar as we understand them correctly, will apply to every' case, whereas in coming to a decision on this matter it will be necessary to take into consideration local conditions, which will necessarily vary in every case. Considering the matter in a rather elementary way, the first point is: What makes coal dust explode? The air of this room is full of dust at the present moment, even though we do not notice it. If you were to close up the room for a week, enough dust would settle on the floor so you could write your name on it. Of course, the dust which is in the air of this room is not coal dust but even if it were coal dust and you were to bring an open light in' here we all know that the dust would not explode. The reason for that is that the dust particles are not close enough together for the flame to propagate itself from one to the next. You will probably be able to see in the demonstration in the gallery on Saturday that the coal dust explosion is practically a very rapid burning of the dust. When you light one piece of dust it burns and gives out heat and if there is another piece of dust close enough to it so that the heat from the first piece is able to bring a second piece up to the burning temperature, then the flame will propagate itself through the mixture. But, if the particles are so far apart in the air that the' heat from one particle is dissipated before it reaches an adjoining one, and is therefore not able to bring the second particle up to the burning temperature, then the flame will not propagate. To get a coal dust explosion, it is not sufficient to have coal dust suspended in the air. The coal dust particles must be close enough together in the air so that the heat from one burning particle is able to heat the adjoining particles up to the point at which they will burn. There has been some discussion of the size of dust particles that will cause an explosion to propagate, and it has been pointed out that the minus 200-mesh dust is the most dangerous, and that particles coarser than 20-mesh will not propagate an explosion. There are two reasons for that. A large particle has very much smaller surface in proportion to its weight than a small particle has. You know the old story about the Indian who when asked how he could go around in cold weather without any clothes on said that an Indian was all face. A 200-mesh particle is almost all surface and there= fore is able to burn almost instantaneously, which means that it gives out its heat almost instantaneously, whereas a large particle with less surface in proportion to its weight burns more slowly, gives out its heat more slowly, and the heat has more time to be dissipated. The second reason why coarse dust does not easily propagate a flame is because it is so hard to make it

Jan 1, 1924

By J. Hanna, H. R. Gonzalez

Discussion by N.L. Hargett Dear Dr. Hanna: As discussed in our phone conversation, we are enclosing a suggested revision to the paper, "Application of Bentonitic Clays in Suspension Fertilizer Formulations," published in MINERAL & METALLURGICAL PROCESSING, May 1990. The revision will replace the second paragraph in the article. The reason for the revision is that suspensions have decreased in market share since 1984 and at present comprise an estimated 25% to 30% of fluid mixtures. This represents 6% to 7% of all fluids and only 2% to 3% percent of all fertilizers. Based on national surveys conducted by the Association of American Plant Food Control Officials (AAPFCO), suspensions comprised an estimated 25% of all fluid mixtures in 1974, 33% in 1976, 35% in 1980, 40% in 1984, and 36% in 1988. Revision to "Application of Bentonitic Clays in Suspension Fertilizer Formulations." Suspension fertilizers are an important segment of the fertilizer industry in the United States. Suspensions allow higher concentrations of nutrients (N, P2O5, K2O) than clear liquid solution fertilizers. K20 content of solutions is limited to about 10%, while suspensions can contain more than 30% K2O. Suspensions can tolerate greater variations in temperature than solutions. In suspensions, if crystallization occurs, the crystals can remain suspended in a clay jell instead of settling as they would in solutions. Solubility is not a limiting factor in suspensions, since complete solubility of phosphate is not required. This allows a wider range of phosphate materials to be used. Also, materials commonly used in formulating suspensions are less expensive and more widely available than those for clear liquids (Getsiner et al., 1984).

Jan 1, 1992

By R. J. Pindred

Discussion by R.J. Pindred In his paper, Davis presents an overview of risk. He also introduces the Capital Asset Processing Model (CAPM) as a foundation for selecting the appropriate discount rate for a mining project. While applying portfolio theory is more defensible than the ad hoc adjustment of discount rates, the CAPM is not a panacea. CAPM shortcomings [The CAPM, as Davis stated, is expressed in the equation: ri=rf+pi4) where ri is the project discount rate rf is the risk free interest rate (3i is the project beta, and 0 is the market risk premium (rm - rf)] Application of the CAPM is more difficult than Davis indicates. Valuation is prospective, while the CAPM parameters are historical. Beta is determined from a regression analysis of historical data, while the beta needed for valuation is the expected beta. Betas are known to be unstable and the regressions that generate them often have low explanatory power. The difficulty of estimating a "project" beta must also be considered. Thus, the beta that is used in the CAPM will be based on the analyst's judgment. Like Cavender's discount rate, this judgment can lead to different project NPVs. Subjectivity in valuation cannot be avoided by a mechanical application of the CAPM. The risk-free rate, which Davis identifies as a short-term real rate of 4%, is also subject to scrutiny. A mining project is not a short-term investment and no single risk-free rate is appropriate for all of the cash flows. The hypothetical mine discussed in Cavender's paper is a six-year project. One might argue for the application of a risk-free rate from the Treasury yield curve at the duration of the project (in a bond-duration sense). This, too, is inappropriate. The risk-free rate should be matched to the timing of the cash flow. These rates can be determined by calculating the implied forward rates from the yield curve using a procedure known as "bootstrapping." It is likely that each of the project's cash flows would be discounted at a different rate. Commodity prices Davis criticizes the "ad hoc adjustment to the discount rate." Yet, in his discussion of the value of stochastic simulation, he suggests that the gold price be modeled as a "random walk, with or without a trend." This is essentially an arbitrary modeling of price risk. Consider that a liquid market in gold futures exists. The futures' price curve, which is closely related to the market's estimate of future spot gold prices, should be used to provide inputs to the model. This is especially true of a relatively short six-year project. Alternatively, as Davis correctly points out, a risk-averse investor can sell the commodity short to hedge price risk. Is it any more correct, in the portfolio sense, to account for price risk at all ?? References Cavender, B., 1992, "Determination of the optimum lifetime of a mining project using discounted cash flow and option pricing techniques," Mining Engineering, Vol. 44, No. 10, pp.1262-1268 Fabozzi, F.J., 1993, Bond Markets, Analysis and Strategies, Second Edition, Prentice Hall, Inc. Higgins, R.C., 1992, Analysis for Financial Management, Third Edition, Richard D. Irwin, Inc. Solnik, B., 1991, International Investments, Second Edition, Addison Wesley Reply by G.A. Davis Pindred discusses two issues related to my paper, the shortcomings of the Capital Asset Pricing Model (CAPM) and which commodity price values to use in the valuation exercise. Even though these topics are not directly related to the use or misuse of Monte Carlo simulation, they are important points to take into consideration in valuation exercises. Since I do not appear to have addressed these issues satisfactorily in my original paper, I will comment on each here. Pindred agrees with me that applying portfolio theory, and specifically the CAPM, to the selection of project discount rates is more defensible than ad hoc methods. But he then points out that the application of the CAPM to project valuation is more difficult that I indicate. It is true that the CAPM is a difficult tool for project valuation in general,. But the application of the CAPM to mining projects is one of the easiest I can think of. The biggest problem with using the CAPM for project valuation is coming up with an expected project beta. I suggest a project beta for gold projects of 0.45. The "true" value might be 0.35, 0.55 or whatever. Pindred correctly notes that the selection of the appropriate project beta is based

Jan 1, 1996

By B. E. Grant

There are several exceptions from the practical operating viewpoint that might be taken with the paper. The author recommends using a "curve" in plotting safety factors of ropes and suggests that the U. S. Bureau of Mines change its present "step curve" method to the "curve" method. From an engineering theoretical point of view that is possible, but from a practical standpoint it is not possible. Rope manufacturers fabricate their ropes in diameter gradients of 1/16 in. to 1/4 in. which automatically sets up a step pattern gradient. If ropes were manufactured in the diameter gradients of 1/1000 of an inch or thereabouts, the curve method could be used. Also, from a practical standpoint, if an operator is to hoist from a depth of 4000 ft, for example, he will determine the size of the rope necessary for that depth, and it is only natural that he will have safety factor to spare at lesser depths, and in plotting this excess capacity the natural end result is a curve. The U. S. Bureau of Mines tables are set up so that the operator will know what size rope at the required safety factor he will need at his maximum depth. He is not concerned with lesser depths. W. A. Boyer (author's reply)-Mr. Grant states, "Rope manufacturers fabricate their ropes in diameter gradients of 1/16" to 1/4" which automatically sets up a step pattern gradient." There is no connection or relationship between the "size" pattern gradient and the present "factor of safety" step pattern gradient. The inherent factor of safety values for any particular size of rope with a given connected load when plotted for various depths is a curve, and with modern rope material and construction the inherent factor of safety curve is quite far removed from the present recommended factor of safety step curve. This is shown quite clearly in Figs. 3 and 4 of my paper.

Jan 11, 1954

Three -dimensional boundary value problems are difficult to solve. Indeed, while the separation of the scalar wave equation can be effected in 11 different coordinate systems, an analytic solution requires that the boundaries, both external and internal, possess the same symmetry as the coordinate system. Numerical methods are thus of great importance for the solution of such problems; however, despite the availability of high-speed, large memory digital computers , the solution to a significant three -dimensional problem is by no means trivial. It is unfortunate that the results reported here by Jones are invalid. Jones discusses a model in which a three-dimensional island lies off a linear coastline where all the interfaces lie in the coordinate planes in a Cartesian system. A downward plane em wave polarized with the electric field parallel to the linear coastline, and two of the islands coast, is incident downward. This, of course, requires that the incident electric field is perpendicular to the other faces of the idealized Cartesian island. The significant equation used by Jones for his finite difference calculations is [ ] where n2 = sµ? . is the conductivity, µ=µo is the permeability of the appropriate medium, and w is the angular frequency of a Fourier component of the incident field. We see that the displacement current term is omitted, which is an excellent approximation for the parameters of interest in this model. In the similar equation for the y- component of the diffusion equation, the same considerations which we are about to discuss are equally valid. It can be seen from the equation that the field components are related and thus any errors introduced will be propagated into all three components. The difficulty arises when Ex or E is perpendicular Y to one or other of the surfaces of discontinuity in s. The continuity of current at the air-earth interface (upper surface), e. g., jy, requires

Jan 1, 1973

By P. J. Ehret

By and large, I agree with many of the precepts discussed in the Saperstein paper. I believe there is much more that can be done on the local level to involve both government and planning in the mine permit review process. Admittedly, much can be done to improve discussion between the state regulatory authority, local authorities, and mine operators. However, I strongly disagree, as suggested in the paper's conclusion, that such a system, as outlined in Saperstein's paper, could result in a reduction of review time or a lowering in the cost of permitting. The implementation of such a system by its very nature would only add a new level of bureaucracy to an already cumbersome process. This system, in my opinion, would only add to the cost of permitting and the time necessary to obtain a permit. Admittedly, it would achieve the result of adding local influence into the permitting process, but at what off-setting cost? It has been my professional first-hand experience and observation in the permitting and lands unsuitable processes to note that under the present system, local governments that do feel they have a stake in local mines and actively advocate their position (not simply comment), have had significant input into the conduct of mining and reclamation plans in their areas. Mining companies that mine in these areas are aware of local interests and attitudes, and do consider them in their plans. Simply put, local governments that want to get involved and want to play significant roles can and do. Those that do not advocate a position or have no input to offer do not. Although the present system is not as direct in effect in regard to local control as Saperstein's, it does work. It does so without the red tape that would result from developing miniature regulatory authorities in every county in which mining occurs. Although local input is important, there are many more sections of the Surface Mining Control and Reclamation Act (SMCRA) than Sections 508(a)(8) and 522(a)(3)(A) that also are important. It appears Saperstein's paper may overindulge these sections. An additional specific comment is made in regard to the statement that while local areas benefit from mining (wages, salaries, and purchases), they also endure environmental deterioration caused by mining affects. It is believed that the whole purpose for which SMCRA and its resulting regulations were enacted and developed was so mining would not be permitted if significant environmental deterioration could occur. Therefore, it could be argued that this point under current law is moot.

Jan 1, 1985

By Wallace W. Wilson

This DaoerAA calls to attention for the first time, to the writer's knowledge, a purported recovery of oil by substantially horizontal gas cycling that is considerably in excess of 50 per cent of the oil originally in place. The average angle of dip along the path of travel of the gas is less than one degree of arc. The best recovery by gas cycling previously brought to the writer's attention1 occurred in the case of the Mile Six Field in Peru, where about 68 per cent of the original oil in place has so far been recovered. In this case, however. the recovery was greatly aided 1)y the high dip angle of about 17.5". An approximate estimate of recovery was made by the method developed by the writer' for the case of the Cook Ranch Field, using an average relative permeability ratio function dictated by past experience with a number of good quality sandstones. A result of about 40 per cent recovery of the original oil in place wa* obtained after the 8.5 pore volumes of cumulative gas cycling undergone by the Cook Ranch Field. Tile writer desires to emphasize what appears to him as the sound engineering practiced at Cook Ranch, and also the care attending the taking of data and the high quality of the calculations made. It is impossible to take issue on factual grounds with any of the conclusions drawn in the paper. It is, however, desired to point out the difficulty of explaining the unexpectedly favorable oil recovery on ordinary theoretical grounds. It would be of interest to know whether relative permeability ratios have been measured on the Cook sand. The operator is inclined to attribute some of the excess benefit to settling of the oil across the bedding planes into the lower part of the sand. It is difficult for the writer to appreciate the importance of such an effect, inasmuch as the concomitant opening of excessive vacated pore space in the upper part of the sand would he expected to cause excessive bypassing of the gas. REFERENCE 1. Welge. H. J.: "A Simplified Method for Computing Oil Recovery by (Gas or Water Drive." Trans. AIME, (1952), 195, 91. AUTHOR'S REPLY TO MR. WELGE Welge has poined out clearly the difficulty of applying material-balance procedures in calculating the Performance of low-pressure, dispersed gas injection operations. There can be no argument that even with the most favorable assumptions of K,/K,, and with cumulative gas cycling in excess of normal expectations, the calculated oil recovery will be considerably less than the 72.5 per cent recovery realized from this project to Jan. 1, 1951. There probably are at least two reasons for the unusually high recovery from this reservoir, the effects of which would be difficult or impossible to calculate by known methods: 1. The operator did not use the same injection wells for the entire operation, and exercised considerable control over producing GOR by changing the locations of injection wells whenever well performance indicated excessive bypassing. Fifty-six wells were used for injection, not over 29 of which were in use simultaneously. This procedure resulted in frequent shifting of the fluid movement patterns in the reservoir and undoubtedly increased the ultimate recovery. The writer knows of no satisfactory procedure for calculating the effect of this practice. 2. Vertical movement of oil across the bedding planes of the reservoir rock hy gravity drainage tended to maintain a relatively high oil saturation, and hence a favorable K,/K, ratio, in the lower sections of the reservoir. As pointed out by Welge, this condition has aggravated the tendency for high producing GOR, but at the same time it has resulted in increased oil recovery. As may be seen from the performance charts for the various battery areas, operating (GOR'S have been quite high for the last several years of reported history.

Jan 1, 1952

By B. Mather

Discussion by B. Mather I was very glad to see an article on alkali-silica reactivity of concrete aggregate in Mining Engineering. It is a topic that should be reexamined every 45 or 50 years. The last paper (that I am aware of) that examined this topic in Mining Engineering was Rexford (1950), who, at the time, was chief of the Petrography Section, South Pacific Division Laboratory, Corps of Engineers, Los Angeles, CA. The paper was presented at the AIME meeting in Los Angeles in October 1948. Professor Leming's discussion generally reflects the current state-of-the-art; but it provides no references for more detailed information. I wrote a paper titled "How to avoid excessive expansion of concrete due to alkali-aggregate reaction," which was included in the Proceedings of the Second International Conference on Alkali-Aggregate Reactions in Hydroelectric Plants and Dams (Mather, 1995). In it, I emphasized that it was the hydroxide ion (not the alkali ion) that caused the pore fluid in the concrete to dissolve reactive silica, producing alkali-silica gel capable of taking up water, thereby, causing swelling and rupturing of the concrete. Excessive expansion will not occur unless there is a sufficient amount of potentially expansive alkali-silica reaction product and water. However, this will not occur unless there is a sufficient quantity of reactive aggregate and unless there is a mechanism by which the pH of the pore fluid can get well above the normal value of about 12.6. In many cases it is sufficient that the cementitious medium not only contains portland cement but also contains ground granulated blast-furnace slag (GGBFS) or a pozzolan such as fly ash, silica fume, metakaolin, volcanic glass or calcined shale - any of which may have been included for economic reasons. If, however, the person selecting the ingredients for concrete knows that the aggregate is reactive (with no alternate nonreactive aggregate available) and if the person knows that the only available portland cement has a high alkali content (i.e., more than 0.6% Na2Oe; Na2Oe = % Na2O + 0.658 x % K20), then it becomes irresponsible not to establish what reasonable precaution one should take. Professor Leming suggests using one of the following: 30% Class F fly ash, 50% GGBFS or 7% to 10% silica fume. My view is that one should prepare test specimens with the cement and aggregate to be used and then expose these both with and without at least two dosages of each of the various alternative materials (slag, pozzolan and lithium salts), so that the required dosage to control the expansion is established. Then, economic considerations can dictate the final selection. I would point out to the author that GGBFS is not a pozzolan and is not an admixture, but, rather, it is a latent hydraulic cement. An admixture is, by definition, a material other than hydraulic cement. I commend him for suggesting that aggregate producers should become more active in knowing about the reactivity of their products and the precautions needed to permit those products to be safely used in concrete. ? References Mather, B., 1995. "How to avoid excessive expansion of concrete due to alkali-aggregate reaction," Proceedings, Second International Conference on Alkali-Aggregate Reactions in Hydroelectric Plants and Dams, US Committee on Large Dams, Denver, CO, pp. 421-439. Rexford, E.P, 1950, "Some factors in the selection and testing of concrete aggregates for large structures," Mining Engineering, Trans. AIME, Vol. 187, pp. 395-402. Reply by M.L. Lerning Mr. Mather has once again provided a cogent and useful discussion. I am grateful to him for bringing these issues to my attention, as well as to the readers, and I am grateful for his discussion of several important factors in alkali-silica reactivity. Mr. Mather's discussion of the mechanism of alkali-silica reactivity (ASR) notes the critical role of the hydroxide ion. While the alkali hydroxides are highly soluble (reducing the solubility of the other primary hydroxide in pore solution, i.e., calcium hydroxide, and providing hydroxide ions that initiate alkali-silica reactivity), it is, as Mr. Mather quite correctly notes, the hydroxide ion that attacks the silica structure. This point was not clearly stated in my original article. As noted in the text, the requirements for mineral admixtures to mitigate or control ASR vary widely depending on the specific characteristics of the materials

Jan 1, 1998

By M. E. Gershon

I would like to congratulate Professor Gershon on an idea that is obviously valuable and is certainly new to me. It has, however, to be further developed, in that one can get fallacious results by limiting oneself to positional weights. In the example problem, the positional weight of one block was 8.6 (the 8.0 value in the paper was incorrect) for 36 units in the cone, for a mean block value of 0.239. The positional weight of the second block was 4.4 for 15 units, for a mean block value of 0.293. Surely one would expect the second block to be rated higher. Additionally, the time value of money adjustment cannot be determined by depth, because the time that a block is mined is determined by the mining sequence, not depth. Figure 1 shows a slightly altered version of Professor Gershon’s pit. The overall value of the pit is the same, as are the total values under each of his blocks. The values in the right-hand cone have been reversed top to bottom, and the values in the rest of the pit adjusted to keep the central cone total unchanged. All values have been multiplied by 10. Figure 2 shows the sequence of mining, at three blocks per year, for Professor Gershon’s pit, and Fig. 3, the sequence for a pit under the other block. Table 1 shows the value and discounted value (157. ) of Professor Gershon’s pit, and Table 2 shows similar values for the second pit. It is apparent that the second pit schedule is more valuable than the first.

Jan 12, 1987

By C. J. Hall

I would like to congratulate Professor Gershon on an idea that is obviously valuable and is certainly new to me. It has, however, to be further developed, in that one can get fallacious results by limiting oneself to positional weights. In the example problem, the positional weight of one block was 8.6 (the 8.0 value in the paper was incorrect) for 36 units in the cone, for a mean block value of 0.239. The positional weight of the second block was 4.4 for 15 units, for a mean block value of 0.293. Surely one would expect the second block to be rated higher. Additionally, the time value of money adjustment cannot be determined by depth, because the time that a block is mined is determined by the mining sequence, not depth. Figure 1 shows a slightly altered version of Professor Gershon's pit. The overall value of the pit is the same, as are the total values under each of his blocks. The values in the right-hand cone have been reversed top to bottom, and the values in the rest of the pit adjusted to keep the central cone total unchanged. All values have been multiplied by 10. Figure 2 shows the sequence of mining, at three blocks per year, for Professor Gershon's pit, and Fig. 3, the sequence for a pit under the other block. Table 1 shows the value and discounted value (15%) of Professor Gershon's pit, and Table 2 shows similar values for the second pit. It is apparent that the second pit schedule is more valuable than the first.

Jan 1, 1988

By E. Tulcanaza

M.S. Azun I do not at all agree with the basic points of the author's conclusion. The use of lognormal or normal model to respond to the attribute distribution function should be carefully questioned. If fitting a distribution function of regionalized variables is the main purpose, one may employ one of certain well-known families of distributions, such as Tukey's lambda functions. For instance, the distribution function of gold assay values studied by Krige (Krige, 1978) may fit the lognormal model well, but one should not exclude that the basic idea of using the distribution function of observed data is to have insight of the physical mechanism that gives rise to the regionalized variables. In Applied Geostatistics, there are many examples considering only lognormal or normal model for the distribution function of geostatistical data even if the observed distribution function has more than one mode. Of course, using lognormal or normal model always brings easy computation for the statistical properties of attribute under investigation. Still, I object to adopting any of those models without having any "statistical inference." As seen from the author's paper, lognormal and normal model produce the same average grade (0.746% copper) and almost the same sample variance. Therefore, which model explains the probabilistic behavior of ore deposit is always an important question that should be replied by the geostatistician using not the common sense but the statistical and probabilistic methods. For kriging procedures, such as linear kriging, one should describe either second order stationary properties or bivariate properties of regionalized variables. Second order properties, such as correlogram function or semivariogram function, are the inputs of the so-called kriging equations. Selecting one of any useable models for the observed second order properties, such as spherical model, is not easy since the models are not based on the correlations structure of the regionalized variables. How those models being used in Applied Geostatistics can be distinguished is another important problem. One may develop many fitting models to respond to the observed second order properties of regionalized variables. However, I suggest that the model should be based on the probabilistic behavior of geologic process. The author used spherical model for copper, wolfram, and silver grades from isotropic sector (Fig. 2, 4, 3b) and exponential model for silver grades obtained along raises in isotropic sector (Fig. 3a). It seems that those semivariograms given in Fig. 3b for isotropic sector and Fig. 4 may be described by the random model implying that regionalized variables do belong to renewal process (Azun, 1983). If I knew the total number of data points used in estimation, I may prove that my conclusion is correct. Recalling the test for the first order sample correlogram value, the test procedure is introduced such that the independence is rejected if [ ] where Za/2 is the a/2 percentile of standard normal random variable and N is the total number of samples used in estimation (Azun, 1983). For the other observed semivariogram functions I suggest the author might try to use the so-called "Markovian model" describing not only the correlation structure of regionalized variables, but also the physical mechanism that produces the regionalized variables. The Markovian model for correlogram function is, [ ] The Markovian model for the other second order properties of ReV are also derived. T ve B, in the above equation, show, for example, structural change and mineralizational variation in a considered deposit, respectively (Azun, 1983). In selecting any model, it is not easy to search for the "best" response to the observed second order properties of ReV's. The Markovian model, based on a theoretical understanding of underlying mechanism, gives more information about the occurrence of regionalized variables and respond to all properties. Random model and the so-called hole effect structure can be easily defined as a special form of the Markovian model (Azun, 1983). An estimation of any second order properties of regionalized variables may be computed through (N-1) lag. However, the dependency between the regionalized variables may be deemed to have "died out" after the so-called stationarity width (range). In practice, the estimation is carried out through 25% of the total number samples available. Therefore, the large fluctuations around the zero level and variance for correlogram or covariogram function and semi-variogram function, respectively, cannot be observed. Since the number of pairs involved in the estimation of higher second order properties is small, the estimation variance at those lags is large. There is no need to show higher order estimated second order properties. After using one of any kriging procedure for block estimation, the distribution function of average grades has the same mean as the drilling sample grades but

Jan 3, 1985

By M. S. Azun

I do not at all agree with the basic points of the author's conclusion. The use of lognormal or normal model to respond to the attribute distribution function should be carefully questioned. If fitting a distribution function of regionalized variables is the main purpose, one may employ one of certain well-known families of distributions, such as Tukey's lambda functions. For instance, the distribution function of gold assay values studied by Krige (Krige, 1978) may fit the lognormal model well, but one should not exclude that the basic idea of using the distribution function of observed data is to have insight of the physical mechanism that gives rise to the regionalized variables. In Applied Geostatistics, there are many examples considering only lognormal or normal model for the distribution function of geostatistical data even if the observed distribution function has more than one mode. Of course, using lognormal or normal model always brings easy computation for the statistical properties of attribute under investigation. Still, I object to adopting any of those models without having any "statistical inference." As seen from the author's paper, lognormal and normal model produce the same average grade (0.746% copper) and almost the same sample variance. Therefore, which model explains the probabilistic behavior of ore deposit is always an important question that should be replied by the geostatistician using not the common sense but the statistical and probabilistic methods. For kriging procedures, such as linear kriging, one should describe either second order stationary properties or bivariate properties of regionalized variables. Second order properties, such as correlogram function or semivariogram function, are the inputs of the so-called kriging equations. Selecting one of any useable models for the observed second order properties, such as spherical model, is not easy since the models are not based on the correlations structure of the regionalized variables. How those models being used in Applied Geostatistics can be distinguished is another important problem. One may develop many fitting models to respond to the observed second order properties of regionalized variables. However, I suggest that the model should be based on the probabilistic behavior of geologic process. The author used spherical model for copper, wolfram, and silver grades from isotropic sector (Fig. 2, 4, 3b) and exponential model for silver grades obtained along raises in isotropic sector (Fig. 3a). It seems that those semivariograms given in Fig. 3b for isotropic sector and Fig. 4 may be described by the random model implying that regionalized variables do belong to renewal process (Azun, 1983). If I knew the total number of data points used in estimation, may prove that my conclusion is correct. Recalling the test for the first order sample correlogram value, the test procedure is introduced such that the independence is rejected if [Ir(1) I > Za/2 N1/2] where Za/2 is the a/2 percentile of standard normal random variable and N is the total number of samples used in estimation (Azun, 1983). For the other observed semivariogram functions I suggest the author might try to use the so-called "Markovian model" describing not only the correlation structure of regionalized variables, but also the physical mechanism that produces the regionalized variables. The Markovian model for correlogram function is, [p (h) = T 1ih , h>1 ,] The Markovian model for the other second order properties of ReV are also derived. [T veß], in the above equation, show, for example, structural change and mineralizational variation in a considered deposit, respectively (Azun, 1983). In selecting any model, it is not easy to search for the "best" response to the observed second order properties of ReV's. The Markovian model, based on a theoretical understanding of underlying mechanism, gives more information about the occurrence of regionalized variables and respond to all properties. Random model and the so-called hole effect structure can be easily defined as a special form of the Markovian model (Azun, 1983). An estimation of any second order properties of regionalized variables may be computed through (N-1) lag. However, the dependency between the regionalized variables may be deemed to have "died out" after the so-called stationarity width (range). In practice, the estimation is carried out through 25% of the total number samples available. Therefore, the large fluctuations around the zero level and variance for correlogram or covariogram function and semivariogram function, respectively, cannot be observed. Since the number of pairs involved in the estimation of higher second order properties is small, the estimation variance at those lags is large. There is no need to show higher order estimated second order properties. After using one of any kriging procedure for block estimation, the distribution function of average grades has the same mean as the drilling sample grades but

Jan 1, 1986

By M. J. McPherson

Having worked on the thermodynamics of air/liquid-water mixtures passing through the surface fans of deep mines, I find this paper of great interest and congratulate the authors on producing it. There are two matters, however, deserving discussion. First, the authors have described the classical theory of fog formation with fogging occurring at "supersaturation." In fact, the process of fog formation begins well below 100% humidity. The more strongly hygroscopic nuclei in the atmosphere will attract water molecules and begin to grow at a relative humidity perhaps as low as 70%. The process is a dynamic one with both condensation and evaporation taking place throughout the mixture on microscopic liquid surfaces. As saturation is approached, the rate of condensation accelerates rapidly producing the familiar reduction in visibility. Hygroscopic nuclei are present in all natural atmospheres and under the appropriate conditions of pressure and temperature, will produce clean fogs. However, if the air is polluted by particulates from combustion or other processes then the resultant coagulation with growing liquid particles may produce dense (and sometimes photochemical) smogs. It is this process that is likely to occur in underground mines when moist air is cooled below dew point. Second, the authors have summarized very well the individual measures that might alleviate the problem. In particular, I agree that neither heating nor refrigerating the air, by themselves, provides a satisfactory solution. However, there is a combination of these that provides a neat and effective control of fog formation. This involves a small, self-contained refrigeration unit within a duct but without the usual external heat rejection facility. The air is cooled below dew point and, hence, dehumidified on passing over the cold evaporator coils. The heat from the condenser is rejected back into the air down-stream from the water eliminator, as shown in [Fig. 1]. The duct configuration can be designed to create good mixing at the outlet.

Jan 1, 1984

By Bunting S. Crocker

Bunting S. Crocker (Executive Vice President, Kilborn Engineering Ltd., Toronto, Ont., Canada)-This is an interesting report on laboratory-scale testing with pebble charges in 12-in.-diam mills, and the author correctly concludes that useful laboratory testing can be done. In the opening paragraphs he states, "laboratory pebble-milling tests have been considered too difficult to perform and not relevant enough to the actual grinding situation"!! This is definitely not so! The writer published detailed tests done in a 12-in.-diam mill comparing pebble and steel balls in 1952 6 and again in 1954 7. We note that our review paper in 1963 was the only one mentioned in the References. The author states that "a smooth lining is more frequently used in commercial-scale pebble grinding." The writer has designed some 22 plants treating 134,000 tpd in North America and Australia. He has also visited pebble installations in Norway, Sweden, Finland, and South Africa. All these plants correctly use heavier lifters than would be considered standard for ball mills of a similar size. Smooth liners are not recommended in wet ore treating plants. In the 1952 paper6 it was clearly demonstrated in the test work that the individual ore pebbles had to be the same weight-not the same diameter or size-as the steel balls if comparable grinding results were to be obtained. In the tests performed in this study most of the pebbles "passed 50-mm round hole screen openings and remained on a 40-mm round hole sieve. Correspondingly most of the ball mill experiments were performed with 50 or 40-mm steel balls". Thus, the balls and pebbles were essentially the same size or diameter. This means we are comparing apples and oranges. When the feed being ground is fine enough the pebbles will be superior. When the feed is coarse the steel balls will be the better. This is similar to grinding tests in which two sizes of steel balls are compared. In the middle of the article, "Crocker5 has reported that equal weights of pebbles and steel balls achieve the same grinding capacities under similar conditions …", then he proceeds to question this postulate. In the eight reports that I have published, I have never made this statement. What I have said is "for comparable grinding, the individual pebbles in the mill charge should be the same weight as the individual steel balls." This is not the first time that I have been misquoted on this point! It is most unfortunate that this test work has been done with a ball and pebble of the same diameter and not the same weight. It is impossible to properly evaluate the results. I also regret the fact that the screen analyses of the tests were not shown. Many of us who have been 40 years or less in the grinding game can easily read screen analyses at a glance. We can tell whether the top meshes are being ground by the ball charge, etc., etc. The graphs are fine for the summary or to illustrate conclusions but are often misleading in the report itself. However, we do agree that useful pebble grinding tests can be done in a 12-in. mill.

Jan 1, 1973

By J. Laskowski

J. Laskowski (Associate Professor, Dept, of Mineral Processing and Coal Preparation, Silesian University of Technology, Gliwice, Poland) -Referring to the papers by J. A. L. Campbell and S. C. Sun, I must say that I am very pleased to see that type of work, because it is the only way to progress towards an understanding of flotation mechanisms. All those interested in the science of the surface chemistry of coals would find these papers a valuable contribution to the literature. In their extensive research, the authors attempt to synthesize their findings into a theory of coal flotation. To do this, the zeta potential of bituminous coals and their lithotypes, as well as that of an anthracite coal, was measured and H+ and OH- ions were found to be potential determining ions. The measured ZPC of bituminous coals was at a pH range from 3.5 to 4.6 and for the anthracites between 2.0 and 4.5. The authors have pointed out that since the hydrophobicity is pH-controlled, the flotation mechanism should also be. The Campbell and Sun papers confirm a good many of the conclusions on surface properties of coals reported in recent years. It is noteworthy to point out that the ZPC found in the paper on bituminous coals was approximately pH 4, similar to the ZPC for the anthracite coal. In fact, we have been investigating different coals from low to high ranks. Thus, the plot of the pH for ZPC against rank could be derived and this curve-an electrocapillary one-indicates a pH near which the hydrophobicity of the coal surface should be at maximum. This can be proved by so-called salt flotation, which, in a sense, can be treated as method of surface charge measurement. It is obvious that the salt flotation experiment is possible when the native hydrophobicity of the mineral surface is high, e.g., when the solid surface does not have an orienting influence on an adjacent water layer and the hydration layers are controlled by the electrical double-layer potential. The addition of electrolyte or a change of pH which decreases the thickness of the stabilizing double-layer "opens up" the hydrophobic surface micro-area mosaic. Thereby the surface becomes hydrophobic and particles become floatable. The orienting influence of surface on hydration layer increases even after very slight oxidation and flotation is affected by it. It seems that such an explanation agrees reasonably well with those of Campbell and Sun.

Jan 1, 1972

By G. F. Raymond

Discussion by G.F. Raymond Knudsen's study presents two curious conclusions: • The kriging of blasthole assays can systematically overstate mill head grades by as much as 21% as a result of unbiased sample variance. • No estimation method is able to reduce this bias to a small margin. The study is based on a simulation using real data and what are presumed to be actual variogram parameters from a real deposit. Although I have no doubt that the first conclusion (21% overestimation) is correct for the author's simulation, I do not believe this represents a realistic mining situation. Over the past 15 years I have done extensive comparisons between exploration drill-hole assays, blasthole assays and mill head grades on seven major open-pit mines, including some very erratic gold deposits. Commonly, nugget effects on blasthole variograms were 10% - 20% higher than on exploration variograms. And in one extreme case, the difference was 50%. Even in the extreme case, ordinary kriging on blastholes agreed well with the mill head grade over the long term. In Knudsen's simulation, the blasthole nugget effect is assumed to be 200% higher than exploration data's. He supports this by variogram plots from each. My guess is that the apparent, large difference between these variograms results from a failure to account for the proportional effect (blasthole assays are likely from a higher-grade area). A simple check would be a comparison of the variance of exploration samples nearest blastholes. As for a nearly conditionally-unbiased estimator of a large random error, the arithmetic mean of all of the data certainly qualifies, provided there is an even data spacing. As a corollary, so does simple kriging, which would include, in this case, a large weighting to the arithmetic mean. Similarly, using a large number of samples with ordinary kriging or indicator kriging would significantly reduce the bias in the case of a large nugget effect for the variogram.[ ] Reply by H. Peter Knudsen Raymond questions two conclusions in the paper. First, he wonders whether a 21 % overestimation represents a realistic mining situation. I agree that 21 % is high, but overestimation in the range of 10% to 15% is certainly common in my experience. Furthermore, several years ago I consulted on a gold mine that was experiencing a 45% overestimation due, predominantly, to poor blasthole samples: In further questioning of the 21 % value, Raymond wonders whether the nugget effect of the blastholes is really so much larger than the nugget values of the exploration data. It is consistently larger throughout the deposit. In my experience with six Nevada gold mines, the high nugget value is not unusual. In fact, for some reason, nugget values for blasthole samples are typically about 0.0005 (opt squared). I am of the opinion that this high nugget effect observed at many gold mines is predominantly due to the inherent inadequacies of the blasthole sample and subsequent sample preparation. The second conclusion Raymond questions is the inability of the estimators tested to reduce the conditional bias. In fact, the conditional bias is extreme with the polygon estimator and greatly reduced by ordinary kriging. However, it was not eliminated. Raymond suggests using simple kriging, or perhaps a larger number of samples, to reduce the conditional bias. The technique of simple kriging may be less affected by the random errors in the data, but I did not test the technique. Using a larger amount of data presupposes that too few samples were used initially. In ordinary kriging and indicator kriging, the screen effect comes into play and ensures that samples beyond the second screen are given zero weight. Hence, increasing the sample size does not change the estimates nor the conditional bias. The main point of my paper is that the random unbiased errors (a fact of life in blasthole samples) cause a conditional bias in our estimates. The mechanics by which the conditional bias is introduced are nicely explained by Springett. My paper simply shows that the bias is also present when working with linear estimators, such as ordinary kriging, and even with nonlinear estimators, such as indicator kriging.[ ]

Jan 1, 1993