RICHARD H. SNOW
Size Reduction Fracture behavior, mill and circuit developments, and industrial classifiers are reviewed and discussed
Fracture Behcrvior
favorite topic this year is the nature of surface scratches in glass, which has practical applications in drilling and wear, and attrition in mills. T h e topic is intriguing, since it is not immediately apparent how a tool can get a start in the initially flat surface. Ernsberger (50) suggests that previously observed microplastic effects d o not actually involve plastic flow,but densification, which he measures. T h e critical stress for densification, rather than the tensile strength, dctermines the scratch hardness. Neely and Mackenzie (749) reach the same conclusion in studies of scratchiiig by a Vickers diamond. Dick ( 3 8 ) and Grosskopf and Scholze (0'8) also measure scratch hardncss. ,\ more familiar fracturc situation arises whcn a particle is pressed against another surface; this is the gcometry of the Hertz experiment, and this is how particles are usually strcssed in mills. Studies of the mechanics of this siruation are continued by Hamilton and Rawson (72j, who interpret their d a t a in terms of a flaw distribution. Billinghurst et al. 1/71 point out that when flat surfaces are placed together, the points of real contact are small, and the prcssures can be very high. If the surfaces are slid over one another, high tangential forccs can greatly incrcase the incidence of brittle fracture. Gilroy and I-Iirsr (65)show that the critical load to produce a ring crack in the Hertz experiment T-arics Lvith the coefficient of friction whcn a sliding load is applied. Examination of the fracture surfaces is one Lvay to gain information about the fracture process. Cottcrcll ( 3 4 ) finds the drnsity of markings proportional to the toughness of organic glasses. T h e electron microscope reveals to Klaus (103)that the mirror zone of a fracture is by no means a flat surface; hacklc marks produced by very high fracture velocities arc discusscd. .\nd Noskins ( 8 0 ) examines rock fracture surfaces Lvith a scanning electron microscope. Tl'iederhorn (273)continues attempts to measure the fracture surface energy of glasses by measuring thc energy required to cleave a specimen. Meloy and Gumtz (738)modify their derivation of a size law based o n raridom fracture to includc the effect of a n imposed stress distribution; this can be substitured into their equation. Articles by Bailey and Hill (7),Kelly (98), and Hasselman (75, 7 6 )investigate the strength of ceramics, which is strongly influenced by their porosity, grain structure: and crack density. Griffith crack theory is invoked to explain why ceramics fail far short of their theoretical strength ; it is hoped that better understanding can show how to improve strength. I n the case of green clays, Foster (57) shows that the variation in dry strength is mainly due to variations in residual stress caused by uneven shrinking during drying. Borchert et al. ( 2 3 ) attempt to classify the strength of types of rock accordillg to their grain structure, and they give data on the measured strengths, Single-particle breakage experiments in the laboratory can be correlated with the grinding behavior of materials. Lyall ( 7 2 6 ) describes a n experimental program a t the Mines Branch, Canada, to determine the impact breakage of numerous materials by a n impact device that can test many particles rapidly. Shand (789) investigates the impact breakage of glass and measures impact
A
36
INDUSTRIAL A N D ENGINEERING CHEMISTRY
energy and velocity, peak impact force, and breaking stress. H e concludes that fracture processes of glass under medium-speed impact are the same as under slow-compression loading. There are experimental pitfalls, since the design of the whole apparatus affects the stress transmittal, not just thc particle and striker. T h e impulse breakage of rock slabs is limited by the energy that can be transmitted by a compressive pulse through the specimen, according to Hakalehto ( 7 0 ) ; thus the damage from a blasting charge may largely be limited to pulverization adjacent to the hole. Hasselman (77) studies thermal shock fracture in ceramics and develops a theory of crack initiation and propagation. Thermal expansion is also the cause of surface spalliiig by a n exotic methodthe use of giant laser pulses-according to Bristow et al. (26). Complicated light absorption processes are involved when a lascr impinges on glass, according to Davit (37). The crushing dynamics of carbon foam is revealed by cincmatography in a n article by Schmitt (179). M i l l and Circuit Simulation and Control
T h e largest concentration of papers on this subject is included in the International Computer Symposium (209)of thc September 1969 -2IME Meeting. T h e proceedings were published beforc the meeting, a notable feat. Each of several investigators develop different mathematical techniques for back-calculating the grinding rate function (previously called the selection function) from mill performance data. Grandy, Gumtz and Fucrstenau (209) formulate a regression procedure in terms of the matrix reprcsentation of the batch grinding cquation, and fit their data from locked-cycle grinding (Bond experiment) to simulate closedcircuit grinding. Schonert (209)combines breakage and grinding rate parameters, and evaluates thc resulting function in a chemical engineering approach. Freeh and Horst (209) demonstrate the possibility of back-calculating the breakage function as well as the grinding rate function. Their results obtained by linear programming show a n unnatural trend, but the principle is sound. Klimpel and Austin (704) present a stepwise regression technique to back-calculate thc grinding rate function from mill data, assuming that the breakage function is given by their own proposed size equation. Mular (745) describes a gradient search method to back-calculate the grinding rate function from a single batch rod mill experiment with a n assumed breakage function. Thirtyfive itera'iions are required to reduce error to 1:&. T h e integro-differential equation describing thc grinding process in terms of the breakage function and the grinding rate function has been solved only for special cases. This is one reason for the slow adoption of modern comminution theory for practical applications. Kapur (95)presents a similarity solution of this equation for batch grinding. Kapur and Agarwal ( 9 6 ) notice that certain elements of the grinding matrix should be negligible, and this allows them to rearrange and simplify the batch grinding equation to obtain a n approximate solution. Harris (73, 7 4 ) criticizes the assumption often made that the grinding rate function does not vary with time. H e shows that some of the best laboratory grind data can better be fitted with a simple equation, Y ( x , t ) = 1 - exp(-bbtP), first proposed by
Alyavdin in 1938. Here Y is the cumulative product size distribution, x is particle size, t is time, a n d 6 and p are constants. Actually Harris' conclusion is not surprising, since constancy of the grinding rate function is only a first approximation, which cannot b e amplified until more reliable comminution d a t a are obtained. T h e disadvantage of Alyavdin's equation is that there seems n o logical way to relate the coefficients 6 a n d p to physical processes occurring in mills, whereas the physical effects governing the grinding rate a n d breakage functions may eventually be elucidated. Turning to other types of mills, Buhlmann ( 2 9 ) presents a stochastic model of impact pulverization. A new type of impact mill with oblique impact adjustable in the range 10' to 90" is useful for grinding of brown coal, according to Tschorbadjiski (200). T h e theory for energy exchange in the case of oblique impact was previously derived by Rumpf. T h e performance of cement mill closed circuits is the subject of several investigations. Krychtin et al. ( 7 7 3 ) statistically analyze experimental d a t a from a pilot cement mill circuit with air separators. T h e resulting equations are used to determine optimum throughput a n d energy requirement to produce cement with a given specific surface. Currier ( 3 5 ) comments o n the use of air separators in closed circuit with ball mills. Pirockij et QI. (763) investigate three mill-separator circuit arrangements, a n d find that the best is to recycle the separator tailings into the first of two mill compartments. Motek (744) discusses 3 circuits used for production of crushed stone by combining impact mills with cone crushers and roll crushers, a n d he compares their costs. I t is not usually recognized that fluctuations in product size distributions occur in impact crushers, as well as with other mills. Maeder (727) shows that such fluctuations c a n be avoided by adjusting the impact bars. H e gives a n example for a 75 kW impact mill. According to a patent issued to Rheinische Kalksteinwerke (768), a hammer mill closed-circuited with a n air separator is controlled by the load o n the separator, similar to the method often used for ball-mill circuits. Schacknies (783)observes that oscillations of ball-mill filling are detrimental to mill performance, and he devises a control scheme to optimize performance. Howard ( 8 7 ) notes that control of oregrinding mills is necessitated by variations in feed grindability. H e proposes to stabilize the feed by splitting the primary crusher discharge into coarse a n d fine fractions. When they are recombined, one is assured of a uniform feed size distribution to the following mill. T h e control of grinding circuits is often limited by lack of instrumentation. Brookes et QZ. (27) install a pilot mill circuit with 3 control loops on mill density, cyclone feed density, and total mass flow. Results are used to derive a mathematical model of the circuit. A Russian study by Icchakin and Chorol'skij ( 8 7 ) describes a circuit with ball-mill and screw classifier; 14 transducers are included. Most of these are standard types, but especially described is a transducer for ore particle size and concentration where each magnetic ore particle induces a n electric impulse whose height indicates the particle size. Another device is a n optical system to indicate the ore color, which is related to grade for this particular ore. Another is a resistance thermometer to indicate the temperature rise of the mill discharge over the feed, thus measuring the power absorbed by the mill. Another is a n inductive device to indicate amplitude and frequency of mill load oscillations owing to ball surging. Another is a n electric ear. All transducers are connected to a central computer. T h e measurement of particle size distribution of the mill output stream is still the limiting quantity in circuit control. A patent issued to Cementfabrik Holderbank ( 3 0 ) describes a n automated method of forming a sample of the dry product into a pellet, measuring its permeability, and relating this to surface area. According to Diener a n d Supp (40, 47), the K r u p p centrifugal classifier c a n be used for on-line monitoring of the performance of a cement mill circuit, when connected to a digital computer. A Russian particle-size on-line detector consists of a n array of piezoelectric crystals upon which the particle stream impinges. T h e amplitude of the electrical signals is converted into a measure a n d count of the particles of each size, and the device is used to control a n autogenous ore mill, according to Davidkovic et al. (36). T h e current industrial methods of solids flow measurement are reviewed by Beck and Wainwright ( 7 7 ) . Adequate devices are available for streams o n a conveyor belt or falling under gravity, b u t not for those in pneumatic or fluidized conveyors.
Mill Performance
T w o reviews of technical progress in crushing a n d grinding equipment appear in German journals: one by Muschelknautz (746) with 48 references, a n d one by Korenkov et ~ l (709) . with 39 references. Pierce (762) comments on 14 recent papers concerning crushing a n d impact milling of coal. Niediek (750) summarizes recent developments in comminution machinery, while North (152)surveys crushing and milling developments. Articles reviewed below include d a t a o n performance of pilot a n d plant mills, grindability measurements, and studies of transport through mills. D a t a from the Mines Branch, Canada, on a n open-circuit pilot ball mill is presented by Kelly (99). T h e statistically designed experiment shows that grindability of the feed is the most important variable, although in a sense this is begging the question since grindability is also measured with a ball mill. T h e mill holdup and mean residence time are affected by solids content of the wet feed; the product size distribution is affected by feed rates, size distribution of feed, and grindability of feed. Pilot mill experiments of Kelsall et QI. were reviewed last year. A study by Balasubramanian a n d H o h m a n n (8) conducted in a batch pilot mill shows the effect of mill holdup and grinding time on the dry grinding of calcite. T h e maximum attainable specific surface is 7000 cm2/g, beyond which ball coating develops. T h e kinetics of wet grinding in a ball mill are reported by Matejka (732), including the effect of peripheral speed. Oeckel (754) finds that the load of media should be 50 to 55Y0 of the mill volume for the fastest grinding rate a n d the least wear; lower media loadings are often used. Rowland and Nealey ( 7 7 7 ) discuss guidelines for application of dry rod mills. Until recently few dry rod mills were used; recent installations, mostly for iron ore, have not operated as expected in accordance with practices learned from wet rod milling. Drying of feed, scaling u p from pilot data, rod sizing, effect of rod length, rod bending, rod breakage, and rod tangling are discussed. T h e German practice reported by Jaeger a n d Ulrich ( 9 0 )also notes the importance of drying. I n a batch ball mill the charge of feed and the residence time are set by the experimenter; in a continuous mill the holdup and residence time are determined by the transport behavior of the mill, a n d only indirectly by the feed rate. Few d a t a on mill residence time a n d holdup have been published, b u t a number of investigators are presently conducting experiments. Kelsall et aZ. (700) report the distribution of residence time in a small continuous wet ball mill. T h e residence time distribution is measured by adding a n impulse of quartz to the calcite feed stream, a n d measuring the appearance of quartz tracer in the product as a function of time. A similar impulse of aluminum shot gives the effect of particle size with n o confusion owing to particle breakage. T h e d a t a show that particlc size has only a 2 0 7 , cffect o n residence time. T h c holdup weight changes drastically with feed-slurry solids content, and somewhat less with mill speed; mean-residence time is directly affected by mill holdup. Molerus and Paulsen (742) report a n extensive experimental investigation of feed transport in a large cement mill and a pilot mill under dry grinding conditions. They also determine the variation of holdup along the length of the mill and its effect o n the powcr uptake of the mill. Sectors of the discharge grate are covered to alter the mill holdup, a n d the small mill is also operated as a batch mill. Maximum power draft for the batch mill with smooth walls occurs a t a holdup of 807, of the ball void volume for a ball loading 387, of the mill volume; it occurs a t llO7, of the void volume for a 30y0 ball loading. These points check a n empirical correlation given by Rose and Sullivan (770). For the continuous mill with step-liners the results were similar, except that power was somewhat lower when the holdup was doubled by blocking the discharge gratc. Molerus and Paulsen (742) determine residence timc distribution, using a n impulse of salt as tracer. (This method is suitablc in dry grinding, b u t it cannot be used in wet grinding, because the residence time of the liquid is known to differ from that of the solids.) T h e results are analyzed in terms of bulk flow with superimposed diffusion. Values of the diffusion coefficient are measured. They increase with feed rate, decrease as the holdup increases, expecially above the void volume, and exhibit a maximum a t a ball loading of 27%. Some reflection occurs a t the end walls; this is unimportant in a long cemcnt mill, b u t it may cause a short mill to act more like a perfect mixer. VOL. 6 2
NO.
1 1
NOVEMBER 1 9 7 0
37
T h e use of residence-time distribution data for estimation of parameters in the axial dispersion model is described by Michelson and Ostergaard (739),and may be applied to tumbling mills. Garside and Wildsmith ( 6 7 ) predict the performance of swing hammer crushers used for fertilizers. Pankratov et al. (758) calculate the power load in a flat cone crusher based o n the breaking forces of the feed particles, the moment angle, and the eccentric load. This crusher is used in processing of iron ore in the KrivoiRog district of Russia. Makarov and Larin (128) investigate the grinding action in pan mills used for refractories. T h e most important factors arc thc downward load exerted on each particle and the frequency with which this load is applied, not the differential speed across the width of the muller wheels. Jet mills are the subject of further investigations. Dobson and Rothwell (45)determine the effect of the feed rate and the size distribution of a n alumina feed in two mills, one with milling and classification in a single chamber, and the other having a n external classifier. Ramanujam and Venkateswarlu (765) report 239 experiments in a n opposed-jet mill grindiiig calcite. This seems to be the first attempt to bring together the variables involved, to suggest conditions for optimum feed and choking, and to develop quantitative relationships concerning feed rate and size, product size distribution, residence time and grinding rate, nozzle pressure and size. Theoretical equations governing particle accclera tioii and particle free path are referred to, however. Vibratory media mills undergo further development. Hey (78) compares the power expenditure required for dry milling quartz sand to cement fineness in a ball mill and in a vibratory mill. T o reduce a 2-mm feed to 807, -0.1 m m requires 30 mill in the vibratory mill and 90 min in the ball mill. To reduce to 99.85y0 -90 pm requires 1 hr in the vibratory mill and 5 hr in the ball mill. T h e energy per unit of surface produced is also higher-by sixfold in the ball mill. Thus the vibratory mill is better suited to very fine grinding. Kalita et al. ( 9 2 ) also compare the performance of a vibratory and a ball mill for grinding of a very hard material, fired alumina. T h e ball mill requires 20 hr dry milling to attain the same mean-particle size that the vibratory mill attains in 2 hr. As the firing temperature is increased from 1450” to 175OoC, the alumina becomes harder, and the preference for the vibratory mill increases. T o produce 90YG particles -2 pm requires 2 h r in a dry vibratory mill, 50 h r in a dry ball mill, and 100 h r in a wet ball mill. Iron contamination is much greater in the ball mill. Hummler (85)reports performance of a 50-kLV vibratory mill installed in a quicklime plant. T h e mill has two grinding chambers in series, and is closed-circuited. T h e energy required per unit surface area produced is slightly higher than that required by a ball mill, and in this respect the soft lime differs from the hard quartz or alumina discussed above. The product size is related to the feed rate and grinding time by a n empirical formula from the Russian literature, similar to that mentioned by Harris. Autogenous mills continue to make inroads in ore processing, because of lower media wear cost. Clement and Kerl (33) find that autogenous milliiig is possible in a 30 X 30 cm (1 X 1 ft) mill, but a fine product is more rapidly attained with steel walls than rubber, and even more rapidly with steel media. Marechal (729) states that quasi-autogenous milling (including some steel balls) is just as advantageous as fully autogenous milling, but is feasible with a wider range of feed materials. H e gives a n example of milling of limestone for cement manufacture. Faucher (53) compares ore-grinding experience in autogenous and conventional ball mills, and he points to lower costs of autogenous milling on a large scale. Digre (42)surveys autogenous milling practice all over the world, from his Scandinavian vantage point. H e feels that the U.S. practice of using mills of greater diameter than length is not economically justified, and he points to Scandinavian and South African practices for the advantages of mills of more nearly equal length to diameter ratio (“square mills”). H e believes that autogenous mills should turn slower than ball mills, so as to reduce impact breakage of the large lumps, and promote cascading and multiple impacts sufficient to break smaller fragments. H e presents n o data to substantiate this view. However, Howard (81) also states that excessive impact conditions should be avoided -why try to destroy the grinding media? Howard finds that feed materials can be tested for their suitability as autogenous media in a mill as small as 10 X 12 in., while a 2 X 2 ft mill is particularly useful for testing larger pieces. Contrary to what others have stated, Howard (87) feels that nearly all ores c a n be ground by autogenous methods. Many failures are due to trying to produce too great a reduction ratio in one mill; often 38
I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y
Figure I . Impact cornminution machine with horizontal comminution rotfir and superFosed dinging-rotor. a , comminution rotor; 6, impact slat; c, slinging-rotor; d , ring go$; e, milling zone; f, slingingplates; g , mill race; h , impact plates. [Courtesy Chem.-Ing.-Tech. ( 72011 two stages of mills would serve better. Where there is difficulty in maintaining enough lump in primary mills, the use of large lumps should bc avoided because they break up very quickly. T h e use of large mills, especially a t high speed, also causes rapid breakage of lumps to the point where they cannot grind the sands. Howard ( 8 7 ) also gives some observations on the selection of ball sizes and media shapes other than spherical. The only definitive data on the effect of media shape that I know of is the work of Norris ( 7 5 7 ) , who shows that spherical balls grind better than any other shape in full-scale tests. Although practically no data have yet been published on the values of the grinding rate function (selection function) in mills, work continues to be done o n the older and cruder concept of grindability. Scheibe et al. (777) compare full-sized mill performance of cement mills with scaled-up design based on 3 grindability methods. T h e Bond and Mittag methods agree within 1 0 % of plant mill performance for cement mills. hTot all users find such good agreement, however. Marechal (730) gives the French experience in grindability measurement in a n English article. Ocepek (753) also compares methods and gives his experience in Yugoslavia. 11 Russian study by likkerman et al. (2) shows that the Bond work index can be applied to predict the behavior of cone crushers for 3 ores, and it gives the observed increase of power required as the product is ground finer. New Mills and Exotic Methods
T h e German patent literature is reviewed by Iohn (89). Subjects covered include stone crushers, mills drying with hot gases, and a jet mill. D u Pont ( 4 7 ) patents a n improvement in a jet mill with a n abrasion-resistant liner of S i c bonded with fused silicon nitride. Production rate is increased, while white pigment contamination is reduced. A new German impact crusher, the “Tonstar” ( 7 0 7 ) , reduces A new 30-cm pieces to 1-3 cm at a production rate of 35 ton’hr. rotary impact mill with an internal air classifier is developed by Leschonski (720). Breakage occurs between stationary and rotating elements, as shown in Figure 1. Attention to physical principles of particle air drag maximizes performance and minimizes interparticle interference. Performance data are given for limestone grinding. Eigner (48; 49) gives a detailed account of the design and testing of a tumbling mill of square cross-section with rounded corners. A spiral lifter bar arrangement tends to segregate balls according to size, and this rather than the shape may be responsible for the claimed improved output per unit power of 3070. Vibratory mills continue to be developed in Germany. Reiners (766) describes a new large mill with two chambers. A central supporting tube gives additional energy-transferring surface, and mechanical rigidity allows operating a t a frequency of 1500 Hz. T h e grinding rate is 2.25 times higher a t this frequency than at 1000 Hz. An English article by Vogeno (202) describes a large German vibratory mill with a throughput rate of 15 to 20 ton,’hr a t a power of 110 k W ; product size is below 10 pm for grinding of materials varying in hardness from clay to silicon carbide.
Pieces of rock can be broken by thermal shock caused by absorption of R F energy. Matthaei (733, 734) describes a Yagi antenna for focussing the energy on the piece, and also describes characteristics of the rock needed to absorb the energy. Results of tests with limestone are presented. Piekarski (767) describes similar experiments, fracturing asbestos ore with R F currents. Yigit et al. (275)continue their studies of electrohydraulic crushing with a study of selective breakage of heterogeneous materials by this method. High-energy explosive shocking can also reduce particle size drastically. Greenham and Richards ( 6 6 )show that alumina particles are sufficiently comminuted to cause X-ray line broadening. A method for producing spherical particles from nuclear fuel materials is patented by Knotic et d . (705). T h e size of ball mills is limited by the power transmitting capacity of the drive gears. Gears c a n be eliminated by making the mill the armature of a n electric motor. Several such mills of 10,000 h p or more are scheduled for startup in European cement plants, according to Hale (77). T h e motor is of low-speed synchronous type, driven by low-frequency current. A converter transforms 60-HZ current to the low frequency. T h e low frequency simplifies design a n d gives a high motor efficiency. T h e converter c a n vary the frequency during startup, avoiding large current inrush. Incidently, the variable-speed capability presents new possibilities for optimizing grinding conditions. Speed of mill is known to be a n important grinding variable, but in conventional installations nothing can be done to change the speed once the mill is installed. A series of articles in Brown Boveri Review ( 2 8 )give details of a 6400 kW installation, and a n earlier article by Kapoor ( 9 4 )describes a 3000 h p dc drive. In more conventional drives, Ackle ( 7 ) describes a 2-stage planetary gear for a 6000-hp mill, which is the present limit for mill gearing.
by Milaplana o n high-temperature abrasion. I t also includes a users report on wear plates by Treguer et d., and articles on hardfacing of wearing parts by Rey, on slotted grates by Courcier, on grinding charges and mill liners by Bodu, on hammers in coal milling by Vial et al., o n abrasion-resistant grinding parts by Milhoud, and on medium- a n d high-manganese steels by Hily. Wear in average French cement mills according to Bombled ( 2 2 ) amounts to 320 g/ton of dry cement raw material, 300 g/ton for clinker, and 90 g/ton for coal grinding. I t can be reduced to 114 or 1/6 as much by use of hard-steel liners and balls. Crushing tests on individual particles, and attrition during annular flow in a rotating-cup apparatus, are two small-scale tests that can be used for preliminary measurements of the relative attrition resistance of various materials of construction, according to Bjorklund and Dygert (78). When a n abrasive grain gains access between two sliding metal surfaces, the wear of each surface depends on their hardnesses, in a way reported by Wahl (207). T h e amount of motion has a greater effect than the velocity. Load, time, a n d temperature have moderate effects. Further research o n this type of wear is needed. For impact wear, Wellinger and Breckel (270) determine the relation between the physical parameters a n d the deformation of metals under repeated impact, such as occurs between balls and liner in a ball mill. Wilman (274)describes the nature of the surface structure of abraded materials, mainly by means of electron diffraction a t grazing incidence. A hardening heat treatment for martensitic alloy cast iron is patented by Greenwood and Westwood (67),while Ksenofontov (174) reports the effects of cooling capacity of the mold on the wear resistance of steel castings. Grinding Aids
Mill liners
Rubber continues to replace steel liners in large autogenous mills grinding iron ore in Canada, according to Pickett (760). These mills are up to 24 ft diameter. Perforated rubber screen is also being used. Diehl a n d Griffiths ( 3 9 )summarize a symposium on rubber mill liners a t which 3 iron ore companies present results from long-time full-scale tests. Rubber linings show longer life and lower costs in all but the biggest mills. Rubber discharge grates are also being successfully used. O n e thing that was not previously known is the effect of rubber linings on the milling rate; all 3 companies report no change in capacity or grind, and 2 report a decrease in power consumption per ton ground. Rubber linings failed tests in the 20’s a n d 30’s mainly because no adequate method of attachment was found. Details of the 2 currently successful methods are given, both based on a rubber lifter bar that secures flat liner sheets. Adding more rubber to the wear areas increases life, and alters bar shape, as shown in Figure 2. Liftcr bar shape greatly affects the wear rate, owing to angle of particle impingement, and it also affects grinding performance. Beebe and Merklin ( 7 2 ) also describe experiments on the shape of rubber and steel lifter bars. Current European experience is described in 3 articles. Gazarek and Milosevic (62) compare side-by-side tests with rubber and steel ball-mill a n d rod-linings in a copper concentrator. Performance is unchanged, b u t life is doubled. Reduction of milling room noise is a decided advantage. A mill grinding a lead-zinc ore failed in 20 days, however. Lange ( 7 78) describes 3 years favorable experience grinding quartz a n d lead-zinc ores in rubber-lined wet ball mills. Mohr (747) reviews the present state of rubber lining of tube mills in Germany. Rubber linings can bc successful if the diameter/length ratio of the mill is a t least 1.2 to 2.5. If rods are used, their diameters must not exceed 1/3 the spacing between lifter bars. T h e rods must be of nearly equal length, and must be removed before they wear thin enough to bend or break. They must not be larger than 10-cm diameter in small mills, or 8 cm in large mills. T h e mills work best with feed below 1.5 cm, a n d feed must not exceed 2.5 cm, except for soft feeds. Urethane rubber (37) extends the life of muller wheels a n d reduces sticking, a t least in the mixing of a corrosive material.
T h e number of articles and patents on grinding aids attests to the interest in this subject, especially for dry grinding in cement manufacture. Scheibe et GZ. (778) review the use of glycols and amines for this purpose. Serafin (786, 787, 788) and Dodson ( 4 6 ) patent a series of organic acetates, amines, and amides to increase grinding efficiency and resistance to pack setting of cement. Amounts from 0.005 to 0.4 wt Yo are used. Bartha et d . ( 7 0 ) patent the use of phosphate-containing ethanolamide compounds for dry-grinding limestone. A comparative test by Opoczky (756)on grinding cement 1 to 90 h r in a laboratory ball rnill shows that triethanolamine is effective in preventing ball coating, b u t not agglomeration of particles. Ball coating impairs productivity of grinding, while agglomeration impairs cement quality. Experience with glycol a n d amine cement grinding aids in Germany, summarized by Schneider (780),shows that throughput is increased 25 to 50yo,classifiers achieve sharper separation, a n d cement is more readily conveyed. Likewise, grinding aids are favored in Russia, a n d Sidocenko et al. (792) report that triethanolamine has the best effect. Plant tests with 3 mills on cement containing 137, slag showed that the optimum dose was 0.020 to 0.02570 additive; this gave the best fineness, increase of strength, and increase of
Lifter
Shell
Rubber vulcanized
Figure 2. T w o currently used methods for clamping rubber b a h n i l i liners t o the mill shell, and an axisymmetric lifter bar design that gices better protection t o clamping device, longer life, and often higher capacity of the mill. [Courtesy Can. M i n i n g J . (39)l
Mill Wear
A review of abrasive wear in general is published by Finkin ( 5 4 , while Gill (64) discusses wear of components in comminution. A French colloquium o n wear published in Review de l’lndustrie Mznerale (767) includes articles by Maratray on wear mechanism of materials subject to abrasion, by Guillon on rock abrasivity, a n d ’401. 6 2
NO. 1 1
NOVEMBER 1 9 7 0
39
mill throughput. Organosilicones are also being tried in Europe. Zadak and Zezulka (276)report on their effect in clinker grinding. I n a 15-1. batch mill Mel’nik et al. (737)find that 0.01 to 0.0570 decreases grinding time by 70%. Quality is not affected in this case. “m extensive investigation of the effect of organic vapors on grinding of cement clinker in hall mills is reported by Seebach (785). T h e dry grinding of alumina for ceramic manufacture is also influenced by reduction of hall coating. Oleic acid, stearic acid, and naphthcnic acid arc invcstigated by Fattah et al. ( 5 2 ) ,while Kukolev ei al. ( 7 7 5 ) investigate the effect of organosilicones. T h e time for grinding in a ball mill is reduced by a factor of 4, and in a vibratory mill by a factor of 6. Concentration is 0,0570and the product has a surface area of u p to 22,000 cm2/g. Kukolev et al. ( 7 7 6 ) also obtains pronounced improvement with the grinding of corundum. For vibratory milling of quartzite, Voznjuk and Dancuk (205)find a sulfonated naphthalein surfactant effective. Grinding aids are also used in w-et grinding. 0.05 to 0.1% polysiloxane increases the rate of wet hall milling of ultraporcelain and talc to a product below 1 0 pm, according to Piven and Naloichenko (764). Kuznetsov and T a u b e (777) discuss the conditions under which a n additive can act by weakening the particles; this can happen with rocks that contain grains cemented together with softer material that the additive can penetrate. Recycled water is said to he less effective than fresh water, b u t surfactants improve the strength-lowering effect of recycled water. A series of experiments by Lin and Mitzmager (723) determines the effect of polar and nonpolar fluids o n the rate of wet hall milling; as was previously known, water has the best effect. According to Zaika and Piven (277) the ball milling of a n alumina slip a t 72 to 74Yc solids content is increased by a silicone dispersing agent, which probably acts by decreasing the viscosity of the slip. Crystal and Chemical Changes in Fine Grinding
Last ycar thc incrcasing popularity of this subject was noted; this year the literature in this area has exploded. Most of the investigators d o not seem to he aware of the large number of others working on the subject, or of its history starting with Beilby ( 7 3 ) in 1921. Lidstrom (722) reviews the subject with 115 references, and points to applications in mineral flotation and water purification, preparation of agglomerates, and a connection with silicosis. O n the other hand, there are so many unknown areas in comminution, that some of this effort might be better redirected. T h e following investigations report crystal deformations as revealed by X-rays, thermal effects, increased solubility, or amorphotization. Dilaktorskii et al. (43) in sand-blasting of quartz ; Koepke and Stokes (706) on magnesium oxide single crystals; Richards and Greenham (769) on the strain energy in garnet particles; Schneider (787, 782) on effects in sand and salt with different mills and milling atmospheres; Ogura and Sobue (755) o n changes in microcrystalline cellulose; Krauth and Tomandl ( 7 7 7 ) on detection of changes in iron oxides by the Mossbauer effect; and Krauth (772) on changes in T a C , VC, and ZrN powders by pressing and grinding. A practical application in connection with changes in the magnetic properties of iron oxide is reported by Karasinska ( 9 7 ) , and Vonszant et ai. (204) report the activation of titanomagnetite by vibrational grinding. I n addition to crystal structure changes, there are numerous examples of chemical reactions promoted by comminution; a few of these appear to have some practical value. ilugustat (5) reports degradation of starch to polysaccharides in a vibratory grinder; Pazonyi et al. (759) report degradation of poly(methy1 methacrylate), PVC, and polystyrene in a hall mill; Scalan (776) patents the decarboxylation of fatty acids by hall milling; O p r e a et al. (757) destruct polyethylene terephthalate by vibratory milling; Schrader et al. (784) prepare methylchlorosilancs by disintegrating silicon alloys in a ball mill with methyl chloride; Fitton and Gill ( 5 5 )report oxidation of ferrous iron in rocks during grinding; Gusev et ai. ( 6 9 ) report crystal distortion and release of argon from micas in a mill with steel shot; Lasiewick et al. (719) find that comminution of silica increases its porosity as a vanadium catalyst carrier; and Slawek (794) finds that wet comminution of metal oxides and hydroxides increases surface activity. Sadahiro and Shimizu (774) find a similar effect with nickel oxide leading to a change in catalytic activity, and Bowden
AUTHOR Richard H. Snow zs a Senior Engineer at the I I T Research Institute, Chicago, Ill. 60676.
40
INDUSTRIAL AND ENGINEERING C H E M I S T R Y
et al. (24)observe gases from thermal decomposition produced by fracture of M g O , calcite, and cerussite. T h e question is often asked what is the finest particle size that can be produced by grinding. I n this connection Sanner (775) reports experiments to determine the finest dry grinding of anthracite, and achieves 2.2 pm mean size with a batch ball mill, and 6 Mm with a continuous ball mill. Finer sizes can be produced by wet grinding or jet milling. Applications
App!ications of size reduction in the following industries are noted in this section: cement, ceramics, coal, ferrites, pigments, metal powders, plastics, cereals, fertilizers, iron ore, pharmaceuticals, chemicals, safety, and laboratory arts. In the cement industry impact crushers are often used for limestone. ,\ndreas (3) describes modern rotary crushers that have full-width impact bars. Since these have more inertia than narrow hammers, they can break rocks u p to 55 in. Flack (56) reports a trial of a n autogenous mill for clinker in a U. S. cement plant. Blazy et al. ( 7 9 ) find that flint rocks can be removed from cement raw materials by selective grinding in a n autogenous mill. At the fine end of Srinding, Beke and Opoczky ( 7 4 ) investigate the grinding behavior of C?S and CIS in cement. They find that the C Z S has the greatest tendency to adhesion and agglomeration, while C3S can readily he ground to fineness. T h e particle size distribution of cement is seldom quoted, since usually only the surface area is determined by permeametry. Blezard (20) investigates size distributions, which can range from 100 down to a fraction of a micrometer. T h e largest contribution to surface area will come from material below 2 pm, where gypsum tends to accumulate, and perhaps C3S. Cement raw matcrials can sometimes be ground in air-swept ring-roll mills by drying with kiln exhaust gases, according t o Loesche (724, b u t control of the feed rate is important. A survey of ball milling procedures in North American vitreous enamel firms (32) reveals that there is no standard set of operating conditions. T o avoid overgrinding for making sintered metal oxide and carbide materials, a test procedure is advocated by Frehn (59). T h e capacities of three types of high-speed impact mills are also reported by Sobolov et al. (795). h review of powder technology in brickmaking by Swallow (797)has 17 references. Levit ( 7 2 7 ) investigates the performance of hammer mills for fine pulverizing of bituminous coals, and relates the optimum mill output to optimum power consumption and condition of coal. Monostory and Pickhardt (743) attempt to separate pyrite from coal dust by electric and magnetic fields, and also investigate the performance of a drum separator o n removal of minerals. T h e removal of fine pyrite by hydrocyclones is investigated by Taylor (798). A literature survey with 52 references is published by LVerkmeister (277) o n the comminution and handling of coal. Although paint pigments are usually precipitated in submicron size, comminution is rcquired to break u p agglomerates, and to disperse the pigment in the vehicle. ‘In electron microscope study of the effectiveness of the dispersion of pigment in paint films is given by Hornby and Murley (7.9). Kogan and Bazilevich (707) study the basic relations existing between critical pigment volume concentration, oil absorption, and specific surface of micronized iron oxide and talc pigments. T h e differences in behavior require paint formula adjustments. Kafarov et al. ( 9 1 ) report on the use of multistage sand mills for pigment grinding. r\ process by Arias ( 4 )for milling chromium in hydrogen halide gas, followed by hydrogen reduction, give3 submicron metal powders. Powdered polyethylene can be made by atomizing the liquid with hot water through a nozzle, according to a n Italian patent (796). Fillers for P V C sheets are made of very fine silica, as described by Belimova and Khodakov (76). A lime and quartz mixture is ground in a vibratory mill, after which the lime may be dissolved, leaving the fine silica. Several articles discuss the milling of cereals. T h e roller milling of corn dried a t various moistures for degerminating is presented by Brekke ( 2 5 ) . Articles by Hutchinson a n d Martin ( 8 6 ) , Kohler et al. (708),and Shoup et al. (190) discuss the effects of fine grinding and separation on the nutritive quality of wheat bran, o n the digestibility of aleurone in wheat fractions, and the amino acid composition of sorghum fractions. Problems of fine milling of grain are discussed by Vonderoh (203), while Tschiers (799) discusses the connection of the milling process to the quality of flour. T h e effect on dough properties owing to damage of starch by hall milling is discussed by Lorenz and Johnson (725).
A new wet cone classifier invented by Hukki (83, 84) is said to have a nearly ideal separation curve. A plastic hydraulic cyclone is reported (740) to cost half as much as a steel cyclone. Kosoi (710) uses a computer to calculate from theory the fluid velocity field in a hydrocyclone as a function of its dimensions. Badeev a n d Kurilkov (6) compare industrial d a t a o n the performance of a spiral classifier and a flotation classifier (presumably a n elutriator). Less fines are carried into the coarse stream of thr flotation classifier for the same yield of the coarse fractions. Reduction of the cost of electronics has made optical sorting of individual particles practical, if the particles are not too fine. T h e cost may be as little as 10 cents/ton. T h e methods are surveyed by King a n d B a h t (702), who give photos a n d diagrams. A German survey is given by Iohn (88), a n d a n application to gypsum beneficiation is described by French (60). Figure 3 shows a production-size air-brush sieve that is claimed to continuously separate particles, using sieves varying in size from 10-mesh to 8 - I m opening.
REFERENCES
Figure 3.
Blower sifter ( Taikosha and Co., Ltd., Toyko, Japan)
Fagerberg a n d Fahlstrom ( 5 1 )survey the crushing and grinding circuits used for processing iron ores in Sweden. Meisel (736) discusses the way in which use of larger mills has affected design of ancillary equipment for ore processing. T h e important? of size reduction a n d powder behavior in pharmaceutical engineering is discussed in a review by Fowler (58). Particle size affects the absorption of drugs, according to Marshall (131),and tableting affects the particle size of aspirin, as reported by Vanooteg et al. (201). A Russian book o n grinding in the chemical industry is published by Sidenko (791). Muschelknautz e t al. (147) explains uses of jet mills in the chemical industry. Grinding and mixing analytical samples can pose problems if one wishes the sample to be representative of a batch of material. Russian practice is described by Vulfson et al. (206). T o prevent silicosis in a quartz grinding works, all equipment is operated under reduced pressure, as described by Skudelny (193). Classifiers
An annual review of sieving, classifying, a n d separation is presented by Bohnet (21). T h e mechanism of screening is investigated by Kanzleiter ( 9 3 ) a n d by Hudson et al. (82). T h e former studies the feed motion o n vibrating sieves, a n d the latter investigates the behavior of deep particulate beds on sieves. T h e particle behavior is related to the selection of best operating conditions, which is studied by Djabin a n d Diomidov (44). T h e experimental d a t a for a continuous horizontal sieve show a mavimum throughput of fines with a certain fixed ratio of feed layer thickness to the sieve opening. Selection of large-capacity sieving systems is discussed by Matthews (735), while R u e b (773) reviews German screens for wet and dry operations. A Russian sieving machine for hard-tosieve materials has 3 racks, one of which is not connected to the drive mechanism, b u t oscillates with greater amplitude than the others, according to Rudin (172). A d a m p ore produced 18 ton/hr o n a sieve 0.5 X 1.0 m . T h e Morgensen Sizer, introduced in the U. S. from Germany, and described as the HI-Prob Sizer in this review 2 years ago, is discussed by Wessel (212). H e presents d a t a from industrial plants a n d shows a small plexiglass model. T h e separation behavior of any classifier can be represented by a curve of classifier function us. size of particle being separated. Geisel ( 6 3 ) gives such curves for the Bahco air separator, a n d correlates their behavior with settings of the apparatus. Muschelknautz a n d Krambrock (748) measure the aerodynamic parameters of a cyclone separator. Barskii a n d Shteinberg (9) investigate the effect of particle concentration o n the effectiveness of a gravity classifier. T h e effectiveness may reach 50 to 70y0 a t low stream density. A practical density range is 0.5 to 1.5 kg/m3, where the effectiveness is somewhat lower. At higher densities the effectiveness falls off.
(1) Ackle W., “Heavy-duty lanetary gears for driving largc ball mills,” Zem.KaIk-Giir, 22 (3), 125-31 (1f69) (Ger.). (2) Akkerman J. E Zvereva T. V. and Kazennov, M. N., “Dependence of the Bond work (ndex & the ch&acteri&ics of crushers,” Obognsfich. R u d , 12 (6), 20-4
(1967) (Russ.). (3) Andreas, A , , “Impact crushers in the cement industry,” Ilack Prod., 73 (51, 98108 (1970). (4) Arias, A , , “Feasibility of producing dispersion-strengthened chromium by ballmilling in hydrogbn halides,” Powder M e t . , 12 ( 2 3 ) , 45-78 (1969). (5) Augustat, S. “Production of polysaccharide preparations of various molecular weights by mechanical degradation in the dry state,’’ Emnehrungsforschung, 13 (2), 475-83 (1968) (Ger.); CA, 70, 21149r. (6) Badeev, Y. S . and Kurilkov B. R. “Comparative tests of a flotation classifier in a closed grinding cycle,” ‘?suet. ,$eta.?,, 41 (91, 21-4 (1968) (Russ.); CA, 69, 107892~. (7) Bailey, J. E. and Hill, N. A , , “The rffect of porosity and microstructure on the mechanical properties of ceramics,” Proc. Brit. Ceram. Sac., (151, 15-36 (1970). (8) Balasubramanian, M . and Hohmann E. H. “Studies on grinding in ball mill on pilot plant scale,” Chem’. Age India, i9 (10),’878-83 (1968). (9) Barskii, M. D. and Shteinberg, A . M., “Effect of the concentration of material in the feed stream on the effectiveness of classification by gravity,” Izu. Vyssh. Ucheb. Zaaed., Khim. Khim. Tekhnol., 11 ( 6 ) , 721-4 (1968) (Russ.); C A , 69, 9 7 9 5 1 ~ . (10) Bartha, Bela, et d l . , “Improvement of grinding,” Hungarian Patent 155,524 (Jan. 25, 1969); CA, 70, 116626h. (11) Beck, M. S. and Wainwright. N. “Current industrial methods of solids flow detection and measurement,” Po$de; Tmhnol., 2 (4), 89-98 (1969). (12) Beebe, R . R . and Merklin, K . E., “Rubber versus steel in ball mill liners,” Mining Congr. J.,55 (12), 54 (1969). (13) Beilby, G., “Aggregation and flow in solids,” Macmillan and Co., Ltd., London, 1921. (14) Beke, B. and Opoczky, L., “Very fine grinding of cement clinkers,” Zem.-KalkGips, 22 (12), 541-6 (1969) (Ger.). (15) Belichenko, A. G., Elenskii, F. Z . , and Chernyshov, Y. A , , “Coking properties of blend constituents by size fractions and improvements to the crushing scheme,” CokeChem., USSR (Engl. E d , ) , (12), 7-12 (1968). (16) Belimova, E. N. and Khodakov, G. S., “Preparation and study of properties of highly dispersed silicate fillers,” Sb. Tr,,Vses. Jv’nuch.-Issled, Insl. ~ V o u .Stroit. M a t e r . , (15), 35-43 (1967) (Russ.); CA, 69, 4 9 8 2 ~ . (17) Billinghurst, P. R . , Biookes, C . A . and Tabor D. “Sliding process as a fracture-inducing mechanism,” Phys. Bkis Yield Fr;ctu;e, Conf. Proc., Oxford Univ. Press, 253-8 (19661; CA, 69, 1 0 7 8 8 8 ~ . (18) Bjorklund, I. S. and Dygert, J. C., “Small-scale tests for attrition resistance of solids inslurry systems,” A.I.Ch.E. J.,14 (4), 553-7 (1968). (19) Blazy, P., Houot R., and Marechal, B.,
