B4 MILLING (BALL MILLS) B4.1 Sieve Sizes (ASTM E11) 4# 475mm 5o# — 300p 200# = 75y 8# © 2.36mm 100# = 149 325# 45 16# = 1.18mm 120# = 125 400# 38, 30# —0.60mm 170# = 90p B4.2 Particle Size Analysis There are now a number of laboratory and on-line laser particle size instruments which provide data of particle size distribution in the range 0.1-200u of interest for kiln feed and 0.1-100u for cement (Bumiller & Malcolmson; WC; 6/1998, pg 50). Samples are taken from the separator or classifier product stream. While these instruments directly determine volume proportions, they normally have software to convert the results to a weight basis. Particle size distributions may be plotted: 1. As weight percent of particles within successive increments of particle size (say, 5u groups) against the mid-point of each particle size group. A logarithmic scale is conventional for particle size (Figure B4.1) 2. As cumulative weight percent smaller (or larger) than a particle size against (logarithmic) particle size (Tromp curve). The Tromp curve is a tool for assessing classifier performance and is described in Section B4.5. 3. There are numerous empirical equations which attempt to convert size distribution to a straight line for easier characterization (Perry; Chemical Engineers’ Handbook, 6th Ed, pg 8.5). The Rosin-Rammler function is widely used in the cement industry: In (In(100Wr)) = b (In(0) - In(a)) where Wr = cumulative % retained at size D D = particle size, a &b = constants The relationship is linear and can be described by the slope, ‘b’ (often referred to as the RRSB number), and the y-intercept, ‘-b In(a)’. For Cement Plant Operations Handbook © 233 = fC aa = G) —_— w Pd - = = - = ) bdPortland cement, ‘b’ typically ranges 0.6 for a broad distribution to 1.1 for a narrow range. If D = a, Wr is numerically equal to 36.8% retention. The size distribution can then be described by ‘b’ and the particle size, Du, at which 36.8% of particles are retained. A typical Rosin-Rammler graph is shown in Figure B4.2. Particle Size Distribution Rosin-Rammler Graph \ 0 uy heed i" RSS = 1 i Be Position parameter = 13.8 3° 2| | 210 zt Sali pe ea 33. foe |e 1 2 \ fed |] So] 6 \ 08 70 it c 7 1 3 10. 2. 100 Particle Size (u) particle Size ta) Figure B4.1 Figure B4.2 B4.3 Circulating Load Circulating load refers only to the classifier and is defined as the ratio of the mass of material rejected back to the mill to the mass removed as fine product. Sometimes the ratio of feed to product is used but this should be referred to as Circulating factor. The ratios are defined as mass flows but are usually estimated from sieve residues of the material streams. All of the following analyses require that samples of materials are rep- resentative of the respective streams and taken during stable operation. Each sample should be a composite of at least three grab sub-samples taken of a period of 15-20 minutes. If necessary, the samples should be reduced using a riffle or other valid method of sample division. Feed to classifier Fuh £ % passing Product from classifier Pth P %passing Rejects back to mill Rth 1 % passing 234 © Cement Plant Operations HandbookFeed (F) Rejects back to mill (R) SEEM Product_(P) Figure B4.3: Separator Feed & Discharge Circulating Load C. = R/P = (p-f)/(f-9) Circulating Factor Ce= F/P = (p-1)/(F-1) Samples are taken at the separator during stable operation. Each sam- ple is screened on 170# or 200# for raw milling and 325% for cement milling. The specific surface area of cement samples may also be used. B4.4 Classifier Recovery Classifier recovery is defined as the fraction of fines present in the feed which is recovered with product. This is sometimes referred to as the classifier efficiency but, being determined from a single mesh size, it is an inadequate representation of overall efficiency. Using the nomenclature from B4.3: % Recovery = 100 x p(f-r)/f(p-1) B4.5 Tromp Curve The ‘Tromp is a graphical representation of the probability of a particle in the classifier feed exiting with the rejects. The probability can also be expressed as probability of exiting with the product but this is not con- ventional in cement industry usage. Using particle size distributions of each of the three streams, a mass balance for incremental size fractions from 1 to 100u for cement, or 1 to 200u for raw meal is performed. Typical curves are shown for a mechanical and a high-efficiency separator respectively. Cement Plant Operations Handbook * 235 = | ae aa = G) — ow Po — | ee = = - ,) atal Conventional Lg High-Efficiency a | 2 74 Classifier - 754 Classifier / > JS 0 { s ee 8% 24 ~_/ ae 0 — ———__ o UV 1 10 100 4 10 100 Particle Size (u} Particle Size (u) Figure B4.4: Tromp Curves The selectivity for each increment is plotted against the mid-point of each increment of particle size. This is conveniently plotted on semi-log paper from which the particle diameters corresponding to 25% (S: 50% (Sso) and 75% (S7s) probability of rejection are determined Apparent bypass (Bp) is the proportion of fines which are rejected. A per- fect separator would have 0% bypass and an efficient unit less than 20%. Classifier cut size (Dss) is the particle size corresponding to a 50% prob- ability of rejection. Sharpness index (D25/D7s) represents the sharpness of cut between prod- uct and rejects. The perfect classifier has an index of 1 and the smaller the index the poorer the classification. More simply, the steeper the curve, the narrower the particle size distribution of the product Many Tromp curves exhibit a characteristic tail at the bottom of the curve. This is often an indication of poor dispersion of the feed in the classifying zone which may be caused either by agglomeration of the feed or by non-uniform distribution of feed in the classifying zone. B4.6 Mill Critical Speed Critical speed (C,) is the rotational speed in rpm of the mill relative to the speed at which centrifugal force just counters gravitation and holds the charge against the shell during rotation. Raw mills usually operate at 72-74% critical speed and cement mills at 74-76%. 236 * Cement Plant Operations Handbookwhere D effective mill diameter, M % C= 100 n/G where n actual mill speed, rpm B4.7 Charge Volume Loading Loading, Vi % = 100 [(nr* (6 / 360)) — (h vr? - h?)] / mr? where r= effective mill radius, M H = free height, M hoe Her C = width of charge surface 2VP-h? angle subtended at mill axis by charge surface cosa = bir (Note: only the free height, H, and internal Figure B4.5 diameter, D, need be measured directly), = ze i = CG) — ee > ed =z = | we ee) — For normal charge levels, this gives: nD w% hp VW % 0.211 24% 0.143 32% 0.202 25% 0.135 33% 0.194 26% 0.127 34% 0.185 27% 0.119 35% 0.177 28% 0.110 36% 0.168 29% 0.102 37% 0.160 30% 0.094 38% 0.151 31% 0.086 39% Cement Plant Operations Handbook * 237B4.8 Grace Factor & Other Mill Parameters Instantaneous clinker charge, M: kg = MaR (C+ 100) /6000 Clinker bulk volume, My MB = M/Mo Ball charge volume, Gy MF = mPLVi/100 Ball charge weight, Gw kg = 4550Gy Steel:Clinker ratio, Sc = Gw/M Ball charge void space, Gs MP = 0.42Gv Void fill by clinker, Ve % = 100My/Gs Grace Factor, G kg/min/M? = — Mr(C. + 100) /6000Gs where r = _ Mill inside radius, M L Mill inside length, M VM Charge loading, % Mill production rate, kg/h Mill retention time, min Mill discharge bulk density, kg/M? Mill circulating load, % An analysis of 79 cement mills gave an optimum Grace factor of 263 and an acceptable range of 180-300. Void fill should be close to 100% (Welch; The Grace Factor - A New Tool for Cement Process Control Engineers; ASTM; 1984). Retention time is determined using fluorescein (7g/10t/h of combined mill feed and circulating load - see Mardulier & Wightman; RP; 8/1971). Samples taken at discharge at 30 second intervals are shaken with water, filtered, and the solutions observed under UV light to determine peak concentration. If XRF is available, which is equipped for zinc analysis, ZnO can be used as tracer; addition should be such as to give sample concentrations of greater than 20ppm, more if the milled material contains significant Zn. 238 * Cement Plant Operations HandbookB4.9 Mill Power Bond's Third Theory of Comminution allows an approximate estima- tion of specific power consumption (kWh/Ton) from the work index, Wi of the material being ground (Section B7.2). The value of Wi deter- mined from Bond’s Ball Mill Grindability test conforms to the mill motor output power of an 8ft diameter wet grinding overflow ball mill in closed circuit. kWh/T = (1OWi/ VF - 10Wi/ Va) CC where f; = the u size which 80% of feed passes f= the size which 80% of product passes G_ = 1.3 for dry grinding C= (2.438 10) D_ = effective mill ID, M ‘The Bond work index (Wi) is related to Hardgrove grindability (Hg): Wi, KWhIT = 435/Hg?3! (Duda; Cement Data Book 1, 3rd Ed, pg 129), Approximate power consumption of a mill can also be calculated from: kW = 7.5Gw\D where Gw = ball charge, t D_ = effective mill 1D, M Traditionally, ball mills have been driven by single girth gear and pinion drives, which are satisfactory up to about 5000kW. Larger mills are driven by girth gears with double pinions, central drives with epicyclic gears or ring motors (ICR; 8/1994; pg 57). B4.10 Ball Weight & Surface Area omm kg/ball #ballsit SA Pit 20 0.033 30,600 38.46 25 0.064 15,700 30.77 30 0.110 9,100 25.64 40 0.261 3,830 19.23 50 0.511 1,960 15.38 60 0.882 1,130 12.82 70 1.40 710 10.99 80 2.09 480 9.61 Cement Plant Operations Handbook * 239 ES - = = G) -— w | - = = | om ) eatSteel density is assumed 7.8g /cm?. Bulk density of a mixed ball charge may be taken as 4550kg/M?, B4.11 Maximum Ball Size Ball diameter, mm = 25.4 V(fWi/200C,) V5 /3.28D) where s = specific gravity of feed G = % of critical speed Interestingly, this suggests that the second compartment of a cement mill with f; of 3mm requires a maximum ball size of about 25mm. B4.12 Measurement of Wear The constant task of wear monitoring, ball sorting and charge make-up has been greatly reduced by use of high-chrome alloys with charge loss of approximately 20-50g /t. Ball wear can be measured by weighing an entire charge before and after grinding a known tonnage of material. With new balls a sample of given size can be weighed before and after grinding a given tonnage and, assuming uniform wear per unit of ball surface, the total charge wear can be estimated. More simply, and quite adequate, is to keep a record of material ground and of charge added periodically to recover a given level of mill power. Mill liners should be measured periodically to determine wear rate and, more importantly, to anticipate replacement. Thickness can usually be measured relative to the shell which can be probed between liner plates. Mill diaphragms are prone to excessive localised wear which must be monitored to anticipate replacement. Note that diaphragms normally have about 15% open area (slots plus centre screen) and that the slots are tapered outward on the discharge side to prevent plugging. eee 240 © Cement Plant Operations Handbook