Conventional Clinker Grinding - A New Approach To The Prediction Of Power Consumption

CONVENTIONAL CLINKER GRINDING - A NEW APPROACH TO THE PREDICTION OF POWER CONSUMPTION K.G. Tsakalakis School of Mining and Metallurgical Engineering, National Technical University of Athens, 15780, Zografou, Athens, Greece, e-mail: [email protected] Keywords: Energy consumption in comminution, Clinker grinding, Particle size, Blaine fineness, Process optimization, Modelling

ABSTRACT The current (2004) world cement production was about 2.11 billion tonnes per annum and it is increasing at about 1% per annum. The cement industry, as other mineralogical transformation process industries (lime, glass, ceramics, extractive metallurgy, etc.), is high energy intensive. Production costs and environmental concerns are emphasizing the need to use less energy, and therefore, the development of more energy efficient grinding and classification machines is of great importance. For all dry grinding applications, cement production is certainly the most important. The increasing demand for “finer” cement product and the need for reduction in energy consumption and greenhouse gas emissions (GHGs) reinforce the need for grinding optimisation. There is much potential in optimising conventional cement clinker grinding circuits, and in the last decades significant progress has been achieved. In the present paper we propose a new empirical relationship between the specific grinding energy, the clinker work index and the specific surface (Blaine) of the product of the clinker fine-grinding in ball mills. This model enables the prediction of the electrical power consumption in clinker grinding (cement production), which contributes significantly to the cement production cost. Furthermore, the above proposed model is used for an approximate correlation of the cement fineness (Blaine) and the d80 of the cement produced.

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

1

1. INTRODUCTION The cement industry, as other mineralogical transformation process industries (lime, glass, ceramics, extractive metallurgy, etc.), consumes large amounts of energy. Energy accounts for about one third of the cement production cost. Each tonne of cement produced requires not only large amounts of fuels (coal, fuel oil, natural gas), but additionally a significant amount of electrical power as well. For the reduction of the cement production cost and for environmental concerns, the utilization of alternative fuels (scrap tires, spent solvents, sewage sludge, etc.), which is compatible with the general principles of waste management, is nowadays a common practice in many European and developed countries. The estimate of the world energy consumption for cement production (Tsakalakis, 2003; Tsakalakis, 2005a) and its significance is: 1. 2.11 billion tonnes cement x 4.0 GJ/tonne cement = 8.44 x 109 GJ (energy from fossil fuels) for a mean energy consumption of 4.0 GJ/tonne cement (CEMBUREAU, 2004). 2. Additionally, 110 kWh/tonne cement electrical energy is consumed in cement production (raw meal crushing-grinding, homogenisation, clinker burning and cooling, finish milling, conveying, packing and loading, etc.). This power corresponds to: 2.11 billion tonnes cement x 110 kWh/tonne cement = 232.1x109 kWh = 0.835 x109 GJ and it is produced from fossil fuels with thermal potential 0.835 x109 GJ/0.4 = 2.088 x 109 GJ (assuming fuel combustion efficiency 40%). 3. Thus, the total energy consumed annually for the world cement production is: (8.44 x 109 GJ+2.088 x 109 GJ) = 10.328 x 109 GJ Since, the world total (2004) primary energy consumption was about 420 Quads Btu corresponding to 443.1 x 109 GJ, the percentage of the energy consumed by the cement industry is: (10.328/443.1)x100 = 2.33%, from which 0.47% refers to the electrical energy consumed in the cement making process.

2. CEMENT PRODUCTION

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

2

Approximately 1.5 tonnes of raw materials are required per 1 tonne of finished cement. Cement production process typically involves: • grinding limestone (and other raw materials to achieve the right chemical composition) to about 90% passing 90 microns in a dry circuit, • production of clinker by the chemical reaction (pyro-processing at high temperatures) between the components of the raw meal in rotary kilns, • grinding (finish milling) the clinker nodules to a product (cement) 100% passing 90 microns in a dry circuit. As it is evident in Fig. 1, the grinding process, taking place at the beginning and the end of the cement making process, consumes approximately 56.1% (i.e. 26.1+30) of the total electrical energy used in cement production (Stoiber, 2002). 100%

Conveying, packing, loading, 5.1%

90%

Electrical power distribution % in cement production processes

80%

Clinker burning and cooling, 24.6%

24,6

Raw meal homogenisation, 1.6%

70% 4,4

60% 50% 40%

Fuel (coal) grinding, 4.4%

30

Clinker grinding-cement production

30% 20%

Strip mining, raw material extraction, 5%

Raw meal grinding, drying, 26.1%

26,1

10% 3,2

0%

Raw material crushing, prehomogenisation, 3.2%

1

Figure 1. Electrical power distribution in cement production processes.

3. The electrical energy consumed in the cement making process is in the order of 110 kWh/tonne, about 30% of which is used for the raw materials preparation (crushing, grinding) and about 30% for the finish milling (CEMBUREAU, 2004) of clinker and additives (cement production).

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

3

Production costs and environmental concerns emphasize the need to use less energy and therefore the development of more energy efficient machines for grinding and classification. 3. CLINKER GRINDING PROCEDURE For most of the twentieth century, the dry grinding circuits for the production of finished cement from clinker consist of two-compartment tube mills and the air separators. It is not also uncommon to produce the cement in an open circuit. Advances in cement grinding technology are slow and these are limited to more developed countries. Approximately 95% of the feed to the cement grinding circuit are clinker and the rest of the feed are “additives”, which includes grinding aids. The cement clinker grinding circuit reduces the feed from 80% passing size between 12 and 20 mm to 100% passing 90 microns. The conventional size reduction takes place in a two-compartment ball mill; the first compartment of the mill is shorter for short retention time than the second one.

Feed

1st compartment Size reduction

2nd compartment Size reduction

R1

R2

Discharge

e Diaphragm

Figure 2. Conventional clinker-grinding ball mills with a diaphragm (Source: Jankovic et al., 1995).

The coarse clinker is ground in the first compartment where larger balls (80, 60, 50 mm) are present and the fine grinding is achieved in the second compartment where smaller balls (below 25 mm) are used. A diaphragm (Fig. 2), separating the two compartments, allows only particles below a certain size to pass the second compartment. Ground material leaves the mill through the discharge grate preventing grinding media (balls)

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

4

from going out the mill. A proportion of material, mostly fines, is “air-swept” out of the mill. The final product is the fine fraction of the air classifier and the coarse fraction returns to the mill. Due to the diaphragm separating the two compartments of the mill, it can be, without any significant error, considered that the size reduction process actually takes place in two successive grinding stages. The first size reduction phase happens in the first compartment, where the great balls are present, and the second in the subsequent with the small balls. The reduction ratio prevailing in the first compartment is supposed to be R1 and in the second R2. The overall reduction ratio R is the product of the two (R = R1 x R2) and the product of the first size reduction stage, designated as din, is the feed of the next. 4. PARTICLE SIZE AND CEMENT PERFORMANCE The cement properties and its performance in concrete production are based significantly on the cement fineness. The cement fineness depends on the particle size distribution of the cement and it is measured by the surface area or the Blaine index. The Blaine index is expressed in cm2/g or m2/kg units and it is determined by the Blaine air permeability test (Blaine apparatus). The method relates the fineness of cement to the porosity of a standard specimen of compacted cement. The finer the cement (i.e. smaller particles have larger surface area) the quicker it reacts with water, resulting in faster setting and higher early strength. If the particle size distribution is known, the Blaine index can be successfully predicted (Zhang et al, 1995; Forschungsinstitut der Zementindustrie, 2004), but several researchers have proved that, although the surface area of two cements could be the same, their particle size distribution could be markedly different (Fig. 3).

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

5

Cumulative passing % Particle size x, μm Figure 3. Particle size distributions of three cements of the same approximately fineness (Blaine). Source: Forschungsinstitut der Zementindustrie, 2004

The cement industry has for many years adopted fineness control as the main parameter for the control of the finish grinding process. However, it was noted that cements with the same chemical composition and surface area could perform differently, particularly in early strength and normal consistency (workability). Extensive research also established that the superfine particles (i.e. those less than 3 microns) affected the concrete slump and workability, but contributed little to the strength of cement. Cement particles larger than 32 microns were found to be too large to hydrate completely during the hydration reaction. Thus, it was established that the cement particles belonging in the size range between 3 and 32 microns are the optimum for cement performance (QCL, 2004). In Fig. 4 the particle size distribution of a cement, having Blaine fineness 3587 cm2/g, is shown. From the distribution it is evident that the particle size d80 = 30 μm (80% passing). It is also shown that 15% of the total is finer than 3 μm.

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

6

Figure 4. Cement particle size distribution (Blaine fineness 3587 cm2/g), Source: QCL, 2004.

5. EMPIRICAL MODEL DEVELOPMENT It is well known that the electrical power consumed in the clinker fine grinding for cement production (Stamboltzis and Tsakalakis, 2003; Schnatz, 2004) depends on: 

the size of the clinker particles (feed),

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

7



the mechanical characteristics of the clinker (hardness, work index, density)



the particle size distribution of the cement being in close relationship with its specific surface (Blaine fineness),



the mill dimensions (length, diameter, L/D ratio), and finally



the mill operating conditions (fraction of the mill filling fL , mill critical speed fC, mill grinding load).

For the prediction of the power consumption and the production cost of cement, the use of the specific electrical power consumption (kWh/tonne clinker) is necessary. This specific power consumption is a function (Fuller, Bulletin M-2 5M 6/76) of the cement fineness and the clinker work index (wi, kWh/short ton). In the present paper, applying multiple linear regression to data received from mill manufacturers (Fuller, Bulletin M-2 5M 6/76; MARCY, CATALOG 101-B), the following empirical model (Tsakalakis, 2005b) was developed:

E  101.7410

4

FBl 0.035wi 0.4714

(1)

where, E is the specific grinding energy (kWh/tonne), FBl is the cement fineness (Blaine) in cm2/g and wi is the clinker work index (kWh/short ton) The conversion factor (1 short ton = 0.907 tonne) has been embodied in the coefficients of the model. Applying Eq. (1) for various values of the cement fineness from 3200-5000 cm2/g with an incremental step 200 and for wi 12, 14, 16 and 18 kWh/short ton respectively, the specific grinding energy E was predicted. The graphical representation of Eq. (1) on semi-log paper for various values of clinker work-indexes wi is shown in Fig. 5. The model predicts the specific grinding energy of the clinker undergoing dry grinding in ball mills.

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

8

5200

12

14

16

18

5000

2

Specific surface Blaine, cm/g

4800 4600 4400 4200 4000 3800 3600 3400 3200 3000 20

10

30

40

50

60

80

100

Specific grinding energyE , kWh/tonne

Figure 5. Graphical representation of Eq. (1) for various values of clinker work-indexes wi (dry grinding in ball mills).

In Fig. 6 the experimental values are compared with those predicted from Eq. (1). From this figure, it is shown that the accuracy achieved is very good.

Calculated specific grinding energy E, kWh/tonne

70 Work index Wi =12 kWh/s.t. Work index Wi =14 kWh/s.t Work index Wi = 16 kwh/s.t. Line y=x

60

50

40

30 30

40

50

60

70

Exp erimental sp ecific grinding energy E, kWh/tonne (Fuller Traylor Grinding M ill Sy stems)

Figure 6. Comparison of the experimental and calculated specific grinding energy values of dry clinker grinding in ball mills. Solid line corresponds to y=x. Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

9

6. ESTIMATING THE d80 OF THE CEMENT PRODUCT It has also been shown earlier (Tsakalakis and Stamboltzis, 2004) that, the specific grinding energy E (kWh/tonne) in ball mils is given by:

E  23.7  wi  R0.193  d800.769

(2)

where E, wi are as previously in Eq. (1) defined R = Df /d80 is the particle reduction ratio Df is the particle size of the feed in μm (80% passing) and d80 is the particle size of the product in μm (80% passing). Substituting R for Df /d80 in Eq. (2) yields:

E  23.7  wi  D f 0.193  d80 0.962

(3)

But, the ball mills used for clinker finish milling are equipped with two separate compartments. Due to the diaphragm separating the two compartments of the mill, it can be thought, without any significant error, that the size reduction process takes place actually in two successive grinding stages. The first size reduction phase happens in the first compartment, where the great balls are present, and the second in the subsequent with the small balls. The reduction ratio prevailing in the first compartment is supposed to be R1 and in the second R2. Hence, the overall reduction ratio R is the product of the two (R = R1 x R2) and the product of the first size reduction stage, designated as din, is the feed of the next. Thus, Eq. (3) used for clinker grinding in two compartment ball mills must be applied as follows: E = E1 + E2

(4)

where, E1 is the energy needed for the first size reduction phase and E2 is the energy needed for the second grinding phase.

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

10

E1 is given by:

E1  23.7  wi  D f 0.193  d in 0.962

(5)

and E2 is:

E 2  0.87  23.7  wi  d in 0.193  d 80 0.962

(6)

The factor 0.87, set in Eq. (6), derived from Fig. 7. E2 is, by a percentage 13%, less than the energy calculated from Eq. (5). This percentage represents losses, assigned to the mill rotation mechanism and the friction. These losses, approximately estimated to 13% (i.e., 8.5+4.5), have already been accounted in Eq. (5). It is obvious that, there are not any additional mill rotation mechanism and friction losses corresponding to the second mill

50 45 40 35 30 25 20 15 10 5 0

46,7

31

8,5 4,5

6,3 0,6

2,4

Lo ss e

R

ot at io

n

(tr an

Fr sm ic tio iss n i lo o s n ss th m es ro e ch ug an h m ism Ab ill ) so sh Ab rp el tio l( so he n rp fro at tio ) m n fro ai r( Th m he eo pr at od re ) tic uc al O t (h th gr ea er in di t) (s ng ou nd en ,v er gy ib ra tio n, et c. )

Percentage, %

compartment.

Figure 7. Approximate electrical power distribution in conventional cement mills (ball mills).

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

11

Therefore, the total energy consumption for the clinker grinding in two-compartment mills can be calculated by: E  23.7  wi  D f 0.193  din 0.962  0.87  23.7  wi  din 0.193  d80 0.962

(7)

Application For the cement shown in Fig. 3, the Blaine fineness is FBl = 3587 cm2/g and assuming that the work index of the clinker wi = 14.5 kWh/short ton, Eq. (1) gives: E = 40.09 kWh/tonne Making also the assumption that Df =16 mm = 16000 μm and observing that d80 = 30 μm (Fig. 4), the arising overall reduction ratio is R = 533.33. Hereupon, solving Eq. (3) for d80, yields:

d 80

 23.7  14.5  533.330.193      40.09  

1

0.769

 79.03m

The above calculated value for d80 is far away from that observed in Fig. 4. Applying the procedure defined by the Eq. (7), we have: Let Df = 16000 μm and an equal reduction ratio 22.36 for the two grinding stages (i.e. 22.362 = 500). Thus, the product 80% from the first compartment is 715.56 μm (i.e. din = 715.56 μm). Substituting in Eq. (7) for E = 40.09 kWh/tonne, Df = 16000 μm, din = 715.56 μm, wi = 14.5 kWh/short ton and solving for d80 yields: d80 = 33.66 μm. The above value (33.66 μm) seems to be sufficiently close to that (30 μm) observed in Fig. 4. Thus, the overall size reduction ratio becomes R = 16000/33.66 = 475.34. Alternatively, choosing an arbitrary initial size-reduction ratio R1 = 15 (i.e. din = 1066.67 μm), instead of that R1 = 22.36 which was previously received, the corresponding value, in that case, is d80 = 35.18 μm. The overall reduction ratio becomes now R = 454.82.

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

12

The above assumption seems to be reasonable, due to the great size of the particles fed in the first compartment and its shorter length resulting in short retention time. Hence, the initial reduction ratio R1 should be less than R2, prevailing in the second compartment. Similarly, applying the same procedure for the different cements shown in Fig. 8, the

Cumulative passing %

results obtained for d80 are presented in Table 1.

Particle size x, μm Figure 8. Particle size distribution of various cements. Source: Forschungsinstitut der Zementindustrie, 2004 Table 1. Comparison of the data observed in Fig. 8 with those predicted from the proposed procedure Data from Fig. 8 Blaine fineness, FBl, cm2/g

d80 observed, μm

2500 3000 4000 5000

55 41 30 23

Overall reduction ratio, R calculated 290.9 390.24 533.33 695.7

Equal reduction ratio r = R0.5 (assumed) Overall d80 reduction Initial predicted, ratio R, R1 = R0.5 μm calculated 17.06 280 57.15 19.75 357.94 44.70 23.06 576.37 27.76 26.38 919.01 17.41

Assumed (arbitrary) initial reduction ratio R1 Overall d80 reduction predicted, ratio R, μm calculated Initial 277.30 57.70 R1 = 15 349.04 45.84 546.45 29.28 846.11 18.91

From Table 1 it is observed that, the proposed procedure predicts satisfactorily the d80 values corresponding to the fineness of the various cements. There is no significant error Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

13

when an arbitrary initial reduction ratio is chosen. These values are indicative, due to the fact previously explained, that cements of the same fineness could have quite different particle size distributions.

CONCLUSIONS In the present work, an empirical model was derived, which gives the specific grinding energy E as a function of the cement fineness FBl (Blaine) in cm2/g and the clinker Bond work index wi (kWh/short ton). This equation was connected with a previously proposed model giving the specific grinding energy E as a function of the Bond work index wi, the size reduction ratio R = Df /d80, and the d80 size of the product. From the procedure developed, it is possible to correlate the cement fineness FBl with the size d80 of the cement product. The calculated d80 values are indicative due to the fact that cements of the same fineness could have quite different particle size distributions. However, for the examples presented here, the predicted d80 values are close to those derived from the cement particle size distributions. This work shows that the proposed equations approach successfully the values given from mill-manufacturers. The whole methodology contributes to the specific grinding energy prediction, the calculation of the cement grinding cost, and serves successfully in the modeling of the cement grinding process. Additionally, it offers acceptable indications for the expected d80 size of the product, and is helpful in foreseeing the performance of the cement produced. ACKNOWLEDGEMENTS The author is greatly indebted to Profs. E. Mitsoulis and G. Stamboltzis for helpful discussions and suggestions.

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

14

REFERENCES 1. Tsakalakis K.G., The Greek cement industry sector and its potential towards sustainable 2.

3.

4. 5. 6.

7.

8. 9. 10. 11. 12.

13.

14.

development, Sustainable Development Indicators in the Mineral Industry (SDIMI 2003), May 21-23, 2003, Milos Island, Greece. Tsakalakis K.G., Scrap tyres management in the EU cement industry - an economic/environmental approach, 2005a (submitted for publication to WASTE MANAGEMENT). CEMBUREAU, COMPETITIVENESS OF THE EUROPEAN CEMENT INDUSTRY, The European Cement Association, 2004, http://www.cembureau.be/Cem_warehouse/2COMPETITIVENESS%20OF%20THE%20EUROPEAN%20CEMENT%20INDUSTRY.PDF Stoiber W., 2002, Comminution Technology and Energy Management, LAFARGE Cement Division in Verein Deutscher Zementwerke (VDZ) Kongress, September 2002. Jankovic, A., Valery, W. and Davis E., Cement grinding optimization, 2004, Metso Minerals Asia-Pacific, Perth, Australia. Zhang, Y. M., Napier-Munn, T. J., Effects of particle size distribution, surface area and chemical composition on Portland cement strength. Powder Technology 83, 1995, pp 245252. Forschungsinstitut der Zementindustrie, Zemente mit mehreren Hauptbestandteilen; Untersuchungen zur Optimierung von Herstellung und Eigenschaften, AIFForschungsvorhaben 13198N, 2004, http://www.vdz-online.de/downloads/aif13198n/13198n.pdf QCL Group Australia, Particle Size Distribution of Cement, Technical Note, August 1995. Fuller Traylor Grinding Mill Systems, GATX-FULLER, Bulletin M-2 5M 6/76, PENNSYLVANIA, U.S.A. MARCY, Ball and Rod Mills, MINE AND SMELTER SUPPLY Co., CATALOG 101-B. Stamboltzis G.A. and Tsakalakis K.G., Tumbling Mills Power Requirements, Mining and Metallurgical Annals, Vol. 3, Issue 1, 1993, pp.17-26, (in Greek with English abstract). Schnatz, R., Optimization of continuous ball mills used for finish-grinding of cement by varying the L/D ratio, ball charge filling ratio, ball size and residence time, Int. J. Miner. Process. 74S, 2004, pp. S55-S63. Tsakalakis K.G., Relationship between electrical energy consumption and cement fineness in clinker grinding, 5th Panhellenic Congress of the Panhellenic Society of Chemical Engineers, pp. 445-448, 2005b, Thessaloniki, Greece (in Greek). Tsakalakis K.G. and Stamboltzis G.A., Modelling the specific grinding energy and ball-mill scale up, 11th IFAC Symposium on Automation in Mining, Mineral and Metal Processing, 2004, Nancy, France.

Presented at the Ultrafine Grinding 06 (UFG 06), June 12-13, Minerals Engineering International (MEI), Falmouth-Cornwall, U.K, http://www.min-eng.com/ultrafinegrinding06/paps.html

15