2 Manufacture of Portland Cement Peter Del Strother

2 Manufacture of Portland Cement Peter del Strother

2.1 RAW MATERIALS FOR CLINKER MANUFACTURE 2.1.1 Chemical Targets for Raw Meal Lime saturation factor (LSF) is a measure of the ratio of limestone to other recipe components, normally expressed as a percentage.

LSF ¼ 100∗CaO=ðÞ2:8∗SiO2 +1:18∗Al2O3 +0:65∗Fe2O3 When clinker LSF is 100, the proportion of alite, the principal strength giving calcium silicate, is maximised. In practice clinker LSF is typically in the range 94–98. Clinker LSF depends on the raw meal LSF and the contribution of fuel ash, which is normally low LSF. Silica modulus (SM) and alumina modulus (AM) influence both clinker quality and burnability, the ease with which alite forms in the hottest part of the kiln, the ‘burning zone’.

SM ¼ SiO2=ðÞAl2O3 +Fe2O3

AM ¼ Al2O3=Fe2O3

2.1.1.1 Quarried Raw Materials Limestone The fundamental requirement for clinker manufacture is a source of lime (CaO) and sources of silica, alumina and iron oxide. In almost all cases the lime comes primarily from limestone. Limestone is a sedimentary rock composed of the hard parts of once living organisms. If raw material limestone has a LSF of more than about 200 it should not be difficult to devise a recipe with LSF, SM and AM suitable for clinker production. If LSF in limestone is lower than about 200 it becomes harder to find suitable non- calcareous components. Limestone is rarely pure calcium carbonate; deposits often incorporate siliciclastic components, such as silt from rivers or even volcanic ash. Limestone may also contain compounds of zinc, lead and fluorine, for example. Contaminants in limestone and other raw meal components may have a significant impact on clinker quality, kiln operation and emission to atmosphere. Of special relevance to quality and kiln operation are:

MgO: Dolomite CaMg(CO3)2 in limestone may arise from the original depositional environment or from diagenesis, the process of change to the sediment since deposition. It may be dispersed very unevenly within a limestone deposit. The ratio of MgO to CaO in dolomite is about 70% by weight. The European Standard EN 197 stipulates that ‘The magnesium content (MgO) shall not exceed 5.0% by mass’ in cement. The US ASTM C150-07 allows 6% MgO. Only a small proportion of dolomite in the raw materials recipe would be required to breach these limits. Chlorides: Limestone, especially porous limestone found near the sea, may be contaminated with sodium chloride (NaCl). Chloride is commonly found in alternative fuels, such as domestic-refuse-derived fuels. Chloride in cement is limited by most national cement standards and a kiln bypass may be required to remove chloride from the kiln system. Fluorides: Limestone altered by mineralising solutions may contain a suite of minerals including fluorine as fluorite, CaF2. The presence of fluoride acts as ‘mineraliser’. It also decreases the viscosity of the liquid phase in the sintering section of the kiln and if present in sufficient quantity may result in what should be nodular clinker becoming more like lava, resulting in serious damage to the clinker cooler. Alkalis, compounds of Na and K: The alkalis are found both in the clay (or shale) and the calcareous components of the recipe, but at higher levels in the clay or shale. If alkali levels in clinker are higher than required, selective use of low alkali clay or shale may overcome the problem. Increasing SM by addition of quartz sand also reduces clinker alkalis by reducing the proportion of clay or shale in the recipe. If those options are not available, non-alkali chloride can be added

to the kiln system and alkali removed as potassium chloride (KCl) in bypass dust. For example with increased use of alternative fuels containing PVC (polyvinyl chloride), Na2Oequiv in clinker can be reduced by between 0.05% and 0.1%. 2.1.1.2 Non-Calcareous Components Quarried non-calcareous components include shale and clay, which are silicate minerals containing compounds of alumina and iron. Quartz sand, a source of silica, and iron ore are also used.

2.1.2 Secondary Raw Materials The use of secondary raw materials is usually driven by cost, to which transport is often a major contributor. Waste-derived raw materials may have sustainability benefits. PFA (pulverised fuel ash), arising from the burning of coal in coal-fired power stations is widely used; it usually contains unburnt carbon, which may give ‘free’ heat input to the kiln system but may also give rise to CO emissions. In countries with emission trading schemes the organic carbon in waste raw materials may provide a financial benefit as it is considered to reduce CO2 emissions through a concomitant reduction in the use of conventional fuels.

2.2 RECIPE In the cement industry the following abbreviations are commonly used: C for CaO S for SiO2 A for Al2O3 F for Fe2O3 In this chapter these abbreviations will be used sparingly. However, they will be used in the Bogue formulae, see Section 2.4. An example of a recipe made from a small selection of raw materials is given in Table 2.1. LSF, SM and AM are also indicated for each component as these give an indication of what that material contributes to the recipe. The data in Table 2.1 is set out with the bold cells being adjustable and the percentage limestone being 100 less the sum of the clay and sand percentages. An LSF of 100 has been set as a raw meal target. Clinker LSF would typically be at least 2% points lower because fuel ash is normally low in CaO. With three components it is possible to control only two variables. In this example if the SM target is reduced to 2.75, equivalent Na2O in raw meal rises from 0.41% to 0.44%, see Table 2.2. It is possible to calculate the exact percentage of each component using the inverse matrix method, but the spreadsheet method is easily used and the natural variability of raw materials does not justify greater precision. On cement production works substantial investments are made to homogenise raw materials in order to minimise variability in raw meal chemistry. Variability in raw meal chemistry leads to kiln instability, variability in clinker quality and hence cement performance, especially strength, in concrete.

2.3 KILN SYSTEMS 2.3.1 Preparation of Feedstock and Grinding The quality of raw meal has a direct impact on clinker quality and on the production process. It is necessary to meet raw meal chemistry targets and minimise standard deviation. Methods of achieving these outcomes vary from slurry basins for the wet process to stacker-reclaimer stores and raw meal homogenising silos for the dry process. Representative sampling is a challenge and for most dry process systems cumulative raw meal samples are collected over an hour or so and control decisions made following analysis of those samples. Increasingly online cross belt analysers are being used for raw meal chemistry control. A significant advantage is that deviations from target are detected in just a few minutes and automatic corrective action taken. The need for investment in raw meal homogenisation is thereby reduced. Cross belt analysers are able to measure chemistry of crushed rock in the 0–75mm size range. The technology used is Prompt Gamma Neutron Activation Analysis. The neutrons penetrate the full cross-section of the material on the belt so the analysis is of the bulk material, not just of the surface. Feedstock preparation technology tends to be site specific and individual manufacturers offer different solutions. The operational principles are introduced in a number of readily available cement making manuals, such as The Cement Plant TABLE 2.1 Example of Recipe (% Dry Basis) at LSF 100 and SM 2.9, Using Three Raw Materials

TABLE 2.2 Example of Recipe (% Dry Basis) at LSF 100 and SM 2.75, Using Three Raw Materials auatr fPrln Cement Portland of Manufacture 33 34 Lea’s Chemistry of Cement and Concrete

Operations Handbook.1 Research has shown that for complete combination in the kiln, the particle size of calcium carbonate should be <125mm and that of quartz <45mm.2 Calcium carbonate loses 44% of its mass during calcination, resulting in a large specific surface area available for reaction. Quartz (SiO2) is resistant to both mechanical and chemical attack and has a much lower specific surface area than a lime particle of the same size. The melting point of SiO2 is about 1700°C and that of CaO >2500°C, both considerably above the temperature achieved in the kiln burning zone. Given an appropriate particle size and time, both CaO and quartz dissolve in the liquid phase.3 Silicate minerals have lower melting points than quartz and are readily incorporated into the liquid phase in the burning zone of the kiln. Coarse quartz in raw meal reduces burnability because belite clusters form around the quartz grains, inhibiting conversion of belite to alite. For explanation of alite and belite see Section 2.4. Quartz is harder to grind than limestone and it is not possible to grind both quartz finely and limestone less finely at the same time in a single raw mill. The target for raw meal fineness is a compromise typically in the range 10%–15% residue on a 90mm sieve. Grinding more finely increases raw mill electrical consumption and hence kiln specific power consumption. Kiln specific fuel consumption is also increased because the smaller the particle size the lower the preheater cyclone efficiency and the higher the dust recirculation from kiln to preheater. Raw mills are fitted with high-efficiency separators to minimise overgrinding. A small number of cement works grind quartz sand in a separate mill. Finely ground quartz sand is a significant health hazard, so improved particle size distribution (PSD) of raw meal incurs considerable extra cost in safe grinding, storage and transport equipment. Most benefit from separate grinding is derived when the raw meal recipe contains a high proportion of coarse quartz.

2.3.2 Thermal Process: Meal Chemical Reactions Edition 4 of FM Lea3 includes a description of wet process kiln systems. Few wet process kilns are now in operation and the wet process will not be described in this edition. Many preheater kilns are still in operation but, except for the lack of a calciner, the principles are similar to those for a precalciner kiln system. Reference will only be made to preheater kilns where significant differences from precalciner kiln systems exist. The first manufacture of ‘modern cement’ by William Aspdin in the mid-19th century used vertical kilns. Although there has recently been a renaissance in the use of vertical kilns4–7 for production rates of 50–300 mt/day, vertical kiln processes will not be described here either. The manufacture of white cement, which is of considerable interest, is also not addressed here. It presents special challenges in terms of minimising the iron oxide content of raw materials and fuels together with operation with a low level of liquid phase, reducing conditions in the kiln burning zone and a low oxygen atmosphere during the rapid early stage of clinker cooling. White cement manufacture is very costly and the achievement of ‘maximum whiteness’ a serious technical challenge. For this reason there is little published about the production process, the details of which tend to be closely guarded secrets of the manufacturing companies involved. The flow diagram in Fig. 2.1 is an example of a four-stage preheater with precalciner kiln system, a description usually abbreviated to a four-stage precalciner. A gas residence time of only about 2 seconds at about 880°C is required for calcination. Calciner geometry and residence time is therefore determined by the need to fully combust the fuels used. When designed for alternative fuels, calciner gas residence time is likely to be between 4 and 6 seconds. For greater thermal efficiency it is common to have a five-stage or even six-stage preheater. For kilns with outputs of 6000t/day or more it would be usual to have two or even three preheater strings feeding a single calciner and kiln. Plants using wet chalk as the main calcareous feedstock tend to have a lower number of cyclone stages and other modifications to the flow diagram in Fig. 2.1 to reflect the need for increased water evaporation and the small particle size of kiln feed. Raw meal entering the preheater is rapidly dried. Chemical reactions commence with dehydration of clay minerals. Natural raw materials may contain organic matter in the form of fossil carbon and hydrocarbons. PFA, the ash from coal-burning power stations, also contains carbon. Volatile organic matter breaks down in the upper stages of the preheater at a temperature too low to burn and the volatile fraction leaves the preheater unburnt, finally being emitted to atmosphere. Refractory organic carbon will begin to burn in the middle stages of the preheater and may give rise to CO (carbon monoxide). Oxidising CO to CO2 (carbon dioxide) is a slow chemical reaction under mid and upper preheater conditions, so CO may leave the preheater in addition to any CO from incomplete combustion in the calciner. All CO leaving the preheater will be emitted to atmosphere. Environmental legislation limits emissions of both volatile organic matter and CO. Calcination, the process by which CaCO3 breaks down into CaO and CO2, uses almost half the heat input to the kiln system. In a precalciner kiln between 95% and 97% of calcination is complete before the meal leaves the lowest stage Manufacture of Portland Cement 35

Stack

Raw meal Filter Raw Exhaust fan elevator meal silo Kiln feed Stack elevator Coal silo

Coal mill filter

Weigh bin

Coal Fine coal mill pump

Cooling Fine tower coal silo

Raw material silos Gas bypass Stack - Mill off Fine coal feed Calciner Cooler filter

Raw mill Tertiary air duct ID fan

Rotary kiln Clinker store Bypass filter Quench Cooler air

FIG. 2.1 Line diagram of a four-stage preheater kiln with calciner. cyclone; this meal is known as ‘hot meal’ and the percentage calcination is known as degree of calcination (DOC). Even in a preheater kiln the DOC of hot meal is likely to be close to 50%. Measurements of unreacted CaO (known as free lime (FL) or CaOfree) in hot meal demonstrate that some CaO reacts with silicate minerals in the calciner. This is especially the case when the calcareous raw material is impure and each hot meal particle contains both CaO and silicates in intimate contact. Calcination in the calciner and preheater takes place over a temperature range, as is seen in Fig. 2.2. At zero partial pressure of CO2 calcination temperature is 800°C but, as gas at calciner exit will contain about 30% CO2 dry basis, actual calcination temperature will be higher. Gas temperature control at the exit of the lowest stage cyclone is critical. After 100% calcination has been achieved further heat input results in a rapid rise in temperature and melt phases with the hot meal will become sticky and build-up on the walls of the cyclone, ultimately leading to a cyclone blockage. The reactions in the kiln forming alite and belite depend on temperature and residence time. In the burning zone, clinker temperature reaches a peak of between 1350°C and 1450°C. The reactions do not go to completion and clinker normally contains between 1% and 2% FL. For that percentage of FL, ‘hard burning’ recipes require a clinker temperature of >1400°C and ‘easy burning’ a lower temperature. Coarse quartz in raw meal makes it hard to burn and the impacts on clinker quality can be seen with optical microscopy. 36 Lea’s Chemistry of Cement and Concrete

Equilibrium degree of calcination of limestone versus temperature in an atmosphere of 100% CO2 at 1 bar pressure 100%

80%

60%

40% Degree of calcination

20%

0% 600°C 700°C 800°C 900°C 1000°C Temperature FIG. 2.2 Graph of degree of calcination against temperature.

2.4 CLINKER MINERALS Fig. 2.3 illustrates where the principal chemical reactions occur in a precalciner kiln and Fig. 2.4 the ranges of temperatures at which these reactions take place. In 1929 Dr. Robert Herman Bogue, at that time Director of the Portland Cement Association Fellowship (United States), published a landmark paper, Calculations of compounds in Portland cement.8 He made the assumption that four compounds C3S, C2S, C3A and C4AF exist in clinker. The abbreviated symbols are C for CaO, S for SiO2, A for Al2O3 and F for Fe2O3. Calculation of the amounts of these compounds from clinker oxide analysis is

Kiln feed Gas to ID fan

=> CaCO3 CaO + CO2

=> 2CaO + SiO2 (CaO)2·SiO2

=> 4CaO + Al2O3 + Fe2O3 (CaO)4·Al2O3·Fe2O3 => 3CaO + Al2O3 (CaO)3·Al2O3

=> CaO + (CaO)2·SiO2 (CaO)3·SiO2

94%–98% DOC

Cooler

1%–3% free lime FIG. 2.3 Diagram of precalciner kiln system showing locations of principal chemical reactions. Manufacture of Portland Cement 37

CaCO3 => CaO + CO2 ~ 800ºC–900ºC ⋅ 2CaO + SiO2 => (CaO)2 SiO2 800ºC–900ºC ⋅ 3CaO + Al2O3 => (CaO)3 Al2O3 900ºC–1300ºC ⋅ ⋅ 4CaO + Al2O3 + Fe2O3 => (CaO)4 Al2O3 Fe2O3 900ºC–1300ºC ⋅ CaO + (CaO)2.SiO2 => (CaO)3 SiO2 >1300ºC FIG. 2.4 Temperature ranges of principal chemical reactions.

straightforward. All F is in C4AF. Remaining A is in C3A. Two simultaneous equations can then be solved to apportion the remaining C and S into C2S and C3S, respectively. Bogue’s calculation still remains very widely used, although advances in XRD (X-ray diffraction) technology and more affordable XRD analytical equipment have made it possible for quality control laboratories on cement works to measure clinker mineralogy directly rather than calculate it. The compositions of the clinker phases measured by XRD differ from Bogue. The terms alite and belite are used for the mineral phases which approximate to C3S and C2S. As a consequence of impurities and solid solutions in the mineral phases, alite measured by XRD tends to be greater than Bogue C3S and belite measured by XRD less than Bogue C2S. Tricalcium aluminate measured by XRD is gen- erally less than Bogue C3A. Despite its limitations it seems likely that the Bogue calculation will still be used, but that XRD analysis will increasingly replace it. The Bogue calculation for clinker can be reduced to the following formulae3:

C4AF ¼ 3:043∗F

C3A ¼ 2:650∗A 1:692∗F

C2S ¼3:071∗ðÞC FL 0:7∗SO3 +8:602∗S+5:068∗A+1:079∗F

C3S ¼ 4:071∗ðÞC FL 0:7∗SO3 7:602∗S 6:719∗A 1:430∗F

In the above formulae FL is % free lime and all oxides are percent. For clinker the inclusion of the FL term makes good sense, see Figs 2.6 and 2.7, because otherwise the Bogue analysis is only a limited indication of clinker quality. The factor 0.7 is applied to clinker SO3 on the basis that all SO3 in clinker is in the form of CaSO4 (calcium sulfate). This term is normally omitted because in most circumstances a large proportion of clinker SO3 is found in alkali sulfates. In some National Standards a slightly different approach is taken, for instance in the United States FL is not used in the calculation. Fig. 2.5 shows indicative proportions of the principal minerals present as the temperature increases through the kiln system. The quantity of liquid phase can be estimated using the following formula:

° Liquid phase at 1450 C ¼ 3∗Al2O3 +2:25∗Fe2O3 + MgO + K2O+Na2O:

MgO is included up to a maximum of 2%. Other versions of this formula add SO3, although it isn’t made clear whether this is clinker SO3 or SO3 in the early part of the burning zone, which may be twice as much because of sulfur recirculation. The amount and viscosity of the liquid phase is key to clinker formation. The liquid phase formula is of most use in a comparative sense when assessing a potential change in recipe on an operating cement works. At a temperature in excess of about 1100°C silicate mineral species containing iron oxide and alumina begin to melt and the hot meal powder agglomerates into nodules. The liquid fraction is crucial to the rate of reaction of CaO with SiO2 as pure CaO is solid at 1400°C but dissolves in the melt. The liquid phase provides mobility so the diffusion rate determined reaction time is minimised. It also leads to the formation of the coating on the refractory lining of the burning zone. The coating pro- tects the lining and provides a short-term buffer against LSF variations in kiln feed. When LSF is high, for instance, coating is incorporated in clinker and its thickness reduced. Apart from its impact on FL and clinker quality, changes in LSF are dam- aging to the kiln refractory lining as large pieces of coating may be shed, resulting in poor clinker cooler performance and potential blockages at the clinker crusher. The clinker cooler is designed to achieve rapid cooling of clinker to a temperature less than about 1200°C. This prevents a reduction in C3S and increase in FL through the reaction

C3S ) C2S + CaO:

Control of FL is fundamental to clinker quality. 38 Lea’s Chemistry of Cement and Concrete

CO2

CaCO3

Free lime Alite Raw meal

Belite Proportions by weight

Low quartz High quartz Liquid Clinker

Clay minerals Cr C12A7 C3A Liquid Loss of Fe O 2 3 C2(A,F) H2O C4AF 0 300 500 700 900 1200 1300 1400 Temperature (∞C)

FIG. 2.5 Indicative proportions of minerals present during conversion of raw meal to clinker. Cristobalite (Cr) is a high-temperature quartz polymorph. (Based on Arnold.3)

Clinker C3S, C2S, and C3S+C2S, vs LSF for fixed SR, AR, FL and SO3 90 6.0

80 5.0 70 S 2 60 4.0 , FL 3 S+C

3 50

C3S C2S 3.0 S, C

3 40 C3S+C2S SM AM Free lime S, C SR, AR, SO 2 30 2.0

C SO3 20 1.0 10

0 0.0 90 91 92 93 94 95 96 97 98 99 100 101 102 LSF FIG. 2.6 Bogue for fixed SM and AM and 1% FL (free lime).

Fig. 2.6 shows C3S and C2S versus LSF for a typical clinker with fixed SM, AM and 1% FL. At a target LSF of 97, C3S content is 63%. The need for a high degree of precision in recipe control is apparent as the C3S content, the principal contributor to early strength falls to 57% at an LSF of 94. In Fig. 2.7 the same ratios have been used as in Fig. 2.6. At an LSF of 97, with FL increased to 3%, C3S content falls from 63% to 55%, a reduction of more than10%. XRD analysis will give somewhat different results, but it remains a fundamental truth that FL represents CaO not incorporated in C3S. Manufacture of Portland Cement 39

Clinker C3S, C2S, and C3S+C2S, vs LSF for fixed SR, AR, FL and SO3 90 6.0

80 5.0 70 S

2 60 4.0 , FL 3 S+C

3 50

C3S C2S 3.0 S, C

3 40 C3S+C2S SM AM Free lime S, C SR, AR, SO 2 30 2.0

C SO3 20 1.0 10

0 0.0 90 91 92 93 94 95 96 97 98 99 100 101 102 LSF FIG. 2.7 Bogue for the same SM and AM as in the previous figure, but with FL increased to 3%.

2.5 FUELS AND COMBUSTION: INFLUENCES ON THE MANUFACTURING PROCESS 2.5.1 Coal Except in locations with a cheap supply of oil or gas, fine coal has been the traditional fuel used in cement kilns. Borman and Ragland9 state that when coal is ground to a fineness of 10% residue on a 90mm sieve and encounters a 1400°C flame, the mass of a 100mm coal particle is reduced by 50% in 10ms. Calculated burn out times for fine coal particles are seen in Fig. 2.8. For a 100 mm particle the burn out time is about 100ms. Kiln burner flames are short with high-velocity primary air used to achieve rapid mixing of fuel with hot secondary air from the clinker cooler. A short flame improves burnability by reducing the residence time in the kiln between completion of calcination and liquid formation. There is no benefit in giving time for growth of belite crystals, inhibiting their later con- 10 version to alite. A short flame also minimises volatilisation of SO2, see Section 2.7. Long residence times are also to be

Single coal particle burn out in air at 1 atm and 1500 K (The range of results falls between the two lines) 100,000

10,000

1000

100

10 Initial particle diameter ( m m)

1 0.01 0.1 1 10 100 1000 Particle burn out time (s) FIG. 2.8 Calculated burn out times for single coal particles. (Based on data from Borman and Ragland.9) 40 Lea’s Chemistry of Cement and Concrete avoided as annealing reduces the number of imperfections in alite crystals and reduces their reactivity during hydration. This is one reason why more modern precalciner kilns operate at higher rotational speeds than was customary in the past. Rapid combustion is essential in the main burner as a high flame temperature is required to achieve the clinkering reactions. Heat transfer from flame to the clinker is mainly radiation, proportional to absolute temperature to the power of four. There is positive feedback, in that the higher the flame temperature the hotter the clinker and the hotter the clinker the greater the back radiation to the flame. Rapid combustion requires high-temperature combustion air and intimate mixing of fine coal with oxygen. This is achieved by supplying a small volume of high-momentum air through the burner, causing intense mixing of ambient temperature burner air with high-temperature secondary air and fine coal. Secondary air from the cooler has a temperature circa 1000°C. Modern burner designs achieve sufficient momentum with a reduced air volume at a higher velocity, thereby also reducing the total volume of ambient air required, which includes transport air used to convey fuels into the flame. In modern burners primary air velocity may exceed 250m/s, Mach 0.7. A reduced volume of primary air reduces kiln specific heat consumption. Combustion of fine coal in the calciner is much slower because the temperature at which it takes place is limited to about 900°C by the calcination reaction. Nevertheless a gas residence time of a little over 2 seconds is sufficient. Fine coal is normally introduced into the calciner by transport air, with no burner and no primary air. Coal contains ash and so kiln feed LSF is higher than that in clinker. Coal also contains sulfur either in organic molecules or as pyrite, FeS2. This sulfur becomes volatile and contributes to the sulfur cycle, see Section 2.7.

2.5.2 Alternative Fuels Alternative fuels are now widely used to reduce use of coal in kiln systems. Their use is likely to reduce manufacturing cost and may also give rise to environmental benefits. Put at its simplest, waste organic matter in landfill ultimately breaks down and the carbon is oxidised to CO2 by thermal or biological processes. If waste is combusted in a cement kiln, the use of coal is reduced and thereby net CO2 production is also reduced. Landfill sites are also filled up more slowly. Commonly used alternatives to coal include: petroleum coke, domestic-refuse-derived fuel, chipped car tyres, waste sol- vents and animal meal, which is the dry ground waste from meat processing. The concept of adiabatic flame temperature is a useful one when considering alternative fuels for the main burner. Adi- abatic flame temperature is the temperature attained after complete combustion of fuel in stoichiometric air without heat loss to the surroundings. With reactants at 1bar and 298K, the adiabatic flame temperature for bituminous fine coal is over 2000°C and that for wood chips at 25% moisture 1700°C. Both temperatures are reduced by >200°C in conditions of 20% excess air.9 Fig. 2.99 shows the time for combustion of a 10mm cube of pine wood. These times are extremely long compared to the residence times available in kiln main burner flame and calciner.

Mass loss of 10 mm pine cube in 1000 K airstream 1

0.9

0.8 0% moisture 15% moisture 0.7 200% moisture

0.6

0.5

0.4

Normalised mass 0.3

0.2

0.1

0 0 20406080100120 Time (s) FIG. 2.9 Mass loss of pine cube in an airstream at 1000K. (Based on data from Borman and Ragland.9) Manufacture of Portland Cement 41

Wood chips are particularly hard to burn but low adiabatic flame temperature is a property of many alternative fuels. Real flames are not adiabatic as radiation heat loss begins as soon as the flame is initiated. The slower the burning rate the lower the actual flame temperature achieved. It has already been stated that radiation heat transfer is proportional to the fourth power of the absolute temperature, so flame temperature matters. Combustion air is provided to the main burner at a high temperature, arising from a mixture of secondary from the clinker cooler with primary and transport air from the burner. When burning alternative fuels it is advantageous to use the minimum transport air flow, as this air replaces high-temperature secondary air. In solid alternative fuels particle size is much larger than fine coal and the combustion rate is much reduced. Moisture in fine coal may be 1% but it is commonly much higher in waste-derived fuels. Particle moisture has to be evaporated and pass out through the particle wall before the particle surface temperature is high enough for combustion to commence. For particles several millimetres thick in their shortest dimension, combustion is a complex process, influenced by moisture and volatile content as well as by thermal conductivity. Use of alternative fuels on the main burner therefore presents major challenges. If solid fuels introduced in the calciner fall into the kiln inlet, the oxygen required to burn this fuel has to pass through the burning zone, further reducing main burner flame temperature by dilution. In the main burner, liquid alternative fuels are relatively easily accommodated. These can be introduced into the centre of the flame through a spray nozzle. Flame temperature is reduced but the flow of the liquid fuel can be controlled to ensure that the clinker FL target is met. When the liquid contains particulates problems arise with nozzle blockage, solutions based on an open pipe fitted with a venturi have proved successful. Domestic-refuse-derived fuel is successfully burned in the main burners of some kilns.11 The drier the fuel and the smaller the constituent particles, the simpler this is to achieve. Shredded plastic bags and paper have one dimension comparable to the diameter of a fine coal particle and are likely to burn rapidly regardless of the size. Fragments of thick solid plastic do not burn rapidly. The main burner requires rapid-burning fuel but slower-burning fuel can be burnt in the calciner. Shredding domestic-refuse-derived fuels to smaller sizes may have a limited benefit, as Table 2.3 demonstrates. For this example a sheet of A4 paper has been cut into four and each piece cut into four and so on; the surface area of the total number of pieces of paper is a guide to speed of combustion. It is clear that the energy expended in shredding finer does not give significant increase in speed of combustion, although shredding to <25mm makes fuel transport more straightforward. The specific surface area of fine coal is >10 times that of the paper in this example. The fuel has to be pneumatically transported into the burner, so this increases ambient air input. Transport air replaces hot secondary air and moisture in the fuel increases gas volume leaving the preheater, so an increase in kiln specific thermal consumption is unavoidable. Steady mass flow and control of fine coal feed can be achieved by a number of well-proven technologies. Control of alternative fuel feed presents challenges particularly for solid fuels, where physical properties, bulk density and net calorific value tend to be variable and blockages may occur in the transport system. The impact of blockages is not trivial; loss of fuel may lead to kiln flushes, fugitive dust emissions and loss of clinker quality. In the traditional calciner, flameless combustion takes place at a temperature of about 900°C. Over recent years there have been huge changes in calciner design to accommodate use of alternative fuels.12 These changes are evident from designs shown on the websites of manufacturers of precalciner kilns. Gas residence time has been increased, 5 or 6 seconds

TABLE 2.3 Specific Surface Area of Shredded Paper

80 gsm 5 80 g/m2 Thickness 5 0.1 mm A Single Sheet of 80 gsm A4 Paper Density 5 800 kg/m3

Sides Area Number of Pieces of Paper Produced From One A4 Sheet Side (mm) Height (mm) Face Area (m2) (m2) (kg) Surface Area (m2/kg)

1 210 297 0.062 0.0001 0.005 12.5 4 105 149 0.062 0.0002 0.005 12.5 16 53 74 0.062 0.0004 0.005 12.6 64 26 37 0.062 0.0008 0.005 12.7 256 13 19 0.062 0.0016 0.005 12.8 1024 7 9 0.062 0.0032 0.005 13.2 4096 3 5 0.062 0.0065 0.005 13.8 42 Lea’s Chemistry of Cement and Concrete

Fuel A proportion of meal from Meal and Calciner second to lowest products of stage cyclone combustion

Tertiary air at 21% oxygen Gas temperature Combustion 800ºC–900ºC chamber Flame temperature up to 1100ºC Remainder of Meal curtain meal from second to lowest stage cyclone

Restrictor Kiln gas at 3% oxygen To bypass

Hot meal from lowest stage cyclone Kiln

FIG. 2.10 Diagrammatic representation of a combustion chamber connected to a calciner. being not uncommon. The geometry has also been changed to improve mixing of kiln gas, fuel and tertiary air. The riser duct above kiln inlet may be fitted with a restriction so that most coarse fuel particles remain in the calciner rather than fall into kiln inlet. This is critically important for fuels such as chipped tyres, as burn out time is likely to be tens of seconds and the tyre chips can only burn in the calciner if suspended in the gas flow from kiln inlet. Understanding the characteristics of fuels is the key to successful calciner design. Increasing the residence time of an existing calciner may be very expensive, so novel designs have been developed to improve waste fuel burn out without increasing the size of the calciner vessel. One design approach is to incorporate a combustion chamber, sometimes known as a ‘hot spot calciner’. The description is misleading as its function is to improve combustion. Fig. 2.10 is a diagrammatic representation of a combustion chamber. The combustion chamber is in the form of a cyclone with tertiary air entering tangentially. Raw meal is introduced into the tertiary air just before it enters the chamber. Fuel is introduced into fully oxygenated tertiary air where it burns at a temperature significantly above 900°C. Two benefits arise, firstly more rapid combustion than at 900°C and secondly, because some combustion has already taken place, the fuel particles are smaller when they enter the ‘lower oxygen and lower temperature’ zone of the calciner. Combustion in a calciner is normally flameless, but in this arrangement a flame forms in the combustion chamber and radiation from it is directed towards the refractory lining which, if no amelioration was provided, would soon exceed its maximum operating temperature. The meal introduced into the tertiary air remains around the circumference of the cylindrical chamber and does not become entrained in the flame. In this position it provides a ‘meal curtain’ to protect the lining. Calcination will take place in the meal curtain and further down the combustion chamber as meal and gas mix. There is extra investment cost for a preheater tower with this calciner design, as the preheater must be taller. Meal from the second to lowest stage cyclone has to pass through a meal splitter where the meal is divided between calciner and combustion chamber, which is much higher than the meal feed to the calciner. Sufficient meal must be diverted to the combustion chamber to prevent the meal becoming fully calcined, after which rapid temperature rise would make the meal sticky and build-ups would occur. Manufacture of Portland Cement 43

Waste fuels may contain for instance, chloride from plastics and phosphate from animal meal. A bypass may be necessary to accommodate high levels of chloride input to the kiln system. Phosphate in clinker does not impact on clinker quality if it is below about 2% in clinker.13

2.6 ENVIRONMENT: EMISSIONS TO ATMOSPHERE OF SO3, NOX, VOC, CO, DUST, Hg, Cd AND Tl Environmental regulations limit emissions to atmosphere. Those of the United States Environmental Protection Agency and the Directive 2008/1/EC of the European Parliament and of the Council of 15 January 2008, concerning integrated pollution prevention and control, will be widely known.

2.6.1 SO3 In the cement industry analytical results are normally expressed in terms of oxides. This is adequate for most purposes but not for raw material assessment. Limestones, particularly the dark coloured varieties, and many clays contain pyrite FeS2. Pyrite decomposes in the upper part of the preheater tower to release SO2, much of which is not scrubbed out even when the raw mill is in operation. SO2 emissions arising from pyrite may exceed the limit set by national environmental standards. The problem may be solved by the installation of an exhaust gas scrubber, but this is expensive both in terms of capital and operational costs.

2.6.2 NOx High-temperature combustion processes, such as take place in the kiln burning zone, oxidise atmospheric and fuel nitrogen into several nitrogen oxides, collectively known as NOx. Combustion of nitrogenous species in waste fuels used in the calciner may also generate NOx. NOx are acid rain gases, so their emission to atmosphere needs to be minimised. Emissions can be reduced by selective non-catalytic reduction, a process in which are ammonia is injected into the calciner.

2.6.3 VOC and CO Limestone, particularly the dark coloured variety, may contain organic carbon. The form of the carbon has a significant impact on CO and/or volatile organic carbon (VOC) emissions. Organic carbon in limestone which has been deeply buried tends to be refractory, that is, have a low proportion of hydrocarbons. Carbon only partially oxidises in the upper preheater and CO emission will result. CO emissions are subject to environmental controls because they are a symptom of incomplete combustion. CO arising from carbon in raw materials cannot be distinguished from that arising from incomplete combustion in the calciner. In limestone which has not been deeply buried, organic carbon tends to be in the form of hydrocarbons, which are volatilised in the upper preheater and a proportion of which are emitted unaltered to atmosphere via the exhaust stack.

2.6.4 Dust The use of bag filters rather than electrostatic precipitators for the main kiln system exhaust has reduced dust emissions to levels which have minimal impact on the environment.

2.6.5 Metallic Minerals

Sphalerite (ZnS), galena (PbS) and fluorite (CaF2) are sometimes found within limestone deposits. Small amounts of these contaminants can be tolerated. The presence of F increases the amount of liquid phase in the burning zone of the kiln and is a potential mineraliser. Excessive fluoride may lead to increased liquid phase so ‘what should leave the kiln as clinker’ does so as a viscous liquid, like lava. Sphalerite may contain cadmium (Cd). This highly volatile and toxic metal is emitted to atmosphere and, because of potential damage to the environment, emission limits are low in all national standards. Thallium (Tl) may be found in sulfide ores. Thallium is also highly volatile and toxic, but instances where cement plants have been significant emitters are few. According to the US Environmental Protection Agency, cement plants and coal burning power stations are significant con- tributors to national thallium emissions. Mercury (Hg) is the most volatile metal and is also toxic. Although rare in natural raw materials it may be present in waste fuels. Emission limits are necessarily very low. 44 Lea’s Chemistry of Cement and Concrete

2.7 CIRCULATION PHENOMENA AND BYPASS 2.7.1 Sulfur and Chloride Cycles Volatile compounds in the kiln and preheater arise from sulfides, sulfates and chlorides. Sulfur and chlorine compounds are found in raw materials and in fuels. Fig. 2.11 shows the vapour pressures of key volatile species. These are indicative only, as clinker contains a mixture of these species. At burning zone temperature, the volatility of alkali chlorides is high and that of alkali sulfates low, potassium chloride (KCl) for instance vaporises at a temperature close to 1400°C. At preheater gas exit temperature the volatility of all species is low, so volatilisation in high-temperature regions of the kiln system will be matched by condensation in cooler regions. This sets up cycles of evaporation and condensation. Fig. 2.12 illustrates the principle locations in a kiln system in which volatilisation of chloride species and condensation of sulfates and chlorides takes place. Simplified reactions for compounds of sulfur are given in Fig. 2.13. Subtleties such as the intermediate formation of calcium sulfite are omitted from the following explanations. SO2 is produced by thermal decomposition of CaSO4 in the burning zone, but decomposition is much increased when conditions in the burning zone are reducing. Decomposition of CaSO4 with respect to temperature and oxygen concentration is shown in Fig. 2.14. At 0% oxygen and normal burning zone temperature breakdown of CaSO4 is close to 100%. What this means is that only a small proportion of CaSO4 in meal approaching the burning zone leaves it in clinker. Sulfur circulation will increase until the sulfur mass balance is established, at which time CaSO4 in hot meal is much higher than that in clinker. Sulfur will continue to leave the kiln system in alkali sulfates, as these are of low volatility. Scrubbing of SO2 in the calciner is a highly efficient process as shown in Fig. 2.15. This graph shows how SO2 scrubbing is efficient at high temperatures when the active chemical compound is CaO. It is least efficient in the upper cyclones of the preheater, which is why SO2 arising from oxidation of pyrite (FeS2) in raw meal gives rise to SO2 emissions. Scrubbing is also efficient at a temperature of about 60°C, which is the operating temperature of a wet gas scrubber where the active chemical compound is Ca(OH)2 or CaCO3. The above chemical reactions give rise to a sulfur cycle, shown in Fig. 2.16. Note that half way up the preheater is effectively a ‘non-return valve’ for SO2 gas. Below this point SO2 can only exit the kiln system as sulfates mainly in clinker. When a bypass is in operation, a small proportion of sulfates leave the kiln system in bypass dust. The behaviour of chloride is more straightforward. Chloride as KCl is evaporated in the burning zone, passes through the kiln in vapour phase and condenses on meal and on cool surfaces in the calciner and preheater.

800 KOH NaOH KCl NaCl

KOH 700 NaOH 600 KCl

500 NaCl

Na2SO4 400 K2SO4 300

Na2SO4 Vapour pressure (mmHg) 200

100

K2SO4 0 700 800 900 1000 1100 1200 1300 1400 1500 Temperature (°C) FIG. 2.11 Vapour pressures of volatile species present in clinker. 760mmHg is atmospheric pressure. (Based on an image from Holderbank.14) Manufacture of Portland Cement 45

FIG. 2.12 Locations in kiln system prone to build up from condensing volatile species.

FIG. 2.13 Simplified reactions for compounds of sulfur.

Sulfur evaporation factor E, of Ca(SO)4 (in 70% nitrogen and circa 30% carbon dioxide) 1 0.9 0.8 0.7 E at 1400 ЊC 0.6 E at 1200 ЊC 0.5 0.4 E at 1000 ЊC 0.3 Evaporation factor 0.2 0.1 0 012345 % oxygen

14 FIG. 2.14 Decomposition of CaSO4 with respect to temperature and oxygen concentration. (Based on data from Holderbank. ) 46 Lea’s Chemistry of Cement and Concrete

Efficiency of sulfur dioxide reduction using CaO / Ca(OH)2

100

30 Reduction in sulfur dioxide (%) 20

0 0 100 200 300 400 500 600 700 800 Reaction temperature (°C)

FIG. 2.15 Variation in scrubbing effectiveness of SO2 by quicklime/hydrated lime.

Sulfur cycle in a precalciner kiln Raw material (raw mill in operation) Main stack

Raw SO2 mill

(thickness of blue-coloured bar Main filter is approximately proportional to the sulfur mass flow)

Sulfide Bypass oxidation to SO2 stack Bypass filter Preheater and calciner SO2 Calciner fuel

Bypass dust

Kiln fuel In

SO2 Out Kiln

Sulfates

Clinker

FIG. 2.16 Diagram representing the sulfur cycle in a precalciner kiln system. Manufacture of Portland Cement 47

Cooling tower

Bypass filter

Bypass dust

Quench air

Rotary kiln Cooler

FIG. 2.17 Typical arrangement of a bypass installed on a precalciner kiln.

The limit on chloride in cement is typically 0.1% so, if total chloride input, clinker basis, is more than about 0.07%, a bypass will be required. Fig. 2.17 shows a typical arrangement. Hot gas is withdrawn from the kiln riser and quenched with cold air to a temperature <400°C. At this temperature, KCl vapour is condensed to KCl solid, predominantly on the surfaces of dust particles. The gas and accompanying dust are further cooled and the dust collected in a filter. As a result of both these volatile cycles, the proportion of chloride and sulfate in hot meal is higher than that in clinker. An excess of chloride and sulfur salts has an impact on the stickiness of hot meal and hence on the propensity for cyclone build-up and blockages. Fig. 2.1814 shows initial melting points for mixtures of the three sulfate salts found in hot meal. In the absence of chloride (left hand diagram) most mixtures have initial melting points above hot meal temperature. The addition of chloride lowers the initial melting points significantly, as can be seen in the right hand diagram. In both diagrams the analysis is normalised so that the sum of the three sulfates is 100%. One way of quantifying the impact of initial melting on hot meal stickiness is to calculate the quantity of salt liquid phase. Table 2.414 shows the calculation of salt liquid phase. This example, from an operating kiln, shows that salt liquid phase is above the guideline value. The amount of sodium is small and has been neglected in this calculation. 48 Lea’s Chemistry of Cement and Concrete

Melting temperature of mixes of Na2SO4 K2SO4 and CaSO4 Na SO Na2SO4 2 4

800ºC–850ºC <700ºC 700ºC–800ºC

850ºC–900ºC <800ºC

900ºC–950ºC >950ºC >800ºC

CaSO4 K2SO4 CaSO4 K2SO4 With 8% KCl FIG. 2.18 Initial eutectic melting points of mixtures of sulfates found in hot meal. (Based on Holderbank.14)

TABLE 2.4 Calculation of Salt Liquid Phase in Hot Meal (Based on Hot Meal Temperature of About 800°C)

Critical Limits for Salt Liquid Phase in Hot Meal Hot Meal Analysis (%)

SO3 % 2.06 Na2O eq. % 2.39 Na2O % 0.13 K2O % 3.43 Cl % 1.99 Calculation of salt liquid phase 1 All Cl as KCl KCl % 4.18 2 Remaining K with sulfate K remaining % 0.66

K2SO4 % 1.46 3 Remaining sulfate with Ca S remaining % 0.56 CaSO4 % 2.36 Total Salt liquid phase % 8.00 Typical values for trouble free operation 5% to 7%

(Based on a methodology from Holderbank.)

Although the science is precise when working with a limited number of components, the real situation is much more complicated. Fig. 2.19 shows a practical guideline used by many cement manufacturers. Some kilns operate successfully outside the guidelines, especially preheater kilns where hot meal temperature is well below 800°C. Using XRD it is possible to measure the amount of each salt directly. It may therefore be possible to use a direct measurement of salt liquid phase for control purposes. When build-up occurs on a refractory surface in the preheater, the surface temperature decreases as build-up thickens because of increased insulation. Ingress of false air, however, follows the inner surface of the build-up, which then continues to thicken. Chemical analysis of build-up provides little useful data. The material which ‘sticks’ is hot meal. The build-up will also contain chloride, not necessarily evidence of a cause relationship because KCl permeates build-up (and refractories) and condenses. Fig. 2.20 is a picture of a lowest stage cyclone in a kiln with chloride in hot meal of about 2%. Corrosion caused by KCl permeating the refractory lining and condensing on the inside of the steel casing reduced its thickness by corrosion to a few millimetres, as a result of which the steel buckled. Refractory anchors at their contact with the steel casing were also severely corroded. The use of high-chloride alternative fuels has consequences beyond process chemistry.

2.7.2 Potential to Control the Sulfur Cycle

It is clear from Section 2.7.1 that the amount of CaSO4 in clinker is a critical parameter. If there were no CaSO4 in clinker the sulfur volatile cycle would be minimised. Sulfur/alkali molar ratio (MRs/a) is a measure of the quantity of CaSO4 in clinker. Manufacture of Portland Cement 49

Operating regime based on SO3 and Cl in hot meal

Frequent blockages

Intensive cleaning

% Cl 1 Normal cleaning

Trouble free operation

0 021 3 45

% SO3 FIG. 2.19 Guideline for expected cleaning regime related to hot meal chloride and sulfate content.

FIG. 2.20 Example of repair to cyclone wall damaged by chloride corrosion.

MRs=a ¼ ðÞSO3=80 =ðÞK2O=94 + Na2O=62 Cl=71 This equation, based on the reactivity of the components, is derived as follows. All chloride is assumed to be combined with alkalis. Residual alkali is combined with SO3 to form alkali sulfates. Thus MRs/a would be unity if there were no SO3 remaining to combine with CaO. When MR ¼ 1 all SO3 leaves the kiln in the form of low volatility alkali sulfates. If inputs of sulfur compounds, whether in raw materials or fuels can be controlled to give a MRs/a between 0.8 and 1.2 it is likely that the sulfur cycle will be within manageable limits. If MRs/a is <0.8 there may be alkali build-ups within the kiln in the form of rings located in the upper third of the kiln.

2.8 CLINKER COOLING If clinker is cooled slowly the following reaction may take place.

C3S ) C2S + CðÞ where C is FL 50 Lea’s Chemistry of Cement and Concrete

Clinker coolers are designed to minimise this reaction by cooling clinker rapidly to a temperature below 1200°C. Clinker cooler design has advanced considerably since Lea edition 4.3 Design principles and examples can be found on the websites of manufacturers of clinker coolers.

2.9 MASS AND HEAT BALANCE Preparation of a mass and heat balance on an operating kiln is a useful aid for diagnosing reasons for non-optimal operation, such as high specific fuel consumption. It is an especially useful tool when used periodically, as data such as heat of formation and preheater and clinker cooler radiation heat loss can be assumed to be constant unless there is good reason not to do so. Suppliers of kiln systems have their own mass and heat balance methodologies, used for kiln design. It is necessary to define a system boundary, such as shown in Fig. 2.21. Enthalpies of reaction of combustion and the heat of formation of clinker are required, as well as all mass and enthalpy flows across the boundary. If 20°C is taken as the base for all enthalpy flows, then for most kilns false air enters the kiln system with zero enthalpy, making it unnecessary to take account of false air mass flow, the estimation of which presents significant difficulties.

Stack

Raw meal Exhaust fan Filter Raw elevator meal silo Kiln feed Stack elevator Coal silo

Coal mill filter

Weigh bin

Coal Fine coal mill pump

System Cooling Fine boundary for tower coal silo mass and heat balance Raw material silos Stack Gas bypass—mill off Fine coal feed Calciner Cooler filter

Raw mill Tertiary air duct ID fan

Rotary kiln Clinker store Bypass filter Quench Cooler air

FIG. 2.21 System boundary for a mass and heat balance of a precalciner kiln. Manufacture of Portland Cement 51

TABLE 2.5 Example of a Mass and Heat Balance for a Precalciner Kiln

Heat

% of Total Description kJ/kg Clinker kcal/kg Clinker Input % by Group

Input

Fuel combustion Main burner 1573 376 38.3 Calciner 2277 544 55.4 93.7 Burnable matter in kiln feed 162 39 4.0 4.0 Sensible heat Raw meal 82 19 2.0 Fuel 4 1 0.1 Water 0 0 0.0 Primary air 1 0 0.0 Cooler air 0 0 0.0 False air + cooler imbalance 0 0 0.0 air Airlift air 5 1 0.1 2.2 Non-carbonatic CaO in kiln feed and 5 1 0.1 0.1 fuel Total input 4109 981 100 100 Fuel alone 3850 920 Output Heat of formation 1750 418 42.6 42.6 Water evaporation Kiln feed 25 6 0.6 Water sprays 0 0 0.0 0.6 Exhaust gas Sensible heat gas 1247 298 30.3 Sensible heat dust 57 14 1.4 Dust CaO loss 0 0 0.0 Unburnt CO 18 4 0.4 32.2 Cooler Sensible heat waste air 221 53 5.4 Sensible heat coal mill air 0 0 0.0 Sensible heat clinker 121 29 3.0 8.3 Bypass at 12.0% of flow at kiln inlet Sensible heat gas 124 30 3.0 Sensible heat dust 5 1 0.1 Dust CaO loss 5 1 0.1 Unburnt CO 1 0 0.0 3.3 Radiation and convection Kiln 334 80 8.1 Tertiary duct 49 12 1.2 Cooler and kiln hood 35 8 0.9 Preheater 104 25 2.5 12.7 Unaccounted for ‘input less output’ 14 3 0.3 0.3 Total output 4095 978 100 100 981

Table 2.5 is an example of a mass and heat balance for a precalciner kiln system. The kiln is an early design and is burning a high proportion of alternative fuels. The specific fuel consumption was 3.85 GJ/mt clinker, very high compared with a modern precalciner kiln with a modern clinker cooler design. Heat of formation, the net heat required to make clinker from raw meal, is about 1.75 GJ/mt clinker. From Table 2.5 it can be seen that close to 32% of heat leaving the kiln system does so in gas (and meal) from the top of the preheater. This is always the largest heat loss but is a symptom rather than a cause. Consider induced draft fan flow, and hence also fuel usage, fixed. If the cooler became less efficient or false air entering at kiln inlet or outlet is increased, some fuel heat will be diverted from making clinker. Kiln feed and therefore clinker output would have to be reduced. Gas flow up the preheater would remain much the same but kiln feed would be less, so the preheater exit temperature would rise (and therefore also the heat loss). Heat in the gas from the top of the preheater is required in the grinding mills for drying raw materials and coal. If raw materials contain a high percentage of water the number of cyclone stages would have to be reduced so as to increase preheater gas exit temperature. An increased kiln specific thermal consumption would result. 52 Lea’s Chemistry of Cement and Concrete

2.10 CLINKER GRINDING 2.10.1 Ball Mills Modern ball mills consist of two chambers separated by a diaphragm. In the first chamber the steel-alloy balls (also described as charge balls or media) are about 90mm diameter. The mill liners are designed to lift the media as the mill rotates, so the comminution process in the first chamber is dominated by crushing. In the second chamber the ball diameters are of smaller diameter, between 60 and 15mm. In this chamber the lining is typically a ‘classifying lining’ which sorts the media so that ball size reduces towards the discharge end of the mill. Here, comminution takes place in the rolling point-contact zone between each charge ball. An example of a two chamber ball mill is illustrated in Fig. 2.22.15 Much of the energy consumed by a ball mill generates heat. Water is injected into the second chamber of the mill to provide evaporative cooling. Air flow through the mill is one medium for cement transport but also removes water vapour and makes some contribution to cooling. Grinding is an energy intensive process and grinding more finely than necessary wastes energy. Cement consists of clinker, gypsum and other components mostly more easily ground than clinker. To minimise over-grinding modern ball mills are fitted with dynamic separators (otherwise described as ‘classifiers’ or more simply as ‘separators’). The working principle is that cement is removed from the mill before over-grinding has taken place. The cement is then separatedintoafinefraction,whichmeets finished product requirements, and a coarse fraction which is returned to mill inlet. Recirculation factor, that is, the ratio of mill throughput to fresh feed is up to three. Beyond this, efficiency gains are minimal.

2.10.2 Vertical Mills For more than 50years vertical mills have been the mill of choice for grinding raw materials into raw meal. More recently they have become widely used for cement production. They have lower specific energy consumption than ball mills and the separator, as in raw mills, is integral with the mill body.

Second chamber, range of ball charge from circa 60–15 mm; 'rolling' motion causing grinding

First chamber, circa 90 mm ball charge; 'cascading' motion causing crushing Diaphragm

Clinker input

Cement output, airswept upwards, gravity downwards to elevator (not shown)

Rotational speed depends on mill diameter; typically about Air flow in 15 rev/min Motor; circa 4500 kW for a 100 t/h mill

FIG. 2.22 Diagram of cement ball mill. (Based on an image from ThyssenKrupp AG.15) Manufacture of Portland Cement 53

Finished product to bag filter

Clinker & gypsum input

Separator

Grinding rollers

Grinding table Hot gas to mill body and mill inlet

Mill inlet

FIG. 2.23 Diagram of a vertical mill. (Courtesy of Loesche GmbH, Dusseldorf.)€

In the Loesche mill, Fig. 2.23,16 two pairs of rollers are used. In each pair the first, smaller diameter, roller stabilises the bed prior to grinding which takes place under the larger roller. Manufacturers use different technologies for bed stabilisation.

2.10.3 The Performance of Ball Mill Relative to Vertical Mills Comminution in ball mills and vertical mills differs fundamentally. In a ball mill, size reduction takes place by impact and attrition. In a vertical mill the bed of material is subject to such a high pressure that individual particles within the bed are fractured, even though the particles are very much smaller than the bed thickness. Early issues with vertical mills, such as narrower PSD and modified cement hydration characteristics compared with ball mills, have been resolved. One modification has been to install a hot gas generator so the gas temperature is high enough to partially dehydrate the gypsum. For many decades the two-compartment ball mill in closed circuit with a high-efficiency separator has been the mill of choice. In the last decade vertical mills have taken an increasing share of the cement milling market, not least because the specific power consumption of vertical mills is about 30% less than that of ball mills and for finely ground cement less still. The vertical mill has a proven track record in grinding blastfurnace slag, where it has the additional advantage of being a much more effective drier of wet feedstock than a ball mill. The vertical mill is more complex but its installation is more compact. The relative ‘installed’ capital costs tend to be site specific. Historically the installed cost has tended to be slightly higher for the vertical mill. 54 Lea’s Chemistry of Cement and Concrete

2.10.4 Particle Size Distribution PSD has an impact on workability and rate of strength gain. The PSD of cement can be described to a good approximation by the Rosin–Rammler–Sperling–Bennett distribution (see DIN 66145, 1976-0417). ÀÁ RxðÞ¼exp fgx=x0 n where x is the particle diameter in micrometres, R(x) the proportion by mass of particles larger than x. x0 the position parameter, which corresponds to R(x) ¼ 0.368. The equation reduces to the straight line graph lglgðÞ¼ 1=RxðÞ n lgðÞx + constant where ‘n’ is the slope of the straight line. Special graph paper is used with lglg(1/R(x)) on the abscissa and lg(x) on the ordinate axes. The higher the value of ‘n’, the narrower the particle size distribution. The position parameter is the particle size with the highest mass density distribution, the peak of the mass density distribution curve. Vertical mills tend to produce cement with a higher value of ‘n’. Values of ‘n’ normally lie between 0.8 and 1.2, dependent particularly on cement fineness. The position parameter is, of course, lower for more finely ground cements.

2.10.5 Characterisation of Separator Efficiency in a Ball Mill Circuit18 Separator efficiency is defined as specific power consumption reduction of the mill open-to-closed-circuit with the actual separator, compared with specific power consumption reduction of the mill open-to-closed-circuit with an ideal separator. The reduction of sieve residues by comminution is considered to be a first order process, represented by:

E ¼ W∗logðÞR0=Rf , where E is specific energy consumption W is comminution index, and R0 and Rf are sieve residues of fresh material and at mill outlet, respectively. R0 is taken to be unity. As shown in Fig. 2.24, circulating factor is defined as mill mass flow, that is, fresh feed plus separator returns. The maximum power reduction arising from use of an ideal separator increases non-linearly with circulation factor and is dependent on Rf, normally based on residues in the interval 32–45mm. The value of the comminution index, W, is also a function of Rf. The

C = Circulating load or circulation factor 1, Rf C is calculated from residues (assuming Ro = 100) C = (Rg-Rf)/(Rg-Rm)

Classifier

Grits (C-1), Rg

1, Ro Mill C C, Rm

FIG. 2.24 Line diagram of ball mill with separator. Manufacture of Portland Cement 55

Tromp curve 100 90 80 70 60 50 Slope 40 % to rejects 30 Bypass 20 10 0 1 10 100 Particle size-(μm)

FIG. 2.25 Example of Tromp curve.

finer the cement, the lower Rf and the greater the maximum power reduction. At C ¼ 2 most of maximum power reduction is achieved, but beyond C ¼ 3 there is very little further reduction. Over the last few decades separator efficiency has increased from about 25% to 75% as a result of improvements in separator design. Separator particle separation performance is assessed using the Tromp curve, a graph of percentage separator feed to rejects against particle size range. An example is shown in Fig. 2.25. Data required is the PSD of separator feed material and of rejects and finished product streams. The bypass and slope provide a measure of separator performance. The particle size is plotted on a logarithmic scale on the ordinate axis. The percentage is plotted on the abscissa either on a linear (as shown here) or on a Gaussian scale. The advantage of using the Gaussian scale is that the two parts of the graph can be approximated by two straight lines. The major cement manufacturers have their own calculation methodologies, manual and/or computerised for calculation of separator efficiency and Tromp curves.

2.10.6 Measurement of PSD The measurement of PSD of a sample of cement is carried out using laser-based methodologies. It requires a skilled operator to achieve consistent results. Agglomeration will vary dependent on whether grinding aid is used. Different laser analysis methods may not give the same results, so for comparative purposes the same method must be used.

REFERENCES 1. Allsop PA. The cement plant operations handbook. 5th ed. David Hargreaves, International Cement Review, and Tradeship Publications Ltd; 2007. 2. Christensen NH. Burnability of cement raw materials at 1400°C. Part II, effect of fineness. Cem Concr Res 1979;9:285–93. 3. Lea FM. Chemistry of cement and concrete. Arnold; 1998. ISBN 0 340 56589 6. 4. Glorius T. Production and use of solid recovered fuels—developments and prospects. Ziment Kalk Gips 2014;67(9):72–80. 5. Rajbhundari CD. Modern vertical shaft kiln technology. World Cem Jan. 1995;26(1):65–8. 6. Anon, “Vertical shaft kilns in China” Asian Cem Constr Mater, Jun. 1997, pp 20–2, 25. 7. Ahluwalia C, Raina K. Morphology and microstructure feature of clinker from vertical shaft kiln, In: National Council for Cement and Building Mate- rials India/ninth international congress of chemistry of cement, New Delhi; 1992. p. 146–52. 8. Bogue RH. Calculations of compounds in Portland cement. Ind Eng Chem Anal Ed 1929;1:192–7. 9. Borman GL, Ragland KW. Combustion engineering. WCB/McGraw-Hill; 1998. ISBN 0-07-006567-5. (pine cube figure adapted from Simmons & Ragland, 1986. Published by Gordon & Breach). 10. Taylor HFW. Cement chemistry. 2nd ed. Thomas Telford; 1997. ISBN 0 7277 2592 0. 11. Thomasberger F, Faber H. The practicalities of refuse-derived fuels. World Cem Mar. 2012;43(3):89–90. 92–4. 12. Anon. Efficient use of waste fuels in a cement manufacturing process. Eur Cem Res Acad Newslett 2014;(1):2–3. 56 Lea’s Chemistry of Cement and Concrete

13. Locher FW. Cement: principles of production and use. Verlag Bau + Technic; 2006. ISBN 3 7640 0420 7. 14. Holderbank. Figs 2.11, 2.14, 2.18 and Table 2.4 adapted from Holderbank. 15. Polysius Ltd., ThyssenKrupp AG. For Fig. 2.22. 16. Loesche GmbH, Dusseldorf.€ For Fig. 2.23. 17. DIN 66145. Graphical representation of particle size distributions, RRSB-grid; 1976. 18. Smidth FL. Comminution Manual; FLSmidth & Co. A/S.