Energy saving cement production by grain size optimisation of the raw meal

Resumo The production of cement clinker is an energy consuming process. At about 50% of the energy is associated with grinding and milling of the raw meal, that normally is in the range 100% <200 μm with 90% <90 μm. Question: is it possible to use coarser components of the raw meal without reducing the clinker quality. With synthetic raw meals of various grain sizes the clinker formation was studied at static (1100 1450°C) and dynamic conditions (heating microscope). A routine to adjust the grain size of the components for industrial raw meals is developed. The fine fraction <90 μm should mainly contain the siliceous and argileous components, whereas the calcitic component can be milled separately to a grain size between 200-500 μm, resulting in lower energy consumption for milling. Considering the technical and economical realizability the relation fine/coarse should be roughly 1:1. The energy for milling can be reduced significantly, that in addition leads to the preservation of natural energy resources.


Introduction
Compared to other industries cement production is an energy intensive process.Besides big energy reductions in the last years due to improvements in the preparation of raw materials, kiln development and heat recovery, a further reduction of energy by an extreme coarsening of the raw meal (up to 2000 µm grain size) is discussed.In the proposed process the energy consumption and operational costs shall be reduced by lowering the energy for milling, prolongation of the life time for the mills, lowering the sintering temperatures and shortening the reaction times in the rotary kiln, resulting in at least 30% energy reductions.It was supposed in a patent (Lörke, 2001 [1]), that the raw meal should be milled separated in a fine fraction of grain size < 80 µm (SiO 2 , Al 2 O 3 rich) and a coarse fraction up to 2 mm (CaO rich) in a relation 1,5 : 1 up to 1 : 9. Thus at the sintering process the composition of the fine fraction is only sufficient to form CS and/or C 3 S 2 as well as C 2 AS, C 3 A, C 12 A 7 and C 4 FA.With respect to the system CaO-Al 2 O 3 -SiO 2 (CAS) (figure. 1) a first eutectic with CS (Wollastonite), C 3 S 2 (Rankinite) and C 2 AS (Gehlenite) at 1265°C will be established.A further eutectic would be C 2 AS (Gehlenite), C 3 A and C 12 A 7 .However, the later one is stably Alkemadelines, red, connecting the subsolidus phases, and the resulting Alkemade triangles point to the according eutectics resp.peritectics listed above. of the sample the composition of the "remaining sample" can be calculated and the resulting Alkemade-triangle determined, thus the eutectic/peritectic at a sufficient temperature can be precast (figure 2).

Sample "fine"
The sample "fine" develops in the temperature range 1150 -1450°C in accordance to the phase relations in the CAS system.The freelime calculation as shown in figure 2 leads at 1150°C to a composition CAS 58,09/11,97/29,94 and thus in the Alkemade-triangle Belite -CA -Gehlenite.With increasing temperature and consumption of CaO the Alkemade-triangles Belite -C 12 A 7 -CA; Belite -C 3 A -C 12 A 7 are passed and finally the clinker-assemblage Alite -Belite -C 3 A is established resulting in a homogeneous clinker.

Sample "coarse"
The sample "coarse" composition at 1150°C is CAS 50,7/14,08/35,21 (after free-lime reduction) and thus in the triangle Rankinite -Gehlenite -Wollastonite.However, at 1150°C beside Gehlenite Belite is observed by XRD as a result due to its kinetically favoured formation at low temperatures (Lindner, 1955 [4]).At 1250°C Rankinite and Wollastonite in addition are observed, in especial by EPMA.Finally at 1300°C a CAS 54,09/13,12/32,80 is calculated and due to it's phase relation Belite -Gehlenite -Rankinite a melt is observed (nominally at 1310°C in the eutectic E4, see figure 1).With increasing temperature the melt composition develops to the tributary point G2 (1315°C), then follows the cotectical line Belite -Rankinite and then passes the liquidus of Belite to the cotectical line Belite -Gehlenite.The thermal divide T6 (1545°C) must be "tunnelled" by partial recrystallisation of the melt before it reaches the cotectical line Belite -Gehlenite again, passing the Belite liquidus at the 1450°C to the cotectical line Belite -C 3 A. Further incorporation of CaO finally leads to the formation of Alite.non existent, since the kinetic favoured formation of C 2 S (Belite) establishes the peritectics C 3 S 2 /C 2 AS/C 2 S (1315°C) and C 2 S/C 2 AS/ CA (1380°C), as well as the eutectics C 2 S/C 12 A 7 /CA (1335°C) and C 2 S/C 12 A 7 /C 3 A (1335°C).Right here it should be emphasised, that in a conventional rotary kiln process the first melt formation at about 1280°C (Taylor: "Cement Chemistry", 1997 [3]) occurs, thus in the range of the eutectic CS/C 3 S 2 / C 2 AS.This paper describes experimental results on synthetic and industrial raw meals, where the attempt was made to coarsen one or more components of the raw meal and achieve a lower eutectic.The basics of the experimental results and the thermodynamically background of the observed equilibria and the technical realization is described.

Materials and experimental program
The respective portions were weight for a total of 100g and the samples were then homogenized for 2 hours.Aliquots of 5 g were then pressed to tablets 2 cm in diameter.Two of each mixture were heated up in a high temperature furnace with a rate of 15°C/min and held for 30 minutes at 1150, 1250, 1300, 1350 and 1450°C.The samples were quenched in a stream of compressed air.Sample preparation and experimental procedure for the Heating microscope is described in section 4.

Sample analysis
The samples were microscopically inspected in reflected light (etched samples), by x-ray diffraction (XRD) and by electron probe micro analysis (EPMA).Free lime was analysed wet-chemically.

Calculation of Phase content from free lime content
From the analysed free lime content and the initial composition  The red line is the CaOreaction line.CaO-isopleths are given by dotted lines.
Solid lines are Alkemade-lines.The resulting subsolidus paragenesis and melting temperatures are given.
(The calculation is available as an Excel-worksheet and can be requested from the author).

B. SIMONS
G2 with coexisting Belite -Gehlenite -Rankinite.Further increase in temperature up to 1450°C, consumption of free CaO, leads to the cotectic line Belite -Gehlenite and develops as described above for the sample "coarse".At 1450°C the Clinker-assemblage is observed.

By reflected light microscopy as by EPMA as well typically rounded
Belites in the range of 20 -50µm and prismatic Alites up to 60 µm are observed.The axial ratio is 2:1 for the Alites.A free lime content of 1,45 is within an acceptable range for industrial clinkers.The texture is homogenous with low porosity.It thus seems possible with the above experimental approach to achieve a "lower melting" eutectic (E3: 1265°C) despite intense coarsening of the calcitic component of the raw meal; however, resulting in a technical acceptable clinker.
But it has to be mentioned, that the fine fraction of the above raw meals was < 10µm in the average, thus by one decade lower as for industrial raw meals.

Reaction paths in the System CaO -Al 2 O 3 -SiO 2
The analysed synthetic samples, especially those in the subsolidus region demonstrate that two different primary reactions occur.
At one hand the reaction of Quartz with CaO (figure 7), which leads to the kinetically favoured formation of Belite (see 3.4); at the other hand the reaction of Metakaolinite with CaO (figure 8 and 9), leading over the formation of an anorthitic component to the Alkemade relations shown in figure 10.Assuming a heterogeneous sample, where after decarbonatisation of the calcitic component and dehydration resp.dehydrolisation of the argileous components to in exemplo Metakaolinite, the free SiO 2 reacts with CaO to Belite (kinetically favoured) and Metakaolinite reacts with CaO to an anorthitic phase and consequently to Gehlenite and Belite, a first melt will establish in one of the Alkemade-triangles Belite -Gehlenite -(Wollastonite/Rankinite). A paragenesis Wollastonite -Anorthite -SiO 2 at nominally E2 = 1170°C will never be achieved.However, when reactions of SiO 2 with CaO at higher temperatures lead to the formation of Wollastonite and/or Rankinite occur, local heterogeneities may led to a partial melting with the anorthitic component to E3.

The Belite "problem"
At the binary join CaO -SiO 2 the phases Lime, CaO; Alite, Ca-

B. SIMONS
According to Lindner (1955 [4]) the early occurrence of Belite is thus explained.
In the present study a sample of fine grain size, being with its overall chemistry within the ternary Alkemade triangle Rankinite -Gehlenite -Wollastonite, at 1150°C only Belite and Gehlenite could be detected; at an according coarse grained sample beside Belite, Rankinite and Wollastonite were observed.In fine as well coarse samples, having the overall composition in the ternary Alkemade  It thus has to be pointed out, given a composition above the Alkemade-line Wollastonite -Anorthite (SiO 2 -rich), Wollastonite is formed.Consequently at CaO -contents below the Alkemade-line Wollastonite-Anorthite the formation of Wollastonite follows over the primary formation of Belite and then Rankinite.This is demonstrated in figure 7, where a SiO 2 grain imbedded in the groundmass is surrounded by a Belite rim.

Sample preparation and experimental procedure
Synthetic samples fine and coarse as described in section 2 as well as various industrial raw meals were pressed to cubes with 3 mm length and placed on an Al 2 O 3 sample holder.The sample holder was sled into a tubular furnace preheated to 500°C.The temperature was registered with an EL18 thermocouple placed right under the sample.
The temperature was then linearly raised with 10°C/min up to 1450 ~ 1500°C.The sample was illuminated inline of an optical bench, and its shadow was registered by a video device.Thus with a picture analysing system the surface resp.volume decrease was monitored.Thermal events as sintering, softening and others can be monitored.

Results
Sample "fine" (1) only shows a decrease in volume at about 1200°C and then a drastic decrease at 1463°C -melting.However the coarse sample decreases in volume at about 800 -900°C (decarbonatization, sintering), then shows a slight decrease at about 1230 -1260°C (eutectic E3: 1265°C) and a softening temperature of 1310°C (partial melting in the eutectic E4).This is in total agreement with the static experiments.All industrial raw meals show similar behaviour to the synthetic sample "coarse" (figure 13a) with softening temperatures in the range 1320 -1330°C, clearly indicating partial melting.
Energy saving cement production by grain size optimisation of the raw meal

Conclusions
Concerning the question, is it possible by coarsening the raw meal for cement production (going along with reduced energy consumption of the milling process) to achieve a controlled phase formation and eventually partial melting of a silica rich melt resulting in a "lower melting" eutectic, synthetic raw meals composed from Calcite p.a. resp.Warsteiner Kalk (pure limestone, Germany), quartz powder and pure Kaolinite were calcined.While in a first experimental run the components were in an average grain size of <14 µm, in a second run the Calcite (Warsteiner Kalk) was chosen in the range 200 -500 µm.As expected the "fine" raw meal followed the phase-theoretical path and finally led to a clinker with the according phase assemblage Alite, Belite and Aluminate.Using a coarse calcitic component, however, led to a displacement of the remaining composition (CaO partially unavailable to the reaction) to higher SiO 2 resp.Al 2 O 3 contents.In fact it was possible to shift the remaining composition as such, that for a coarse mixture with a relation fine/coarse of 1:1 (with fine <14µm, and coarse 200-500 µm, Warsteiner Kalk) a low melting peritectic with 1315°C was approached.Macroscopically the sample was already recognised to be partially molten.Increasing the temperature to 1450°C results in the desired clinker assemblage with a homogenous texture.For a so called "super coarse" mixture, where the total calcitic component was added as Warsteiner Kalk with a grain size 200 -500 µm, the eutectic Wollastonite -Gehlenite -Anorthite with 1265°C was established.Thus it seems, that by coarsening the calcitic component of the raw meal a lower eutectic might be achieved.Heating microscopy of the very same synthetic raw meals confirmed these observations.For a fine synthetic sample sintering was observed at 1223°C and weakening of the sample (partial melting) was observed at about 1463°C, whereas the onset of softening for a coarse grained synthetic sample was at 1310°C.The onset of weakening for a coarse grained sample was at 1287°C.Various industrial raw meals were subjected to heating microscopy (static experiments have been performed as well, not described in the present paper), demonstrating similar features to the coarse resp.super coarse synthetic sample of this study: the softening temperatures are in the range 1310 -1330°C).Modifying one of the industrial raw meals with respect to the grain size of the calcitic component finally led to a softening onset at 1279°C.It can thus be stated, that a coarsening of the calcitic component of a raw meal for cement production is advisable and may lead to a significant cost reduction.Anyhow, this is strongly dependant on the available resources.For each location an optimisation routine has to be performed.Some of the given calculations within this project are suitable to give a quick access to the parameters.For a given raw meal a quick test run can be performed showing the onset of weakening.A static clinker calcination and routine reflected light observation reveals the texture of the clinker.Stepwise the optimum of coarsening is detected.
It is proposed to separately grind resp.mill two significantly different grain fractions.A "fine" fraction <90µm should in principal contain the siliceous and argillaceous components, whereas the calcitic component may be according to the observations of this study in the range 200 -500µm.Considering the technical and economical realizability a ratio 4:1 lime-marl/limestone is proposed.However, increasing the pure limestone contend eventually leads to a "lower melting" eutectic.The lime-marl and of course the limestone itself will be fractured during the rotary kiln process due to the decarbonatisation of the calcitic component and thus increase the "fine" fraction.Coarsening the calcite bearing fraction of the raw meals thus results in significantly lower energy consumption for milling.

Figure 1 -
Figure 1 -The system CaO -Al O -SiO , (modified after Rankin and Wright, 1915 [2]) in weight percent.Alkemadelines, red, connecting the subsolidus phases, and the resulting Alkemade triangles point to the according eutectics resp.peritectics listed above.

Figure 3 -
Figure 3 -Sample "coarse" 1300°C.Semi quantitative EPMA.Note the strong maxima for Belite and melt.Gehlenite is weakly developed.At the CaO apex is free lime.Also note the trends following the Alkemade lines to C A and C A 3 1 2 7

Figure 5 -
Figure 5 -sample "super coarse" -1250°C.Ternary plot of semi quantitative EPMA super-positioned on the CAS system.Note the maxima for Belite, Gehlenite and Wollastonite.On the thermal divide T3 the maximum for the melt composition is shown.The maximum in the CaO apex is free lime.

Figure 10 -
Figure 10 -Reaction path in the CAS system (mole %).Solid lines are Alkemade-lines.Black wide dotted lines as indicated.The narrow dotted blue line indicates the reaction of SiO with CaO; the red dotted line indicates the reaction of Metakaolinite with CaO.2

Figure 12 .
Figure 12.The heating microscope.On an optical bench the following major components are mounted (from left to right): Light source, vertical tubular furnace (max.1650°C), video camera.At the lower left the programming and control unit is shown.The picture analysing system is not shown.

Figure 13
Figure 13 -a), b) -Heating curves for synthetic and industrial raw meals.Heating rate: 10°C/min in air.The volume decrease is shown on the ordinate.

Table 1 -
Composition and fine/coarse ratio

saving cement production by grain size optimisation of the raw meal
Figure 2 -CAS-Free-lime relation, modified system CAS.After subtraction of the analysed free-lime content from the initial composition (lower red dot) the remaining CaO reduced composition (upper red dot) results.

Table 2 -
Standard state values

Table 3 -
Free Enthalpy of formation